A sensitive and bright single-cell resolution live imaging reporter of Wnt/ß-catenin signaling in the mouse

Generation of TCF/Lef:H2B-GFP reporter mice

Different promoter elements used in already characterized LacZ reporter lines were considered as drivers for H2B-GFP reporter expression after Wnt signaling activation. We opted for both the BAT-gal and the TCF/Lef-LacZ reporter lines, since both have been extensively used in recent years as reliable read-outs of canonical Wnt activity. In the BAT-gal transgene construct, the promoter elements driving reporter expression consist of 7 TCF/Lef binding sites together with a 130 bp fragment containing the minimal promoter-TATA box of the gene siamois [12]. On the other hand, the TCF/Lef-LacZ construct was generated by fusing six TCF/Lef response elements upstream of the hsp68 minimal promoter (Figure 1A) [11].

We subsequently cloned the TCF/Lef response elements and minimal promoter of each reporter in front of H2B-GFP to create the final construct. Transgenic mouse lines were generated from both constructs. However, since the TCF/Lef-siamois:H2B-GFP transgenic animals gave no readily detectable fluorescent signal, only the TCF/Lef-hsp68:H2B-GFP founder lines were considered for further study, and from now on are referred as TCF/Lef:H2B-GFP strain (Figure 1B).

To establish the specificity of reporter expression, and determine that we did not observe any position or copy-number dependent variability, we confirmed that two independent TCF/Lef:H2B-GFP (F₀) founder animals produced identical patterns of reporter expression in F₁ progeny. We then compared the pattern of expression of the TCF/Lef:H2B-GFP reporter with that observed in stage-matched embryos of the related TCF/Lef-LacZ strain (Figure 2 and 3). Overall, the H2B-GFP-based reporter confirmed all sites of expression observed with the LacZ reporter at all stages and in all tissues examined. This validated the specificity of our TCF/Lef:H2B-GFP reporter as a single-cell resolution live imaging read-out of canonical Wnt signaling activity in mouse embryos and adult tissues. Moreover, as we had predicted, the TCF/Lef:H2B-GFP reporter revealed cells undergoing active Wnt/ß-catenin signaling within the epiblast and visceral endoderm of the early postimplantation stage embryo (Figure 4). Genetic analysis and marker expression data have supported a role for Wnt/ß-catenin signaling in these tissues of the early embryo, but these sites have not been substantiated using previous generation Wnt/ß-catenin signaling reporter strains.

Failure to detect TCF/Lef:H2B-GFP reporter expressing cells at preimplantation stages of embryonic development

Considerable debate has surrounded the issue of Wnt/ß-catenin signaling at preimplantation stages of mouse development, and specifically at the blastocyst stage. In support of a transcriptional read-out of canonical Wnt signaling, previous reports have suggested possible transient nuclear-localization of ß-catenin in a minor population of cells at the blastocyst stage. In our hands and with our TCF/Lef:H2B-GFP reporter, which we believe to exhibit increased sensitivity over existing reporter strains, we were unable to detect convincing reproducible nuclear-localized GFP fluorescence at any stage of preimplantation development in either successively staged or time-lapse imaged embryos. We take this to suggest that either the TCF/Lef:H2B-GFP reporter is not sufficiently sensitive, or that any Wnt/ß-catenin pathway signaling response is non-transcriptional.

TCF/Lef:H2B-GFP reporter reveals sites of Wnt/ß-catenin signaling in the early postimplantation embryo not previously detected with reporter strains

At the late blastocyst stage (E4.5) the embryo implants into the maternal uterus. This event is followed by lineage expansion, which results in formation of a cup-shaped egg cylinder. Close apposition of epiblast, extraembryonic endoderm and visceral endoderm (VE) facilitates signaling cross-talk between these layers, leading to the formation of the distal visceral endoderm (DVE), migration of the anterior visceral endoderm (AVE) and establishment of the anterior-posterior axis of the embryo [13]. Canonical Wnt signaling has been implicated in both formation of anterior-posterior axis and AVE migration. Specifically, a population of cells which are proposed to actively migrate within the VE epithelium, a morphogenetic movement which results in the formation of the AVE, have been proposed to move away from a region of high WNT activity [14]. Although β-catenin localization can be detected throughout VE, Wnt/β-catenin reporter strains generated to date have not exhibited activity at these early postimplantation stages making it difficult to determine the spatiotemporal dynamics of in vivo signaling within the VE.

In TCF/Lef:H2B-GFP embryos GFP reporter expression was detected specifically in the VE tissue layer around the time of DVE specification (E5.5), and throughout the period encompassing AVE migration (E5.75). VE-specific localization was confirmed by colocalization with the pan-VE marker HNF4α [15]. GFP reporter expressing cells were not detected in the epiblast and extraembryonic ectoderm at this stage (Figure 4A, 4A', 4B, 4B'). Initially, only a subset of VE cells expressed detectable and variable levels of GFP expression. After initiation of AVE migration, the number of GFP-positive cells increased, although still exhibiting varying levels of fluorescence intensity (Figure 4B1, 4B'1; Figure 4B5- yellow arrowhead marks migrating DVE/AVE cells). A small group of cells that were robustly GFP-positive were initially observed in a broad region around the embryonic-extraembryonic junction (Figure 4 and 5). In the period leading up to the emergence of the primitive streak, in the majority of embryos analyzed, robustly GFP-positive cells were observed to be predominantly localized to the posterior side of the embryo, likely corresponding to the region of the posterior visceral endoderm (PVE) (Figure 4A1, 4D1, D3; white arrowhead). The localization of this cohort of GFP-positive was designated as putative PVE, since it was contralateral to Cer1-positive cells, which are localized anteriorly (Figure 4C3; white arrowhead). Cer1 antibody specificity in the DVE/AVE was confirmed by colocalization with GFP in Hex:GFP transgenic embryos (Figure 4C, 4C'; yellow arrowhead). Taken together these observations are consistent with previous reports suggesting a repulsive or counteractive effect of high levels of Wnt signaling on cells of the AVE [14]. Further in depth analyses will be required to determine the precise localization of Wnt-responsive cells, combined with a quantitative analysis of levels of reporter activity within the VE, during this critical period of embryonic development.

Tracking TCF/Lef:H2B-GFP reporter expressing cells in the visceral endoderm

One of the advantages of H2B fusions as live imaging reporters is that they facilitate the identification and tracking of single cells while at the same time permitting visualization of an entire population. Indeed, how a group or population of cells can move collectively and in doing so radically change the structure and function of a tissue, is a central question in developmental biology, and underscores many integral morphogenetic cell behaviors driving embryonic development. Since the TCF/Lef:H2B-GFP transgenic is the first Wnt reporter line to reveal the dynamics of Wnt signaling activity in cells of the visceral endoderm of E5.5 embryos, we focused on this stage to study the behavior of cells expressing the reporter. We assumed that GFP-positive cells were either actively signaling, or had recently been transducing a Wnt signal and remained GFP-positive due to perdurance of the GFP protein (Additional file 1).

3D time-lapse movies of embryos pre-streak (PS) stage embryos (Figure 5, Additional file 1) confirmed that the number of GFP-positive cells increased within the VE as development proceeded, and confirmed the heterogeneity of GFP reporter levels among GFP-positive cells, as had been observed in sequentially staged embryos (Figure 4). Tracking of reporter-expressing VE cells in E5.5 embryos for over 9 hours, a period of time during which the AVE would have migrated, revealed extensive cell division (over 50% of cells tracked divided), and the conservation of nearest-neighbor relationships between GFP-positive cells. No change was observed in the relative position GFP-positive cells that were tracked, suggesting that the TCF/Lef:H2B-GFP reporter might not be labeling cells of the DVE/AVE.

In the data depicted in Figure 5, a total of 22 GFP-positive VE cells were identified and tracked. Color-coded open circles identify individual cells in Figure 5, and color-coded closed spheres identify individual tracked cells in Additional file 2. About half the tracked population (12 cells) divided during the 9 hours of time-lapse (Figure 5A - color-coded open circles with white outline indentify cells having divided since previous time-point shown). In nearly all cases, nearest-neighbor relationships were preserved both in regard to individual cells, as well as their relative positions within the group.

We selected 4 cells that did not divide during the time-lapse as a reference set (closed yellow circles in Figure 5, yellow spheres in Additional file 2), and documented their relative distances during the period of the time-lapse (white lines connecting yellow circles in Figure 5A, red lines connecting yellow spheres in Additional file 2). Notably, the distance between these non-dividing reference cells doubled as their individual positions changed relative to each other. Despite this fact, their nearest-neighbor relationships remained predominantly unchanged, as did their position relative to their neighbors, suggesting that the topology of cells within the VE was constant.

We tracked dividing reporter-expressing VE cells. We plotted the orientation of the division planes, as well as the final position of daughter cells at the last time-point for which we generated image data (Figure 5). Our data suggest that any migration of reporter-expressing cells may result from oriented divisions, as well as a general increase in the size of the embryo, rather than by an active movement of cells. However, additional data and statistical analyses will be required to determine if cell divisions within this population exhibit a prevalent orientation, as our data might suggest.

As reporter-expressing VE cells proliferated, their progeny did not reorganize, but retained their relative positions. A minor reorganization or 'jostling' of cells could serve to alter cell geometry, but facilitate conservation of cell topology, within the VE epithelium and could be driven by differential regional proliferation within the VE. Our data support a model whereby active cell migration driving AVE formation [16], and cell proliferation may together result in a reorganization of cells within the VE epithelium. These morphogenetic cell rearrangements are also likely influenced by, and are expected to accommodate, the rapid growth of the embryo, notably the adjacent epiblast, at this stage. Further detailed analyses will be important in extending these observations and determining the respective roles of proliferation and cell signaling within the VE and its neighboring tissues, and importantly how these coordinated cell behaviors in association with cell signaling might direct the morphogenesis of the early mouse embryo.

The TCF/Lef:H2B-GFP reporter marks the primitive streak and nascent mesoderm at gastrulation

Gastrulation is the event that results in the generation of the three primary germ layers (ectoderm, definitive endoderm and mesoderm) from the pluripotent epiblast, and the elaboration of the axes (anterior-posterior, dorsal-ventral and left-right). In the mouse, the onset of gastrulation is marked by the appearance of the primitive streak (the source of mesoderm and definitive endoderm) which represents a morphologically-distinct structure which breaks the bilateral symmetry of the epiblast, and in doing so, defines the posterior of the embryo at E6.5. The site of primitive streak formation has been proposed to be regulated by Nodal and Wnt3 signaling activities likely emanating from the overlying visceral endoderm [17–20]. This data combined with the expression pattern of Wnt3 suggest an essential role for Wnt signaling in primitive streak specification, and the maintenance of gastrulation. By E7.5 Wnt3 is downregulated while the related gene, Wnt3a, is activated. Embryos lacking Wnt3a exhibit a complete absence of paraxial mesoderm, the cell type emerging from the primitive streak starting at E7.5. These genetic data support an essential role for Wnt/ß-catenin signaling associated with the site of the primitive streak, in the initiation and progression of gastrulation in the mouse.

To this end, TCF/Lef:H2B-GFP reporter expression reproduced Wnt3 early expression in the proximal epiblast before the onset of gastrulation (pre-streak (PS) stage, ~E6.0) and later on, in early streak (ES, ~E6.5) stage embryos, became restricted to the posterior side, where single cells within the epiblast became positive for the reporter and could be visualized as they underwent gastrulation traversing the primitive streak (Figure 6A-6C and data not shown). At this stage TCF/Lef:H2B-GFP reporter expression was restricted to cells of the primitive streak. By late streak stages (LS, E6.75), the level of streak-specific expression of the H2B-GFP reporter increased, and by the time the streak had fully elongated and reached the distal tip of the embryo, H2B-GFP fluorescent cells could be seen along the posterior body axis (Figure 6D-6F). Tissue sections of gastrula stage (E7.75) embryos, revealed the presence of individual H2B-GFP-positive cells within the epiblast in the vicinity of (and likely to be ingressing through) the primitive streak, as well as in the wings of mesoderm migrating away from it (Figure 6J).

By late bud/early headfold (~E7.5) stages, a second population of H2B-GFP-positive cells located in the proximal part of the conceptus within the extra-embryonic region emerged. These small patches of H2B-GFP-positive cells resembled the pools of primitive erythroid cells, or "blood islands", a site at which Wnt signaling has been proposed to play a role in specifying hematopoietic cell populations (Figure 6H, 6I; yellow arrowhead) [21–23].

TCF/Lef:H2B-GFP reporter during midgestation

By the time embryos had developed seven to nine somites (E8.5) robust levels of transgene expression were evident along the presomitic mesoderm and newly formed somites (Figure 7A, 7A5, 7A6). This strong posterior reporter expression might reflect the activity of Wnt8a and Wnt3a which are highly expressed at the streak at these stages, the last being required for mesoderm formation and somitogenesis [24–26]. By this stage, additional sites of the H2B-GFP reporter could be detected including the posterior neural plate, which displayed high levels of transgene expression. However, levels of reporter expression decreased in the anterior neuroepithelium (Figure 7A3, 7A4). In the head and branchial arch region, H2B-GFP expression was detected in cranial neural crest cells at the neural folds, as well as in cells migrating into the face and branchial arches (Figure 7A1, 7A2). Indeed, the generation of multipotent neural crest cells has been long associated with early induction through exposure to Wnt signaling [27].

TCF/Lef:H2B-GFP reporter expression at midgestation to later fetal stages

At E9.5 transgene expression persisted in all previously described domains (Figure 7B1, 7B2, 7B5, 7B6). At this stage, expression in the somites reached more anterior levels and increased fluorescence could be seen all along the neural tube, also in the brain. New sites of expression were observed in the otocyst (Figure 7B3), and in restricted regions of the heart tube like the atrioventricular (AV) canal (Figure 7B4). At E10.5 GFP expression was maintained in the neural tube, somites, heart, nasal and pharyngeal region (Figure 8A, 8C). Expression in the otocyst was restricted to the dorsomedial epithelium, where Wnt signaling together with Shh have been shown to regulate dorsal-ventral axis specification (Figure 8D) [28]. By this stage H2B-GFP expression in the tail bud became restricted to a defined population (Figure 8H). In the limb bud, individual H2B-GFP-positive cells within the apical ectodermal ridge (AER) were first seen at this stage (Figure 8B). Expression of different Wnt ligands such as Wnt3a or Wnt6 in the AER has been previously reported, further confirming the specificity of the reporter as a Wnt signaling readout [29–31]. Positive cells were detected in the foregut endoderm, possibly corresponding to lung progenitors (Figure 8E) [32], but requiring further confirmation by marker analysis. In addition, the epithelial cells that form the lining of the mesonephric duct were also H2B-GFP positive as seen in transverse sections at E10.5 (Figure 8F, 8G). Wnt9b is expressed in the duct and is essential for early urogenital system organogenesis supporting a role for Wnt/ß-catenin signaling in this process [33].

Widefield fluorescence imaging at E12.5 revealed high levels of reporter expression within the spinal cord and limbs (Figure 8I). In the brain, expression of the transgene was localized to discrete regions such as the infundibulum, whereas it was broader in the spinal cord (Figure 8M, 8L, 8R). Close-up of the otic region showed increased levels of fluorescence in the developing semicircular canals (Figure 8J). In sections through the inner ear, transgene expression was detected in the dorsomedial otic epithelium that will give rise to the vestibular structures (Figure 8O). Previous studies have also described Wnt activity in the olfactory epithelium (Figure 8N) and oral epithelium (Figure 8P) [34]. High magnification wholemount views of the limb revealed continued expression of the transgene at the AER (Figure 8K), while sections showed regions positive for the transgene outside cartilage primordia (Figure 8T). Moreover, high levels of transgene were also detected in the lung epithelia (Figure 8U), where Wnt signaling is proposed to play a key role in branching morphogenesis [35].

TCF/Lef:H2B-GFP reporter expression at postnatal stages

To further validate the specificity and utility of the reporter, TCF/Lef:H2B-GFP transgene expression was also analyzed at postnatal stages. Robust but restricted reporter expression was observed in many tissues, with only a few documented in detail for this report. At postnatal day 2, H2B-GFP fluorescence was observed in the muscle layer or muscularis externa of the esophagus (Figure 9B). By three weeks of age, almost widespread expression of the transgene was detected in the atrial and ventricular myocardium of the heart (Figure 9A, 9D, 9E). However, restricted expression of the transgene was observed in the heart valves (Figure 9C). In the respiratory tract, reporter expression was restricted to epithelial cells in the lining of the trachea (Figure 9F), bronchi and bronchioles (Figure 9G) as previously described for a LacZ based Wnt/ß-catenin signaling reporter [36]. In the thymus, scattered positive cells were observed in the thymic medulla (Figure 9H), supporting a recent report demonstrating the role of Wnt signaling in epithelial microenvironment maintenance [37]. While in the liver, H2B-GFP fluorescent hepatocytes were only located in a single-cell layer surrounding the central veins (Figure 9I). This hepatocyte subpopulation was also identified using a different GFP-based Wnt/ß-catenin signaling reporter [7]. In this study these reporter-positive cells were shown to co-label with glutamine synthase, a marker for perivenous hepatocytes [7]. In the small intestine, H2B-GFP reporter expressing cells were primarily located in the crypts where they have been shown to represent the progenitor cell population of the intestine (Figure 9J) [38]. In the female reproductive tract, expression of the transgene was found in distinct layers of the oviduct including the muscularis and the lamina propia within the mucosa (Figure 9K). In the uterus, GFP reporter expression was observed in the luminal epithelium (Figure 9L). This observation is in agreement with reports identifying the Wnt/ß-catenin pathway as a major signaling pathway involved in embryo-uterus cross-talk during implantation [39, 40].

TCF/Lef:H2B-GFP reporter expression in the kidney

To validate the TCF/Lef:H2B-GFP reporter as an improved tool for visually dissecting morphogenetic processes driven by Wnt/ß-catenin pathway in any developmental event, we focused on the development of two major organ systems; the kidney and the brain. The mammalian kidney develops from a reciprocal signaling interaction between the epithelial ureteric bud and the adjacent metanephric mesenchyme [41]. By E12.5, distinct expression of the H2B-GFP reporter was observed in the urogenital anlage, where the developing kidneys as well as the gonads were positive for the transgene (Figure 10A). By contrast the mesonephros was negative for the reporter. In the kidney, Wnt reporter expression was restricted to the epithelial cells forming the collective ducts, as observed in both whole mount and transverse sections of an E14.5 kidney (Figure 10B, 10D). Differential expression of the H2B-GFP transgene was observed in the Wolffian and Müllerian ducts. In the mesonephric or Wolffian duct, reporter levels were higher than in the Müllerian duct (Figure 10E). At E14.5, the adrenal gland was also positive for the transgene (Figure 10B). At postnatal stages, expression of the reporter persisted in the adrenal gland, primarily in the zona glomerulosa of the adrenal cortex as evident in transverse sections (Figure 10C, 10F, 10I). Cortical expression of the transgene was in agreement with previous reports highlighting a requirement for Wnt signaling in the region for proper hormone production [42]. Also at postnatal stages, H2B-GFP fluorescence was detected in the collective ducts of the kidney (Figure 10G, 10H). The single-cell resolution of the reporter facilitated the detailed visualization of branches in the renal cortex (Figure 10G). This resolution of reporter if coupled with 3D time-lapse imaging of ureteric bud explants should facilitate the detailed analysis of cell behaviors driving branching morphogenesis [43].

TCF/Lef:H2B-GFP reporter expression in the brain

TCF/Lef:H2B-GFP transgene expression was also analyzed in the brain. During embryonic stages, reporter fluorescence was primarily localized to the olfactory bulbs, mesencephalon and part of the telencephalon (Figure 11A, 11B). At postnatal stages, widefield fluorescent views of the brain revealed increased levels of fluorescence (Figure 11C). However, when examined at high magnification, transgene expression was shown to be restricted to specific neuronal subtypes in different brain regions. In the olfactory bulb, tufted cells in the external plexiform layer and periglomerular cells were positive for the transgene (Figure 11D, 11E). These periglomerular cells might correspond to olfactory ensheathing cells, which have been previously reported as Wnt signaling responsive cells [44]. H2B-GFP fluorescence was widely observed in the cortex (Figure 11F). The transgene was also broadly expressed in the septum and striatum around the subventricular zone (SVZ) (Figure 11G). Section through the SVZ confirmed positive cells that colocalized with GFAP, a marker of SVZ niche astrocytes (Figure 11H and data not shown). Reporter fluorescence was also detected in the astrocytes of another neurogenic zone in the adult, the hippocampus (Figure 11I and data not shown), where a previous study demonstrated a role for Wnt signaling in the regulation of adult neurogenesis [45]. In the cerebellum, H2B-GFP expression was exclusively restricted to the Purkinje cell layer where Wnt3 is expressed in the adult brain (Figure 11J, 11K) [46]. Analysis of the peripheral nervous system was only performed at developmental stages where TCF/Lef:H2B-GFP expression was broadly detected in the spinal cord (Figure 11L). The sensitivity, resolution and localization of the TCF/Lef:H2B-GFP reporter if combined with time-lapse imaging of brain slice cultures should facilitate the detailed analysis of cell dynamics driving a wide variety of processes including neurogenesis.

Generation of Reporter Constructs and Transgenic Animals

To generate the pTCF/Lef:H2B-GFP construct, six copies of the TCF/Lef response elements together with the hsp68 minimal promoter from the TCF/Lef-LacZ reporter construct [11] were inserted into the AseI/Nhe1 sites of pCMV::H2B-GFP. Transgenic mice were generated by pronuclear injection following standard protocols. Animals were genotyped by PCR. Primers used for the PCR reaction were GFPGenotFW: ACAACAAGCGCTCGACCATCAC; GFPGenotRW: AGTCGATGCCCTTCAGCTCGAT. Two transgenic founder lines (TCF/Lef:H2B-GFP #16, TCF/Lef:H2B-GFP #61) were established both exhibiting similar patterns and levels of reporter expression. However, only one line (TCF/Lef:H2B-GFP #61) was further characterized in detail. Subsequent generations exhibited Mendelian transgene inheritance, stable transgene activity and comparable levels of reporter expression. Animals were maintained in accordance with National Institute of Health guidelines for the care and use of laboratory animals and under the approval of the Memorial Sloan-Kettering Cancer Center Institutional Animal Care and Use Committee.

Embryo collection and strains used

Embryos were dissected in DMEM/F12 (1:1) containing 5% calf serum and fixed in 4% paraformaldehyde. Presomitic embryos were staged by morphological landmarks according to Downs and Davies [47]; somite stage embryos were staged according to their somite number. Additional mouse strains used were TCF/Lef-LacZ [11] and TOPGAL [6] reporters.

Embryo culture

For ex utero culture embryos were dissected in DMEM/F12 (1:1) containing 5% calf serum and cultured in medium comprising 50% DMEM/F12 (1:1), 50% rat serum and 1% Penicillin/Streptomycin [48].

Embryo sectioning

For vibratome sections, embryos were embedded 0.5% gelatin/15% BSA/20% sucrose (gelatin/albumin mix) and 4.5% glutaraldehyde in PBS and cut on a Leica vibrating microtome (VT1000) at 35-175 μm. For cryosections, samples were equilibrated in PBS/10% sucrose followed by PBS/30% sucrose overnight and snap-frozen in OCT (Tissue-Tek). Sections were cut in a Leica cryostat at 12 μm and counterstained with Hoechst for nuclei (5 μg/ml, Molecular Probes) and Phalloidin Alexa546 for F-actin (Invitrogen).

ß-Galactosidase staining

Mouse embryos were dissected in PBS, fixed for 20 min in 0.35% glutaraldehyde in PBS and processed as whole mounts or embedded in the gelatin/albumin mix and sectioned. Samples were washed 3 times with X-Gal rinse (0.02% NP-40, 0.01% Sodium deoxycholate, 2 mM MgCl₂, 100 mM sodium phosphate pH 7.3) for 20 min and incubated in 1 mg/ml X-gal, 5 mM K₃Fe(CN)₆, 5 mM K₄Fe(Cn)₆, 0.02% NP-40, 0.01% Sodium deoxycholate, 2 mM MgCl₂, 100 mM sodium phosphate pH7.3 overnight. They were finally rinsed with PBS and fixed for 30 min in 4% paraformaldehyde.

Immunostaining

Embryos were permeabilised in 0.55% Triton X-100 in PBS for 20 minutes and blocked in 10% fetal bovine serum in PBS for 1 hour. Primary antibodies used were: anti-Cer1 (R&D Systems), anti-Eomes (abcam), anti-GFP (Invitrogen), anti-HNF4a (Santa Cruz), and anti-Oct4 (Santa Cruz). Secondary Alexa Fluor (Invitrogen)-conjugated antibodies were used at a dilution of 1/500. DNA was visualized using Hoechst 33342 (5 μg/ml, Molecular Probes).

Image acquisition

Widefield and epifluorescence images were acquired using a Zeiss Axiocam MRc camera coupled to a Leica M165FC dissecting scope or a Zeiss A1 Axioscope. Laser scanning confocal data were acquired using a Zeiss LSM510 META using a PlanApo 20×/NA0.75 objective. Fluorophores were excited using a 405 nm diode laser (Hoechst), 488 nm Argon laser (GFP, GFP) or 543 nm HeNe laser (Alexa 546). Embryos were imaged whole mount in MatTek dishes (Ashland). Spinning disc confocal 3D timelapse data was acquired using Volocity acquisition software http://www.improvision.com/ and a Perkin-Elmer RS3 Nipkow-type scan head mounted on a Zeiss Axiovert 200 M with Hamamatsu C4742-80-12AG camera. GFP was excited using a 488 nm Argon laser. Images were acquired using a Zeiss plan-Neofluar 25×/0.8 DIC korr objective. 10-20 xy planes were acquired, separated by 3-4 μm. Time intervals between z-stacks were 15 minutes. For live imaging experiments, embryos were maintained in a humidified, temperature-controlled chamber with 5% CO₂ atmosphere. Sections were mounted in Fluoromount-G (Southern Biotech) and imaged through glass coverslips. Confocal images were acquired as z stacks of xy images taken at 1 μm z -intervals.