Develop. Growth Differ. (2007) 49, 49–59

doi: 10.1111/j.1440-169x.2007.00906.x

Improved method for visualizing cells revealed dynamic morphological changes of ventral neuroblasts during ventral cleft closure of Caenorhabditis elegans Blackwell Publishing Asia

Zhicen Liu,1,2,* Akira Nukazuka2 and Shin Takagi2 1

Center for Gene Research, Nagoya University and 2Department of Biological Science, Graduate School of Science, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8602, Japan

The formation of intricate and functional biological structures depends on the dynamic changes of cellular morphology. Confocal laser scanning microscopy (CLSM) is a widely used method to reveal the threedimensional (3-D) structure of cells during the development of Caenorhabditis elegans (C. elegans) and other model organisms. Improving the efficiency and image quality of CLSM would benefit studies using this method. We found that CED-10::GFP::CED-10, a green fluorescent protein (GFP) marker, is intensely expressed beneath the cell surface, facilitating visualization of cellular morphology in C. elegans embryos. By combining the unique properties of this marker, and with the help of direct 3-D rendering of images obtained by CLSM, we developed a simple but powerful method for investigating cellular morphology in developing embryos. Using this method we, for the first time, document the dynamic changes in the morphology of ventral neuroblasts in vivo during ventral cleft closure. Key words: Caenorhabditis elegans, cellular morphology, confocal microscopy, embryogenesis, neuroblast.

Introduction Orchestrated morphological changes in individual cells underlie the formation of complex structures during the development of multicellular organisms (Gilbert 2000). Among the model organisms, the morphogenesis of Caenorhabditis elegans (C. elegans) is one of the most systematically studied, because the entire development of the embryo can be observed under a microscope at single-cell resolution owing to its transparency. All of the greater than 500 cells that make up the hatching animal have documented lineages and are generated within 6 h of fertilization. The dimension of the whole specimen (approximately 60 µm × 40 µm × 30 µm) fits not only within a high magnification microscope field, but also within the reachable depth to objective lenses with large numerical aperture (NA). Therefore, the specimen is amenable to detailed observations by confocal laser

*Author to whom all correspondence should be addressed. Email: [email protected] Received 29 May 2006; revised 24 November 2006; accepted 25 November 2006. © 2007 The Authors Journal compilation © 2007 Japanese Society of Developmental Biologists

scanning microscopy (CLSM) or multiphoton laser scanning microscopy (MPLSM). Furthermore, the initial differentiation and morphogenesis of a number of tissues commences before the beginning of relatively rapid muscular movement, facilitating time-lapse recording. To investigate the three-dimensional (3-D) structure of cells during embryonic development, traditionally the C. elegans embryo is fixed, embedded, sectioned, examined by transmission electron microscopy (TEM), and then the overall cellular structure is reconstructed by stacking images of serial sections. Extra procedures are also required to remove the tough egg shell impermeable to fixatives and embedding media (Krieg et al. 1978; Williams-Masson et al. 1998). CLSM provides an alternative means to obtain a 3-D cell structure by stacking sectional images of the cell membrane. Antibodies against proteins on the plasma membrane can be used to label the cell surface of fixed embryos (Sawa et al. 2003), but this method is not applicable to living embryos. Another method uses the aqueous soluble vital probe SynaptoRed, also known as FM 4-46, to stain the plasma membrane of living embryos (Mohler & White 1998; Lee & Goldstein 2003). This method, however, requires special techniques and equipment to permeate the egg shell with laser beams. In addition, the influence

50

Z. Liu et al.

of permeating the egg shell and the phototoxity of dye to normal development remains undocumented. Green fluorescent protein (GFP) markers are widely used to identify and visualize cells in vivo (Mohler et al. 1998; Williams-Masson et al. 1998; Heid et al. 2001; Koppen et al. 2001). Many are expressed in a fraction of the cells in embryos within a limited time. A GFP marker expressed ubiquitously on the cell membrane would be suitable for the analysis of morphological cell changes in living embryos. Through our investigation of the organogenesis of C. elegans (Liu et al. 2005), we found that CED-10, the C. elegans Rac protein, fused to GFP (Lundquist et al. 2001) is strongly expressed beneath the cell boundaries. By integrating this unique property with an image reconstruction technique based on direct 3-D rendering, we developed a method to examine the cellular morphology of C. elegans in vivo with improved quality and efficiency. This method has helped us discover the movement of protrusive neuroblast extensions during early embryogenesis.

Materials and methods Strains and general methods The C. elegans strain MT10865, bearing the transgene ced-10::gfp::ced-10, was provided by Erik Lundquist (Lundquist et al. 2001). ced-10 encodes CED-10, a C. elegans Rac orthologue, and MT10865 expresses the intact CED-10 protein fused with GFP via its native promoter-enhancer. Animals showing strong fluorescence under a dissecting microscope were transferred to new plates to maintain the strain. Basic methods for worm culture and genetics were performed as described by Brenner (1974) and compiled by Wood (1988). Observation and sample preparation

CLSM. To minimize the deformation caused by the pressure from the coverslip, extra distilled water was applied to fill the space between the coverslip and slide glass. Laser scanning was performed with an UplanApo 100 × 1.35 oil iris objective lens (Olympus). For high quality 3-D reconstruction, Z stacks were scanned 0.2 or 0.3 µm apart at 640 × 480 or 800 × 600 resolution. Before being scanned, the embryos were screened for developmental stages under bright field by differential interference contrast microscopy to minimize photobleaching and phototoxity. The focal plane and the scanning depth were also estimated from bright field rather than by scanning. The power and pinhole diameter of the scanning laser was minimized while the photomultiplier tube (PMT) voltage and signal gain were maximized under adequate image quality. Time-lapsed scanning Embryos were mounted on a double-thickness agar pad and then selected under a bright field microscope for their developmental stages. Embryos with their ventral sides facing toward the objective lens were chosen for observation. Distilled water was applied between the coverslip and slide glass during time-lapse observation to compensate for the reduction of the thickness of an agar pad due to evaporation of water. Three-dimensional reconstructions Several commercial and non-commercial packages for 3-D analysis were tested for reconstructing laser scanned images (3D Doctor version 3.5; Amira version 4.0; Image J version 1.34 [free]; Imaris version 4.1.3; Olympus fluoview viewer version 1.4 [free, registration needed]). Amira (TGS) was chosen to render the 3-D reconstructions. Scanned images can be directly rendered by the ‘voltex’ function. Direct rendering minimizes digital distortion due to segmentation or manipulation of the raw data. The software is also capable of interactively turning or slicing the reconstructed stacks, and scanned images can be processed for further analysis. For analyses of the embryos’ ventral side cellular morphology, 15 animals for each stage represented in the figures were reconstructed.

Worms were mounted on 4% agarose. To avoid desiccation, agar pads of twice the usual thickness (c. 0.6 mm) were used. To isolate fertilized eggs from the uterus, an egg-bearing adult hermaphrodite was first transferred to the agar pad on a slide glass and then dissected using the sharp edge of a broken tapered glass pipette under a dissecting microscope. The tapered glass pipettes were drawn to a diameter of around 300 µm while heated by a Bunsen lamp before being broken.

Results

Scanning procedure

CED-10::GFP::CED-10 is localized to the cell boundary

A Fluoview-500 (Olympus, Japan) equipped with an argon-krypton laser source (Melles Griot) was used for

Embryogenesis of C. elegans starts in a hermaphrodite after egg fertilization in spermatheca (Fig. 1).

© 2007 The Authors Journal compilation © 2007 Japanese Society of Developmental Biologists

Dynamic morphology of C. elegans neuroblasts

51

Fig. 1. Timetable of early embryogenesis of Caenorhabditis elegans and the corresponding stages of the embryos shown in figures. Nucleus count versus minutes after first cleavage at 22°C for each stage during embryogenesis is shown by a blue line (adapted and modified from Chisholm & Hardin 2005).

Development proceeds while embryos pass through the uterus to the vulva; the embryos at more advanced developmental stages are positioned nearer to the vulva. The embryos near the vulva assumed stronger fluorescence of the CED-10::GFP::CED-10 marker than their younger, more distant siblings (Fig. 2A,B). The signal was not detectable until the 50-cell stage of the gastrulating embryo, when the marker started to accumulate beneath the cell boundaries (Fig. 2C). The signal gradually strengthened until the three-fold stage, and remained unchanged thereafter until adulthood. The marker was expressed ubiquitously in all cells except the germ line Z2 and Z3 cells (g in Fig. 3C). At the subcellular level (Fig. 2D,E), the signal was absent in the nucleus, easily identified as a circular void within the cell. We found that the distribution of the marker in cytosol sometimes assumed a vesicular structure, probably because the CAAX box within the CED-10 molecule (Lundquist et al. 2001) would associate the marker with the endoplasmic membrane system. This is consistent with the fact that CED-10 is a C. elegans orthologue of the RAC protein, which has been reported to play a role in intracellular vesicular trafficking in other organisms (Ridley 2001). The intracellular vesicles in the scanned images overlapped with most, if not all, the granules observed by differential interference contrast microscopy.

The marker accumulated at the boundaries between neighboring cells, appearing much more sparsely at cell surfaces facing the egg shell where no neighboring cell was present (Fig. 2D), and little accumulation was seen at the boundary facing the germ line cells Z2 and Z3 (Fig. 3C). Large round vesicular structures exhibiting the strongest signals correspond to the cells undergoing apoptosis, indicating the cell boundaries between cells undergoing programmed cell death and the cells engulfing the dying cells (best seen in Fig. 4E, indicated by arrowheads). Direct 3-D rendering technique facilitates morphological analysis of embryonic cells We took advantage of the strong fluorescence of CED-10::GFP::CED-10 at the cell boundaries to visualize the cellular morphology in living embryos using CLSM. Figure 3 shows an animal at the 1.4fold stage of embryogenesis, scanned from the right lateral side. Panels B and C show the scanned images from individual focal planes. We further exploited a direct 3-D rendering technique (Amira software, TGS) to reconstruct the cellular morphology from scanned images, which gives an overall view of the organization of tissues in the embryo (Fig. 3D,E,F). Structures of various cell types like the epidermis, traditionally called the hypodermis in nematodes, covering the surface of the animal; the

© 2007 The Authors Journal compilation © 2007 Japanese Society of Developmental Biologists

52

Z. Liu et al.

Fig. 2. Embryonic expression of CED-10::GFP::CED-10. All animals in the figures in the present paper are MT10865 carrying the transgene ced-10::gfp::ced-10. (A–B) Confocal laser scanning microscopy (CLSM) and differential interference contrast (DIC) image of embryos in the uterus of an adult hermaphrodite, respectively. (C) Embryos in panel A scanned under higher sensitivity are shown at higher magnification. (D–E) CLSM and DIC image of the dorsal side of a 300 min stage embryo, respectively. The accumulation of the marker beneath the cell boundaries first became detectable at the 50-cell stage (white asterisk in panel A and C). The fluorescence became stronger as the eggs passed through the uterus towards the vulva. The peak of the signal was reached at the three-fold stage (black asterisk) and the intensity was kept constant throughout adulthood. (D) Localization of the green fluorescent protein (GFP) marker at the subcellular level. The marker is absent in the nucleus (white arrowhead), but is intensively accumulated at cell boundaries. Weak accumulation is seen under cell membranes where no adjacent cell presents (arrow). The marker also associates with vesicular structures, which correspond to the granules visible under DIC (compare circled areas of D and E). Adult in (A), (B) and (C) are shown anterior towards left, dorsal towards up. Embryo in (D) and (E) are shown anterior towards left, right towards up. In all figures the length of the eggs is ∼50 µm.

huge intestinal cells; excretory cells; body muscle tissue cells beneath the ventral hypodermis and the rectum were clearly visualized. A subcellular vesicle system, to which the marker appeared to associate, was also discernible in 3-D reconstructions. In the following sections, we will describe three developmental events studied with this method: gastrulation; closure of ventral cleft; and ventral enclosure. Emergence of the ventral cleft Little is known about the cellular basis underlying morphological changes of the ventral side compared to the accessible dorsal side during early embryogenesis. Unlike the solely epidermal dorsal side, whose cellular morphology can be visualized with the apical epidermal junction marker AJM-1::GFP, the cellular morphology of the non-epidermal ventral side was hitherto difficult to visualize. We used our method to study the cellular morphology of the ventral side during early embryogenesis. At the 150 min stage (after first cleavage of the fertilized egg in 22°C), the posterior half of the embryos were occupied by the descendants of the P4

(Z2, Z3) and D (Daa, Dap, Dpa, Dpp) cell lineages (white and black asterisks, respectively, in Fig. 4A), which had just started to ingress into the embryo. No fluorescence was observed throughout development in the germ line Z2, Z3 cells, and the signal of the D descendants was still weak at this stage compared to surrounding cells. This gave an empty appearance in the area, serving as a landmark for identifying the cells during gastrulation. Notably, in the 3-D rendered context, these GFP negative cells were recognized as voids beneath Z2 and Z3 cells, easily distinguishable from the GFP negative outside of embryos. At the 170 min stage (Fig. 4B), together with the ancestors of future mesoderm cells (MS lineage), the P4 and D lineage cells were ingressing into the embryo. The space that they had previously occupied was taken over by the descendants of surrounding AB lineage cells, which are the precursors of ventral neuroblasts. After the mass ingression of the MS, P4 and D lineage cells at the 200 min stage (Fig. 4C), a prominent depression was recognized in the central area of the ventral side, which is known as the ‘ventral cleft’. At this stage the germ line Z2 cell could be recognized on the surface as the small void at the posterior end of the cleft.

© 2007 The Authors Journal compilation © 2007 Japanese Society of Developmental Biologists

Dynamic morphology of C. elegans neuroblasts

53

Fig. 3. Cellular morphology of embyros revealed by scanned raw image and 3-D reconstruction. A 1.5-fold embryo is shown head towards up. The embryo is scanned from the right lateral side. (A) A 3-D reconstruction. (B–C) Two focal planes cutting through the animal as shown in (A). (D–E) Reconstructions seen from the angle the embryo is scanned, where (D) is the whole reconstruction and (E) is the reconstruction in which the upper part was removed from the plane of (B). (F) Reconstruction in which the upper part was removed from the plane of (C). Black arrowheads in (D) and (F) indicate green fluorescent protein (GFP) positive vesicular structures. White arrowheads in (C) indicate the cells undergoing programmed cell death. Arrows in (C–F) indicate the position of the anus and rectum. Labels indicate nuclei of dorsal epidermal cells (asterisks); body wall muscles (+); an excretory cell (e); germ line cells (g); and intestine (i).

Process-like structures protruding from ventral neuroblasts during ventral cleft closure Next, we visualized the ventral neuroblasts (VNBs) surrounding the ventral cleft. At the 250 min stage, VNBs are known to move towards the ventral midline and eventually close the ventral cleft in a process called ‘ventral cleft closure’ (Sulston et al. 1983). This is a crucial process in early embryogenesis, as proper positioning of VNBs is necessary for their future function as neurons. In addition, VNBs serve as the substrate for epidermal cell movement during the next stage of development. However, little is known about the cellular mechanisms underlying this event. In a review, Chisholm and Hardin wrote that “the rate and timing of the movements for individual neuroblasts have not been examined in detail, nor is it yet known if specific cells or cell groups play different roles in the movement” (Chisholm & Hardin 2005). We found that GFP-positive, fine process-like structures 0.2–3.0 µm long were protruding from VNBs surrounding the posterior end of the ventral cleft at the 250 min stage. The structures were protruding over the surface of the underlying GFP-negative Z2 cell (arrows in Fig. 4G). Images scanned at different

focal planes revealed that the structures were thick (> 0.5 µm) at their basal sides where they protruded from a VNB, whereas they were flattened or tapered (< 0.3 µm) where they extended over the surface of a Z2 cell. At this stage, we also found thin processlike structures that connected the left and right sides of the cleft (Fig. 4E). These process-like transverse structures had a thicknesses of less than 0.2 µm, because they were captured in a single focal plane under the scanning interval of 0.2 µm. Reconstructed transverse sections confirmed that the process-like structures also originated from VNBs (data not shown). These observations suggest that the GFPpositive process-like structures are VNB filopodia, and the process-like transverse structures are presumably two filopodia extending from VNBs on opposite sides of the cleft and connected to each other over the Z2 cell. While most prominent around the 250 min stage, the structure could be observed as early as the 200 min stage. The single section image over the surface of the embryo also revealed that GFP signals were localized in an irregular pattern around boundaries between VNBs at the 250 min stage. Rather than forming thin sharp lines, the GFP signal at the apical surface was localized in broader areas extending

© 2007 The Authors Journal compilation © 2007 Japanese Society of Developmental Biologists

54

Z. Liu et al.

Fig. 4. Cellular morphology of the ventral side of early embryos until ‘ventral cleft closure’. (A) Reconstructed embryo at the 150 min stage. White asterisks are Z2 and Z3 cells, black asterisks are D lineage cells. (B) Reconstructed embryo at the 170 min stage. (C) Reconstructed embryo at the 200 min stage. Ventral cleft is encircled by the dashed line. Arrows are pointing to some of the protrusive structures at the cell boundaries. (D) Reconstructed embryo at the 250 min stage. (G) Magnified squared area in (D). The green fluorescent protein (GFP)-negative space at the lower (posterior) center of the embryo is a Z2 cell (white asterisk). Arrowheads indicate cells undergoing programmed cell death. Black arrows are pointing to the putative filopodial protrusions from neuroblasts. (E) and (H) Sectional raw image of the embryo in (D) and (G), respectively. The pictures captured putative filopodial protrusions (white arrows) crawling over the surface of a Z2 cell (white asterisk). (F) Sectional raw image at the surface of the embryo in (D). (I) Magnified square area in (F). Putative filopodial or lamellipodial protrusions (arrows) are visible at cell boundaries. Embryos are shown anterior towards up.

from the lines (compare the cell boundaries indicated by arrows in Fig. 4E with those in Fig. 4F). This irregular pattern might indicate that the GFP marker in VNBs is distributed with uneven thickness beneath the cell surface. Scanning VNBs at deeper focal planes, however, showed that the GFP signal was localized in sharp thin lines at cell boundaries (arrows in Fig. 4E), indicating that the irregular pattern is restricted to the apical surface. Assuming that the CED-10::GFP::CED-10 marker is localized within a limited distance from, and consecutive to, the cell boundaries, we think that these irregular patterns represent the laminar process or lamellipodia of VNBs on their apical surface. The apparent laminar processes ranged between 0.2 and 2.0 µm in length and between 0.2 and 0.6 µm in thickness. They were

thicker at the cell boundary, while flattened away from the cell boundary. These features are similar to those of the apparent filopodia extending over the surface of Z2 cells described above. The irregular GFP localization pattern was also observed at the 200 min stage, although less frequently (Fig. 4C). Movement of the process-like structures at cell boundaries To further investigate the nature of the irregular pattern of GFP localization, we monitored their dynamics in living embryos. Time-lapse 3-D reconstruction (4D) of the ventral surface of embryos during the 250– 280 min stage revealed that the GFP positive structures changed their shapes dynamically (Fig. 5). We

© 2007 The Authors Journal compilation © 2007 Japanese Society of Developmental Biologists

Dynamic morphology of C. elegans neuroblasts

55

Fig. 5. Four-dimensional (4-D) reconstruction of the protrusive structures at cell boundaries of ventral neuroblasts (VNBs) in living embryo. An embryo was scanned for its ventral surface at 2-min intervals. The reconstructions are made for the upper 3 µm depth of the ventral surface of the embryo. The panel at the left of the figure is the whole reconstructed ventral surface at the beginning of the 250 min stage. Arrows point to several prominent protruding structures and within the rectangle is the boundary between the neighboring neuroblasts ‘a’ (ABprppppaa) and ‘p’ (ABprppppap). Right-side panels are time-lapse reconstructions at 2-min intervals which show dynamic movements of neuroblasts at the cell boundary. Arrowheads in the figure indicate extending protrusions whereas arrows indicate retreating ones.

reconstructed the upper 3 µm depth of the ventral surface in 3-D, which allowed us to monitor simultaneously both apical movements of putative protruding structures and their relationship to deeper cell boundaries. As an example, we focused on the boundary between neuroblasts ABprppppaa (designated as ‘a’ in Fig. 5) and ABprppppap (designated as ‘p’). At the beginning of the 250 min stage, two GFP positive areas protruding from ‘a’ crossed the cell boundary and extended into ‘p’. Two minutes later, the right protrusion appeared to retreat, while the left one appeared to keep extending. Four minutes from the beginning, the right protrusion disappeared completely and the left one further extended into ‘p’, while at the left end of the boundary, another GFP positive area started to bulge into ‘a’. At 6 min, the right protrusion started to retract, but it still largely remained over the surface of ‘p’. At the same time, the bulging area at the left extended further into ‘a’. These two adjacent but opposite protrusions distorted the cell boundary into a sinuous shape. Then the left protrusion exhibited very dynamic movements: it disappeared from the surface of ‘a’; while at the opposite side, a protruding structure extended over the surface of ‘p’ (8–10 min panels). Because it is unlikely that the GFP marker aggregated de novo during the 2 min interval, we presume that the protrusion over ‘a’ collapsed and then plunged into the surface of the opposite ‘p’ cell. If this is the case, such movements would have a speed approaching 3.0 µm/min (change of the distance from the protrud-

ing end to cell boundary per time interval). Reorganization of GFP-positive structures started to diminish around the end of the ventral cleft closure (22 min panel). These observations indicate that putative lamellar processes on the apical surface of VNBs are dynamic structures, undergoing continuous extension and retraction. Concomitantly, boundaries between VNBs also shift continuously. In all five animals whose development was successfully reconstructed in 4-D at similar stages, we observed similar activity at the apical surfaces of the VNBs. After this stage, the apical GFP signals became confined mostly to a thinner line representing the cell boundary, while some irregular patterns still persisted until the > 330 min stage (Fig. 6A,B). Epidermal ventral closure Finally, we visualized embryos undergoing epidermal ventral enclosure. Figure 6A shows a 330 min embryo, when the ventral cleft closure was complete and the epidermal cells extending from the dorsal side had started to cover the ventral surface of the embryo. The frontline of the protruding edges of epidermal cells (arrows and arrowheads) was circular and shrinking towards the central midline. At the 360 min stage (Fig. 6B-B4), the anterior two pairs of ‘leading cells’ were about to join at the middle line, while the remaining ‘pocket cells’ from the dorsal side were still half on the way to meet at the central midline. Strong signals were seen at the extending edges (arrows) of the uppermost two pairs of ‘leading

© 2007 The Authors Journal compilation © 2007 Japanese Society of Developmental Biologists

56

Z. Liu et al.

Fig. 6. Reconstructed embryos undergoing epidermal ventral enclosure. (A) The 330 min stage. In this and subsequent panels, asterisks indicate ‘leading cells’, the uppermost two pairs of epidermal cells protruding from the dorsal side. Arrows are pointing to their protruding edges. White arrowheads are pointing to the protruding edges of the ‘pocket cells’. (B) The 360 min stage. Strong signals are detected at the protruding tips of leading cells. (B1– B3) Individual scans at the focal depth near putative filopodia/lamellipodia from leading cells in the rectangular area in ‘B’. (B4) Schematic drawing summarizing ‘B1– B3’, which shows green fluorescent protein (GFP)positive lamellar structures extending from leading cells over the apical surface of underlying neuroblasts. (C) The 390 min stage when the epidermal ventral enclosure is almost finished. At the protruding end of the epidermal cells, where the strongest signals are detected, we often see accumulation of vesicle-like structures the marker is associating with (exemplified by ‘v’). (C1) Enlarged rectangular area in panel ‘C’. (C2) Average signal strength of the area between the two dashed lines in ‘C1’. Signals are stronger at the protruding end of leading cells. Embryos are shown head towards up.

© 2007 The Authors Journal compilation © 2007 Japanese Society of Developmental Biologists

Dynamic morphology of C. elegans neuroblasts

cells’ (asterisks). The sections with an interval of 0.2 µm better demonstrated the laminar structure of the extending edges of ‘leading cells’ (Fig. 6B1-B4). This observation is consistent with the CED-10 protein’s role as a regulator of cytoskeletal reorganization during cell movement (Kinchen et al. 2005). The timing and lamellar morphology of the structure match previously described filopodia/lamellipodia extended by leading cells (Mohle et al. 1998; Raich et al. 1999). At the 390 min stage (Fig. 6C), after the leading cells met at the middle line, the extending epidermal pocket cells met sequentially from anterior to posterior, before eventually sealing up the ventral opening. Within these epidermal cells, more vesicle-like structures were found on the extending side (Figs 6C1 and 2). Such distribution may reflect the requirement of adding cytoplasmic vesicles to the protruding cell membrane. The VNBs anterior to the leading cells remained exposed until being covered by head epidermal cells.

Discussion Visualizing cellular morphology of living embryo Although the method we introduce here requires CLSM, it is much simpler than other known methods such as electron microscopy or laser confocal microscopy (Krieg et al. 1978; Mohler & White 1998; Williams-Masson et al. 1998; Sharma-Kishore et al. 1999; Heid et al. 2001; Koppen et al. 2001; Lee & Goldstein 2003; Sawa et al. 2003). Despite its simplicity, our method achieved improved image quality, capturing subcellular structures such as intracellular vesicles and protrusions from living cells. AJM-1::GFP expressed at apical junctions is a widely used marker for visualizing cellular morphology (Simske & Hardin 2001). However, non-epidermal cells either do not have apical junctions (e.g. myoblasts and neuroblasts) or the localization of apical junctions does not always represent the shape of a whole cell (e.g. cells of the alimentary tract). The ubiquitous, strong and membrane-restricted expression made CED-10::GFP::CED-10 a good marker for investigating cellular morphology. While the image of planner expression of AJM-1::GFP can be handled by a conventional fluorescent microscope, imaging of the expression of CED-10::GFP::CED-10 on the entire cell surface requires the more complex CLSM to cut off signals from different focal planes. One or both markers may be chosen according to the investigator’s purpose. To obtain the best image quality, such as the 3-D reconstructions shown in Figures 3, 4 and 6, the samples were intensely scanned under slow mode,

57

at 0.2 µm intervals for about 30 µm in depth (≈150 sectional images for each whole scan). Obvious bleaching occurs with deterred development. Timelapsed imaging of developing animals was achieved by increasing scanning speed, the time interval between observations and the depth interval between focal planes, or by restricting the strength of the laser and the total scanning depth, or by combining all of the above. These modifications come at the cost of reduced image quality. Better quality scans would be possible for longer times by applying MPLSM, 3–6 h of development comprising up to 2800 images have been reported (Mohler & White 1998; Koppen et al. 2001). When the scanned images are stacked in a conventional way, the intricate cellular morphology inside the embryo is often far from obvious. We have chosen direct rendering (Amira, TGS) from raw data to reconstruct the scanned images, because segmentation that eliminates weak signals is avoided. The underlying model is based on the emission and absorption of light by every voxel of the view volume. This makes detailed subcellular structures, such as protrusions from VNB, distinguishable from the adjacent strong signals of cell boundaries. Weak signals can be detected under less intense laser scanning, facilitating in vivo observations. Further information on the 3-D cellular arrangement can be obtained when the models are rotated and sectioned on computer rather than those shown in still figures. Process-like structures of neuroblasts Our time-lapse analyses strongly suggest that VNBs undergo dynamic morphological changes. We found that VNBs appear to extend motile GFP-positive protrusions on its apical surface over neighboring VNBs. We also found that VNBs extend fine GFP-positive protrusions over Z2 cells. These fine cellular processes for VNBs are shown here for the first time, and we propose that they represent VNB filopodia/ lamellipodia. The GFP-positive protrusions are 0.2– 2.0 µm in length and 0.2–0.6 µm in thickness; the size matches filopodia/lamellipodia of other species (Miller et al. 1995; Kinchen et al. 2000; Iioka et al. 2004; Srinivas et al. 2004). They also resemble filopodia/lamellipodia of leading cells of C. elegans (Raich et al. 1999; Chisholm & Hardin 2005), have similar thickness and appear to spread over the surface of neighboring cells when observed with the GFP::CED-10::GFP marker (Fig. 6B1-3), although the protruding structures of leading cells are larger and more spread out than those of VNBs. We confirmed that at least some of the GFP-positive protrusions are

© 2007 The Authors Journal compilation © 2007 Japanese Society of Developmental Biologists

58

Z. Liu et al.

motile. Our preliminary estimation shows that they can move as fast as 3.0 µm/min, which is slower but within the order observed in other systems such as the filopodia of chick retinal and dorsal root ganglion (DRG) growth cones cultured in vitro (≈8.0 µm/min, Gehler et al. 2004); or that of primary mesenchyme cells in gastrulating sea urchin (≈10.0 µm/min, Miller et al. 1995; Jacinto & Wolpert 2001). A slower speed may be expected from the smaller size of VNBs, being less than one-tenth the volume compared to chick retinal and DRG cells or sea urchin mesenchymes. Filopodia and lamellipodia play essential roles in cell migration during the gastrulation of many animal species (Miller et al. 1995; Kinchen et al. 2005; Iioka et al. 2004; Srinivas et al. 2004). In C. elegans embryogenesis, they are important in the processes of dorsal intercalation and ventral enclosure by epidermal cells (Chisholm & Hardin 2005); defective formation of filopodia is known to block the entire enclosure (reviewed in Chisholm & Hardin 2005). VNBs at the left and right side move toward the central line and then join together at the posterior end of the ventral cleft, with closure proceeding anteriorly (Sulston et al. 1983; Chisholm & Hardin 2005). We found that the boundary between VNBs shifted its position continuously, which may be directly relevant to the migratory behavior of VNBs. The widely observed protrusions imply that most VNBs have the potential to send protrusions. The protrusions of VNBs were most frequently observed and behaved most actively during the 200–300 min stages, and this period largely overlaps with the ventral cleft closure of embryogenesis (Fig. 1). These facts suggest that putative filopodia/lamellipodia of VNBs may play roles in ventral cleft closure. For example, lamellipodia may be involved in the migration of VNBs and filopodia may be necessary for closing the cleft. Further analysis of the behavior of VNBs in wild type as well as mutant animals defective in these processes using the presented technique will help to reveal how these structures would participate in ventral cleft closure.

Acknowledgements We thank Professor M. Ishiura of the Center for Gene Research, Nagoya University for discussion, and we acknowledge the kind advice from the editors and anonymous referees. We also thank J. Koerwer of the Graduate School of Science, Nagoya University for his assistance with our English. This work is supported by the Ministry of Education, Culture, Sports, Science and Technology, Japan.

References Brenner, S. 1974. The genetics of Caenorhabditis elegans. Genetics 77, 71–94. Chisholm, A. & Hardin, J. 2005. Epidermal morphogenesis – WormBook. The C. elegans Research Community. [Cited 7 Oct 2006]. Available from URL: http://www.wormbook.org. Gehler, S., Gallo, G., Veien, E. & Letourneau, P. C. 2004. p75 neurotrophin receptor signaling regulates growth cone filopodial dynamics through modulating RhoA activity. J. Neurosci. 24, 4363 – 4372. Gilbert, S. F. 2000. Developmental Biology. Sinauer Associates, Sunderland. MA. Heid, P. J., Raich, W. B., Smith, R. et al. 2001. The zinc finger protein DIE-1 is required for late events during epithelial cell rearrangement in C. elegans. Dev. Biol. 236 (1), 165–180. Iioka, H., Ueno, N. & Kinoshita, N. 2004. Essential role of MARCKS in cortical actin dynamics during gastrulation movements. J. Cell Biol. 164, 169 –174. Jacinto, A. & Wolpert, L. 2001. Filopodia. Curr. Biol. 11, R634. Kinchen, J. M., Cabello, J., Klingele, D. et al. 2005. Two pathways converge at CED-10/Rac to mediate actin rearrangement and corpse removal in C. elegans. Nature 434, 93–99. Koppen, M., Simske, J. S., Sims, P. A. et al. 2001. Cooperative regulation of AJM-1 controls junctional integrity in Caenorhabditis elegans epithelia. Nat. Cell Biol. 11, 983 –991. Krieg, C., Cole, T., Deppe, U. et al. 1978. The cellular anatomy of embryos of the nematode Caenorhabditis elegans. Analysis and reconstruction of serial section electron micrographs. Dev. Biol. 65, 193–215. Lee, J.-Y. & Goldstein. B. 2003. Mechanisms of cell positioning during C. elegans gastrulation. Development 130, 307– 320. Liu, Z., Fujii, T., Nukazuka, A. et al. 2005. C. elegans PlexinA PLX-1 mediates a cell contact-dependent stop signal in vulval precursor cells. Dev. Biol. 282 (1), 138–151. Lundquist, E. A., Reddien, P. W., Hartwieg, E., Horvitz, H. R. & Bargmann, C. I. 2001. Three C. elegans Rac proteins and several alternative Rac regulators control axon guidance, cell migration and apoptotic cell phagocytosis. Development 128, 4475–4488. Miller, J., Fraser, S. E. & McClay, D. R. 1995. Dynamics of thin filopodia during sea urchin gastrulation. Development 121, 2501–2511. Mohler, W. A. & White, J. G. 1998. Multiphoton laser scanning microscopy for four-dimensional analysis of Caenorhabditis elegans embryonic development. Optics Express 3, 325– 331. Mohler, W. A., Simske, J. S., Williams-Masson, E. M., Hardin, J. D. & White, J. G. 1998. Dynamics and ultrastructure of developmental cell fusions in the Caenorhabditis elegans hypodermis. Curr. Biol. 8, 1087–1090. Raich, W. B., Agbunag, C. & Hardin, J. 1999. Rapid epithelialsheet sealing in the Caenorhabditis elegans embryo requires cadherin-dependent filopodial priming. Curr. Biol. 9, 1139 –1146. Ridley, A. J. 2001. Rho proteins: linking signalling with membrane trafficking. Traffic 2, 303 – 310. Sawa, M., Suetsugu, S., Sugimoto, A., Miki, H., Yamamoto, M. & Takenawa, T. 2003. Essential role of the C. elegans Arp2/3 complex in cell migration during ventral enclosure. J. Cell Sci. 116, 1505–1518. Sharma-Kishore, R., White, J. G., Southgate, E. & Podbilewicz, B. 1999. Formation of the vulva in Caenorhabditis elegans:

© 2007 The Authors Journal compilation © 2007 Japanese Society of Developmental Biologists

Dynamic morphology of C. elegans neuroblasts

a paradigm for organogenesis. Development 126, 691– 699. Simske, J. S. & Hardin, J. D. 2001. Getting into shape: epidermal morphogenesis in Caenorhabditis elegans embyros. BioEssays 23, 12–23. Srinivas, S., Rodriguez, T., Clements, M., Smith, J. C. & Beddington, R. S. P. 2004. Active cell migration drives the unilateral movements of the anterior visceral endoderm. Development 131, 1157–1164.

59

Sulston, J. E., Schierenberg, E., White, J. G. & Thomson, J. N. 1983. The embryonic cell lineage of the nematode Caenorhabditis elegans. Dev. Biol. 100, 64–119. Williams-Masson, E. M., Heid, P. J., Lavin, C. A. & Hardin, J. 1998. The cellular mechanism of epithelial rearrangement during morphogenesis of the Caenorhabditis elegans dorsal hypodermis. Dev. Biol. 204 (1), 263 – 276. Wood, W. B. 1988. The Nematode Caenorhabditis elegans. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.

© 2007 The Authors Journal compilation © 2007 Japanese Society of Developmental Biologists