QUICK SEARCH:   [advanced] Author: Keyword(s):
Home     Help     Feedback     Subscriptions     Archive     Search     Table of Contents

doi: 10.1242/10.1242/dev.00314

This Article Summary Full Text Full Text (PDF) Movies Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Glickman, N. S. Articles by Adams, R. J. Search for Related Content PubMed PubMed Citation Articles by Glickman, N. S. Articles by Adams, R. J.

Shaping the zebrafish notochord

Nathalia S. Glickman^1,*, Charles B. Kimmel¹, Martha A. Jones¹ and Richard J. Adams^2,,

¹ Institute of Neuroscience, University of Oregon, Eugene, OR 97403-1254, USA
² Department of Biology and Biochemistry, University of Bath, Bath BA2 7AY, UK
^* Present address: Developmental Genetics Program and Department of Cell Biology, Skirball Institute of Biomolecular Medicine, New York University School of Medicine, New York, NY 10016, USA
Present address: Department of Anatomy, University of Cambridge, Downing Street, Cambridge CB2 3DY, UK

View larger version (18K): [in a new window]   Fig. 1. The mediolateral intercalation behavior (MIB) hypothesis. (A) Early protrusive activity is random. (B) The cells take on a bipolar shape as their protrusions are restricted to the ML axis. Protrusive activity is not present at the notochord/somite boundaries (to the sides). (C) The field narrows and elongates (converges and extends) as the cells exert traction upon one another and pull together. Adapted, with permission, from Shih and Keller (Shih and Keller, 1992a).

View larger version (78K): [in a new window]   Fig. 3. Convergence and extension of zebrafish dorsal mesoderm. Time sequences taken from a single 4-hour 4D recording of a wild-type embryo, beginning at midgastrula stage (7.3 hpf) (Kimmel et al., 1995) (see Movies 2-5 at http://dev.biologists.org/supplemental/). The AP axis is vertical (anterior towards the top), and the ML axis horizontal, here and in Figs 5,6,7, Fig. 9, Fig. 11. The field is centered approximately on the dorsal midline. (A) Views from the recording, at 7.3 (A1), 8.8 (A2), 9.3 (A3) and 11.3 (A4) hpf (hours postfertilization). Negative confocal microscope images of the BODIPY-ceramide labeled cellular field are shown. At the first time point (A1) the blastoderm margin is evident, separating the cellular blastoderm (upper) from the yolk syncytial layer (YSL, lower). Subsequently, the blastoderm comes to cover the yolk completely by the spreading movement of epiboly. Brachet's cleft, which appears as a hazy ring (arrows, A2), separates ectoderm to the outside, and mesoderm to the inside. The notochord/somite boundaries (axial/paraxial boundaries: arrowheads, A3) appear in the mesoderm. They are barely visible in A2 and then become prominent. Convergence narrows the notochord domain to about 2 cells wide at the last time point (A4). At the lateral side of each boundary lies a distinctive row of somitic adaxial cells that will form slow muscle (Blagden et al., 1997; Devoto et al., 1996). In addition, the somite boundaries are visible and are marked by arrows. (B) About 200 cells, represented as spheres, were tracked from the recording and are shown at 7.3, 8.3, 8.8, 9.3, 10.3 and 11.3 hpf. Cells are color-coded according to their eventual fates: notochord-forming cells (green), adaxial cells (dark blue), cells that form somite 2 (yellow) and other somite-domain cells (red). Nearly all of the cells divided during course of the 4 hour recording. Almost without exception, both siblings ended up in the same domain, and we color-coded the mother cell identically to its daughters. The notochord, and somite domains are spatially separate from the outset of the recording, even though distinctive boundaries are not yet present (A1). (C) Notochord-forming cells (green) and overlying floorplate-forming cells (magenta) shear relative to one and other during extension (48 floor plate-forming cells were tracked). The two cell types were collected from separate focal planes of the original 4D recording, and a 30 µm stripe cut at a single AP level is shown at 7.3, 8.3, 9.3 and 11.3 hpf. (D) The notochord domain (green) extends more than the somite domains (red). A 30 µm horizontal stripe of mesoderm cut from the field and is shown at 7.3, 9.3 and 11.3 hpf. Scale bar: 50 µm.

View larger version (52K): [in a new window]   Fig. 2. Predicted rearrangements and movement kinetics in a field of hypothetical `cells' uniformly expressing MIB (see Movie 1 at http://dev.biologists.org/supplemental/). (A-D) Time course of rearrangements of the cells (spheres) undergoing convergence and extension. The field halves in width and doubles in length over the period of A to D; the intermediate stages (B,C) represent the field at one- and two-thirds of completion of the round of MIB. The area and thickness of the field does not change, and MIB is constant with respect to cell position and time. (E,F) Close-up views near the field center (*) at the beginning (E) and end (F) of the same sequence. See the text for description of the local cellular reorganizations. (G) The cellular flows resulting from intercalations occurring between times B and C. The most rapidly moving cells (longest lines) are those most distant from the center. (H,I) Linear gradients relate ML velocity with ML position (H, convergence) and AP velocity with AP position (I, extension). The red arrows indicate the directions and rates of cell movement (ML movement is inwards, AP movement is outwards) such that the gradient slopes are opposite in sign. (J,K) The field width decreases exponentially (J), and the field length increases exponentially (K). Cell position at a given time t can be predicted by the relationships xt=x0ekct and yt=y0eket.

View larger version (15K): [in a new window]   Fig. 4. Cellular kinetics during convergence and extension. (A) Kinetics of notochord domain convergence follows exponential decay. The rate of narrowing is represented by a single exponential fit to the data. The rate constant kC is -0.0064. (B,C) Velocity versus position is linear for both convergence (B; kC=-0.0059) and extension (C) (kE=0.0062 for the notochord, green circles; kE=0.0013 for the somite domain, red circles). The individual points represent velocities and locations of the entire field of tracked cells from one time sample, at 9 hpf (midgastrula stage). At this time point, kC, is slightly lower than the value determined from the decreasing width of the field in A (-0.0059 min-1 versus -0.0064 min-1). We observed that under the conditions of our recordings, kC, measured as the slope of the gradient, is not stable time point to time point, rather the slope is `jittery'. An `average'/integrated estimate (see Materials and Methods) over the entire time period is -0.0073 min-1, agreeing within 15% with the calculation from the notochord's change in width (A). The line of best fit was determined using linear regression. Gradients were assumed to fit when r2 showed evidence for substantial explanation of variance and the F-statistic exceeded the P=0.05 significance threshold. The blue arrow in B indicates the position of the midline; here, the ML velocity is approximately zero.

View larger version (25K): [in a new window]   Fig. 5. MIB can account for all or most organized cellular movement within the wild-type notochord domain. Each purple line shows the observed movement pathway of individual cells over a 16 minute interval. The first 16 minutes of cell movement (A) and the last 16 minutes of cell movement (B, 4 hours later) of the recording used for Fig. 3 are shown. The yellow lines show the essentially random movements remaining after using the values of the convergent and extension rate constants to subtract the components of the cell movements due to MIB at every time point. At the beginning of the recording, before significant tissue shape changes have occurred (A), there is little difference between the location of the observed positions (purple) and the calculated positions with MIB removed (yellow). At the end of the 4 hour period, the observed field has undergone a significant shape change, convergence and extension (B, purple) but with MIB removed, the field has hardly changed shape at all. See Movie 6 at http://dev.biologists.org/supplemental/.

View larger version (116K): [in a new window]   Fig. 6. Morphogenesis of the dorsal mesoderm is disrupted in the no tail/Brachyury mutant. Negative images of BODIPY-ceramide labeled preparations of wild type (A-C) and ntl mutants (E-G) at 9.5 (A,E), 10.2 (B,F) and 11 hpf (C,G). D1-5 show a through focus series (3 µm steps, shallow to deep) at the level of the second somite in another wild-type embryo at 11 hpf. D1 and D2 pass through the floor-plate region of the neural tube. D3 passes through the notochord-floor plate boundary. D4 and D5 pass through the notochord. In wild-type embryos, the notochord/somite boundary is apparent at 9.5 hpf (A), but in the ntl mutant a boundary between axial and paraxial domains begins to form only at 10.2 hpf (F). Cells become oriented along the ML axis, and exhibit a wedge-shape in the wild-type notochord-domain (B,C,D4,D5) but not in the mutant axial domain (E,F). Scale bar: 25 µm.

View larger version (49K): [in a new window]   Fig. 7. Loss of function of ntl disrupts convergence but not extension. See Movies 7-12 at http://dev.biologists.org/supplemental/. The wild-type data are from the same 4D recording used for Fig. 4. Time points shown across A-C are 8, 8.8 and 9.5 hpf. (A) Cellular movement pathways in the notochord/axial (green) and somite domains (red). The arrows show the directions of cell movements and their lengths show the speed (the movement history during 16 minutes). Disruption of convergence (i.e. failure of the field to narrow) is evident in the ntl- axial domain: kC for the axial domain in the mutant is -0.0017, when compared with -0.0073 for the wild-type notochord domain over the same time interval (8-9.5 hpf). Horizontally (ML) oriented tracks are largely missing in ntl mutant. However, extension, vertical lengthening of the field, is prominent in the mutant. (B) Intermixing does not occur between the notochord/axial domain (green) and the overlying midline epiblast (floorplate domain, magenta) in either the wild-type or ntl mutant (48 and 84 cells traced in wild-type and ntl epiblast, respectively). Side views made by 90° rotation of vertical strips of cells in the tracked data sets that are present at the dorsal midline at each time point. Accompanying these images, similar rotations from the original recorded BODIPY images reveal the boundary (Brachet's cleft) between the two layers of cells in the wild type (notochord and floorplate), but this boundary is less apparent in the mutant. (C) The thickness of the region of tracked dorsal mesodermal cells does not change greatly during convergence and extension in the wild type, and may decrease slightly during extension without convergence in the ntl mutant axial domain (green). Scale bar: 50 µm.

View larger version (11K): [in a new window]   Fig. 8. Quantitative comparisons of features of morphogenesis of the wild-type notochord domain (blue) and the ntl mutant axial domain (red) during gastrulation and early segmentation stages. Values in A-C are normalized to 1.0 at the first time point. (A) Convergence: the domain widths are computed from values of kC, estimated as in Fig. 3 at each time point (see Materials and Methods). Convergence is markedly decreased in the mutant until about 9.5 hpf. (B) Extension, domain lengths are computed from values of kE, estimated as in Fig. 3 at each time point. In ntl mutants the axial domain extends rapidly during the same interval when the convergence is occurring slowly or not at all. (C) Area=widthxlength. The area is not expected to change if MIB alone underlies convergence and extension. Area only slightly decreases in the wild type, but increases in the mutant. (D) Cellular densities (number of cells per unit area). The density in wild type remains approximately constant during gastrulation and then increases. In the mutant, the density decreases, as would be expected if the field is increasing in area, extending but not converging.

View larger version (49K): [in a new window]   Fig. 9. (A) Extensive cellular mixing along the ML axis occurs in ntl-, but not in wild-type embryos. Fields are shown at 8, 8.8 and 9.5 hpf, with the spheres representing tracked cells color-coded to produce vertical stripes (30 µm wide) at the first time (8 hpf). To facilitate comparison with ntl-, only the medial portion (size matched with the ntl- field) of the tracked field is shown for the wild type. The stripes all narrow in the wild type and very little mixing is occurring between the stripes. In ntl-, the stripes mix rather than narrow. (B) Cells move toward the midline in wild type, but move in a disorderly way with respect to the midline in ntl-. Red and blue circles show cells moving to the right and left, respectively; green circles show cells with no left-right component to their movements at the representative time point illustrated (8.8 hpf). The center of each field corresponds the approximate midline for both A and B. See Movies 13-16 at http://dev.biologists.org/supplemental/.

View larger version (28K): [in a new window]   Fig. 10. Cells intercalate mediolaterally in both the wild-type and ntl mutant embryo (A,B), but only during brief intervals in the mutant, when compared with the wild type (C). (A) Intercalating cells become neighbors (blue, neighbor gains), pushing apart old neighbors (red, neighbor losses). Hence, for cells undergoing MIB, neighbor gains are ML (horizontal in the diagram) and neighbor losses AP (vertical; see also Fig. 2). (B) The polar plots show that these predicted behaviors are observed; the distributions are evidently broadened for the mutant. All four distributions are nonrandom (P<0.05; Watson's U2n test). The distributions are based on 623 gains and 633 losses for the wild type, and 1473 gains and 1822 losses for the mutant. (C) Colored bars show the durations of time periods when neighbor gains and losses are significantly oriented along any axis (Watson's test for nonrandom data). In the wild type, these periods of highly oriented activity are sustained during most of the time interval sampled. In the mutant, pulses of oriented and random behaviors are interspersed.

View larger version (66K): [in a new window]   Fig. 11. Local cellular movement correlates of (A) convergence and extension in the wild-type embryo, and (B) of extension without convergence in the ntl mutant. The sets of panels show time sequences, at 4 minute intervals, for the notochord (WT)/axial (mutant) domains. For every cell in the field in turn (`reference cell', the point at the center of each plot), we calculate the displacements of all other notochord-domain cells relative to this reference cell, i.e. the cell movement toward or away from the reference cell. This is repeated for every cell in the notochord domain, each cell in turn taking the role of a reference cell. The data are then averaged for each time point, and the resulting averaged plots are shown. The arrows represent the average directions and speeds (length of arrow) of these relative movements; their placements on the grid are according to the positions of the cells relative to the reference cell. Inward cellular movements (i.e. horizontal arrows point towards the reference cell) correlate with convergence of the field, and outward movements (vertical arrows point away from the cell) correlate with extension. Inward movements are not sustained in the ntl mutant. Positions along the x-axis reflect positions along the AP axis relative to the reference cell while positions along the y-axis reflect positions along the ML axis relative to the reference cell. See Movies 17 and 18 at http://dev.biologists.org/supplemental/.