Regulated tissue fluidity steers zebrafish body elongation

Abstract

The tailbud is the posterior leading edge of the growing vertebrate embryo and consists of motile progenitors of the axial skeleton, musculature and spinal cord. We measure the 3D cell flow field of the zebrafish tailbud and identify changes in tissue fluidity revealed by reductions in the coherence of cell motion without alteration of cell velocities. We find a directed posterior flow wherein the polarization between individual cell motion is high, reflecting ordered collective migration. At the posterior tip of the tailbud, this flow makes sharp bilateral turns facilitated by extensive cell mixing due to increased directional variability of individual cell motions. Inhibition of Wnt or Fgf signaling or cadherin 2 function reduces the coherence of the flow but has different consequences for trunk and tail extension. Modeling and additional data analyses suggest that the balance between the coherence and rate of cell flow determines whether body elongation is linear or whether congestion forms within the flow and the body axis becomes contorted.

Fig. 1.

Cell movement in the wild-type zebrafish tailbud. (A) Confocal image stacks of RFP-labeled nuclei were taken through time and nuclei were tracked in 4D. (B) Independent tracking of red and green double-labeled nuclei revealed cell tracking to be 85% accurate. (C) Cell tracks were divided into regions of interest: dorsal medial zone (DM; red), progenitor zone, (PZ; green), presomitic mesoderm (PSM; cyan). (D) A lateral view of the image shown in C. (E) Examination of mean track speeds. P-values calculated using ANOVA with additional validation by permutation tests. The DM and PZ speeds are equivalent (DM vs PZ, P>0.05) and are higher than PSM (DM vs PSM, P<0.05; PZ vs PSM, P<0.05). (F) The means of the coefficient of variation (CV) of track speed show that the CV increases from the DM to PZ (P<0.05) and from the PZ to PSM (P<0.05). (G-I) Two-photon labeling of DM cells at the ten-somite stage. (G) Dorsal view. (H) Lateral view. (I) After completing elongation, the uncaged cells differentiated as myofibers. A-C and G are dorsal views. In all panels, anterior is left the right except panel I in which anterior is up.

Fig. 2.

Cell flow in the zebrafish tailbud. Wild-type (A,D,G,J), notum1a-overexpressing (B,E,H,K) and SU5402-treated (C,F,I,L) embryos. (A-C) The vector displacement map averages cell motion in discrete sectors. The heat map indicates mean speed with warmer colors indicating higher speeds and arrows signifying averaged 3D velocity vectors. See also supplementary material Fig. S3. (D-I) The top 10% of individual PZ tracks exhibiting the greatest displacement in each direction (green, dorsal to ventral; yellow, medial to lateral; red, posterior to anterior; blue, ventral to dorsal). See also supplementary material Fig. S4. (J-L) 3D finite element method was used to measure the local rotation in the flow of cells. Arrows slanting rightward indicate a dorsal-to-ventral local rotation whereas leftward-slanting arrows signify a ventral-to-dorsal local rotation.

Fig. 3.

Quantitative metrics reveal changes in tissue fluidity. (A) A vector map of cell velocities. Warmer colors represent higher velocities and arrows indicate the 2D projection of the averaged 3D velocity vectors. (B) A vector map of local tissue velocity averaged over a 15 μm radius. (C) A vector map of cell velocity variation from local tissue velocity. (D-O) Data for four wild-type zebrafish embryos (D-G), three notum1a-overexpressing embryos with elongation defects (H-K) and three SU5402-treated embryos with elongation defects (L-O). (D) The MSD exponent indicates that the movement is a combination of ballistic and Brownian motion for all regions with the DM being the most directed (DM vs PZ, P<0.05; DM vs PSM, P<0.05). Motion in the DM is strongly polarized (DM vs PZ, P<0.05; DM vs PSM, P<0.05) (E), has the largest length scale over which direction of motion is correlated (DM vs PZ, P<0.05; DM vs PSM, P<0.05) (F) and exhibits highest neighbor similarity (G). (H) MSD is not significantly changed in notum1a-overexpressing embryos (DM, P>0.05). However, notum1a reduces the polarization (P<0.05) (I), correlation length (P<0.05) (J) and neighbor similarity (K) of the DM. (L) MSD in the DM is reduced in SU5402-treated embryos (P<0.05). In addition, SU5402 treatment reduces the polarization (P<0.05) (M), correlation length (P<0.05) (N) and neighbor similarity (O) of the DM. P-values calculated using Student's t-test.

Fig. 4.

Ectopic notum1a expression or SU5402 treatments perturb axis elongation. (A-M) Wild-type (A-C,F,K), notum1a-overexpressing (D,E,G,L) and SU5402-treated (H-J,M) zebrafish embryos. (C) notum1a is transcribed in the posterior DM. (D) Ectopic notum1a gives rise to a shortened axis and malformed tailbud. (E) Half of these embryos develop severely misdirected body axes. Spatial expression of tbx6 (F-H) and neurogenin1 (K-M) remains relatively normal despite notum1a overexpression (G,L) or SU5402 treatment (H,M). (I,J) SU5402 treatment gives rise to a mild axis elongation defect. (N) qPCR of dissected tailbuds after notum1a overexpression reveals reductions in transcription of dkk1 and tbx6. Student's unpaired t-test. *P≤0.05. n=4. Error bars represent s.e.m.

Fig. 5.

Loss of cadherin 2 reduces the effectiveness and coherence of cell motion in the zebrafish tailbud. (A,B) cdh2 mutants have morphologically abnormal tailbuds (A) and tails (B). (C-J) Quantitative analysis of cell motion in the tailbud of three wild-type embryos (C-F) and three cdh2 mutants (G-J). (G) MSD is reduced in the DM of cdh2 mutants (P<0.05). (H) Polarization is reduced in the DM (P<0.05), PZ (P<0.05) and PSM (P<0.05) of cdh2 mutants. (I) Correlation length is lower in the DM (P<0.05). (J) Neighbor similarity is diminished in the DM of cdh2 homozygotes. (K-N) Adherens junctions in the PZ are visualized by immunohistochemistry for β-catenin. All panels are at the same magnification; scale bar: 5 μm. The lattice of adherens junctions among cells is present in wild-type embryos (K) but is disrupted in cdh2 mutants (L). The pattern of adherens junctions in notum1a-overexpressing embryos (M) and SU5402-treated embryos (N) resembles the wild-type pattern. P-values calculated via Student's t-test.

Fig. 6.

Simulation of cell flow during trunk elongation. (A-C) Sample snapshots of cell positions obtained using computer simulations of a 2D model system of self-propelled particles with soft interactions. The upper portion of the vertical column of the ‘T’ represents the DM whereas the intersection between the vertical and horizontal sections is representative of the PZ. Arrows denote the instantaneous velocity of each cell. Cell density is displayed as a heat map with warmer colors indicating higher density. See also supplementary material Fig. S8. (A) Wild-type simulation. (B) In ectopic notum1a embryos, cells are introduced at the lower column at the same rate as in wild type, but ectopic notum1a causes cells to move with higher angular noise (now the same as that of the PZ). (C) To model inhibition of Fgfr signaling by SU5402 treatment, the rate of addition of cells to the lower column is reduced to half the rate in A and B. Cells move with same randomness as in the ectopic notum1a embryos, but the attenuated cell flow into the DM reduces the number of cells in the DM. The polarization (Φ), standard deviation (std) of angles of posterior migration and cell number of the DM (vertical column) are the averages from 100 simulations for each of the three conditions.

Fig. 7.

Axis extension requires balance of cell flow rate and flow coherence. Wild-type (A,D,G,J,M), notum1a-overexpressing (B,E,H,K,N) and SU5402-treated (C,F,I,L,O) zebrafish embryos. In A-I, values for ADM and DM are connected by a line, in order to compare the two domains in the same embryo. Thus, each line represents a single embryo. (A-C) Polarization (Φ). (D-F) Wild-type ADM and DM (D) show low variance in the angle of posterior migration (ϕ). The variance is not significantly changed in notum1a-overexpressing embryos (E) but the ADM variance is increased in SU5402-treated embryos (F) relative to wild type (P<0.05), indicating progressively larger deviation from direct posterior migration. (G-I) The MSD is similarly high in the wild-type DM and ADM (G). notum1a-overexpressing DMs and ADMs have relatively high MSD, statistically indistinguishable from wild type, indicating continued effective migration (H). MSD is reduced in SU5402-treated embryos in the ADM compared with wild type (P<0.05) (I). (J) The wild-type rate of ADM convergence and extension. (K,L) notum1a-overexpressing ADMs (K) display wild-type levels of convergence and extension, but SU5402-treated ADMs (L) show reduced extension (P<0.05). P-values calculated using Student's t-test. (M) In wild type, a strong flow in the ADM and collective migration in the DM places cells at the posterior tip of the embryo. In the PZ, the coherence of the flow diminishes, which facilitates cell mixing and bilateral distribution of cells in the PZ. (N) In notum1a-overexpressing embryos, the coherence in the DM is prematurely lost but the flow rate in the ADM is maintained. Cells are unable to move through the tailbud and a ‘traffic jam’ forms leading to a bump and a turning of the axis. (O) In SU5402-treated embryos, the coherence in the DM is strongly reduced, as is the flux in the ADM. The latter reduction curtails the propensity of the flow to jam, thus cells pass through the PZ, albeit at a reduced rate, leading to a straight but shorter axis than that of wild type.