May 9, 2006;
Integrin alpha5beta1 and fibronectin regulate polarized cell protrusions required for Xenopus convergence and extension.
Integrin recognition of fibronectin
is required for normal gastrulation including the mediolateral cell intercalation behaviors that drive convergent extension and the elongation of the frog dorsal axis; however, the cellular and molecular mechanisms involved are unclear.We report that depletion of fibronectin
with antisense morpholinos blocks both convergent extension and mediolateral protrusive behaviors in explant preparations. Both chronic depletion of fibronectin
and acute disruptions of integrin alpha5beta1 binding to fibronectin
increases the frequency and randomizes the orientation of polarized cellular protrusions, suggesting that integrin-fibronectin
interactions normally repress frequent random protrusions in favor of fewer mediolaterally oriented ones. In the absence of integrin alpha5beta1 binding to fibronectin
, convergence movements still occur but result in convergent thickening instead of convergent extension.These findings support a role for integrin signaling in regulating the protrusive activity that drives axial extension. We hypothesize that the planar spatial arrangement of the fibrillar fibronectin
matrix, which delineates tissue
compartments within the embryo
, is critical for promoting productive oriented protrusions in intercalating cells.
[+] show captions
Figure 1. FN Antisense Morpholino “Knocks down” FN Protein Synthesis, Reduces Fibril Formation, and Reduces Dorsal Axis Extension(A) Western blot analysis of morpholino (MO)-injected embryos harvested at stage 10.5 and processed for detection with the anti-FN mAb 4H2 and the anti-integrin β1 PcAb 363. The two FN bands correspond to two major alternatively spliced FN subunits expressed during development. The control MO has no affect on the levels of expression of either FN or integrin β1 at up to 15 μMoles MO injected per embryo. A dose-dependent reduction in FN protein expression is noted in the presence of the combined (50:50) FNMOs. Levels of integrin β1 protein expression are unaffected by the FN antisense MO.(B) Expression of chordin, a marker of dorsal axial tissues, shows that embryos injected with 15 μMoles FNMO have wide dorsal axial tissues. (Arrows mark the width of the chordin expression.)(C) Transverse confocal sections of embryos stained for FN fibrils show a typical pattern in both control morpholino and uninjected control embryos while fibril formation is almost completely abolished (see arrowheads for an example of faint labeling of blastocoel roof ectoderm in FNMO-injected embryos).(D) FNMO-injected embryos show defects in both dorsal and ventral elongation by early tadpole stages and severe anterior mesoderm defects by later tadpole stages. At early tadpole stages, the dorsal axial extension is moderately reduced while ventral extension is severely reduced (asterisk). Late-stage FNMO-injected tadpoles show defects in ventral mesoderm morphogenesis, ectodermal lesions, and failure of heart (arrow) and gut (arrowheads) formation.(E) Pooled data from three batches of embryos were observed over several days and their progress was assessed. None of the embryos injected with 15 μMoles FNMO develop beating hearts.
Figure 2. FN Knockdown Retards Blastopore Closure, Slows Convergence and Extension, and Produces Tissue Thickening and an Excess of Actin-Rich Cellular Extensions in Whole Embryos
(A) Frames from a representative time-lapse sequence show that blastopore closure is delayed in 15 μMoles FNMO-injected embryos until sibling control embryos reach equivalent midgastrula stage 13.
(B) Anterior-posterior length to the mediolateral width ratios of the noninvoluting marginal zone of control morpholino-injected embryos (open circles), uninjected embryos (filled circles), or FNMO-injected embryos (filled triangles) show that FNMO-injected embryos lag 2 hr behind control embryos.
(C) Transverse confocal section through the dorsal axis of an embryo injected contralaterally with FNMO along with a FITC-dextran lineage tracer shows the absence of fibrils is accompanied by tissue thickening on the injected side.
(D and E) A maximal projection of a 30 μm thick sagittal confocal stack of a region just animal of the blastopore lip at stage 11.5 show that actin-rich cellular extensions are abundant in FNMO-injected embryos (D) but far less common within uninjected embryos (maximal projection of a 60 μm sagittal confocal stack; [E]).
Figure 3. FN Knockdown Retards Convergence and Extension in Sandwich Explants
(A) Representative 6-hr-old Keller sandwich explants from 15 μMoles FNMO-injected embryos show convergent thickening but little extension, while uninjected sandwiches and control morpholino-expressing explants (COMO) converge and extend.
(B) Individual explants showing convergence of the mesoderm with respect to the ectoderm (arrowheads versus arrows) were stained for fibronectin fibrils.
(C) Montages of en face maximal projections of confocal z-series of FN fibrils from the same explants shown in ([B]; dashed boxes) show that fibrils do not assemble in FNMO-expressing explants while fibrils assemble in control explants.
(D) XZ-projections analogous to transverse projections of confocal z-series (shown by white boxes in [C]) show that fibrils assemble within control explants but not within the thicker sandwich explants made from FNMO-injected embryos. Arrow-ended lines indicate thickness of explant.
(E) Measurements of explant thickness from five batches of sandwich explants demonstrate FNMO-expressing explants are significantly thicker than sibling control or COMO-injected explants.
Figure 4. Mediolaterally Oriented Cell Elongation Can Be Rescued by Planar FN
(A–D) Mesoderm cells within confocal sections collected at 6 μm depth from 6-hr-old control uninjected or COMO-injected explants expressing a membrane localizing GFP exhibit characteristic bipolar shape with length to width ratios (LWR) greater than 2.0 when cultured either on BSA (A and C) or on FN (B and D). Cells that exhibit high LWRs align strongly to the mediolateral axis.
(E) Explants from embryos injected with 15 μMoles of FNMO do not develop bipolar cell shapes, exhibit a low LWR when cultured on BSA, and do not align to the mediolateral axis.
(F) However, when presented with a FN-coated glass coverslip cells intercalate, become elongated with LWRs similar to control explants on FN, and align strongly to the mediolateral axis.
Rose diagrams in each panel show the angular distribution of cells' long axes with respect to the mediolateral axis. The ellipse in each panel represents the average cellular LWR. The error is the standard deviation for the number of cells measured. Yellow arrows indicate orientation of the long axis in each individual cell. In each case, analyzed cells come from a single representative explant.
Figure 5. Two-Plane Confocal Time-Lapse Sequence of MZ Explants Cultured on FN Show Cell Behaviors during Mediolateral Cell Intercalation
(A–C) Optical sections from the plane of the FN-coated coverslip (A) and from 5 μm deeper in the explant (B) were collected at 15 s intervals and combined into a two-color time-lapse sequence (C) where cell identities and cell shapes were correlated with protrusions. Our dual-level imaging technique is required for the analysis of cellular behaviors, for instance confirming the identity of cellular protrusions in the neighborhood of triple-cell junctions, and is critical to our interpretation of cellular protrusions over the time course of an acute experiment, for instance in determining the onset of explant shear, which can produce shingled cells that are easily mistaken for monopolar protrusive cells.
(D) Single lamellipodia can be identified (arrowhead), their duration measured, and their orientation determined with regard to both the cell and the mediolateral axis of the explant.
(E) A frequency distribution of 166 lamellipodia extended on the FN-coated coverslip is shown for 32 cells in the reference explant shown in (A).
(F) The life-time distribution, from initiation to retraction, of these lamellipodia.
(G and H) Rose diagrams demonstrate that most lamellipodia at the surface (G) are mediolaterally oriented, as are deep cell extensions shown in (B) at the 5 μm level (H).
(I) Long-lived lamellipodia (black) are found at the mediolateral ends of cells whereas short-lived lamellipodia (gray) are more randomly oriented. Scale bar in (C) is 25 μm and in (D) is 10 μm (a, anterior; p, posterior; ml, mediolateral). This time-lapse sequence can be viewed in Movie S2.
Figure 6. Mediolateral Cell Intercalation Behaviors in FN Knockdown Explants Are Rescued when Cells Provided with a Planar FN-Substrate while Blocking Integrin α5β1 Stimulates Cell Protrusive Activity
(A and B) Two frames from representative confocal time-lapse sequences of 15 μMoles FNMO-injected explants expressing GAP43-GFP and their calculated “difference” images were calculated for explants cultured on BSA ([A]; see Movie S3) and on fibronectin ([B]; see Movie S4). A “difference time-lapse” sequence shows regions of cell protrusion and retraction as solid white and black regions, respectively, while regions where little change occurs are typified by gray.
(C) Culture of FNMO-injected explants on FN significantly represses the rate of protrusions.
(D–F) Two frames from a representative confocal time-lapse sequence and their calculated “difference” image before (D) and after ([E]; see Movies S5 and S6 for difference time-lapse sequence) addition of 0.1 μg/ml of the integrin α5β1 function-blocking antibody P8D4 to the explant show that protrusive behaviors are stimulated and the rate of protrusive activity increases (F).
The time-lapse sequences, in two-level form, and the difference time-lapse sequence can be viewed in Supplemental Data.
Figure 7. Integrin “Superactivation” Represses Protrusive Activity
(A and B) Two frames from a representative confocal time-lapse and their calculated difference image before (A) and after (B) superactivation of integrins after addition of 1 mM MnCl2 demonstrate repression of protrusive activity.
(C) Rates of protrusive activity drop 3-fold after addition on MnCl2.
(D and E) Two frames from a representative confocal time-lapse sequence and their calculated difference image before (D) and after (E) superactivation of integrin after addition of 1 mM MnCl2 demonstrate repression of protrusive activity in explants already incubated for 1 hr in 0.1 μg/ml of the integrin α5β1 function-blocking antibody P8D4.
(F) Addition of MnCl2 results in a 2-fold reduction in rates of protrusions.
Figure 8. (A) Control of protrusive rates and orientation by FN/integrin interactions. Superactivation of integrins with Mn2+ results in reduced protrusive rates whereas removal of FN fibrils or inhibition of α5β1 enhances the frequency of protrusions and results in loss of mediolateral orientation.
(B) Effective convergence and extension requires maintenance of a two-cell layered mesoderm. Closer inspection of mediolateral cell intercalation (within circle) reveals an underlying cellular mechanism.
(C) Tractive protrusions (red arrows), balanced along mediolateral contacts with neighboring cells, directs cell intercalation and separates neighboring cells along the anterior-posterior axis (a, anterior; p, posterior). Balanced, directed intercalation maintains the two-cell layers of the mesoderm and leads to the efficient conversion of convergence into axial extension.
(D) Without a balance of forces, tractive protrusions may be directed either over the top of neighboring cells or underneath neighboring cells (single red arrows). In this case, cells converge to the midline but become multilayered. Without maintenance of two-cell layers, cell intercalation results in tissue thickening.
All views are “cut-away” transverse perspectives, centered on the notochord where the mediolateral axes run left and right and the anterior posterior axis runs toward and away (no, notochord; so, somitic mesoderm; ne, neural ectoderm; en, endoderm; d, dorsal; v, ventral).