In vivo topology converts competition for cell-matrix adhesion into directional migration

Abstract

When migrating in vivo, cells are exposed to numerous conflicting signals: chemokines, repellents, extracellular matrix, growth factors. The roles of several of these molecules have been studied individually in vitro or in vivo, but we have yet to understand how cells integrate them. To start addressing this question, we used the cephalic neural crest as a model system and looked at the roles of its best examples of positive and negative signals: stromal-cell derived factor 1 (Sdf1/Cxcl12) and class3-Semaphorins. Here we show that Sdf1 and Sema3A antagonistically control cell-matrix adhesion via opposite effects on Rac1 activity at the single cell level. Directional migration at the population level emerges as a result of global Semaphorin-dependent confinement and broad activation of adhesion by Sdf1 in the context of a biased Fibronectin distribution. These results indicate that uneven in vivo topology renders the need for precise distribution of secreted signals mostly dispensable.

Fig. 1

Sema3A and 3F are co-expressed with Sdf1 and restrict NC migration in vivo. a In situ hybridisation for neural crest markers (st17, slug; st21–28, twist), Sdf1, Semaphorin-3A and 3F. NC cells migrate as streams numbered 1–4, anterior to posterior. Asterisk marks the eye. b, c Diagrams depicting the relative distribution of Sdf1 (grey), Sema3A (red) and Sema3F (blue) with respect to NC cells (green) at stage 17 in wholemount (b) or transversal section (c). d Loss-of-function with antisense Morpholinos for Sema3A and 3F analysed by in situ hybridisation for twist, embryos st25. Arrows indicate cells from stream 1 that did not reach the area ventral to the eye. Black arrowheads indicate shorter streams. Red arrowheads, cells accumulated in dorsal region. Asterisks mark the eye. Note cells migrating over the eye on the injected sides. e Anterior view of a representative embryo injected with both MOs against Sema3A and 3F. Dotted line marks the midline. f Loss-of-function with CRISPR/Cas9 and gRNAs for Sema3A and 3F analysed by in situ hybridisation for twist, embryos st25. g Anterior view of a representative embryo injected with gRNAs against Sema3A and 3F together with dCas9. h Percentages of embryos with symmetrical migration along the dorso-ventral axis on non-injected and injected side. i Percentages of embryos with NC cells in ectopic locations (over the eye, in between streams, caudal expansion, between midline and NC streams). Proportions for each experimental conditions were compared to control uninjected embryos. 327 embryos were used obtained from three independent experiments, n of embryos per conditions are indicated on the figure. Comparisons of proportions were made using contingency tables. Null hypothesis rejected if T > 3.841 (*α = 5%); T > 6.635 (**α = 1%); T > 10.83 (***α = 0.1%). Note that normal NC migration in control embryos displays some level of randomness. Around 15% of non-injected embryos had noticeable differences between their left and right sides. The front of migration was more ventral on one side than the other. Also, about 20% of non-injected embryos had some cells that would be counted as ectopic in experimental embryos, mainly cells located dorsally to the streams, seemingly late. Scale bars are 500 μ, except for e and g, 50 μ. Source data are provided as a Source Data file

Fig. 2

Neural Crest migration in control and sema3A-KD environments. a Procedure for grafting of NC cells. b Images at t0h and t6h of the time lapse movies corresponding to 4 h and 10 h after the graft, respectively. c Projections of all NC domains at t0h and t6h for all experimental conditions, note that overall distribution and dorso-ventral migration of control NC cells are affected in sema-KD background. d Graph plotting the distance migrated by NC cells in all conditions shown in b, c. 35 embryos (n_CMO = 10, n_Sema3AMO = 11, n_{Sema3A-gRNA-aCas9} = 7, n_{Sema3A-gRNA-dCas9} = 7), from five independent experiments. ANOVA followed by multiple comparisons. ****p < 0.0001; ***p < 0.001. Scale bar 150 μ. Box and whiskers plot: the box extends from the 25th to the 75th percentile; the whiskers show the extent of the whole dataset. The median is plotted as a line inside the box. Source data are provided as a Source Data file

Fig. 3

Sdf1 and Sema3A have antagonistic effects on cell dispersion. a Diagram explaining how NC explants were prepared. b Representative examples of explants at t0 (one hour after plating on Fibronectin) and +8 h. Note that cells exposed to Sema3A have a round morphology and tend to stay as small clusters, even when dispersion is rescued by Sdf1. c Distribution of explants areas per hour per experimental condition. A total of 229 explants (n_controls = 50; n_Sdf1 = 32; n_{Sema3A-15ng.mL} = 29; n_{Sema3A-15ng.mL+Sdf1} = 17; n_{Sema3A-30ng.mL} = 24; n_{Sema3A-30ng.mL+Sdf1} = 14; n_{Sema3A-60ng.mL} = 47; and n_{Sema3A-60ng.mL+Sdf1} = 16) from five independent experiments were used. Two-way ANOVA, matching: stacked, pairwise multiple comparisons. *p value < 0.05; **p value < 0.01; ***p value < 0.001; ****p value < 0.0001. Dotted line on the graphs represents the mean value for controls at 8 h, provided as a visual reference for comparison with other conditions. Scale bar on low magnification, 100 μ. Scale bar on zooms, 20 μ. Box and whiskers plot: the box extends from the 25th to the 75th percentile; the whiskers show the extent of the whole dataset. The median is plotted as a line inside the box. Source data are provided as a Source Data file

Fig. 4

Sdf1 and Sema3A have opposite effects on spreading, adhesions and protrusions. a DAPI (blue) and Phalloidin staining (green) on NC cells cultured in control conditions or exposed to Sdf1 and/or Sema3A. b Colour-coded time projection of time-lapse movies of cells transfected with life-Act-GFP (see Supplementary Movie 8). c–e Area occupied by the cells, normalised to control conditions (c), circularity (d) and aspect ratio (e), n = 105 cells (n_controls = 25; n_Sdf1 = 21; n_{Sema3A15ng.mL} = 25; n_{Sema3A15ng.mL+Sdf1} = 34) from two independent experiments, from fixed cells counterstained with Phalloidin. c–e ANOVA followed by multiple comparisons; ****p < 0.0001. f Average area occupied per protrusion (from Life-Act movies), normalised to controls, n = 122 protrusions (n_controls = 24; n_Sdf1 = 23; n_{Sema3A15ng.mL} = 35; and n_{Sema3A15ng.mL+Sdf1} = 40) analysed from two independent experiments. ANOVA followed by multiple comparisons ** (ctl vs sdf), p = 0.0014, ** (ctl vs Sema), p = 0.0048, ***p = 0.0009. g Phospho-paxillin (PPax) immunostaining, cells adhering on Fibronectin, Sema3A/Sdf1 were added in solution 30 min before fixation. h Ratio of area occupied by PPax divided by area of the cell, n = 51 cells from one experiment. Student’s t-test; *p = 0.0427. **p = 0.0144. i PPax intensity plot from cell tip to cell centroid (mean+s.e.m). Colour code corresponds to conditions shown in graphs e and h. j Frequency distribution of FA length (main axis). Note that exposure to Sema3A reduces the number of large FAs. Colour code corresponds to conditions shown in graphs e and h. k Average FA length, ANOVA followed by multiple comparisons; ****p value <0.0001. Data shown in h–k were gathered from two independent experiments, n = 30 cells per conditions. Scale bars 10 μ. Box and whiskers plot: the box extends from the 25th to the 75th percentile; the whiskers show the extent of the whole dataset. The median is plotted as a line inside the box. Source data are provided as a Source Data file

Fig. 5

Sdf1 counteracts Sema3A’s effects on single cell motility. a Tracks of single cells on Fibronectin or Fibronectin plus Sema3A. b, c Speed and Directionality in cells shown in a, n = 188 cells (n_controls = 64; n_Sema3A = 53; n_Sema3F = 71). ANOVA, followed by multiple comparisons. d Tracks of single cells cultured on Fibronectin or Fibronectin plus Sema3A before and after Sdf1 was added in solution. e Speed of cells in conditions shown in d, ANOVA, followed by multiple comparisons; ns, p = 0.621; *p = 0.0188; ****p < 0.0001. f Directionality in cells shown in d, ANOVA, followed by multiple comparisons; *p = 0.0331. Note that Sdf1 reduces slightly speed and increases directionality whereas Sema3A strongly reduces both. Sdf1 restores control values when added to the Sema3A condition. n = 190 cells (n_controls = 60; n_Sdf1 = 29; n_Sema3A = 73; n_Sema3A+Sdf1 = 28). Box and whiskers plot: the box extends from the 25th to the 75th percentile; the whiskers show the extent of the whole dataset. The median is plotted as a line inside the box. Source data are provided as a Source Data file

Fig. 6

Rac1 activation is sufficient to rescue exposure to Sema3A. a Rac1, RhoA and Cdc42 activity assessed by FRET in cells from explants cultured in control, Sdf1 and/or Sema3A conditions, n = 118 cells (Rac1: controls(21), Sdf1(9), Sema3A(9), Sema3A + Sdf1(4); RhoA: controls(19), Sdf1(6), Sema3A(8), Sema3A + Sdf1(2); Cdc42: controls(14), Sdf1(9), Sema3A(8), Sema3A + Sdf1(9)) from three independent experiments. For each FRET probe, Sdf1, Sema3A and Sdf1+Sema3A conditions were compared to their cognate controls via ANOVA followed by multiple comparisons, individual p values are indicated on the figure. b Rac1 FRET in single cells under control conditions (Fibronectin) or with Fibronectin plus Sema3A coated at 15 or 60 ng/mL, n = 33 cells (n_controls = 11, n_{Sema3A-15ng.mL} = 14 and n_{Sema3A-60ng.mL} = 8), ANOVA followed by multiple comparisons, p values indicated on the graph. c Immunofluorescence against Rac1-GTP. d Distribution of rac1 intensity from cell protrusions to cell centroid. Rac1 staining intensity in the cell centroid was measured in each condition and subtracted from each data set, n = 50 cells from one experiment. e Rac1 staining intensity in protrusions for each experimental condition, n = 50 protrusions from one experiment, ANOVA, followed by multiple comparisons; *p < 0.05; ***p < 0.001. f Photoactivation experiment with single cells transfected with CIBN-Caax-GFP and Tiam1-CRY2-mCherry under control conditions. g Photoactivation experiments with single cells transfected with CIBN-Caax-GFP and Tiam1-CRY2-mCherry cultured on Fibronectin, plus Sema3A coated at 60 ng/mL. h Normalised cell area for experimental conditions displayed in f and g, n = 71 cells (n_{controls-PA/OFF} = 16, n_{controls-PA/ON} = 15, n_{Sema3A-PA/OFF} = 20, and n_Sema3A-PA/ON = 20) from seven independent experiments, ANOVA followed by multiple comparisons, p value indicated on the graph. i Cell area overtime (mean+s.e.m) for cells under sustained photoillumination after being transfected with CIBN and Tiam on Fibronectin or Fibronectin plus Sema3A coated at 60 ng/mL or cells transfected with CIBN only on Fibronectin plus Sema3A coated at 60 ng/mL, n = 21 cells from one experiment (n_CIBN/FN = 3, n_CIBN+Tiam/FN = 4, n_CIBN/Sema = 8, and n_{CIBN+Tiam/Sema} = 6). j Size of protrusions from cells used in f and g, n = 105 protrusions (n_CIBN/FN = 18, n_CIBN+Tiam/FN = 21, n_CIBN/Sema = 29, n_{CIBN+Tiam/Sema} = 37). ANOVA followed by multiple comparisons, ****p < 0.0001. Scale bars 10 μ, except for zooms in panel c, 3 μ. Box and whiskers plot: the box extends from the 25th to the 75th percentile; the whiskers show the extent of the whole dataset. The median is plotted as a line inside the box. Source data are provided as a Source Data file

Fig. 7

Activation of adhesion and actin dynamics are required for rescue of exposure to Sema3A. a Effects of Manganese (Mn²⁺) and CucurbitacinE (CuE) on adhesion and actin. b Rac1 activity (FRET) in control/treated cells with Mn²⁺ (2 mM) for 2 h, n = 24 cells (n_controls = 11, n_Mn2+ = 13) from two independent experiments, unpaired Student t-test, p value on the graph. c Cells cultured on Fibronectin with/without Mn²⁺ (2 mM) or CuE (1 nM), Phalloidin. d Normalised cell area, n = 104 cells (n_controls = 15, n_Mn2+ = 17, n_CuE = 14, n_Sema = 20, n_Sema+Mn2+ = 20, n_Sema+CuE = 18), ANOVA, multiple comparisons; **p_{(ctl vs sema)} = 0.0036; **p_{(sema vs sema+Mn)} = 0.0053; and *p_{(ctl vs sema+CuE)} = 0.0210. e PPax immunostaining, cells cultured on Fibronectin with/without Sema3A at 60 ng/mL with/without Mn²⁺ (2 mM) and/or CuE (1 nM). f Focal adhesion per cell per condition depicted in e, n = 178 cells (n_controls = 29, n_Mn2+ = 29, n_Sema = 30, n_Sema+Mn2+ = 30, n_Sema+CuE = 30, and n_{Sema+Mn2++CuE} = 30) from two independent experiments; ANOVA, multiple comparisons; *p_{(ctl vs sema)} = 0.0348; **p_{(ctl vs Mn)} = 0.0053; **p_{(ctl vs sema+Mn)} = 0.0026; ****p_{(sema vs sema+Mn)} < 0.0001. g, h FA length and frequency distribution from e, n of individual FA analysed: n_controls = 2197, n_Mn2+ = 1688, n_Sema = 3101, n_Sema+Mn2+ = 2708, n_Sema+CuE = 2469, n_{Sema+Mn2++CuE} = 2328, from two independent experiments. Colour code same as in g. i Explants cultured on Fibronectin or Fibronectin plus Sema3A at 15 ng/mL with/without Mn²⁺ (2 mM), stained with DAPI. j Mean distance between nearest neighbours from i, n = 106 explants (n_controls = 47, n_Mn2+ = 26, n_Sema = 17, and n_Sema+Mn2+ = 16), from two independent experiments; ANOVA, multiple comparisons, ***p = 0.0005. k Percentage of adherent and dispersing explants after a 3-hour culture on Fibronectin or Fibronectin plus Sema3A at 15, 30 or 60 ng/mL with/without Mn²⁺ (2 mM) and/or CuE (1 nM), n = 268 explants (n_controls = 52, n_Mn2+ = 26, n_CuE = 26, n_Sema15 = 17, n_Sema15+Mn2+ = 17, n_Sema30 = 22, n_Sema30+Mn2+ = 22, n_Sema60 = 43, n_Sema60+Mn2+ = 17, and n_{Sema60+Mn2++CuE} = 26) from five independent experiments. Comparisons of proportions were made using contingency tables. Null hypothesis rejected if T > 3.841 (*α = 5%); T > 6.635 (**α = 1%); T > 10.83 (***α = 0.1%). l Normalised area overtime for explants cultured on Fibronectin or Fibronectin plus Sema3A at 60 ng/mL with/without Mn²⁺ (2 mM), n = 145 explants (n_controls = 38, n_Sema = 33, n_Sema+Mn = 42, and n_Sema+Mn+CuE = 32) from four independent experiments. m Normalised explant areas after 8 h for conditions shown in l; ANOVA, multiple comparisons, ***p = 0.0005; ****p < 0.0001. Scale bars (c, e), 10 μ. Scale bar in i, 100 μ. Box and whiskers plot: the box extends from the 25th to the 75th percentile; the whiskers show the extent of the whole dataset. The median is plotted as a line inside the box. Source data are provided as a Source Data file

Fig. 8

Sdf1 does not need to interact with Fibronectin to rescue the effect of Sema3A. a Xenopus NC cells cultured on Fibronectin or Matrigel with or without Sema3A coated at 60 ng/mL and/or Sdf1 added in solution at 0.5 μg/mL, stained with DAPI and Phalloidin. b–d Normalised cell area (b), aspect ratio (c) and circularity (d) per cell for each condition shown in (a). For b, c and d, n = 160 cells (FN: n = 30, Matrigel, n_controls = 38, n_Sdf = 69, n_Sema = 7 and n_Sdf+Sema = 16). Since only seven cells remained attached in the Sema3A condition we did not perform statistical analysis. e Mouse neural crest cell line, 09–1, cultured on Matrigel with or without Sema3A coated at 60 ng/mL and/or Sdf1 at 0.5 μg/mL and/or Mn²⁺ (2 mM) added in solution, stained with DAPI and Phalloidin. f–h Normalised cell area (f), aspect ratio (g) and circularity (h) per cell for each condition shown in e. n = 1334 cells (n_controls = 370, n_Sema = 268, n_Sdf = 96, n_Sema+Mn = 224 and n_Sema+Sdf = 376) from four independent experiments. ANOVA followed by multiple comparisons, ****p < 0.0001; ***p < 0.001; **p < 0.01. i–k Normalised cell area (i), aspect ratio (j) and circularity (k) per cell for mouse NC cells cultured on Fibronectin with or without Sema3A coated at 60 ng/mL and/or Sdf1 (0.5 μg/mL and/or Mn²⁺ (2 mM) added in solution corresponding to experimental conditions shown in Supplementary Fig. 8. n = 3421 cells (n_controls = 653, n_Sema = 804, n_Sdf = 675, n_Sema+Mn = 591 and n_Sema+Sdf = 698). ANOVA followed by multiple comparisons, ****p < 0.0001; ***p < 0.001. Scale bars in a, 10 μ; in e low magnification, 50 μ; in e high magnification, 10 μ. Box and whiskers plot: the box extends from the 25th to the 75th percentile; the whiskers show the extent of the whole dataset. The median is plotted as a line inside the box. Source data are provided as a Source Data file

Fig. 9

Global activation of adhesion in vivo is sufficient to rescue Cxcr4 loss-of-function. a, b Fibronectin immunostaining on cryosections through the cephalic region of St17 (a) and St20 (b) Xenopus laevis embryos. c Diagram depicting the grafting procedure. d Representative images of the three types of grafts that were performed. Controls NC cells (FDx), cells injected with Cxcr4MO with or without prior exposure to Mn²⁺ were grafted into control non-injected hosts embryos. e Normalised net distance of migration along the dorso-ventral axis of grafted cells after an overnight incubation following the graft, n = 43 grafted embryos from four independent experiments. ANOVA followed by multiple comparisons, p values are indicated on the graph. f Still images from time-lapse movies for Cxcr4MO cells co-injected with CIBN-GFP grafted into control embryos. g Still images from time-lapse movies for Cxcr4MO cells co-injected with CIBN-GFP and photoactivatable Tiam1 grafted into control embryos. Arrows indicate NC cells migrating along the dorso-ventral axis. h Normalised net distance of migration along the dorso-ventral axis of grafted cells after 6 h of time-lapse imaging, n = 33 grafted embryos from five independent experiments. Student’s t-test, two tailed, p value indicated on the graph. FDx, fluorescein dextran; NC, neural crest; No, notochord. Scale bar in a, 50 μ. Scale bars in d and f, 500 μ. Box and whiskers plot: the box extends from the 25th to the 75th percentile; the whiskers show the extent of the whole dataset. The median is plotted as a line inside the box. Source data are provided as a Source Data file

Fig. 10

Model. a Diagram summarising the expression of Sema3A (red), Sema3F (blue), Sdf1 (grey, stripped), and the distribution of Fibronectin (green fibres) on a transversal section at the onset of NC cell migration (NC cells are in green). b Diagram of the proposed signalling events taking place. Sdf1 activates Rac1 whereas Sema3A inhibits Rac1. Medio-dorsally, Fibronectin and Sdf1 are lacking. Thus, the effect of Sema3A on Rac1 dominate and NC cells cannot adhere to the matrix. Ventro-medially, all players are present. Sdf1 counterbalances Sema3A, NC cells can adhere to Fibronectin. No, notochord; Plx/Nrp, plexins/neuropilins