Fine-tuning of Fgf8 morphogen gradient by heparan sulfate proteoglycans in the extracellular matrix

Abstract

Embryonic development is orchestrated by the action of morphogens, which spread out from a local source and activate, in a field of target cells, different cellular programs based on their concentration gradient. Fibroblast growth factor 8 (Fgf8) is a morphogen with important functions in embryonic organizing centers. It forms a gradient in the extracellular space by free diffusion, interaction with the extracellular matrix (ECM), and receptor-mediated endocytosis. However, morphogen gradient regulation by ECM is still poorly understood. Here, we show that specific heparan sulfate proteoglycans (HSPGs) bind Fgf8 with different affinities directly in the ECM of living zebrafish embryos, thus affecting its diffusion and signaling. Using single-molecule fluorescence correlation spectroscopy, we quantify this binding in vivo, and find two different modes of interaction. First, reducing or increasing the concentration of specific HSPGs in the extracellular space alters Fgf8 diffusion and, thus, its gradient shape. Second, ternary complex formation of Fgf8 ligand with Fgf receptors and HSPGs at the cell surface requires HSPG attachment to the cell membrane. Together, our results show that graded Fgf8 morphogen distribution is achieved by constraining free Fgf8 diffusion through successive interactions with HSPGs at the cell surface and in ECM space.

Graphical abstract

Figure 1

Fgf8 directly binds to glypican and syndecan HSPG in the extracellular space. (a) Illustration of constructs used for FCCS. GPI-anchor from Gpc and cytoplasmic and transmembrane domains from Sdc were replaced with mRFP, and subsequently used for FCCS with Fgf8-eGFP. Ss, signal sequence; Gpc, glypican domain; GPI, GPI-anchor; Sdc-ex, syndecan extracellular domain; Sdc-tm, transmembrane domain; cy, cytoplasmic domain. (b) Schematic showing the site of measurement in the ECM and the potential interaction between Fgf8-eGFP (green) with extracellular HSPG (red). (c–f) Fusion proteins localize to the ECM in embryos. One member from each HSPG family is represented here: (c) Fgf8-eGFP, (d) Gpc4^exmRFP (Gpc family 1/2/4/6), (e) Gpc3^exmRFP (Gpc family 3/5), and (f) Sdc2^exmRFP. mRNA was injected at 32-cell stage and embryos imaged from the animal pole after 5 hpf. Scale bar, 20 μm. (g–i) Representative FCCS curves and schematic of molecular interactions in the Gaussian volume. Red and green curves represent the autocorrelation amplitude in red and green channels, respectively. Blue curves represent cross correlation amplitude. (j) FCCS curve obtained for maximum cross correlation (max. CCR) with tandem-eGFP-mRFP; schematic depicts strong association of fluorescent tags. (k) FCCS curve for Gpc4^exmRFP versus Fgf8-eGFP interaction; schematic represents partial binding. (l) Absence of cross correlation between sec-eGFP and sec-mRFP, schematic represents no molecular interaction. (j) Percentage of cross correlation for Fgf8-eGFP and various HSPGs and controls. The different HSPG families are indicated. A higher percentage of cross correlation was measured for Sdc, followed by Gpc family 1/2/4/6 followed by family 3/5. Bar graph represents mean with SD. The significance of the data was inferred using one-way ANOVA. All data points were highly significant compared with sec-eGFP versus sec-mRFP control, with p < 0.0001. (k–m) Exemplary scatterplots for evaluation of effective dissociation constant (K_d) for binding between Fgf8-eGFP and representative HSPGs: (k) Gpc4^exmRFP, (l) Gpc3^exmRFP, and (m) Sdc2^exmRFP. K_d values are indicated in red. (n) Table showing FCCS measurements and corresponding K_d values for interaction of Fgf8-eGFP with different HSPGs in the ECM.

Figure 2

Binding to extracellular HSPG reduces Fgf8 diffusion. (a–c) Localization of Fgf8-eGFP in wild-type embryos (a) and embryos injected with Gpc4^exmRFP mRNA (b and d) and Sdc2^exmRFP mRNA (c and e). Note the extracellular retention of Fgf8-eGFP in b and c. Scale bar, 10 μm. (f) Autocorrelation curve of Fgf8-eGFP in wild-type and embryos overexpressing Gpc4^exmRFP and Sdc2^exmRFP. The curve shifts toward longer diffusion times upon binding to Gpc4^exmRFP and Sdc2^exmRFP, indicating a reduction in mobility. (g) Autocorrelation of sec-eGFP is not affected by the presence of excess Gpc4^exmRFP and Sdc2^exmRFP in embryos. The autocorrelation function was fit to a two-component model for Fgf8-eGFP and a one-component model for sec-eGFP. (h) Diffusion coefficient of Fgf8-eGFP (gray bars) and sec-eGFP (white bars) in embryos overexpressing Gpc4^exmRFP and Sdc2^exmRFP. Diffusion coefficient of Fgf8-eGFP is reduced, but sec-eGFP does not change. Bar graph represents mean with SD. Statistical significance was inferred using one-way ANOVA.

Figure 3

Fgf8 gradient is influenced by changing the composition of extracellular matrix. (a) Confocal image showing extracellular Fgf8-eGFP gradient from a restricted source of cells (white circular arc). The source was created by injecting Fgf8-eGFP mRNA in one blastomere at the 32-cell stage and confocal images acquired after 4 h. The fluorescent intensity was measured at different positions (white arrows) away from this source in the free extracellular space. Scale bar, 20 μm. (b–e) Normalized concentration gradient binned in 20 μm intervals and fit to a radial symmetry model to obtain the decay length λ (3). Comparison of Fgf8-eGFP gradient profile in wild-type embryos (black) and embryos expressing Gpc4^exmRFP (b, blue) and Sdc2^exmRFP (c, blue). The gradient becomes steeper due to extracellular binding and reduction of Fgf8-eGFP diffusion. (d) The gradient assumes a shallower profile after morpholino (MO)-mediated knockdown of Gpc4 (red). (e) Injection of a control morpholino does not influence Fgf8-eGFP distribution (gray, n = 14). (f) Comparison of decay length (λ) under specified conditions of ECM modification. Graph represents mean with SD; n = 17, 18, 25, 20, and 14 (left to right, respectively). One-way ANOVA was used to evaluate statistical significance.

Figure 4

Fgf8 and Fgf receptor interaction is regulated by cell membrane bound and not extracellular HSPG. In dual color scanning FCS fluorescent intensity fluctuations of membrane localized molecules are monitored and used for cross correlation analysis. The Gaussian focal volume is scanned across the membrane to obtain photon counts. (a–c) Membrane localization of Fgf8-eGFP (a), FgfR1-mRFP (b), and merged image (c) are shown for illustration. Scale bar, 10 μm. (d and e) Exemplary cross correlation curves obtained after scanning FCS on the membrane. (d) High amplitude of cross correlation (blue) was observed for FgfR1-mRFP and Fgf8-eGFP binding. (e) Cross correlation was negligible for mRFP-GPI control and Fgf8-eGFP interaction on the membrane. (f) Schematic showing aim of the experiment to directly monitor Fgf8-eGFP and FgfR1-mRFP interaction after addition or removal of Gpc4 in the ECM. (g) Fgf8-eGFP affinity for FgfR1-mRFP did not change significantly upon addition of excess Gpc4^ex (n = 21) or morpholino-mediated knockdown of Gpc4 (n = 26), as compared with wild-type control (n = 21). Graph represents mean with mean ± SE. One-way ANOVA was performed on the data set to infer statistical significance. (h) Illustration depicting aim of the experiment to measure direct binding between Fgf8-eGFP and cell-surface attached HSPGs. (i–l) Membrane localization of different HSPG constructs. mRFP was attached at the N-terminal for Gpc4 (i) and C-terminal for Sdc2 (j), Sdc3 (k), and Sdc4 (l). Scale bar, 10 μm. (m) Scanning FCS revealed that Fgf8-eGFP directly binds cell-surface attached mRFP-Gpc4, Sdc2-mRFP, Sdc3-mRFP, and Sdc4-mRFP with similar affinities. Graph represents mean with mean ± SE. One-way ANOVA and Dunnett’s multiple comparison tests were performed to test for significance. (n) Cross correlation values obtained from membrane scanning FCS between Fgf8-eGFP and the indicated constructs.

Figure 5

Electron microscopy-based visualization of Fgf8 around the cell membrane. (a–f) Correlative light electron microscopy to study localization of Fgf8-eGFP at high resolution around cell membranes. (a–c) Light microscopy image of a 70 nm cryosection, immunostained with rabbit anti-eGFP protein A gold, goat anti-rabbit Alexa488, and DAPI (nucleus). From fluorescent images, areas showing strong GFP labeling (filled arrowhead) and intact membrane/extracellular space were imaged further in the TEM. (d–f) Extracellular space enclosed by cell membranes on four sides, revealed localization of 10 nm gold-labeled Fgf8-eGFP molecules. (e) Arrows point at Fgf8-eGFP particles located ∼1 μm away from cell membranes. (f) A close-up of the membrane (highlighted in blue) shows that, even in the vicinity of the membrane, particles are distributed at a distance of 10–50 nm away from the membrane (arrows), presumably representing both membrane and ECM bound molecules. Scale bars, 50 μm (a), 10 μm (b), 10 μm (c), 2 μm (d), 500 nm (e), and 100 nm (f). (g) Proposed model for the regulation of extracellular Fgf8 gradient by HSPG. (i) Most Fgf8 molecules (green) traverse the embryo by free diffusion. (ii) Glypicans and syndecan heparan sulfate proteoglycans (blue) are constitutively shed from the cell surface into the ECM. (iii) Due to binding with extracellular HSPGs, Fgf8 gets trapped near the cell surface. HSPGs bind Fgf8 mainly via their HS side chains. This transient binding forms a relatively immobile Fgf8 fraction and is essential to obtain the optimal gradient length. (iv) Formation of a ternary complex with Fgf receptor (red), Fgf8, and HS sugar chains, involves cell-surface bound HSPGs. The extracellular secreted HSPGs do not contribute to cell-membrane interactions.