Analysis of the molecular interactions and local molecular environment of fibroblast growth factors (FGFs)

Abstract

Fibroblast growth factors (FGFs) regulate essential processes in development and tissue homeostasis by engaging their cognate tyrosine kinase FGF receptors (FGFR) on target cells thereby eliciting changes in intracellular phosphorylation. Importantly, as with the majority of growth factors, cytokines, chemokines and morphogens, FGFs also bind the glycosaminoglycan heparan sulfate (HS) in the extracellular matrix (ECM). This interaction governs their localisation, diffusion, and promotes the formation of the FGF-FGFR-HS ternary complex essential for receptor activation. FGF4 regulates many key steps in embryonic development and is involved in stem cell maintenance, yet relatively little is known about its interactions in the ECM. The related FGF2, which binds the same FGFRs is often used as a model, including for intracellular signalling elicited by FGF4, although evidence for the appropriateness of this model has not been established.

To investigate the local molecular environment of FGF4 in ECM, fusion proteins were generated in which an engineered peroxidase (APEX) was linked to the N- terminus of the FGF, to enable proximity labelling of neighbouring proteins. The APEX-FGF4 fusion was successfully expressed in E. coli and purified. The APEX moiety of the fusion protein possessed peroxidase activity. Moreover, the fusion protein was functionally equivalent to native FGF4 in terms of binding heparin and stimulation of the phosphorylation of the adaptor fibroblast growth factor receptor substrate-2 (FRS2) and of mitogen-activated protein kinases MAPK1/3 in rat mammary (Rama) 27 fibroblasts. Compared to FGF2, however, FGF4 had a distinct concentration dependence. Phosphorylation events stimulated by FGF4 in Rama 27 fibroblasts were therefore explored in more detail. Not only was FGF4 less potent than FGF2 but, its stimulation of phosphorylation of FRS2 and MAPK1/3 were oscillatory, and, unlike FGF2, there was no strong evidence for a bell-shaped dose response curve. These data suggest that in development, oscillations in FGF4 stimulated intracellular signals may contribute to oscillatory systems, such as those involved in somitogenesis.

Reproducible conditions were established for biotin-tyramide proximity labelling in Rama 27 fibroblasts and HaCaT keratinocytes. Under optimised conditions (300 ng/µL APEX-FGF, 5 min reaction), APEX-FGF4 generated a distinctive labelling profile, in which discrete polypeptides were observed by western blot. This contrasted sharply with APEX-FGF2 and APEX-FGF10, which yielded broad smears of labelled proteins. This demonstrated that FGF4

vii bound HS in ECM in locations that, at molecular length scales, differed from those occupied by FGF2 and FGF10. The specificity of FGF4 binding to HS in the matrix was demonstrated by competition assays involving FGF4 and heparin.

Given the differences in proximity labelling, the location of proximity labelled proteins and their dynamics were examined using fluorescein tyramide and fluorescence microscopy techniques. In addition, advantage was taken of the fact that APEX-FGF4 could auto label itself in vitro. This afforded, using the same reagent, an opportunity to determine in parallel the distribution of APEX-FGF4 itself. Confocal microscopy demonstrated that APEX-FGF4 proximity labelled proteins had a heterogeneous and punctate distribution in both Rama 27 fibroblasts and HaCaT keratinocytes. In general, auto-labelled APEX-FGF4 exhibited a similar distribution to its proximity-labelled proteins, but with lower overall abundance and reduced labelling intensity. Moreover, Z-stack imaging showed that the signal was restricted to the extracellular matrix at the cell periphery, with no labelling detected in the nucleus. Comparison with APEX-FGF2 revealed stronger and more extensive labelling for FGF2, as confirmed by quantitative fluorescence intensity analysis. Together, these observations indicated that FGF4 bound to HS in the ECM to sites that were spatially heterogeneous and distinct from those occupied by FGF2. Fluorescence recovery after photobleaching (FRAP) was used to quantify the mobility of APEX-FGF4 itself, and of its proximity labelled proteins in the extracellular matrix of Rama 27 fibroblasts and HaCaT keratinocytes. Analysis of half-time recovery and diffusion behaviour showed that APEX-FGF4 was mobile, but its diffusion speed was considerably slower than that of other FGFs studied previously. Intriguingly, the APEX-FGF4 proximity labelled proteins were also mobile indicating that, at least these endogenous ECM proteins, while trapped in the ECM, are nonetheless dynamic.

Together, these findings demonstrate that FGF4 differs to a far greater extent from other FGFs than previously considered: it has unique intracellular signalling properties, occupies binding sites on ECM HS that differ in their local environment from other FGFs and that the proteins in this local environment are mobile. This work provides new insights into the molecular specificity of FGF4–HS interactions, revealing how FGF4 may impact embryonic development and suggests that our current concepts of ECM as a fairly homogenous and, in molecular terms, static gel, needs to be revisited.