Changye Sun and Yong Li, who successfully defended their PhD theses in November have published a paper each in Open Biology on the interactions of fibroblast growth factors (FGFs) with glycosaminoglycans:
and
They are co-authors on each other’s papers, since their thesis work was very much a parallel and integrated activity, that aimed to shed light on the molecular basis of these interactions and their functional significance. Both papers rely on the work we published previously in PeerJ that described how HaloTag enhanced the expression of recombinant FGFs in E. coli. Without that discovery, our FGF cupboard would have been rather bare!
Yong’s paper just about completes the partial earlier analysis we published in 2012 on six FGFs on the specificity of the interactions of FGFs with a library of model poly- and oligosaccharides, and of the location of the heparin binding sites on the FGF proteins. We use a library of model sugar structures, because these are available in large quantities. This means that the structure of the polysaccharides in solution can be probed using a variety of spectroscopies. So we have some understanding of the solution structures of our model compounds. This is important, since the solution structure of FGFs clearly deviates significantly from the crystal structures, though much less so when the FGFs bind glycosaminoglycans, e.g., here.
The data highlight a correlation between the diversification of the FGF family, based on phylogeny established from amino acid sequence, and the structures the FGFs recognise in glycosaminoglycans, as well as the number and type of secondary binding sites for heparin on the surface of the FGFs. Since the specificity of the FGFs for their tyrosine kinase receptors follows the same pattern, this suggests that these different molecular features may have co-evolved. Or perhaps they didn’t, because the phylogeny of FGFs based on chromosomal location differs somewhat from that derived from amino acid sequence. Only one of these at best can represent divergence form a common ancestor.
The bottom line is that the interactions of FGFs with glycosaminoglycans are not simple 1 protein : 1 oligosaccharide structure. This is understandable, since molecular interactions depend on surfaces of macromolecules, which themselves depend on the solution structure of the macromolecules. For sulfated polysaccharides, saccharine sequence does not equate simply to solution structure and activity, a point made very elegantly in a previous paper of Ed Yates’
So specificity is definitely there, but it isn’t simply a matter of reading a linear sequence of saccharides.
Changye’s paper describes the analysis of the binding and movement of a selection of FGFs in fibroblast pericellular matrix. The Halo-FGF fusion proteins allowed simple and stoichiometric labelling. The choice of FGFs was part rational and part expediency. For example, Halo-FGF7 was rather sticky in our hands, so we used the related Halo-FGF10 instead.
Some surprises. The affinity for FGF1 for HS on these cells (and for HS and heparin in a range of studies) is lower than FGF2’s, but a lot more FGF1 binds. Our explanation is that FGF1 is less selective in its choice of binding sites in HS, so more binds. At the other extreme was Halo-FGF20, which didn’t bind at all. We had noticed that the capacity of heparin affinity columns for FGF20 was lower than for most other FGFs, suggesting that it bound to a rare/unusual saccharide structure. This is demonstrated in the first paper. So it seems as if our fibroblasts do not make structures appropriate for FGF20 binding, or if they do, these are occupied by endogenous HS-binding proteins.
The other striking features of the binding experiments include (i) the heterogeneous distribution of the fluorescence, which reports on binding sites on HS; Halo-FGF10 binds pretty much equally HS and CS/DS in this particular pericellular matrix; the increased binding to HS for some FGFs after CS/DS are digested with chondroitinase ABC. Quite a lot to follow up there.
The analysis of average movement parameters using fluorescence recovery after photobleaching (FRAP) of Halo-FGF1, -2, -6 and -10 showed very different behaviours. At one extreme was Halo-FGF10, which didn’t move over the course of the measurement. This does not mean it is entirely immobile, as the FGF10 molecules may all be stuck in confinements. However, we would need the much higher resolution of single molecule tracking, as we did previously with nanoparticle-FGF2 conjugates to see if this was the case. In the middle sits FGF1, and fastest were FGF2 and FGF6.
The data in the two papers look at protein-glycosaminoglycan interactions from two very different perspectives. The clean, pure molecules in Yong’s paper provide essential biomolecular information, without which we operate in a vacuum. The pericellular matrix in Sun et al., gives an insight into what happens in a real matrix. It is clear that there is a gap between the two and bridging it is not always easy. I think there is a piece we can place in between the two, which will provide us with a deeper understanding: the HS brush, which has already revealed unsuspected consequences of protein-HS interactions namely that at least one FGF with multiple HS binding sites, FGF2, is able to cross link HS chains.
So two steps forward in understanding structure-function relationships of the extracellular matrix, but still a good way to travel.
Superb, insightful stuff!!!!