Interaction of Fibronectin with Glycosaminoglycans

Heparin is the only glycosaminoglycan (GAG) whose interaction with fibronectin has been extensively characterized. It is a highly sulfated but heterogeneous GAG analogous to those found on cell surfaces and in the extracellular matrices of a wide variety of animal tissues. Heparin is best known for its high-affinity interaction with and stimulation of the serine protease inhibitor, antithrombin, which serves as the basis for its common use as a "blood thinner" to prevent the formation of blood clots during surgery. However, heparin also interacts with many other proteins including fibronectin whose precipitation it induces at low temperature [ref]. In fluid phase, the interaction is dominated by the so called hep-2 domains located in the C-terminal third of each polypeptide chain. The interaction is sensitive to ionic strength and pH but relatively insensitive to temperature. A fragment of modular composition III_12-14 is almost indistinguishable from the parent protein with respect to affinity for fluorescein-labeled heparin in solution . Deletion analysis ([ref1 , ref2]) and direct binding of proteolytic subfragments [ref] implicated module III-13 as critical for binding. This module contains a so-called heparin-binding consensus sequence, RRAR, similar to ones found in other heparin-binding proteins [ref] (but also in many other proteins). Isolated III-13 binds directly to heparin, in both fluid and solid phase, with an affinity about 3-fold lower than fragments containing both III-13 and III-14 [ref]. Site directed mutagenesis experiments identified a cluster of six basic residues, including RRAR, that form a "cationic cradle" on one face of module III-13 which accounts for almost all of the hep-2 binding [ref]. Additional cationic residues in module III-14 that had been implicated in the context of synthetic peptides [ref] appear to be of minor importance in the context of larger fragments but some of them may contribute to the slightly higher affinity of larger fragments compared to III-13 alone.

Reported values of K_d for binding of fibronectin to heparin under physiological conditions vary from 2 mM [422] to 0.1 mM [ref] with some evidence for a minor population of even higher affinity sites (see [ref] and [ref]). Some of this variation could be due to differences in the heparin preparations. The higher affinity of antithrombin for a subpopulation of heparin molecules is well documented and requires the presence of a pentasaccharide unit with a specific pattern of sulfation that occurs more frequently in higher molecular weight fractions [ref]. Although there is some evidence for subpopulations of heparin ([ref1 , ref2]) and heparan sulfate [ref] having different affinities for fibronectin, the basis for those differences has not been elucidated. Ogama et al [ref] fractionated unlabeled heparin on Fn-Sepharose by stepwise elution with NaCl. Careful analysis of the fractions by size-exclusion chromatography revealed a direct correlation between molecular size and the concentration of salt required for elution. In contrast, Ingham et al [ref] fractionated fluorescein-labeled heparin by size-exclusion chromatography and found the K_d, determined by direct titration with Fn, to be insensitive to size above ~6kDa with a value of ~2mM, in good agreement with that obtained by Evington et al. [422] under similar conditions with the same method. However, it is significantly larger than the value of 0.2 mM reported by Benecky et al [ref] who also used a similar approach with the exception that their fluorescein-heparin was first applied to Fn-Sepharose in 3-fold excess over the capacity of the column, possibly selecting for a higher-affinity fraction. Likewise, Bentley et al [ref] obtained a C₅₀of 0.2 mM by titrating fluorescein-heparin that had been "affinity fractionated" by co-precipitation with Fn at 2^o. Finally, San Antonio et al [ref] used a gel-retardation assay in which ¹²⁵[I]-labeled heparin fractions of 6 kDa or less were electrophoresed in the presence of different concentrations of Fn to yield a range of affinities between 0.2 and 5mM. Clearly there is more to learn about the strength and specificity of the interaction of heparin with the hep-2 region of fibronectin.

The N-terminal region of fibronectin also interacts with heparin but less is know about the nature of the binding. When a thermolytic digest of fibronectin is applied to a column of immobilized heparin, one obtains in addition to the hep-2 fragments, an N-terminal Fib-1/hep-1 fragment that contains the first five type I modules. The concentration of NaCl required for its elution (0.4 - 0.5M) is only slightly less than that required to elute hep-2 (0.5 - 0.6M). Yet, in fluid phase it reacts poorly with heparin with a K_d at least 20-fold higher than that of hep-2 [ref]. Again, a potential explanation is that the hep-1 site recognizes only a subpopulation of heparin molecules that is able to capture hep-1 fragments when coupled to a solid phase but whose titration as a fluorescent derivative in fluid phase is masked or damped by the signal from the unreactive majority. This would be consistent with the fact that only a small portion of heparin binds to immobilized hep-1 whereas all of it binds to hep-2 [ref]. In any case, it seems that both sites could contribute to interaction with heparin-like GAGs present in the extracellular matrix, which after all can be viewed as a solid phase. The same may be true for other much weaker heparin-binding sites located in module III-1 [ref] and in the N-terminal half of 110 kDa CBF [ref]. Fragments containing these latter sites bind heparin-sepharose only at sub-physiological ionic strength but may add to the overall strength of interaction of Fn with an extracellular matrix that is rich in heparan sulfate proteoglycans.

Heparan Sulfate and other GAGs

The choice of heparin to probe GAG binding sites on fibronectin is one of convenience; the two molecules are unlikely to encounter each other in vivo except in the blood of patients on heparin therapy. It is heparan sulfate that is the more relevant GAG as far as Fn is concerned since it is the one covalently attached to proteoglycans that codistribute with fibronectin in the extracellular matrix. The main difference between heparin and heparan sulfate is that the latter is not as highly sulfated and therefore its affinity for fibronectin and other heparin binding proteins, to the extent that it has been measured, is lower than that of heparin. The early literature on the interaction of fibronectin with heparan sulfate, as summarized by Hynes [1808] in 1990 is contradictory and not much has happened in the interim to clarify the situation. Clearly, heparan sulfate preparations will bind to fibronectin affinity columns at low ionic strength but unlike heparin, they tend to elute in a NaCl gradient at ionic strength near physiological or below depending on the degree of sulfation [ref]. Heparan sulfate that was labeled with fluorescein showed no detectable binding to Fn at physiological ionic strength in solution [422].

Other GAGs that have been reported to bind to fibronectin under various conditions include dermatan sulfate, condroitin sulfate, and hyaluronic acid. In most cases their affinities tend to be qualitatively lower than for heparin and little quantitative information is available. Binding of fibronectin to immobilized chondroitin sulfate (at low ionic strength) appears to be mediated by the same Hep-2 determinants that mediate binding to heparin [ref]. In our hands, Hep-2 showed no affinity for fluorescein labeled chondroitin sulfate in solution at physiological ionic strength (unpublished data).

Some of the discrepancies in the literature regarding the interaction of fibronectin with GAGs may relate to the different types of assays used as well as differences between fibronectin preparations. Laterra & Culp [ref] noted that cell-surface fibronectin bound hyaluronate much better than did plasma fibronectin and attributed the difference to the tendency of the former to aggregate. It should be mentioned that the form of fibronectin that is most prevalent in the extracellular matrix is the fibrillar form which may thus be expected to bind GAGs more efficiently than would be inferred from in vitro studies with plasma fibronectin. Fibrillar fibronectin may present an array of positively charged domains to the polyanionic GAGs leading to a cooperative multipoint attachment.

see also Proteoglycans