|Home | About | Journals | Submit | Contact Us | Français|
The ninth and tenth type III domains (FIII9–10) in the central cell binding domain of human fibronectin contain integrin receptor binding sites, including RGD in FIII10 and a synergy site, PHSRN, in FIII9. The specific amino acids that contribute to cell binding have been identified by the use of wild-type and mutant fragments of human fibronectin containing the FIII9–10 domain pair. At high concentrations FIII9–10 mimics, to a large extent, the biological activity of the full-length fibronectin molecule. However, FIII9 is conformationally unstable, even in the context of the FIII9–10 pair. Here we report the construction of a series of hybrid mouse–human FIII9–10 pairs that confer varying degrees of conformational stability to FIII9. The conformational stability of the human FIII9 module was increased 2–3-fold by substitution of Leu1408 with Pro. We demonstrate that the biological activity of this mutant is enhanced. The resulting FIII9–10 mutant has good solution properties and will provide a template into which further mutations can be incorporated in order to probe the structure–function relationship of the cell binding module of fibronectin.
Fibronectins are extracellular matrix molecules that elicit diverse cellular responses, such as gene activation and changes in cell shape, by virtue of their ability to bind to members of the integrin family of cell surface receptors. The regulation of fibronectin–integrin interaction is thus critical for many biological processes including development and tissue homeostasis. The primary cell binding site in fibronectin is the RGD motif in the tenth type III module (FIII10) (Pierschbacher and Ruoslahti, 1984). Cell adhesion in response to FIII10, however, is significantly less than to the native fibronectin molecule. The addition of the FIII9 module containing a site (PHSRN) that acts synergistically with FIII10 results in near-maximum cell adhesion activity (Aota et al., 1994; Bowditch et al., 1994), albeit at relatively high protein concentrations. Recent studies have shown that additional residues besides the RGD and PHSRN sites can contribute to the adhesion activity of FIII9–10 (Redick et al., 2000; Altroff et al., 2001).
The type III modules of fibronectin all share a common framework (Huber et al., 1994; Potts and Campbell, 1994). The crystal structure of human FIII7-10 reveals that the four FIII modules assume an extended rod-like structure with the RGD and synergy sites 34 Å apart (Leahy et al., 1996). The solution structure for human FIII10 suggests that the conformation of the RGD site, which resides on a loop between the β-strands F and G, is largely disordered and mobile (Main et al., 1992; Dickinson et al., 1994). Most of the residues of the PHSRN synergy site in FIII9 also reside on a loop (between the β-strands C′ and E) on the same surface of the molecule as the RGD site. Both loops protrude some distance away from the molecule. The calculated tilt and rotation angles between FIII9 and FIII10 are small compared with those of the FIII7-8 and FIII8-9 pairs (Leahy et al., 1996). The buried interdomain surface area for human FIII9–10 (333 Å2) is also lower than for the other domain pair interfaces, suggesting a certain degree of mobility between the two domains. Indeed, the solution structure of the mouse recombinant FIII9–10 indicates a high degree of intermodule flexibility (Copié et al., 1998).
Although the importance of the precise spatial relationship between the RGD loop and the FIII9 synergy site is not yet fully established, it is clearly critical for biological activity (Grant et al., 1997). Copié et al. have suggested that FIII9–10 acts as a flexible scaffold upon which the RGD and synergy sites exist in an ensemble of biologically active conformations (Copié et al., 1998). However, the conformations of the RGD and synergy loops of human FIII9–10 in solution remain poorly defined.
A previous report (Copié et al., 1998) demonstrates that mouse FIII9–10 has increased conformational stability compared to the human module pair (Spitzfaden et al., 1997) and is sufficiently stable for resolution of the solution structure by NMR, even though the sequence identity between mouse and human FIII9 is high (83%). The series of mutant human FIII9–10 proteins we describe here contain amino acid substitutions in FIII9 that were introduced according to differences between the mouse and human FIII9 amino acid sequences. The mutations were chosen with the aim of defining specific residues that can confer thermodynamic stability on FIII9 in the FIII9–10 pair and that have a key function in folding of the domain pair. Alignment of primary sequences highlighted five non-conservative and 10 semi-conservative or conservative amino acid mismatches that exist between the mouse and human FIII9 modules (Figure 1a). Of these, three non-conservative amino acid substitutions, Ser for Phe1335, Ile for Arg1358 and Pro for Leu1408, were introduced either alone or in combination into the human FIII9 module (Figure 1b) to produce seven mutants: L1408P, R1358I, F1335S, L1408P + R1358I, L1408P + F1335S, R1358I + F1335S and L1408P + R1358I + F1335S. Mutations were made following the Quickchange protocol (Stratagene). Residues Phe1366–Ser1367 were not substituted with Ser–Val from mouse because these mutations were predicted to represent a minimal change. The semi-conservative and conservative amino acid differences between the mouse and human sequences were similarly ignored.
SDS–PAGE analysis of the soluble cell fraction showed that the relative expression level of mutant human FIII9–10 constructs without an F1335S substitution (L1408P, R1358I and L1408P + R1358I) was similar to or higher than that of native FIII9–10 (Figure 2). Proteins expressed from constructs containing the F1335S mutation, however, remained in the insoluble cell fraction (pellet) (Figure 2). Those mutants could therefore not be assessed for stability or biological activity. Since poor expression of a protein in the soluble cell fraction reflects to a certain extent poor solution properties, the F1335S mutation was assumed to result in low conformational stability of FIII9. Indeed, the crystal structure of FIII7–10 shows that Phe1335 occupies a position on β-strand A of FIII9, with the aromatic side chain lying away from the solvated surface towards the hydrophobic core of the module (Figure 1b). It is thus reasonable to assume that substitution of Phe1335 with Ser may disrupt, and therefore weaken, the hydrophobic interactions within the module core, leading to a loss of conformational stability and solubility.
In order to compare the conformational stabilities of mutant FIII9–10 pairs, equilibrium unfolding experiments were performed. Protein samples were rapidly diluted 11-fold in GdnHCl [0 to ~8 M GdnHCl in 20 mM NaOAc, pH 4.8, molarity calculated by the weight of the solution (Pace and Scholtz, 1997)] and allowed to equilibrate for 10 min at 25°C before measuring the fluorescence emitted at 350 ± 3 nm, using an excitation wavelength of 278 nm on a Shimadzu RF5001PC spectrofluorimeter, at 25°C. All unfolding experiments were repeated independently and the data were fitted for a two-state unfolding mechanism according to the method of Pace and Scholtz (Pace and Scholtz, 1997). Evidence for a two-state transition for both FIII9 and FIII10 has been provided previously by the observation that the kinetics of the recovery of the native fluorescent and circular dichroism signals were effectively identical (Plaxco et al., 1997).
Figure 3 shows the two-step equilibrium denaturation curves for wild-type human FIII9–10 and mutants L1408P, R1358I and L1408P + R1358I. The results are in good agreement with previous studies which have demonstrated that the initial step represents the unfolding of FIII9 and the second step the unfolding of FIII10 (Spitzfaden et al., 1997; Altroff et al., 2001). A large difference in thermodynamic stability was observed between the native FIII9 and FIII10 modules (ΔGH2O being 4.9 and 12.6 kcal/mol, respectively) (Table I). None of the mutations effectively altered the stability of FIII10: the differences in the values of ΔGH2O and m for the FIII10 module can be accounted for by the error associated with the linear extrapolation of ΔG back to zero concentration of GdnHCl.
Most noticeable was the increase in the free energy (ΔGH2O) of FIII9 unfolding observed between the wild-type protein and the L1408P mutant (Table I). The measurements for ΔGH2O indicated that the L1408P mutation increased the conformational stability of FIII9 by 1.5-fold (from 4.9 to 7.2 kcal/mol). The R1358I mutation also resulted in a modest increase in the conformational stability of FIII9 (ΔGH2O = 6.3 kcal/mol). When the two substitutions were combined in the L1408P + R1358I double mutant, the increase in conformational stability was cumulative (ΔGH2O rising to 8.3 kcal/mol). However, the values for FIII9 in the L1408P and L1408P + R1358I mutants were both 2.6 M. Since the value for FIII9 in the L1408P mutant was 2–3-fold higher than for the wild-type, the denaturant clearly had less of an effect on the transition between the folded and unfolded states of this mutant than would be predicted from its conformational stability. This discrepancy may be explained by taking into account the dependence of the free energy change on denaturant concentration (m). The value of m [which reflects the surface area exposed to solvent in the unfolded module and may vary noticeably between single amino acid mutants (Shirley et al., 1989)] dropped from 4.3 kcal/mol.M in the wild-type FIII9–10 pair to 2.8 kcal/mol.M in the L1408P mutant and to 3.2 kcal/mol.M in the double mutant. This suggests that substitution of Pro for Leu1408 restricts the extent of unfolding of the FIII9 module and is in contrast to the substitution of Ile for Arg1358, which resulted in only a modest loss of sensitivity to denaturant (m = 4 kcal/mol.M).
The adhesion-promoting activity of the mutant L1408P was determined in cell attachment and spreading assays. In comparison with wild-type FIII9–10, both cell attachment and spreading were enhanced on surfaces coated with mutant L1408P (Figure 4). At a relatively low coating concentration (0.1 μM) the increase in attachment and spreading activities was ~25 and ~50%, respectively, in accordance with the increased conformational stability observed for the L1408P mutant.
We have reported the use of a rational approach to engineer human FIII9–10 mutants with minimal amino acid substitutions that confer increased conformational stability on FIII9. The main findings are that a single substitution of Leu1408 with Pro in human FIII9–10 increases both protein expression and conformational stability of the FIII9 domain and enhances the biological activity of FIII9–10. Our data indicate that Pro at this position has a key function in protein folding.
The solution properties of the L1408P mutant were greatly improved over those of wild-type FIII9–10. The Pro1408 residue is predicted to lie at the boundary between β-strand G and the F–G loop (Leahy et al., 1996). Sequence alignment of the FIII7, FIII8, FIII9 and FIII10 modules shows that a Pro residue is commonly found at the beginning or end of a predicted β-strand and could thus have a role in the folding of the common type III fibronectin module.
Interestingly, three Pro–Pro pairs exist in the mouse FIII9–10 pair, while none are present in the human FIII9–10 pair. Such Pro–Pro pairs would be expected to restrict severely the local flexibility of the protein backbone because of the narrow range of dihedral angles allowed by the Pro residue. However, refolding of the human FIII10 module, which is rich in Pro residues, proceeds very rapidly (Plaxco et al., 1996), and this module has good solution and stability properties (Spitzfaden et al., 1997). The presence of Pro–Pro pairs may therefore contribute to the stability of mouse FIII9–10 and also explain why the L1408P substitution, which creates the Pro1407–Pro1408 pair, confers enhanced solubility and stability on human FIII9–10. In addition, the presence of the Pro–Pro pair in the synergy loop of mouse FIII9 may result in reduced flexibility compared with the human counterpart, although the solution structure of the latter has yet to be determined.
In correlation with improved structural stability, the biological activity of the L1408P mutant was likewise increased in comparison with wild-type human FIII9–10. This was not expected to result from a direct interaction between the introduced Pro residue and the FIII9 synergy site, since they lie on the opposite sides of the module. It should also be noted that neither the L1408P nor the R1358I mutation was expected to alter the FIII9/FIII10 interface and thereby the interdomain mobility, which has been shown to affect the domain pair’s biological function (Grant et al., 1997). This study therefore provides an example of how single residue substitution distant from the binding site appears to alter biological activity via a long-range conformational change.
A modest increase in conformational stability of FIII9 was detected on substitution of Ile for Arg1358. The aliphatic side chain of Ile would not be expected to occupy the same position as the guanidium group of Arg, which, in the crystal structure, forms part of a cluster of four Arg and two Glu residues at the module surface. Despite this, the packing of the seven-β-strand module does not appear to be significantly altered, because the R1358I mutant exhibits the same extent of unfolding (m) as the wild-type. The increase in stability could therefore be attributed to more extensive hydrophobic interactions within the Ile1358–Ile1359 pair, assuming that any alteration to the electrostatic interactions within the cluster of Arg/Glu residues does not affect module packing or solvation.
In conclusion, we have designed a mutant human FIII9–10 domain pair containing a single amino acid substitution that has increased conformational stability of the FIII9 module. The enhanced solubility, stability and function of L1408P suggests that the proline in this position affects the global conformational stability of FIII9, including the synergy site. The L1480P substitution in the human FIII9–10 pair provides a further tool for the dissection of the structural properties that confer functional activity on FIII9–10.
We thank Iain D.Campbell and Laurence Choulier for critical discussions. This work was supported by the Wellcome Trust.