Our study has provided the first high-resolution structure and possible evolutionary origin of a hybrid domain, one of the major disulphide-rich domains of the fibrillin/LTBP superfamily. Overall the fibrillin-1 hyb2 domain has a globular shape, resembling a TB domain, and is involved in extensive interactions with its neighboring domains, cbEGF9 and cbEGF10. A mechanism for the evolution of the hybrid domain as a fusion of TB and cbEGF domains was first proposed by Pereira et al. (1993)
. They showed that the N-terminal sequence of hyb2, encoded by exon 21, is homologous to the sequence encoded by the upstream exon of a 2-exon TB domain and that the C-terminal part of domain hyb2, encoded by exon 22, has a similar spacing between the cysteines as the C-terminal third of a cbEGF domain. A comparison of the structures of the hyb2, TB, and cbEGF domains indicates that this similarity in sequences is conserved at the structural level, suggesting that these domains arose from the fusion of a TB-cbEGF pair (D). The functional significance of the rearrangement of disulphides in a hyb domain compared with a TB domain is unclear. It has been proposed that the C-terminal linker region of fibrillin domain TB4 could act as a molecular spring (Lee et al., 2004
), with application of a tensional force causing unfolding of the C-terminal β-hairpin that could allow the region to extend by up to 5 nm. This would not be possible in a hybrid domain, with the C-terminal β-hairpin constrained by a disulphide bond, but a different mechanism of extension might be possible in this case. An unstructured loop between C6 and C7 (890
) of hyb2 separates the N-terminal TB-like region from the C-terminal cbEGF-like region. Under tension, this loop could allow the disruption of the hyb-cbEGF interface, extending the length of the structure by approximately 3.5 nm. Recoil of the system would be achieved by the reformation of the hyb-cbEGF interface, which is stabilized by the binding of Ca2+
binding to cbEGF10. Such dynamic behavior could facilitate the structural role of fibrillin in connective tissues that are subjected to high tensional forces.
The organization of fibrillin in the 10–12 nm diameter microfibrils is still a topic of controversy, with several models having been proposed based on a wide range of experimental techniques (Baldock et al., 2001, 2006; Knott et al., 1996; Kuo et al., 2007; Lee et al., 2004; Reinhardt et al., 1996; Smallridge et al., 2003
). An understanding of the interdomain interactions within fibrillin is required for the development of any model of microfibril organization. Binding of calcium to the cbEGF domains rigidifies the structure of tandem cbEGF repeats and has been shown to determine the shape of fibrillin and protect it against proteolytic cleavage (Reinhardt et al., 1997a, 1997b
). High-resolution structures of cbEGF domain pairs have shown that calcium binding to cbEGF domains stabilizes hydrophobic interdomain contacts (Knott et al., 1996; Smallridge et al., 2003
). Solution and X-ray crystal structures of a number of TB domain-containing fragments (Lack et al., 2003; Lee et al., 2004; Yuan et al., 1997
) have shown that these domains are stabilized by disulphide bonds in a C1-3, 2-6, 4-7, 5-8 arrangement, and contain a conserved aromatic residue that forms the hydrophobic core of the domain. In fibrillin-1, extensive interactions at cbEGF-TB and TB-cbEGF interfaces have been observed in an integrin-binding fragment consisting of domains cbEGF22-TB4-cbEGF23. The high Ca2+
affinity of the TB4-cbEGF23 pair (Kd
= 16 nM) in this fragment was in contrast to previous observations of a TB6-cbEGF32 fragment (Kd
= 1.6 mM), suggesting that Ca2+
affinities across the molecule could vary substantially. This was found to be the case in a study of all fibrillin-1 TB-cbEGF domain pairs, where it was also shown that Ca2+
affinity could be correlated with the degree of interdomain interaction. Given the relatively high affinities seen for all the TB-cbEGF pairs except TB6-cbEGF32, it was suggested that these regions of fibrillin form rigidified, linear structures at Ca2+
concentrations found in the extracellular matrix (Jensen et al., 2005
The overall structure of the Ca2+
-saturated fibrillin-1 cbEGF9-hyb2-cbEGF10 fragment determined here generally resembles the cbEGF22-TB4-cbEGF23 integrin-binding fragment described previously (Lee et al., 2004
), with extensive interdomain interactions that are Ca2+
-dependent at the hyb2-cbEGF10 interface. As in the case of the TB4-cbEGF23 domain pair (Jensen et al., 2005
), the Ca2+
affinity of hyb2-cbEGF10 was found to be much higher than would be required to reach saturation under the Ca2+
concentrations found in the extracellular matrix. The buried surface area at the hyb-cbEGF interface was slightly less than the 670 Å2
seen at the TB4-cbEGF23 interface (Lee et al., 2004
). The significance of these very high Ca2+
affinity sites is not understood, but might be related to the proposed extensible nature of the microfibrils in dynamic connective tissues. Extensive hydrophobic interactions at hyb-cbEGF and TB-cbEGF interfaces, stabilized by Ca2+
, could provide the driving force for recoil after the application of a tensile force. An extensive protein-binding interface between hyb1 and cbEGF1, as suggested by the high Ca2+
affinity observed for this domain pair, would also explain the context-dependent interactions of domain hyb1 with the fibulins seen by El-Hallous et al. (2007)
. Domain hyb1, when placed in the position of domain hyb2, was unable to “drive” the binding of recombinant fibrillin-1 fragments to fibulins, suggesting that a specific molecular surface formed by the interactions of hyb1 with its adjacent domains is required for these intermolecular interactions.
Structural studies on fibrillin fragments are also highly relevant to the closely related LTBPs, each of which contains one hyb domain in the context of a hyb-cbEGF-TB fragment bordered by relatively long stretches (89–148 residues) of cysteine-free sequence. LTBPs 1, 3 and 4 have been shown to bind covalently the latency-associated peptide of TGFβ through the formation of an intermolecular disulphide bond, with the interaction mediated by the rearrangement of a disulphide bond in the second TB domain of the LTBPs. The structural basis of this rearrangement has been well-characterized and is due to the presence of a two amino acid insertion between C6 and C7 of the TB domains, leading to the increased solvent accessibility of the C2-C6 disulphide bond. In the fibrillin-1 hyb2 structure, no comparable arrangement of disulphides was observed, suggesting that this domain is unlikely to be involved in disulphide rearrangements. In contrast, domain hyb1 has been shown to be involved in the formation of intermolecular disulphides as a result of the presence of a conserved 9th cysteine (Reinhardt et al., 2000
). Our model of domain hyb1, based on the structure of hyb2 determined in the cbEGF9-hyb2-cbEGF10 structure, is consistent with a surface-exposed position for Cys204, near the N-terminal end of the central α helix of the domain (A), and suggests that both TB and hyb domains contain potential protein binding surfaces distinct from their intramolecular interfaces.
A fibrillin fragment spanning from cbEGF7-cbEGF11 has recently been studied by SAXS (Baldock et al., 2006
). Rigid body modeling to the SAXS data, using known cbEGF and TB domain structures, produced a V-shaped structure with the bend at the hyb domain (C[i]), in contrast to the model based on X-ray crystallographic data (C[ii]). In these previous studies, modeling of the hyb domain region was possible only for the N-terminal region homologous to TB domains, but without high-resolution structural data it would not have been possible to model accurately the interaction of hyb2 with the adjacent domains. The Ca2+
-saturated cbEGF9-hyb2-cbEGF10 structure presented in our study shows that domains cbEGF9 and cbEGF10 pack against domain hyb2 to form a near-linear structure. Ca2+
binding data for the hyb2-cbEGF10 pair shows that an extensive interface is formed between these domains, as seen previously in TB-cbEGF pairs where specific hydrophobic contacts between domains were found to influence Ca2+
affinity (Jensen et al., 2005
). These differences in the observed and modeled hyb domain structures highlight the importance of high-resolution data in the development of models to explain the organization of fibrillin microfibrils. Variations in the way domains pack against each other can lead to substantial differences in the predicted dimensions and shape of full-length fibrillin monomers (C), which can have a significant impact on how these data are used to model microfibril organization. Although dissection approaches for structure determination have limitations in that studies can be time-consuming and might not detect long-range interactions involving flexible regions, the identification of interdomain interactions at atomic resolution provides an essential constraint for microfibril modeling.
In summary, the data presented here extend our knowledge of the structural properties of fibrillin-1 and provide further evidence for an extended model of microfibril organization. In addition, our data suggest a mechanism by which microfibril extensibility could be achieved that does not disrupt potential binding sites for microfibril accessory proteins.