Assembly assay identifies a critical region of human fibrillin-1 required for 10-12 nm diameter microfibril biogenesis

Abstract

The human FBN1 gene encodes fibrillin-1 (FBN1); the main component of the 10-12 nm diameter extracellular matrix microfibrils. Marfan syndrome (MFS) is a common inherited connective tissue disorder, caused by FBN1 mutations. It features a wide spectrum of disease severity, from mild cases to the lethal neonatal form (nMFS), that is yet to be explained at the molecular level. Mutations associated with nMFS generally affect a region of FBN1 between domains TB3-cbEGF18-the "neonatal region". To gain insight into the process of fibril assembly and increase our understanding of the mechanisms determining disease severity in MFS, we compared the secretion and assembly properties of FBN1 variants containing nMFS-associated substitutions with variants associated with milder, classical MFS (cMFS). In the majority of cases, both nMFS- and cMFS-associated neonatal region variants were secreted at levels comparable to wild type. Microfibril incorporation by the nMFS variants was greatly reduced or absent compared to the cMFS forms, however, suggesting that nMFS substitutions disrupt a previously undefined site of microfibril assembly. Additional analysis of a domain deletion variant caused by exon skipping also indicates that register in the neonatal region is likely to be critical for assembly. These data demonstrate for the first time new requirements for microfibril biogenesis and identify at least two distinct molecular mechanisms associated with disease substitutions in the TB3-cbEGF18 region; incorporation of mutant FBN1 into microfibrils changing their integral properties (cMFS) or the blocking of wild type FBN1 assembly by mutant molecules that prevents late-stage lateral assembly (nMFS).

Fig 1. FBN1 domain organisation and sites of neonatal Marfan syndrome mutations.

A) The structure of FBN1 is dominated by cbEGF domains interspersed with TB domains. The majority of nMFS-associated mutations affect the central TB3- cbEGF18 domains, which have traditionally been referred to as the "neonatal region". B) A model of the neonatal region (i) was created using Modeller software [35] and coordinates from the structures of domains TB4 [4], cbEGF12-13 [36] and cbEGF32-33 [37]. TB3 is coloured blue, cbEGFs are coloured green with Ca²⁺ ions shown as grey spheres. Substitutions associated with cMFS (panel ii, orange) are found evenly distributed across the neonatal region while those associated with nMFS (panel iii, red) show a clustering around cbEGF11-12. C) Plotting the positions of known nMFS-associated substitutions on a schematic representation of a cbEGF domain shows that in many cases the effect of the substitution is structural, affecting either cysteines involved in disulphide bond formation (yellow) or Ca²⁺-binding consensus residues (red). Where nMFS and cMFS substitutions have been found at the same position, the cMFS version is shown in brackets. Data extracted from the Universal Mutation Database (http://umd.be/FBN1) [38].

Fig 2. Construction of the NterPro-TB3cbEGF19 mini-gene and secretion of cMFS and nMFS variants from fibroblasts.

A) NterPro-TB3cbEGF19 is a fusion of the N-terminal domains of FBN1 up to the proline-rich region (orange) with domains TB3 to cbEGF19, which encompass the neonatal region. Western blotting using an antibody directed against the proline-rich domain (orange) is used to distinguish the fusion construct from endogenous FBN1 expressed by fibroblasts on the basis of size. B) Model of the TB3-cbEGF19 domains highlighting the positions of the nMFS (red spheres) and cMFS (orange spheres) substitutions described in this work, and the position of domain cbEGF16, which is deleted in one of the nMFS variants. Domain TB3 is coloured blue, cbEGF domains are coloured green and Ca²⁺ ions bound to the cbEGF domains are shown as grey spheres. C) Secretion profiles of the neonatal region mutants associated with nMFS and cMFS. Medium samples from untransfected MSU-1.1 fibroblasts (MSU-1.1) and fibroblasts transfected with the wild-type (WT) construct were used as controls to allow the identification of the recombinant construct (MW ~130kDa, arrowed). Full-length FBN1, expressed endogenously by the fibroblasts, functions as a loading control (MW ~350kDa, arrowed).

Fig 3. Secretion of full length GFP-FBN1 fusion constructs from HEK293T cells.

Mutations associated with nMFS (neonatal) or cMFS (classical) were engineered into GFP-tagged FBN1 cDNA construct and transiently transfected into HEK293T cells. After 3 days of culture, samples of cells and medium fraction were analysed by Western blotting with an antibody directed against the GFP epitope. Empty vector (pcDNA) and wild-type construct (GFP-FBN1) were used as negative and positive controls. The majority of mutant constructs were easily detected in the culture medium. The C1086Y variant was consistently seen at reduced levels in the medium compared to the other constructs. Cell fraction samples indicated that expression levels were similar across the different variants and that the reduction seen for the C1086Y variant was not due to a lack of expression. *The G1127S substitution, which was identified in cases of isolated aortic aneurysm rather than cMFS, is associated with a milder form of disease and so is grouped with the cMFS variants here.

Fig 4. Microfibril incorporation of GFP-FBN1 nMFS variants.

FS2 fibroblasts were co-cultured for 5 days with HEK293T cells transiently transfected to express GFP-FBN1 (WT) or variants associated with nMFS. Co-cultures were then fixed and stained with anti-GFP and anti-FBN1 antibodies without permeabilisation [16, 26]. Although the nMFS variants I1048T, E1073K, C1111R, N1131Y and ΔcbEGF16 were clearly detectable in medium samples when expressed as GFP-FBN1 fusions, microfibril networks labelled with the recombinant GFP-FBN1 (arrow in WT panel) were rarely seen. Bar = 100 μm.

Fig 5. Microfibril incorporation of GFP-FBN1 cMFS variants.

FS2 fibroblasts were co-cultured for 5 days with HEK293T cells transiently transfected to express GFP-FBN1 (WT) or variants associated with cMFS. Co-cultures were then fixed and stained with anti-GFP and anti-FBN1 antibodies without permeabilisation [16, 26]. In contrast to the nMFS variants, co-cultures expressing GFP-cMFS variants produced microfibril networks containing readily detectable recombinant material (white arrows). Bar = 100 μm.

Fig 6. Model of the FBN1 neonatal region’s roles in microfibril assembly and nMFS pathogenesis.

A) In normal FBN1 assembly, secretion of the protein is coupled to the proteolytic cleavage (1) of the C-terminal propeptide (red circle). FBN1 interactions with heparan sulphate proteoglycans (HS), at sites including the N- and C-termini and neonatal region (other sites not shown), limit diffusion from the cell surface during assembly. Multimerisation of C-terminal domains, previously blocked by the propeptide, initiate the assembly process (2) and promote intermolecular avidity-driven interactions with N-terminal domains (3) that result in a head-to-tail alignment of monomers. The neonatal region (green) may be involved in the higher order, lateral assembly of FBN1 to form the mature microfibril (4). B) Lateral assembly (panel A, step 4) would occur between FBN1 molecules (black and grey) that have already undergone C-terminal multimerisation and head-to-tail alignment. Compared to the wild type case (i), cMFS-associated mutations in the neonatal region (ii; orange) appear to incorporate into microfibrils at near-normal levels. In contrast, mutations affecting the neonatal region that lead to nMFS (iii; red) are unlikely to affect N- to C-terminal interactions but would disrupt later stages of assembly such as lateral association. Mutants such as ΔcbEGF16 may not directly affect the neonatal assembly site, but would affect the position and register of the region in the maturing microfibril. In addition to preventing the incorporation of mutant molecules into microfibrils, the interaction between wild type and mutant variants at step 3 would reduce the amount of wild type FBN1 deposited into the matrix, resulting in microfibril levels that are less than expected for haploinsufficiency.