Prodomain-growth factor swapping in the structure of pro-TGF-β1

Abstract

TGF-β is synthesized as a proprotein that dimerizes in the endoplasmic reticulum. After processing in the Golgi to cleave the N-terminal prodomain from the C-terminal growth factor (GF) domain in each monomer, pro-TGF-β is secreted and stored in latent complexes. It is unclear which prodomain and GF monomer are linked before proprotein convertase cleavage and how much conformational change occurs following cleavage. We have determined a structure of pro-TGF-β1 with the proprotein convertase cleavage site mutated to mimic the structure of the TGF-β1 proprotein. Structure, mutation, and model building demonstrate that the prodomain arm domain in one monomer is linked to the GF that interacts with the arm domain in the other monomer in the dimeric structure (i.e. the prodomain arm domain and GF domain in each monomer are swapped). Swapping has important implications for the mechanism of biosynthesis in the TGF-β family and is relevant to the mechanism for preferential formation of heterodimers over homodimers for some members of the TGF-β family. Our structure, together with two previous ones, also provides insights into which regions of the prodomain-GF complex are highly structurally conserved and which are perturbed by crystal lattice contacts.

Keywords: Arg-Gly-Asp-Leu-any-any-Leu/Ile (RGDLXX(L/I)); Protein Data Bank (PDB); activin; bone morphogenetic protein (BMP); crystal structure; dimerization; glycoprotein-A repetitions predominant protein (GARP); growth factor (GF); heterodimer; latency-associated peptide (LAP); latent TGF-β-binding proteins (LTBPs); prodomain; proprotein convertase; swapping; transforming growth factor beta (TGF-β).

Figure 1.

SDS-PAGE of purified pro-TGF-β1 and crystal. A, non-reducing (left) and reducing (right) SDS 4–20% PAGE of purified WT and R249A mutant pro-TGF-β1. B, reducing SDS-PAGE (10%) of a crystal of pro-TGF-β1 R249A mutant. Gels were stained with Coomassie Brilliant Blue.

Figure 2.

Crystal structures, lattice contacts, and structural comparisons. A and B, crystal structure of pro-TGF-β1 R249A mutant (A) and re-refined crystal structure of WT cleaved porcine pro-TGF-β1 (B). A ribbon cartoon is colored differently for straitjacket, arm, and GF domains. Disulfide bonds (yellow) are shown in stick representations. C termini of the prodomains and the N termini of the GFs are shown as spheres. C–E, lattice contacts in the crystal structures of uncleaved pro-TGF-β1 R249A PC mutant (C), cleaved pro-TGF-β1 WT (D), and uncleaved pro-TGF-β1 R249A PC mutant in complex with integrin α_Vβ₆ (E). Crystal lattice contacts are shown as transparent surfaces with their outsides white and insides black. F, comparison of five representative pro-TGF-β1 prodomain monomers: the human R249A mutant 2.9 Å structure (PC mutant), WT porcine (cleaved chains C and D), and integrin-bound and unbound human R249A mutant monomers in a pro-TGF-β1 complex structure (19) (unbound chain H and bound chain G).

Figure 3.

Prodomain–GF connections. A–C, overlays of the PC cleavage region in different crystal structures: R249A PC mutant 2.9 Å structure chain A (PC mutant), WT porcine cleaved chains C and D (cleaved monomer C and cleaved monomer D), and integrin complex R249A PC mutant chains G and H (complex monomer G and complex monomer H). A, all five structures; B and C, individual comparisons of uncleaved and cleaved monomers. D, views of the two possible arm domain–GF connections in the TGF-β1 R249A PC mutant 2.9 Å structure, from terminal residues in one arm domain monomer (C-ter 1) to terminal residues in one GF monomer (N-ter 1) or the other monomer (N-ter 2). The gaps to be spanned are dashed. Spheres mark the carbon atom of C-ter 1 and nitrogen atoms of N-ter 1 and N-ter 2. Terminal residues are labeled. Prodomain and GF domains are shown in ribbon cartoon and surface representations, respectively. E and F, two possible arm domain–GF connections in integrin complex uncleaved pro-TGF-β1 R249A dimers. E, chains G+H; F, chains C+D. Views are identical to those in the bottom view in D, and other details are as in D.

Figure 4.

Modeling and mutagenesis of the arm domain–GF linker region. A, representative arm domain–GF connections (residues 243–250) built by MODELLER between the C-ter 1–N-ter 1 connection are shown as loops of different colors with Cα atoms of the RHRR PC cleavage site shown with small spheres. Residues 61 and 72, adjacent to the residues missing in density between the straitjacket and arm domain, are also marked with spheres (see “Discussion”). The remaining portions of pro-TGF-β1 are shown as surface representations. B, PC site mutations. C, WT and mutant pro-TGF-β1 co-expression with soluble GARP. Culture supernatants were incubated with StrepTactin-Sepharose (GE Healthcare) for 1 h. StrepTactin beads were washed and heated at 100 °C in reducing SDS sample buffer, and eluates were subjected to SDS-PAGE and stained with Coomassie Brilliant Blue. D, transient expression of WT and mutant pro-TGF-β1. Culture supernatants were subjected to reducing SDS-PAGE and Western blotting with antibody to the TGF-β1 prodomain (BAF246, R&D Systems).

Figure 5.

Prodomain–GF swapping. A–C, straitjacket, arm domain, and GF connectivity in pro-TGF-β1 (A), pro-activin A (B), and pro-BMP9 (C). Straitjacket, arm, and GF elements are shown in the same color if they belong to the same precursor monomer; elements are numbered 1 or 2 according to their monomers. They are labeled 1 or 2 when the monomer from which they are derived is unknown and are given a distinctive color (straitjackets in pro-TGF-β1 and GFs in pro-BMP9).