Crystal structure of a Serpin A1 trimer. The diagram features three copies of Serpin A1 in assembly
Hsp47 is an ER-resident unique molecular chaperone crucially involved in the correct folding of procollagen 2, 3. Hsp47 was characterized as being a basic protein with an isoelectric point of 9.0 15. Hsp47 bears an N-terminal signal sequence, two N-glycosylation sites as well as a C-terminal ER-retention signal, RDEL 28, 29. All serpins share a common secondary structure consisting of three β-sheets (sheet A – C) and at least seven α-helices as well as a reactive center loop (RCL). The RCL represents an exposed loop which is linked to the fifth strand position of β-sheet A (s5A) at the N-terminus and the first strand position of β-sheet C (s1C) at the C-terminus. In inhibitory serpins, this loop has been identified to contain the primary recognition site for attacking proteinases 52. Biophysical analysis of mature recombinant mouse Hsp47 revealed the existence of a hyperstable, biologically active trimeric variant of Hsp47 together with an – also biologically active – structurally mesostable monomer with a 5-strand A-sheet 35. This is in contrast to previous observations postulating that monomeric Hsp47 shows a conformation comparable to latent non-inhibitory Serpin E1 (PAI‑1), featuring a six-stranded A-sheet 140. Formation of trimeric Hsp47 obviously occurs via a linkage formed by the insertion of the RCL from one monomer into the vacant fourth strand position (s4A) of β-sheet A of a second one 35. More recent observations on a domain-swapped Serpin C1 (antithrombin) dimer suggested an alternative model for serpin oligomerization. Herein, the intermolecular contact was achieved by the insertion of two long β‐strands (s4A and s5A) of one molecule into the β‐sheet A of another 141. However, the mechanism of pathological polymer formation seems to be distinct from the previously described s4A/s5A domain swap. Analysis of the crystal structures of pathological Serpin A1 (α1-antitrypsin) trimers revealed a polymeric linkage mediated by domain swapping of the C‐terminal 34 amino acid residues rather than the s4A/s5A domain swap previously postulated (Figure 3) 142.
The first crystal structures of Hsp47 from canine, either in its free form and in complex with different synthetic collagen model peptides able to form triple helices, were determined by Widmer and colleagues who demonstrated that Hsp47 exhibits the serpin fold consisting of three β-sheets, nine α-helices and the RCL (Figure 1 and 2) 24. Analysis of the crystal structures revealed that Hsp47 undergoes no signiﬁcant conformational change upon procollagen binding. Furthermore, the Hsp47-procollagen complex is formed by two Hsp47 molecules that attach in a head-to-head fashion via β-sheet C to two strands of one procollagen trimer (Figure 2). The consensus motif for binding of Hsp47 to the procollagen triple-helix has been identified previously as being Gly – Xaa – Arg 143, with Arg playing an important role in the Hsp47-procollagen interaction 25, 144, 145. Widmer and colleagues additionally determined the Hsp47 residues interacting with Arg at the Yaa0 position and the Yaa−3 position of the Yaa−3 – Gly – Xaa – Arg sequence of procollagen as being Asp385 and Arg222, respectively 24. Moreover, the critical amino acid residues Leu381 andTyr383 in Hsp47 have been confirmed to play a pivotal role in several hydrophobic interactions. The first structural investigation of the procollagen binding site in Hsp47 using NMR and mutational analysis proposed a possible contribution of Met223, Met363 and Tyr230 in the B/C β-barrel domain to procollagen binding 25. Of note, Hsp47 has been shown to preferentially recognize the folded triple-helical conformation of procollagen 144-149. In contrast to the critical Arg in the consensus binding motif Gly – Xaa – Arg, the nature of the residue in the Xaa position is negligible; Hsp47 also recognizes the amino acid in the Yaa−3 position in the sequence Yaa−3 – Gly – Xaa – Arg. Explicitly, Hsp47 predominantly recognizes Thr and Pro at Yaa−3, followed by Ser, Hyp, Val, and Ala, but it does not recognize Lys, Gln, or Glu 23. Adapted from the characterization of Hsp47-binding motifs, the number of Hsp47-binding sites in native procollagens was estimated by the group of Kouki Kitagawa, revealing that the triple-helical region (1029 amino acid residues) of human type III procollagen harbors 29 putative Hsp47-binding sites 23. In a more recent study using canine Hsp47 further important residues could be determined. Herein, His238 located at the interface of Hsp47-procollagen binding, was described to form hydrogen bonds with Asp220 and Ser305. By forming a stable salt bridge to Arg222 and Tyr383, Asp220 then enables binding of Hsp47 to procollagen 27.