Modeling the three-dimensional structure of serpin/molecular chaperone HSP47

Joseph W. Davids,[1] Talal S. H. El-Thaher,[1] Akira Nakai,[2] Kazuhiro Nagata,[2] and Andrew D. Miller,[1*]

[1] Imperial College of Science, Technology & Medicine, Department of Chemistry, South Kensington, London, SW7 2AY, UK telephone: (44) (171) 594 5773; telefax: (44) (171) 594 5803 [2] Department of Cell Biology, Chest Disease Research Institute, Kyoto University, Kyoto 606, Japan



Heat Shock Protein 47 (HSP47) is a molecular chaperone which assists procollagen triple helix assembly and secretion. As a molecular chaperone it is unique in that it binds specifically to a very narrow range of protein "substrates" (i.e., procollagen and collagen only), and it is also a member of the well characterised serine protease inhibitor (serpin) superfamily. The following paper describes how in the absence of any X-ray crystallographic data, a novel tandem-modeling procedure is used to obtain three-dimensional structural information on mature recombinant mouse HSP47 (mrmHSP47). MrmHSP47 is shown to have 30 % amino acid sequence identity and 70 % sequence similarity with human protein C inhibitor (hPCI). Therefore, molecular models of inhibitory and latent states of hPCI are generated, using the X-ray crystal structure coordinates of proteolytically cleaved hPCI, and used as templates for the homology modeling of inhibitory and latent states of mrmHSP47. The validity of the models is discussed and the latent state model of mrmHSP47 is shown to have a suitable candidate binding groove for procollagen/collagen peptides which appears to account for experimental observations made with amino acid deletion experiments.[1]

Key words: molecular modeling/serpins/HSP47/protein C inhibitor


Return to index

INTRODUCTION

Molecular chaperones are a ubiquitous, abundant and highly conserved group of proteins which assist protein folding/refolding in vitro and in vivo (1, 2) as well as protecting proteins from stress induced unfolding (1). They first came to attention because of their specific induction during the cellular response of all organisms to heat shock (3, 4) but are now known to be constitutively and abundantly expressed in the absence of any stress. Heat Shock Protein 47 (HSP47) is one such molecular chaperone. Whilst the vast majority of molecular chaperones interact with a large variety of globular proteins (1, 2, 5, 6), HSP47 (also known as colligin [7] or J6 protein [8]) is specific only to polypeptides of procollagen and the fibrous protein collagen. There is now considerable evidence that HSP47 is a molecular chaperone for the formation of procollagen triple helices and also for the secretion of procollagen from endoplasmic reticulum to the Golgi (9, 10, 11). Both bone and connective tissue are ultimately formed from these procollagen triple helices.

In order to understand the underlying mechanism of HSP47-assisted formation of procollagen triple helix, knowledge about the three dimensional structure of this protein is essential. However, there is no X-ray crystal structure of HSP47 currently available which led us to consider an alternative molecular modeling approach to obtaining structural information with which to understand structure/function relationships in this unique molecular chaperone. Molecular modeling appeared to be a valid approach to obtaining realistic structural information since in addition to being a functional molecular chaperone, HSP47 is known (9, 10, 12) to be a member of the serpin (serine protease inhibitor) superfamily of proteins which is one of the most widely and best structurally characterised protein families. There are at least 30 serpin proteins, including HSP47, which have been identified on the basis of their primary amino acid sequences (12, 13) and where X-ray crystal structure information is to hand the three dimensional structures are surprisingly consistent (13, 14). As a result, it is not unreasonable that HSP47 should adopt a similar protein fold and three-dimensional structure. Not all these serpins are functional serine protease inhibitors, but many do contain a ~ 28 amino residue loop (serpin loop) which is responsible for specific protease inhibition. Binding of this loop to a specific serine protease is followed by eventual dissociation of the serpin in a cleaved form where the amino acids to the N- and C-terminal sides of the cleaved, scissile peptide bond (i.e., respectively the P1 and P1' amino acid residues according to the Pn nomenclature of Schechter & Berger [15]) become separated by about 70 Å. Indeed, many of X-ray crystal structures of serpins which have been determined (14) are of such proteolytically cleaved serpins. Exceptions to this are the recent X-ray crystal structures of uncleaved ovalbumin (16), antichymotrypsin (ACHY) (17), plasminogen activator inhibitor-1 (PAI-1) (18) and antithrombin III (AT III) (19, 20). These recent structures have revealed that uncleaved serpins may adopt one of two main conformations. In one, the uncleaved serpin loop projects out of the protein in a conformation accessible to a serine protease active site. This inhibitory state is typified by a conformation observed in the X-ray crystal structure of AT III (19, 20). In the other, the serpin loop does not protrude, but is rendered inaccessible to a serine protease active site. This so called latent, non-inhibitory state is typified by the X-ray crystal structure of PAI-1 (18) and an alternative conformation of AT III (19).

As a result of this wealth of structural data and consistency within the serpin superfamily, several studies have been published (21, 22, 23) in which molecular modeling techniques have been used to successfully predict the structures of serpin proteins, for which no X-ray crystal structures were then available, using the X-ray crystal structure coordinates of other serpin family members. Therefore in the light of this literature precedent, the following paper outlines the application of molecular modeling, including the homology model approach, to derive appropriate three-dimensional structural information on serpin/molecular chaperone HSP47. Recently, the gene for mature recombinant mouse HSP47 (mrmHSP47) was cloned and over expressed in Escherichia coli (9), therefore we elected to model this protein specifically.


Return to index

METHODS

Human protein C inhibitor coordinates

The X-ray crystal structure coordinates of mature, cleaved human protein C inhibitor (hPCI) were available from the Brookhaven Protein Data Bank under the code 2pai. The structure begins at amino acid residue number 5.

Sequence alignment of human protein C inhibitor and mature recombinant mouse HSP47

The gene derived amino acid sequence of mature recombinant mouse HSP47 (mrmHSP47) (9) was optimally aligned with the amino acid sequence of hPCI using an initial conventional pairwise sequence alignment involving the Needleman-Wunsch algorithm (24) followed by sequence core analysis, using the Dayhoff mutation (PAM250) matrix (25), and finishing with a review of residue hydrophobicities (26).

Molecular modeling of human protein C inhibitor

Model structures of uncleaved hPCI were generated from the known X-ray crystal structure coordinates (2pai) on a Silicon Graphics Iris 3000 model workstation using the molecular modeling facilities of the protein imaging software package QUANTA 3.3. Two models of uncleaved hPCI were generated corresponding to a putative inhibitory and a putative latent, non-inhibitory state respectively. The putative inhibitory state of hPCI was modeled by rotating the φ and ψ backbone dihedral angles of both P10 and P12 (residues Ala349 and Ala347) in order to bring the P1 and P1' residues (Arg358 and Ser359) into close proximity. A peptide bond was then created between P1 and P1' residues, after which the structure was subject to energy minimisation using CHARMm (27) to assist in the packing of side chain atoms while harmonically constraining the alpha-carbon backbone. The putative latent state of hPCI was modeled by rotating the φ and ψ backbone dihedral angles of P12' (residue Asn370) in order to bring the P1 and P1' residues (Arg358 and Ser359) into close spatial proximity once more. Once again, a peptide bond was then created between P1 and P1' residues, after which the structure was subject to energy minimisation using CHARMm.

Homology modeling of mature recombinant mouse HSP47

Two structural models of mrmHSP47 were generated on a Silicon Graphics Iris 3000 model workstation using the homology modeling facilities of the protein imaging software package QUANTA 3.3. A putative inhibitory state homology model of mrmHSP47 was generated using the inhibitory state model of hPCI (described above) as a template. Firstly, the alpha-carbon backbone coordinates of inhibitory hPCI were assigned to corresponding residues of mrmHSP47, which had been found to correspond through the sequence alignment (see above). Undefined mrmHSP47 amino acid side-chain coordinates were then generated with idealised geometries using CHARMm. Subsequently, energy minimisation, using a "Steepest Decents" algorithm (100 cycles) followed by an "Adopted Basis Set Newton Raphson" algorithm (100 iterations), gave the completed putative inhibitory state homology model of mrmHSP47. A putative latent state homology model of mrmHSP47 was also generated by homology modeling in the same way using the latent state model of hPCI (described above) as a template.

Serine protease inhibition studies with mature recombinant mouse HSP47

MrmHSP47 was purified from recombinant Escherichia coli DH1λind-/pKS26 according to previously published protocols (28) and protein concentrations were estimated by the A280/A260 absorption method of Warburg and Christian (29). Standard assays (30, 31) for thrombin and elastase were performed in the presence of mrmHSP47 at pH 7.5-7.8 using human thrombin (5 μg) and porcine pancreatic elastase (2 μM). The highest mrmHSP47 concentrations used were 38 μM and 51 μM respectively. Standard trypsin, kallikrein and tonin assays (32) were carried out in the presence of mrmHSP47 at pH 8.0 with bovine trypsin (100ng), rat submandibular kallikrein (25ng), and rat submandibular tonin (1.5 μg). The highest mrmHSP47 concentration used was 26 μM in all cases.
Return to index

RESULTS

Sequence alignment

Mature recombinant mouse HSP47 (mrmHSP47) is comprised of a sequence of 400 amino acids with a molecular mass of 46 589 Da (9). Figure 1 shows the optimal sequence alignment of mrmHSP47 with human protein C inhibitor (hPCI). The sequence identity between the two proteins was 30 % and the similarity 70 % over 393 aligned residues. Figure 1 also indicates the serpin loop residues of hPCI and their equivalent in mrmHSP47 (from residues P12 through to P12') using the Pn nomenclature of Schechter and Berger (15).

Molecular modeling of human protein C inhibitor

The X-ray crystal structure coordinates of hPCI in a cleaved form were readily available from the Brookhaven Protein Data Bank. This structure (Figure 2a) has been orientated and coloured to explain the molecular modeling procedures by which models of inhibitory and latent states of hPCI were generated. Owing to the fact that all the currently solved three dimensional X-ray crystal structures of serpins have the same basic secondary structural features (13, 14), Huber & Carrell (13) devised a careful labelling system so as to identify each of the secondary structural components individually (i.e., there are three beta-sheets labelled A, B, and C with each strand separately numbered e.g. s4A is strand 4 of beta-sheet A; s1C is strand 1 of beta-sheet C). The molecular modeling procedures revolved around beta-sheet A (in yellow, Figure 2a), strand s4A (in red, Figure 2a) and strand s1C (in green, Figure 2a). The inhibitory state of hPCI was modeled by perturbing the φ and ψ backbone dihedral angles of both residues P10 and P12 (Ala349 and Ala347) so that the P1 residue (Arg358) at the free end (Carboxyl terminus) of beta-strand s4A (in red, Figure 2a) could be lifted out from beta-sheet A and brought into close proximity with the P1' residue (Ser359) at the free end (Amino terminus) of strand s1C (in green, Figure 2a). After the creation of a peptide bond between P1 and P1' residues, energy minimisation was used to optimise the inhibitory state model of hPCI (Figure 2b) with the serpin loop (in yellow, Figure 2b) projecting away from the remainder of the protein thereby adopting a conformation accessible to a serine protease active site. The latent state model of hPCI was modeled by perturbing the φ and ψ backbone dihedral angles of residue P12' (Asn370) only, so that the P1' residue (Ser 359) at the free end (Amino terminus) of strand s1C (in green, Figure 2a) could be brought into close proximity with the P1 residue (Arg358) at the free end (Carboxyl terminus) of beta-strand s4A (in red, Figure 2a). This process involved the introduction of a little disruption to beta-sheet C, whilst maintaining the integrity of beta-sheet A. Once more, after the creation of a peptide bond between P1 and P1' residues, energy minimisation was used to optimise the latent state model of hPCI (Figure 2c) with the serpin loop (in red, Figure 2c) closely associated with the remainder of the protein thereby rendered inaccessible to a serine protease active site. Both inhibitory and latent state models of hPCI were superimposed on the X-ray crystal structure of the cleaved form of hPCI and the R.M.S. deviations recorded (Table 1).
Table 1. R.M.S. deviation of uncleaved hPCI molecular models from X-ray crystal structure of cleaved hPCI.
Latent state hPCI model Inhibitory state hPCI model
Residue range R.M.S. deviation (Å) Residue range R.M.S. deviation (Å)
Pre-serpin loop region (10-356) 0.11 Pre-serpin loop region (10-345) 0.08
Post-Serpin loop region (368-391) 0.26 Post-Serpin loop region (362-391) 0.12
Table 2. R.M.S. deviation of mrmHSP47 homology models from corresponding parent models of hPCI.
R.M.S. deviation (Å)
hPCI structure element 1Residue range in mrmHSP47Latent state mrmHSP47Inhibitory state mrmHSP47
Helices
A1-72.10.8
B12-261.41
C21-421.40.9
D53-671.20.9
E69-781.20.8
F87-1031.20.8
G126-1361.30.9
H148-16321.1
I260-2791.21.3
Sheet A111-1201.21.1
137-1421.31
180-19220.9
294-29611.1
329-3392.10.9
345-3582.40.9
Sheet B49-521.80.9
214-2291.31.1
235-2431.31.1
246-2541.31.1
371-3731.81.2
382-38821.1
Sheet C283-2892.11.6
1 Huber and Carrell (13) secondary structure nomenclature.

Homology modeling of mature recombinant mouse HSP47

Inhibitory and latent state models of mrmHSP47 were generated from the corresponding models of hPCI (see above) by means of homology modeling. By using the mrmHSP47/hPCI sequence alignment (Figure 1), alpha-carbon backbone coordinates of either inhibitory or latent state models of hPCI were easily assigned to corresponding residues of mrmHSP47. Thereafter, mrmHSP47 amino acid side chains were set, using the idealised geometries of CHARMm (27), and then energy minimisation protocols were followed to generate the desired inhibitory (Figure 3a) and latent (Figure 3b) state homology models of mrmHSP47. Both inhibitory and latent state homology models of mrmHSP47 were superimposed on the corresponding parent models of hPCI and the R.M.S. deviations between the major secondary structure elements recorded (Table 2).

Serine protease inhibition studies

Serine protease inhibition studies with purified, homogeneous mrmHSP47 were carried out in parallel with the molecular modeling studies so as to gain additional insight into the mrmHSP47 structure. In a series of standard assays (pH 7.5-7.8) for thrombin (30, 31) mrmHSP47 failed to inhibit this serine protease. Moreover, mrmHSP47 also failed to inhibit the serine proteases elastase, trypsin, kallikrein or tonin in other standard assays (pH 7.5-8.0) (30, 31, 32). To be completely sure that mrmHSP47 was not acting as a functional serine protease inhibitor towards any of the serine proteases tried, the highest concentrations of mrmHSP47 used in all assays were always in considerable excess over the Km values measured for the chromogenic substrates used with each protease enzyme.
Return to index

DISCUSSION

Previously, mrmHSP47 had been reported (9) to have an identity of 27 % with the prototypical serpin human alpha1-proteinase inhibitor (alpha1-PI) and predicted to have essentially the same secondary structure. However, the amino acid sequence identity of mrmHSP47 was found to be 30 % (Figure 1) with the serpin human protein C inhibitor (hPCI), therefore the structure of mouse HSP47 was modeled using the X-ray crystal structure coordinates of hPCI in preference.

The only currently available X-ray crystal structure of hPCI (2pai, Brookhaven Protein Data Bank) is of a cleaved structure (Figure 2a) in which the peptide bond between the P1 and P1' residues (Arg358 and Ser359) (Figure 1) is proteolytically hydrolysed leaving these residues separated in space by the expected distance of 70 Å. Therefore a necessary first step in the mrmHSP47 modeling procedure was to model the uncleaved hPCI. In line with the fact that both inhibitory and latent state conformations had been observed in the recently solved X-ray crystal structures of uncleaved serpins (33), both inhibitory and latent states of hPCI were also modeled. The modeling procedure adopted to generate the inhibitory state of hPCI was based upon a previous procedure used to model the inhibitory state of alpha1-PI from the known X ray crystal structure of the cleaved form (21). The features of the resulting structure, in particular the conformation adopted by the serpin loop (yellow strand, Figure 2b), agree well with the inhibitory structure of AT III (19, 20). Also, measurements of the R.M.S. deviations (Table 1) demonstrate that the inhibitory state model of hPCI superimposes on the X-ray crystal structure of cleaved hPCI with little notable structural difference. The modeling procedure used to obtain the latent, non-inhibitory form of hPCI was designed after close inspection of the X-ray crystal structure of latent, non-inhibitory PAI-1 (18). As a result, the conformation of the serpin loop (red strand, Figure 2c) in the model is very similar to the conformation displayed in the X-ray crystal structure of latent PAI-1 (18). Once more, the latent state model of hPCI superimposes very closely on the X-ray crystal structure of cleaved hPCI (Table 1).

The approach used to model the inhibitory and latent states of mrmHSP47 was drawn from an original homology modeling procedure devised to model the inhibitory state of ACHY (23) on the refined atomic coordinates of ovalbumin (16). Homology modeling is an appropriate and effective method for determining the three-dimensional structure of a protein provided that the three-dimensional structure of a protein related by at least 25 % amino acid sequence identity is available as a template (34). Clearly hPCI and mrmHSP47 satisfied this requirement (Figure 1) thereby justifying this approach to modeling the inhibitory and latent states of mrmHSP47. The resulting homology models of inhibitory (Figure 3a) and latent (Figure 3b) states of mrmHSP47 were found to superimpose well on their respective parent models of hPCI as shown by the data in Table 2 which reports the R.M.S. deviations of individual secondary structure elements between superimposed mrmHSP47 and hPCI model structures. Generally speaking, R.M.S. deviations of <= 2.0 Å are indicative of close superposition. Table 2 shows that the R.M.S. deviations between the superimposed latent models of HSP47 and hPCI were generally larger than those between inhibitory models, especially in the region of beta-sheet A.

Whilst both latent and inhibitory states of mrmHSP47 may exist, protease inhibitor assays were carried out with purified mrmHSP47 to determine which state or conformation was more likely to exist at physiological pH-values (pH 7.5 to 7.8). The close structural relationship between hPCI and mrmHSP47, particularly the identical P1 and P1' residues (Figure 1), suggested that mrmHSP47 would behave as a functional serine protease inhibitor with the same specificity against the serine protease thrombin as hPCI (35, 36). However, in a series of standard assays (pH 7.5-8.0), mrmHSP47 not only failed to inhibit thrombin but also the serine proteases elastase, trypsin, kallikrein and tonin. Therefore the latent, non-inhibitory state of mrmHSP47 appears to predominate at standard physiological pH-values. Circular dichroism spectroscopy of mrmHSP47 at neutral pH (Miller, unpublished data) revealed that the protein contains approx. 25 % alpha-helix and 36 % beta-sheet. Gratifyingly, the latent state model of mrmHSP47 was estimated to contain approx. 26% alpha-helix and approx. 33% beta-sheet which agrees very well with the percentage levels of secondary structure experimentally determined.

The latent state of mrmHSP47 possesses a long, deep cleft (seen in side view [Figure 3b] and top view [Figure 3c]) which appears to offer an explanation for the ability of mrmHSP47 to bind specifically and tightly to procollagen and collagen (Types I to V) peptides at standard physiological pH (11, 28, 37). The base of this cleft is formed by beta-sheet B with sides formed by helices hA and hG/hH (the helix and sheet nomenclature is derived from the Huber & Carrell labelling system [13]). Helices hA, hG/hH project hydrophilic amino acid residue side chains in towards the cleft whilst beta-sheet B projects hydrophobic amino acid residue side chains up from the bottom. This cleft is a very promising procollagen/collagen peptide binding groove for several reasons. Firstly, two deletion mutants of mrmHSP47, NΔ1 and CΔ3, have been prepared (Nagata, unpublished data) which were unable to bind collagen. NΔ1 is missing the first N-terminal 32 amino acids (Ala1-Leu32; shown in orange in Figure 3c) and CDelta 3 the last C-terminal 34 amino acids (Ala366-Leu400 or beta-strands s4B and s5B; shown in purple in Figure 3c). In both cases, the deleted peptide lengths comprise substantial parts of the putative procollagen/collagen binding groove. Secondly, this putative binding groove also bears reasonable similarity to the peptide binding regions of human class I histocompatibility glycoprotein HLA-Aw68 (38) and the human class II histocompatibility protein HLA-DR1(39). Finally, the corresponding region in hPCI has already been found to have a binding interaction with heparin (35, 40).

In conclusion, until the three-dimensional structure of mrmHSP47 is determined and refined at atomic resolution, the homology models of mrmHSP47 (Figure 3) serve as the next best structural reference for understanding structure/function relationships in serpin/molecular chaperone HSP47. This is the first time that a tandem-modeling procedure (i.e., using molecular models of an uncleaved serpin as a template for homology modeling) has been used to model the three-dimensional structure of a serpin.


Return to index

ACKNOWLEDGEMENTS

We thank the BBSRC, The Royal Society, The British Council and Roche Products Ltd for financial support.

REFERENCES

1. Hendrick, J. P., and Hartl, F.-U. (1993) Annu. Rev. Biochem. 62, 349-384.

2. Gething, M.-J., and Sambrook, J. (1992) Nature 355, 33-45.

3. Stress Proteins in Biology and Medicine (1990) (Morimoto, R. I., Tissieres, A., and Georgopoulos, C., Eds.), Cold Spring Harbor Laboratory Press, NY

4. Stress Proteins: Induction and Function (1990) (Schlesinger, M. J., Santoro, G., and Garaci, E., Eds.), Springer-Verlag, Heidelberg

5. Miller, A. D., Maghlaoui, K., Albanese, G., Kleinjan, D. A., and Smith, C. (1993) Biochem. J. 291, 139-144.

6. Hutchinson, J. P., El-Thaher, T. S. H., and Miller, A. D. (1994), Biochem. J. 302, 405-410.

7. Jain, N., Brickenden, A., Lorimer, I., Ball, E. H., and Sanwal, B. D. (1994), Biochem. J. 304, 61-68.

8. Wang, S.-Y. (1994)J. Biol. Chem. 269, 607-613.

9. Takechi, H., Hirayoshi, K., Nakai, A., Kudo, H., Saga, S., and Nagata, K. (1992), Eur. J. Biochem. 206, 323-329.

10. Hirayoshi, K., Kudo, H., Takechi, H., Nakai, A., Iwamatsu, A., Yamada, K. M., and Nagata, K. (1991) Mol. Cell. Biol. 11, 4036-4044.

11. Nakai, A., Satoh, M., Hirayoshi, K., and Nagata, K. (1992) J. Cell Biol. 117, 903-914.

12. Potempa, J., Korzus, E., and Travis, J. (1994) J. Biol. Chem. 269, 15957-15960.

13. Huber, R., and Carrell, R. W. (1989) Biochemistry 28, 8951-8966.

14. Schulze, A. J., Huber, R., Bode, W., and Engh, R. A. (1994), FEBS Lett. 344, 117-124.

15. Schechter, I., and Berger, A. (1967)Biochem. Biophys. Res. Commun. 27, 157-162.

16. Stein, P. E., Leslie, A. G. W., Finch, J. T., Turnell, W. G., McLaughlin, P. J., and Carrell, R. W. (1990) Nature 347, 99-102.

17. Wei, A., Rubin, H., Cooperman, B. S., Schechter, N., and Christianson, D. W. (1994) Nature Str. Biol. 1, 251-257.

18. Mottonen, J., Strand, A., Symersky, J., Sweet, R. M., Danley, D. E., Geoghegan, K. F., Gerard, R. D., and Goldsmith, E. J. (1992)Nature 355, 270-273.

19. Carrell, R. W., Stein, P. E., Fermi, G., and Wardell, M. R. (1994), Structure. 2, 257-270.

20. Schreuder, H. A., de Boer, B., Dijkema, R., Mulders, J., Theunissen, H. J. M., Grootenhuis, P. D. J., and Hol, W. G. J. (1994) Nature Str. Biol. 1, 48-54.

21. Engh, R. A., Wright, H. T., and Huber, R. (1990) Prot. Eng. 3, 469-477.

22. Jarvis, J. A., Munro, S. L. A., and Craik, D. J. (1992) Prot. Eng. 5, 61-67.

23. Katz, D. S., and Christianson, D. W. (1993) Prot. Eng. 6, 701-709.

24. Needleman, S. B., and Wunsch, C. D. (1970) J. Mol. Biol. 48, 443.

25. Atlas of Protein Sequence and Structure (1978) (Dayhoff, M. O., Ed.) 5(3), NBRF, Silver Spring, MD.

26. Feng, D. F., and Doolittle, R. F. (1987)J. Molec. Evol. 25, 351-360.

27. Brooks, B. R., Bruccoleri, R. E., Olafson, B. D., States, D. J., Swaminathan, S., and Karplus, M. (1983)J. Comp. Chem. 4, 187-217.

28. Saga, S., Nagata, K., Chen, W.-T., and Yamada, K. M. (1987), J. Cell Biol. 105, 517-527.

29. Warburg, O., and Christian, W. (1941)Biochim. Z. 310, 384-421.

30. Lottenburg, R., Christensen, U., Jackson, C. M., and Coleman, P. L. (1981), Meth. Enzymol. 80, 341-361.

31. Castillo, M. J., Nakajima, K., Zimmerman, M., and Powers, J. C. (1979), Anal. Biochem. 99, 53-64.

32. El-Thaher, T. S. H. & Bailey, G. S. (1993) Int. J. Peptide Protein Res. 41, 196-200.

33. Goldsmith, E. J., and Mottonen, J. (1994) Structure 2, 241-244.

34. Blundell, T. L., Sibanda, B. L., Sternberg, M. J. E., and Thornton, J. M. (1987), Nature 326, 347-352.

35. Whinna, H. C., and Church, F. C. (1993) J. Prot. Chem. 12, 677-688.

36. Carrell, R. W., and Boswell, D. R. (1986) in Serpins: The superfamily of plasma serine proteinase inhibitors. Proteinase Inhibitors , pp 403-420, Elsevier Science BV.

37. Natsume, T., Koide, T., Yokota, S., Hirayoshi, K., and Nagata, K. (1994), J. Biol. Chem. 269, 31224-31228.

38. Silver, M. L., Guo, H.-C., Strominger, J. L., and Wiley, D. C. (1992), Nature 360, 367-369.

39. Stern, L. J. , Brown J. H., Jardetsky, T. S., Gorga, J. C., Urban, R. G., Strominger, J. L., and Wiley, D. C. (1994) Nature 368, 215-221.

40. Shirk, R. A., Elisen, M. G. L. M., Meijers, J. C. M., and Church, F. C. (1994), J. Biol. Chem. 269, 28690-28695.


Return to index

FIGURE LEGENDS

Fig. 1. Amino acid sequence alignment of mature recombinant mouse HSP47 (mrmHSP47) with human protein C inhibitor (hPCI). The serine protease inhibitor loop region of hPCI, and the corresponding region of mrmHSP47, is indicated using the Pn nomenclature of Schechter & Berger (15). Identical residues are shown against a black background and similar residues against a hatched background.

Fig. 2 a. X-ray crystal structure of human protein C inhibitor (hPCI). beta-Sheet A is shown in yellow, beta-strand s4A in red and beta-strand s1C in green. Huber and Carrell (13) nomenclature applies. 2b. Inhibitory state model of hPCI. Modeled serpin loop is shown in yellow. 2c. Latent state model of hPCI. Modeled serpin loop is shown in red.

Fig. 3 a. Side view of inhibitory state homology model of mature recombinant mouse HSP47 (mrmHSP47). alpha-Helices are coloured blue, beta-strands red and loops yellow. 3b. Side view of latent state homology model of mrmHSP47. alpha-Helices are coloured blue, beta-strands red and loops yellow. 3c. Close-up, top view of the putative procollagen/collagen binding groove. The groove has been coloured coloured to show the peptide lengths of mrmHSP47 deleted in mutant NDelta 1, (orange segment) and CDelta 3 (purple segment).


Return to index
keyword search Home page