research papers\(\def\hfill{\hskip 5em}\def\hfil{\hskip 3em}\def\eqno#1{\hfil {#1}}\)

IUCrJ
Volume 6| Part 2| March 2019| Pages 317-330
ISSN: 2052-2525

Structure of mammalian plasma fetuin-B and its mechanism of selective metallopeptidase inhibition

aProteolysis Laboratory, Department of Structural Biology, Molecular Biology Institute of Barcelona, CSIC, Barcelona Science Park, Helix Building, c/o Baldiri Reixac 15-21, E-08028 Barcelona, Catalonia, Spain, bInstitute of Molecular Physiology, Cell and Matrix Biology, Johannes Gutenberg-University Mainz, Johann-Joachim-Becher-Weg 7, D-55128 Mainz, Germany, cDepartment of Biosciences and Nutrition and Center for Innovative Medicine, Karolinska Institutet, Blickagången 16, SE-141 83 Huddinge, Sweden, dBiointerface Laboratory, Helmholtz Institute for Biomedical Engineering, RWTH Aachen University Medical Faculty, Pauwelsstrasse 30, D-52074 Aachen, Germany, eESRF – The European Synchrotron, 71 Rue Jules Horowitz, F-38000 Grenoble, France, and fInstitute of Molecular Pathobiochemistry, Experimental Gene Therapy and Clinical Chemistry, RWTH Aachen University Hospital, Pauwelsstrasse 30, D-52074 Aachen, Germany
*Correspondence e-mail: stoecker@uni-mainz.de, luca.jovine@ki.se, xgrcri@ibmb.csic.es

Edited by J. L. Smith, University of Michigan, USA (Received 13 November 2018; accepted 28 January 2019; online 28 February 2019)

Mammalian fetuin-A and fetuin-B are abundant serum proteins with pleiotropic functions. Fetuin-B is a highly selective and potent inhibitor of metallo­peptidases (MPs) of the astacin family, which includes ovastacin in mammals. By inhibiting ovastacin, fetuin-B is essential for female fertility. The crystal structure of fetuin-B was determined unbound and in complex with archetypal astacin, and it was found that the inhibitor has tandem cystatin-type modules (CY1 and CY2). They are connected by an exposed linker with a rigid, disulfide-linked `CPDCP-trunk', and are followed by a C-terminal region (CTR) with little regular secondary structure. The CPDCP-trunk and a hairpin of CY2 form a bipartite wedge, which slots into the active-site cleft of the MP. These elements occupy the nonprimed and primed sides of the cleft, respectively, but spare the specificity pocket so that the inhibitor is not cleaved. The aspartate in the trunk blocks the catalytic zinc of astacin, while the CY2 hairpin binds through a QWVXGP motif. The CY1 module assists in structural integrity and the CTR is not involved in inhibition, as verified by in vitro studies using a cohort of mutants and variants. Overall, the inhibition conforms to a novel `raised-elephant-trunk' mechanism for MPs, which is reminiscent of single-domain cystatins that target cysteine peptidases. Over 200 sequences from vertebrates have been annotated as fetuin-B, underpinning its ubiquity and physiological relevance; accordingly, sequences with conserved CPDCP- and QWVXGP-derived motifs have been found from mammals to cartilaginous fishes. Thus, the raised-elephant-trunk mechanism is likely to be generally valid for the inhibition of astacins by orthologs of fetuin-B.

1. Introduction

Fetuin was discovered in 1944 as the most abundant globulin in fetal calf serum (Pedersen, 1944[Pedersen, K. O. (1944). Nature (London), 154, 575.]). Today, fetuins constitute a broad group of liver-derived plasma and matrix proteins (Lee, 2009[Lee, C. (2009). Front. Biosci. 14, 2911-2922.]). These include fetuin-A, also known as α2-Heremans–Schmid glycoprotein, and fetuin-B, which was discovered as a paralog of fetuin-A in 1999 (Olivier et al., 2000[Olivier, E., Soury, E., Ruminy, P., Husson, A., Parmentier, F., Daveau, M. & Salier, J.-P. (2000). Biochem. J. 350, 589-597.]; Denecke et al., 2003[Denecke, B., Gräber, S., Schäfer, C., Heiss, A., Wöltje, M. & Jahnen-Dechent, W. (2003). Biochem. J. 376, 135-145.]). Fetuins belong to the cystatin family of proteins (Turk & Bode, 1991[Turk, V. & Bode, W. (1991). FEBS Lett. 285, 213-219.]), which matches family I25 of the MEROPS database (https://www.ebi.ac.uk/merops). This family is derived from the archetypal reversible inhibitor specific for cysteine peptidases, the monomeric 116-residue chicken egg-white cystatin (ovocystatin; Bode et al., 1988[Bode, W., Engh, R., Musil, D., Thiele, U., Huber, R., Karshikov, A., Brzin, J., Kos, J. & Turk, V. (1988). EMBO J. 7, 2593-2599.]; Stubbs et al., 1990[Stubbs, M. T., Laber, B., Bode, W., Huber, R., Jerala, R., Lenarcic, B. & Turk, V. (1990). EMBO J. 9, 1939-1947.]). Among the cystatins, fetuins belong to subfamily I25C, also referred to as type-3 cystatins, which groups glycosylated proteins with two or three cystatin-like repeats (Rawlings & Barrett, 1990[Rawlings, N. D. & Barrett, A. J. (1990). J. Mol. Evol. 30, 60-71.]; Lee, 2009[Lee, C. (2009). Front. Biosci. 14, 2911-2922.]). Fetuins consist of tandem cystatin domains (hereafter referred to as CY1 and CY2) followed by a C-terminal region (hereafter referred to as CTR). Another member of this subfamily is histidine-rich glycoprotein, a two-repeat plasma protein from vertebrates that is involved in the vascular, coagulation and immune systems and is not a peptidase inhibitor (Turk & Bode, 1991[Turk, V. & Bode, W. (1991). FEBS Lett. 285, 213-219.]). Yet other members are the kininogens, which are the precursors of the kinin hormones and are initiators of the surface-activated blood-coagulation cascade. They comprise three cystatin modules, the last two of which inhibit cysteine peptidases (Salvesen et al., 1986[Salvesen, G., Parkes, C., Abrahamson, M., Grubb, A. & Barrett, A. J. (1986). Biochem. J. 234, 429-434.]). Finally, the subfamily also includes endogenous inhibitors of snake-venom metallopeptidases (MPs) from the ADAM/adamalysin family (Gomis-Rüth, 2013[Gomis-Rüth, F. X. (2013). Encyclopedia of Metalloproteins., edited by V. N. Uversky, R. H. Kretsinger & E. A. Permyakov, pp. 2473-2479. Heidelberg: Springer-Verlag.]). These inhibitors contain two cystatin modules and operate through an unknown mechanism.

Fetuin-A serum levels are high throughout life and are even higher during fetal development; fetuin-B is less abundant than fetuin-A (Floehr et al., 2016[Floehr, J., Dietzel, E., Neulen, J., Rösing, B., Weissenborn, U. & Jahnen-Dechent, W. (2016). Hum. Reprod. 31, 630-637.], 2017[Floehr, J., Dietzel, E., Schmitz, C., Chappell, A. & Jahnen-Dechent, W. (2017). Mol. Hum. Reprod. 23, 34-44.]). Despite their common phylogenetic origin and close chromosomal location within all studied vertebrate genomes, fetuin-A and fetuin-B are chemically different and have separate biological functions. An important role in mineralized matrix metabolism has been demonstrated for fetuin-A, among several other moonlighting functions (Jahnen-Dechent et al., 2011[Jahnen-Dechent, W., Heiss, A., Schäfer, C. & Ketteler, M. (2011). Circ. Res. 108, 1494-1509.]), while fetuin-B is essential for fertilization (Dietzel et al., 2013[Dietzel, E., Wessling, J., Floehr, J., Schäfer, C., Ensslen, S., Denecke, B., Rösing, B., Neulen, J., Veitinger, T., Spehr, M., Tropartz, T., Tolba, R., Renné, T., Egert, A., Schorle, H., Gottenbusch, Y., Hildebrand, A., Yiallouros, I., Stöcker, W., Weiskirchen, R. & Jahnen-Dechent, W. (2013). Dev. Cell, 25, 106-112.]; Stöcker et al., 2014[Stöcker, W., Karmilin, K., Hildebrand, A., Westphal, H., Yiallouros, I., Weiskirchen, R., Dietzel, E., Floehr, J. & Jahnen-Dechent, W. (2014). Biol. Chem. 395, 1195-1199.]). It blocks the proteolytic activity of ovastacin, a member of the astacin family (Gomis-Rüth, Trillo-Muyo et al., 2012[Gomis-Rüth, F. X., Trillo-Muyo, S. & Stöcker, W. (2012). Biol. Chem. 393, 1027-1041.]) within the metzincin clan of MPs (Gomis-Rüth, 2009[Gomis-Rüth, F. X. (2009). J. Biol. Chem. 284, 15353-15357.]; Cerdà-Costa & Gomis-Rüth, 2014[Cerdà-Costa, N. & Xavier Gomis-Rüth, F. (2014). Protein Sci. 23, 123-144.]). Ovastacin mediates the hardening of the zona pellucida (ZP), a glycoprotein matrix surrounding the oocyte, by cleaving the ZP component ZP2 after gamete fusion (Burkart et al., 2012[Burkart, A. D., Xiong, B., Baibakov, B., Jiménez-Movilla, M. & Dean, J. (2012). J. Cell Biol. 197, 37-44.]) and thus preventing further sperm from penetrating the oocyte. By blocking ovastacin, fetuin-B counteracts premature ZP hardening before fertilization and therefore maintains female fertility by keeping the ZP penetrable to sperm (Körschgen et al., 2017[Körschgen, H., Kuske, M., Karmilin, K., Yiallouros, I., Balbach, M., Floehr, J., Wachten, D., Jahnen-Dechent, W. & Stöcker, W. (2017). Mol. Hum. Reprod. 23, 607-616.]).

Detailed biochemical analysis revealed that mammalian fetuin-B inhibits its physiological target ovastacin at picomolar concentrations and most other astacins, for example crayfish astacin, zebrafish nephrosin and human meprins α and β, at nanomolar concentrations (Karmilin et al., 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]). As no other peptidase type was inhibited, fetuin-B is considered to be an endogenous and specific inhibitor of non-BMP-1-like astacins (Karmilin et al., 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]). It complements the broad-spectrum pan-peptidase inhibitor α2-macroglobulin (Goulas et al., 2017[Goulas, T., Garcia-Ferrer, I., Marrero, A., Marino-Puertas, L., Duquerroy, S. & Gomis-Rüth, F. X. (2017). Biol. Chem. 398, 975-994.]) and Xenopus laevis Sizzled/Ogon, which blocks BMP-1-like astacins (Lee et al., 2006[Lee, H. X., Ambrosio, A. L., Reversade, B. & De Robertis, E. M. (2006). Cell, 124, 147-159.]). In addition, a carp ortholog of fetuin strongly inhibited carp nephrosin, an endogenous astacin family member (Tsai et al., 2004[Tsai, P.-L., Chen, C.-H., Huang, C.-J., Chou, C.-M. & Chang, G.-D. (2004). J. Biol. Chem. 279, 11146-11155.]). Finally, unlike fetuin-B and despite conflicting reports (Yamamoto & Sinohara, 1993[Yamamoto, K. & Sinohara, H. (1993). J. Biol. Chem. 268, 17750-17753.]; Yoshida et al., 1996[Yoshida, K., Suzuki, Y., Yamamoto, K. & Sinohara, H. (1996). Biochem. Mol. Biol. Int. 39, 1023-1028.]; Hedrich et al., 2010[Hedrich, J., Lottaz, D., Meyer, K., Yiallouros, I., Jahnen-Dechent, W., Stöcker, W. & Becker-Pauly, C. (2010). Biochemistry, 49, 8599-8607.]), fetuin-A is not a peptidase inhibitor (Lee, 2009[Lee, C. (2009). Front. Biosci. 14, 2911-2922.]; Dietzel et al., 2013[Dietzel, E., Wessling, J., Floehr, J., Schäfer, C., Ensslen, S., Denecke, B., Rösing, B., Neulen, J., Veitinger, T., Spehr, M., Tropartz, T., Tolba, R., Renné, T., Egert, A., Schorle, H., Gottenbusch, Y., Hildebrand, A., Yiallouros, I., Stöcker, W., Weiskirchen, R. & Jahnen-Dechent, W. (2013). Dev. Cell, 25, 106-112.]; Karmilin et al., 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]; Galembeck & Cann, 1974[Galembeck, F. & Cann, J. R. (1974). Arch. Biochem. Biophys. 164, 326-331.]): depending on the analyzed peptidase, the apparent inhibitory activity of fetuin-A was owing to contaminating fetuin-B or to chromogenic substrate competition by fetuin-A.

Currently, more than 200 sequences, exclusively from vertebrates, have been annotated as fetuin-B in the UniProt database. This supports a ubiquity and physiological relevance of this protein that is probably associated with the selective and pleiotropic inhibition of a family of MPs. To understand this function at the molecular level and to guide the rational design of small-molecule regulators, we solved the structure of mouse fetuin-B both unbound and in complex with the archetypal MP astacin, which is a valid model for the physiological target ovastacin. We verified the findings using inhibitory assays of recombinant protein variants, and further discuss the implications of this novel inhibition mechanism.

2. Materials and methods

2.1. Protein production and purification

Crayfish astacin was purified from the digestive fluid of the European freshwater crayfish Astacus astacus L. as described by Gomis-Rüth et al. (1993[Gomis-Rüth, F. X., Stöcker, W., Huber, R., Zwilling, R. & Bode, W. (1993). J. Mol. Biol. 229, 945-968.]). Recombinant mouse pro-ovastacin was obtained as reported by Dietzel et al. (2013[Dietzel, E., Wessling, J., Floehr, J., Schäfer, C., Ensslen, S., Denecke, B., Rösing, B., Neulen, J., Veitinger, T., Spehr, M., Tropartz, T., Tolba, R., Renné, T., Egert, A., Schorle, H., Gottenbusch, Y., Hildebrand, A., Yiallouros, I., Stöcker, W., Weiskirchen, R. & Jahnen-Dechent, W. (2013). Dev. Cell, 25, 106-112.]) and was proteolytically activated with plasmin (Karmilin et al., 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]). Human meprin β was obtained as described by Becker et al. (2003[Becker, C., Kruse, M. N., Slotty, K. A., Köhler, D., Harris, J. R., Rösmann, S., Sterchi, E. E. & Stöcker, W. (2003). Biol. Chem. 384, 825-831.]) and was activated with trypsin (Fridrich et al., 2016[Fridrich, S., Karmilin, K. & Stöcker, W. (2016). Curr. Protoc. Protein Sci. 83, 21.16.1-21.16.20.]). Mouse fetuin-A and mouse fetuin-B, as well as the mouse fetuin-B loop-swap variant mFBH1-swap, in which the sequence Q199WV­SGP204 is replaced by Q78EDMGP82, were hexahistidine-tagged at the C-terminus, cloned in pFASTBac1 vector and expressed in baculovirus-transduced High Five cells as described for meprin (Becker et al., 2003[Becker, C., Kruse, M. N., Slotty, K. A., Köhler, D., Harris, J. R., Rösmann, S., Sterchi, E. E. & Stöcker, W. (2003). Biol. Chem. 384, 825-831.]; Becker-Pauly et al., 2007[Becker-Pauly, C., Höwel, M., Walker, T., Vlad, A., Aufenvenne, K., Oji, V., Lottaz, D., Sterchi, E. E., Debela, M., Magdolen, V., Traupe, H. & Stöcker, W. (2007). J. Invest. Dermatol. 127, 1115-1125.]). Bovine and human fetuin-B orthologs, as well as the CY2 (residues K147–F265; mouse fetuin-B residue numbers are shown as superscripts; see UniProt Q9QXC1) and CTR (residues F266–P388) domains of mouse fetuin-B and point mutants (P155A, D156A, P155A+D156A, C154S+C157S and C39S+C374S) of mouse fetuin-B, were expressed in ExpiCHO-S cells (ThermoFisher Scientific, Waltham, USA) according to the manufacturer's specifications (Karmilin et al., 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]). For the generation of point mutants of mouse fetuin-B, the original pcDNATM3.4 TOPO_mouseFetuB construct (Karmilin et al., 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]) was used as a template for site-directed mutagenesis. All variants were generated by the QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, USA) using the standard protocol and specific primers. Following mutagenesis, parental DNA was digested by DpnI. Afterwards, XL10-Gold ultracompetent cells were transformed with the resulting PCR products according to the manufacturer's instructions. The gene sequences were confirmed by DNA sequencing (Eurofins Genomics, Ebersberg, Germany). The chimeric mutant mFBABA comprised the CY1 (residues A19–P136; mouse fetuin-A residue numbers are shown in italics, see UniProt P29699) and CTR (residues P254–I345) domains of mouse fetuin-A plus an intercalated mouse fetuin-B CY2 domain (residues V145–Q269) and was obtained from adenovirus-infected COS-7 cells (Karmilin et al., 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]). All proteins carry a cleavable C-terminal hexa­histidine tag and were purified by nickel–nitrilotriacetic acid (Ni2+–NTA) affinity chromatography as described previously (Dietzel et al. 2013[Dietzel, E., Wessling, J., Floehr, J., Schäfer, C., Ensslen, S., Denecke, B., Rösing, B., Neulen, J., Veitinger, T., Spehr, M., Tropartz, T., Tolba, R., Renné, T., Egert, A., Schorle, H., Gottenbusch, Y., Hildebrand, A., Yiallouros, I., Stöcker, W., Weiskirchen, R. & Jahnen-Dechent, W. (2013). Dev. Cell, 25, 106-112.]; Karmilin et al. 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]). Finally, cyclized peptides including the sequences CPDC (full sequence Ac-VSKRKTHTTCPDCPSPIDL) and CPRC (full sequence Ac-DSAEDVRKLCPRCPLLTPFN) were purchased from JPT Peptide Technologies GmbH, Berlin, Germany.

Mouse fetuin-B was also produced in HEK293S cells, which synthesize proteins carrying endoglycosidase H-sensitive Man5GlcNAc2 N-glycans (Reeves et al., 2002[Reeves, P. J., Callewaert, N., Contreras, R. & Khorana, H. G. (2002). Proc. Natl Acad. Sci. USA, 99, 13419-13424.]). For this, a cDNA encoding full-length mouse fetuin-B (GenScript) was cloned in frame with the 3′ hexahistidine tag-encoding sequence of pHLsec2, a mammalian expression vector derived from pHLsec (Aricescu et al., 2006[Aricescu, A. R., Lu, W. & Jones, E. Y. (2006). Acta Cryst. D62, 1243-1250.]). The resulting construct was used for transient transfections using 25 kDa branched PEI (Sigma–Aldrich; Aricescu et al., 2006[Aricescu, A. R., Lu, W. & Jones, E. Y. (2006). Acta Cryst. D62, 1243-1250.]). The secreted fetuin-B was purified by Ni2+–NTA chromatography, deglycosylated with endoglycosidase H and purified by size-exclusion chromatography essentially as described by Bokhove et al. (2016[Bokhove, M., Sadat Al Hosseini, H., Saito, T., Dioguardi, E., Gegenschatz-Schmid, K., Nishimura, K., Raj, I., de Sanctis, D., Han, L. & Jovine, L. (2016). J. Struct. Biol. 194, 1-7.]), except that the deglycosylation step was carried out overnight at 4°C. Finally, the protein was concentrated to 7.5–30 mg ml−1 in 150 mM sodium chloride, 20 mM sodium HEPES pH 7.8.

2.2. Crystallization and diffraction data collection

Crystallization assays of the astacin–fetuin-B complex were performed using the sitting-drop vapor-diffusion method. Reservoir solutions were prepared by a Tecan robot and 100 nl crystallization drops were dispensed onto 96 × 2-well MRC plates (Innovadyne) at the joint IBMB/IRB Automated Crystallography Platform at Barcelona Science Park using a Cartesian MicroSys 4000 XL robot (Genomic Solutions) or a Phoenix nanodrop robot (Art Robbins). Plates were kept in Bruker steady-temperature crystal farms at 4 or 20°C. Crystals were obtained by incubating the mouse inhibitor with a sixfold molar excess of the crayfish peptidase. The best crystals were obtained at 20°C in 0.1 µl:0.1 µl drops with protein solution at a concentration of 6 mg ml−1 with 200 mM sodium chloride, 10 mM Tris–HCl pH 7.8 as the buffer and 0.05 M ammonium sulfate, 20%(w/v) polyethylene glycol 2000, 0.1 M sodium acetate pH 4.6 as the reservoir solution. Crystals were cryoprotected by rapid passage through drops containing increasing concentrations of glycerol [up to 10%(v/v)]. It is noteworthy that properly diffracting crystals appeared after 4–5 days and rapidly lost diffracting power after a further 1–2 weeks. SDS–PAGE and N-terminal Edman degradation of washed and dissolved complex crystals revealed that fetuin-B was cleaved at positions S296–S297 within the CTR, which did not alter its inhibitory properties (data not shown).

Crystals of intact unbound mouse fetuin-B produced in HEK293S cells were grown using a Mosquito robot (TTP Labtech) by sitting-drop vapor diffusion at room temperature against a mother liquor consisting of 100 mM sodium acetate, 25%(w/v) PEG 4000, 8%(v/v) isopropanol.

Diffraction data for the astacin–fetuin-B complex crystals were collected at 100 K from liquid-nitrogen flash-cryocooled crystals (Oxford Cryosystems 700 Series Cryostream) using a PILATUS 6M detector (Dectris) on the XALOC beamline (Juanhuix et al., 2014[Juanhuix, J., Gil-Ortiz, F., Cuní, G., Colldelram, C., Nicolás, J., Lidón, J., Boter, E., Ruget, C., Ferrer, S. & Benach, J. (2014). J. Synchrotron Rad. 21, 679-689.]) of the ALBA synchrotron, Cerdanyola, Catalonia, Spain. Diffraction data were processed with XDS (Kabsch, 2010a[Kabsch, W. (2010a). Acta Cryst. D66, 125-132.]) and XSCALE (Kabsch, 2010b[Kabsch, W. (2010b). Acta Cryst. D66, 133-144.]), and were transformed using XDSCONV to a format suitable for the CCP4 suite of programs (Winn et al., 2011[Winn, M. D., Ballard, C. C., Cowtan, K. D., Dodson, E. J., Emsley, P., Evans, P. R., Keegan, R. M., Krissinel, E. B., Leslie, A. G. W., McCoy, A., McNicholas, S. J., Murshudov, G. N., Pannu, N. S., Potterton, E. A., Powell, H. R., Read, R. J., Vagin, A. & Wilson, K. S. (2011). Acta Cryst. D67, 235-242.]). The crystals belonged to space group P212121, contained two peptidase–inhibitor complexes per asymmetric unit (molecules A/B and C/D, respectively) and diffracted to 3.1 Å resolution.

Diffraction data for the free fetuin-B crystals were collected at 100 K from liquid-nitrogen flash-cryocooled crystals using a PILATUS 6M-F detector (Dectris) on beamline ID29 (de Sanctis et al., 2012[Sanctis, D. de, Beteva, A., Caserotto, H., Dobias, F., Gabadinho, J., Giraud, T., Gobbo, A., Guijarro, M., Lentini, M., Lavault, B., Mairs, T., McSweeney, S., Petitdemange, S., Rey-Bakaikoa, V., Surr, J., Theveneau, P., Leonard, G. A. & Mueller-Dieckmann, C. (2012). J. Synchrotron Rad. 19, 455-461.]) at the European Synchrotron Radiation Facility (ESRF), Grenoble, France. Three 360° wedges collected from different positions of a single crystal at 7 keV (1.7712 Å) were first integrated and scaled with XDS and then merged with XSCALE, resulting in a 31-fold redundant data set. The crystals belonged to space group P412121, contained one inhibitor molecule per asymmetric unit and diffracted to 2.3 Å resolution (see Table 1[link] for data-processing statistics).

Table 1
Crystallographic data

Values in parentheses are for the outermost resolution shell.

Data set Astacin–fetuin-B Fetuin-B
Space group P212121 P41212
a, b, c (Å) 83.4, 85.8, 168.7 67.7, 67.7, 197.8
Wavelength (Å) 0.9792 1.7712
No. of measurements 264256 384912
No. of unique reflections 22647 19993
Resolution range (Å) 84.4–3.10 (3.28–3.10) 47.2–2.30 (2.38–2.30)
Completeness (%) 100.0 (100.0) 93.6 (63.0)
Rmerge 0.153 (1.499) 0.109 (1.626)
Rmeas 0.160 (1.567) 0.112 (1.782)
CC1/2 0.998 (0.771) 0.999 (0.521)
Average intensity 14.1 (2.0) 20.3 (1.4)
B factor (Wilson) (Å2) 78.8 54.2
Avererage multiplicity 11.7 (11.6) 19.3 (6.0)
No. of reflections used in refinement 21981 19989
No. of reflections in test set 643 1000
R factor/free R factor 0.216/0.270 0.224/0.254
Correlation coefficient, FobsFcalc 0.931 0.931
Correlation coefficient, test set 0.882 0.939
No. of protein residues 1127 299
No. of atoms 7899 2344
No. of solvent molecules 46 85
No. of covalent ligands 5 NAG 2 NAG
No. of noncovalent ligands 2 Zn2+, 3 glycerol 1 acetate
R.m.s.d. from target values
 Bonds (Å) 0.010 0.002
 Angles (°) 1.14 0.59
Average B factors (Å2)
 Overall 101.2 67.1
 Molecule A 79.2 67.7
 Molecule B 109.0
 Molecule C 89.6
 Molecule D 117.1
All-atom contacts and geometry analysis
 Residues in favored regions 908 [91.6%] 280 [96.7%]
 Outliers 7 0
 All residues 991 290
 Outlying rotamers 56 [6.4%] 5 [1.8%]
 Outlying bonds 0 0
 Outlying angles 0 0
 Outlying chirality 0 0
 Outlying planarity 0 0
 All-atom clashscore 4.8 3.2
N-acetyl-D-glucosamine.

2.3. Structure solution and refinement

The structure of the astacin–fetuin-B complex was solved by maximum-likelihood-scored molecular replacement using Phaser (McCoy et al., 2007[McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. & Read, R. J. (2007). J. Appl. Cryst. 40, 658-674.]) with the coordinates of the protein part of unbound astacin (PDB entry 1ast; Bode et al., 1992[Bode, W., Gomis-Rüth, F. X., Huber, R., Zwilling, R. & Stöcker, W. (1992). Nature (London), 358, 164-167.]; Gomis-Rüth et al., 1993[Gomis-Rüth, F. X., Stöcker, W., Huber, R., Zwilling, R. & Bode, W. (1993). J. Mol. Biol. 229, 945-968.]) as a search model. These calculations were followed by automated density modification and model building with the AutoBuild protocol of the PHENIX suite (Terwilliger et al., 2008[Terwilliger, T. C., Grosse-Kunstleve, R. W., Afonine, P. V., Moriarty, N. W., Zwart, P. H., Hung, L.-W., Read, R. J. & Adams, P. D. (2008). Acta Cryst. D64, 61-69.]), which included twofold averaging. The resulting Fourier map initiated 12 rounds of manual model building with Coot (Emsley et al., 2010[Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. (2010). Acta Cryst. D66, 486-501.]), which alternated with crystallographic refinement with PHENIX (Afonine et al., 2012[Afonine, P. V., Grosse-Kunstleve, R. W., Echols, N., Headd, J. J., Moriarty, N. W., Mustyakimov, M., Terwilliger, T. C., Urzhumtsev, A., Zwart, P. H. & Adams, P. D. (2012). Acta Cryst. D68, 352-367.]) and BUSTER/TNT (Smart et al., 2012[Smart, O. S., Womack, T. O., Flensburg, C., Keller, P., Paciorek, W., Sharff, A., Vonrhein, C. & Bricogne, G. (2012). Acta Cryst. D68, 368-380.]) with the inclusion of translation/libration/screw-rotation refinement and noncrystallographic symmetry restraints. The final refined model was comprised of residues A1–H202 and A1–L200 [mature astacin residue numbers are shown in subscript, as in Gomis-Rüth et al. (1993[Gomis-Rüth, F. X., Stöcker, W., Huber, R., Zwilling, R. & Bode, W. (1993). J. Mol. Biol. 229, 945-968.]) and Guevara et al. (2010[Guevara, T., Yiallouros, I., Kappelhoff, R., Bissdorf, S., Stöcker, W. & Gomis-Rüth, F. X. (2010). J. Biol. Chem. 285, 13958-13965.]); add 49 for residue numbers in the full-length sequence; see UniProt P07584] from astacin molecules A and C, respectively, plus a zinc cation each, residues R29–P388 (except for T218–Q227, L246–L251 and P273–A302) plus S501–A502 from the C-terminal purification tag from molecule B and Q28–P388 plus S501 (except for T218–H228, H248–L251, S268–A302, D314–P326 and P354–G355) from molecule D. One N-acetyl-D-glucosamine (NAG) moiety was attached to N40 and N139 of molecule B, and one and two NAG moieties were linked to N40 and N139 of molecule D, respectively. Three glycerols plus 46 solvent molecules completed the model. Some regions of molecule D were only poorly defined in the final Fourier maps and were modeled based on molecule B to preserve the overall chain continuity, as this led to lower free R-factor values. The satisfactory quality of the final model was assessed with the wwPDB Validation Service (https://www.wwpdb.org/validation; Berman et al., 2003[Berman, H., Henrick, K. & Nakamura, H. (2003). Nat. Struct. Biol. 10, 980.]).

The structure of isolated fetuin-B was independently solved by sulfur SAD using the CRANK2 (Pannu et al., 2011[Pannu, N. S., Waterreus, W.-J., Skubák, P., Sikharulidze, I., Abrahams, J. P. & de Graaff, R. A. G. (2011). Acta Cryst. D67, 331-337.]) experimental phasing pipeline of CCP4 Cloud's jsCoFE (Krissinel et al., 2018[Krissinel, E., Uski, V., Lebedev, A., Winn, M. & Ballard, C. (2018). Acta Cryst. D74, 143-151.]). Using an initial high-resolution cutoff of 4.5 Å, this procedure located a substructure that included 12 sites corresponding to five disulfide bonds and three methionine residues (mean phasing figures of merit of 0.1084 before density modification and 0.4363 after density modification with Fourier recycling) and produced an initial model consisting of 285 amino acids (Rwork = 30.9%, Rfree = 33.8%). This model was expanded using AutoBuild within PHENIX. It was then manually rebuilt with Coot and refined against the first wedge of diffraction data with PHENIX. The final model included residues L31–P388 (except for the segments T218–H228, S244–V255, A270–T295 and E316–S325), one NAG attached to N40 and N139, one acetate and 85 solvent molecules. Structure validation was performed with MolProbity (Williams et al., 2018[Williams, C. J., Headd, J. J., Moriarty, N. W., Prisant, M. G., Videau, L. L., Deis, L. N., Verma, V., Keedy, D. A., Hintze, B. J., Chen, V. B., Jain, S., Lewis, S. M., Arendall, W. B. III, Snoeyink, J., Adams, P. D., Lovell, S. C., Richardson, J. S. & Richardson, D. C. (2018). Protein Sci. 27, 293-315.]), Privateer (Agirre et al., 2015[Agirre, J., Iglesias-Fernández, J., Rovira, C., Davies, G. J., Wilson, K. S. & Cowtan, K. D. (2015). Nat. Struct. Mol. Biol. 22, 833-834.]) and pdb-care (Lütteke & Lieth, 2004[Lütteke, T. & von der Lieth, C. W. (2004). BMC Bioinformatics, 5, 69.]). Data-collection, refinement and validation statistics are reported in Table 1[link].

2.4. Inhibition assays in vitro

The inhibitory capacities of fetuin variants towards mouse ovastacin, human meprin β and crayfish astacin were determined in vitro by means of a fluorogenic enzyme-activity assay monitored using a Varioskan Flash 3001 spectral plate reader equipped with the Skanit 2.4.3.RE software (Thermo Scientific, Dreieich, Germany). Enzyme concentrations for astacin and meprin β were determined from the absorbance at 280 nm (astacin = 42 800 M−1 cm−1, meprin β = 106 520 M−1 cm−1) and for ovastacin via the IC50 calculation of wild-type murine fetuin-B. Assays were performed in triplicate at 37°C in a 100 µl final volume, with 150 mM sodium chloride, 50 mM Tris–HCl pH 7.4, 0.01% Brij 35 as the buffer. Enzyme-activity measurements were started by the addition of 20–30 µM Ac-RE(Edans)DRNleV­GDDP­YK(Dabcyl)-NH2 (Biosyntan GmbH, Berlin, Germany) for ovastacin and meprin β or 70–80 µM Dansyl-PKRAPWV-OH (PANATecs GmbH, Heilbronn, Germany) for astacin dissolved in dimethyl sulfoxide (final concentration 0.4%). Initial velocities were recorded for at least 600 s (100 times for 100 ms at intervals of 15 s). Thereafter, 1.5 µl proteinase K (at 20 mg ml−1; Sigma–Aldrich, Taufkirchen, Germany) or 1 µl astacin (at 200 µM) were added to reach complete substrate turnover, which was monitored and subsequently calculated using the formula v = [S] × m/ΔF, where [S] is the substrate concentration, m is the (F/t) slope of initial linear substrate turnover and ΔF is the maximal fluorescence intensity corresponding to complete turnover. Kinetic parameters of inhibition (Ki) were determined using Morrison's equation (Morrison, 1969[Morrison, J. F. (1969). Biochim. Biophys. Acta, 185, 269-286.]).

2.5. Bioinformatics

A computational homology model of human fetuin-A was calculated using the automodel routine of the Modeller v.9.20 suite (Fiser & Šali, 2003[Fiser, A. & Šali, A. (2003). Methods Enzymol. 374, 461-491.]). The coordinates of mouse fetuin-B and a curated structure-based sequence alignment of the two proteins were used for these calculations, and the disulfides were fixed according to Kellermann et al. (1989[Kellermann, J., Haupt, H., Auerswald, E.-A. & Müller-Ester, W. (1989). J. Biol. Chem. 264, 14121-14128.]). The resulting model was subjected to visual inspection with Coot to remove cis-peptide bonds and gross Ramachandran outliers, and was regularized with the phenix.geometry_minimization protocol of PHENIX with restraints to the starting model. The final model was assessed for the correct assignment of secondary-structure elements by comparison with an independent comparative model that was automatically generated using the multiple-template threading program RaptorX (http://raptorx.uchicago.edu/StructurePrediction/predict; Källberg et al., 2012[Källberg, M., Wang, H., Wang, S., Peng, J., Wang, Z., Lu, H. & Xu, J. (2012). Nat. Protoc. 7, 1511-1522.]). The quality of the final model is reflected by the following statistics according to PHENIX: r.m.s.d.s from target values for bonds, angles, chirality and planarity of 0.002 Å, 0.57°, 0.04° and 0.01°, respectively, an all-atom clashscore of 9.7, 1.4% rotamer outliers and no cis-peptide bonds, and 12% and 88% of residues in allowed and favored Ramachandran regions, with 0% outliers. The coordinates are available from the last corresponding author upon request.

Sequences were aligned with MUSCLE (Edgar, 2004[Edgar, R. C. (2004). Nucleic Acids Res. 32, 1792-1797.]) as implemented in SeaView (Gouy et al., 2010[Gouy, M., Guindon, S. & Gascuel, O. (2010). Mol. Biol. Evol. 27, 221-224.]). The alignment was refined based on the structures of mouse fetuin-B overlaid with stefin-B (PDB entry 1stf; Stubbs et al., 1990[Stubbs, M. T., Laber, B., Bode, W., Huber, R., Jerala, R., Lenarcic, B. & Turk, V. (1990). EMBO J. 9, 1939-1947.]), cystatin-C (PDB entry 3gax; Kolodziejczyk et al., 2010[Kolodziejczyk, R., Michalska, K., Hernandez-Santoyo, A., Wahlbom, M., Grubb, A. & Jaskolski, M. (2010). FEBS J. 277, 1726-1737.]) and the aforementioned model of human fetuin-A using UCSF Chimera (Pettersen et al., 2004[Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C. & Ferrin, T. E. (2004). J. Comput. Chem. 25, 1605-1612.]). Structural figures were prepared with UCSF Chimera. Secondary-structure predictions were made with JPred4 (Drozdetskiy et al., 2015[Drozdetskiy, A., Cole, C., Procter, J. & Barton, G. J. (2015). Nucleic Acids Res. 43, W389-W394.]). Structural superimpositions were performed with SSM (Krissinel & Henrick, 2004[Krissinel, E. & Henrick, K. (2004). Acta Cryst. D60, 2256-2268.]) within Coot. Protein interfaces were analyzed with PISA (Krissinel & Henrick, 2007[Krissinel, E. & Henrick, K. (2007). J. Mol. Biol. 372, 774-797.]; http://www.ebi.ac.uk/pdbe/pisa). The area of the complex interface was taken as half of the sum of the buried surface areas of the molecules. Sequence-similarity searches were performed with the PSI-BLAST protocol at NCBI (https://blast.ncbi.nlm.nih.gov/Blast.cgi) or the BLAST protocol at UniProt (https://www.uniprot.org/blast) using default parameters. Sequence identities were calculated using SIM with default parameters (https://web.expasy.org/sim/). The final coordinates of the crayfish astacin–mouse fetuin-B complex and free fetuin-B are available from the PDB at http://www.rcsb.org (PDB entries 6ht9 and 6hpv, respectively).

3. Results and discussion

3.1. Crystallization of mouse fetuin-B

Several groups have unsuccessfully tried to crystallize fetuins since their discovery (Pedersen, 1947[Pedersen, K. O. (1947). J. Phys. Colloid Chem. 51, 164-171.]) and `a number of obvious and potentially serious problems that may forestall attempts to crystallize fetuin' were identified (Dziegielewska & Brown, 1995[Dziegielewska, K. M. & Brown, W. M. (1995). Fetuin. Heidelberg: Springer-Verlag.]). Eventually, we obtained crystals of recombinant mouse fetuin-B from insect cells by incubation with an excess of the readily crystallizable MP astacin from crayfish (33% sequence identity with ovastacin; Gomis-Rüth, Trillo-Muyo et al., 2012[Gomis-Rüth, F. X., Trillo-Muyo, S. & Stöcker, W. (2012). Biol. Chem. 393, 1027-1041.]) as we failed to obtain sufficiently large amounts of ovastacin for structural studies. Astacin cleaved the inhibitor at S296–S297 within the CTR without affecting its inhibitory properties (data not shown). The resulting astacin–fetuin-B complex structure was solved by maximum-likelihood-scored molecular replacement, followed by density modification and noncrystallographic symmetry averaging, and was refined to 3.1 Å resolution (Table 1[link]). Once the complex structure had been solved, we managed to obtain crystals of intact mouse fetuin-B from HEK293S cells, and the structure was independently solved by native sulfur phasing and refined to 2.3 Å resolution (Table 1[link]). Examination of the crystal packing of the protein expressed in mammalian cells explains why it crystallized in its unbound form, whereas the material produced in insect cells did not: endoglycosidase H processing of the high-mannose N-glycans of the HEK293S-derived protein (a step performed during purification) left a single N-acetyl­glucosamine moiety attached to N40 that stacks against W200 of a symmetry-related molecule (Supplementary Fig. S1). This introduces a crucial crystal-packing contact that would most likely be hindered by the presence of additional sugar residues attached to the core N-acetylglucosamine, such as those normally found in proteins expressed in insect cells.

3.2. Structure of fetuin-B

The structure of fetuin-B (residues R29–P388) in its astacin complex includes the domains and segments CY1 (R29–S146), linker (LNK; K147–P158) and CY2 (S159–E267, except for the loops T218–Q227 and L246–L251), and S268–V272 plus P303–P388 from the CTR (S268–P388). The N-terminal 118-residue CY1 domain exhibits the cystatin fold and consists of a twisted and curled five-stranded antiparallel β-sheet (strands β1–β5) with simple up-and-down connectivity. An α-helix (α1) is inserted between β1 and β2 through short linkers and nestles perpendicularly to the strands in the concave face of the sheet [Figs. 1[link](a) and 1[link](b)]. Strand pairs β2/β3 and β4/β5 are connected by short loops and are arranged as hairpins I and II, respectively. In contrast, β3 and β4 are connected by a long 21-residue loop (Lβ3β4), which was found in several conformations in ovocystatin (Engh et al., 1993[Engh, R. A., Dieckmann, T., Bode, W., Auerswald, E. A., Turk, V., Huber, R. & Oschkinat, H. (1993). J. Mol. Biol. 234, 1060-1069.]) and is involved in the inhibition of legumain (Dall et al., 2015[Dall, E., Fegg, J. C., Briza, P. & Brandstetter, H. (2015). Angew. Chem. Int. Ed. 54, 2917-2921.]). This `legumain-binding loop' is internally cross-linked by the disulfide C96–C107 (`type-A disulfide' according to Kellermann et al., 1989[Kellermann, J., Haupt, H., Auerswald, E.-A. & Müller-Ester, W. (1989). J. Biol. Chem. 264, 14121-14128.]). A second disulfide staples β4 and β5 together, roughly at half the strand length (C120–C140; type-B disulfide). A fifth cysteine (C39) in Lβ1α1 is bound to the CTR through a disulfide (see below), and two N-glycans are attached to N40 and N139, respectively. The C-terminal strand of CY1 (β5) contains a bulge and leads to the 12-residue LNK, which consists of a short α-helix (α2) and the segment C154PDCP158 arranged in a tight 1,4-turn of type I, hereafter referred to as the `CPDCP-trunk' [Fig. 1[link](c)]. Owing to the prolines and the cysteines, which are disulfide-linked (C154–C157), this structure forms a rigid 14-atom ring that protrudes from the molecular surface [Fig. 1[link](b)]. In the absence of structural information, this disulfide had been previously termed type-C and included within the CY2 domain in fetuins and within the CY2 and CY3 domains in kininogen (Kellermann et al., 1989[Kellermann, J., Haupt, H., Auerswald, E.-A. & Müller-Ester, W. (1989). J. Biol. Chem. 264, 14121-14128.]; Lee, 2009[Lee, C. (2009). Front. Biosci. 14, 2911-2922.]).

[Figure 1]
Figure 1
Structure of fetuin-B. (a) Topology scheme of mouse fetuin-B depicting strands as arrows (labeled β1–β13) and helices as rods (α1–α3), each with their delimiting residues. Domain CY1 is in orange, the linker (LNK) is in red, CY2 is in turquoise and the CTR is in purple. N-glycans are marked with blue lollipops and labeled. Disulfides are pinpointed as pale green lines and labeled: [\bigcirc \kern -.7em \scriptstyle 1], C39–C374; [\bigcirc \kern -.7em \scriptstyle 2], C96–C107; [\bigcirc \kern -.7em \scriptstyle 3], C120–C140; [\bigcirc \kern -.7em \scriptstyle 4], C154–C157; [\bigcirc \kern -.7em \scriptstyle 5], C217–C224; [\bigcirc \kern -.7em \scriptstyle 6], C237–C263. Missing segments are shown as dotted lines, and hairpins I and II of each cystatin domain are further labeled. (b) Ribbon-type plot of mouse fetuin-B, with each domain colored as in (a). N-glycans are labeled with an encircled `A' (N40) and `B' (N139); P155 and N156 from the CPDCP-trunk are labeled and their side chains are shown. The N- and the C-termini of the protein are labeled and residues from the purification tag are shown in pink. (c) The CPDCP-trunk featuring a rigid 14-atom ring (C atoms in gray). (d) Superposition of CY1 (orange), CY2 (cyan) and ovocystatin (purple; PDB entry 1cew; Bode et al., 1988[Bode, W., Engh, R., Musil, D., Thiele, U., Huber, R., Karshikov, A., Brzin, J., Kos, J. & Turk, V. (1988). EMBO J. 7, 2593-2599.]) as Cα plots. The respective type-A and type B-disulfides are shown in red and labeled, as are the legumain-binding loops in the three structures, hairpins I and II, and the N- and C-termini. (e) Superposition of the Cα traces of mouse fetuin-B: unbound (CY1 in yellow, LNK in green, CY2 in dark blue and CTR in pink) and in complex with astacin [domain colors as in (b)]. The inset shows a close-up showing the CPDCP-trunk, hairpin I and hairpin II in cross-eyed stereo. The observed differences may be authentic or may be the result of crystal contacts.

The LNK is followed by the 109-residue CY2 domain, which likewise adopts the cystatin fold and is topologically equivalent to the CY1 domain. It interacts through the convex face of its β-sheet with Lα1β2 and the legumain-binding loop from the CY1 domain [Fig. 1[link](b)]. Interestingly, the latter loop encompasses the segment L92ETDCHVLSRKA103, which has been identified as a characteristic signature of fetuins (Olivier et al., 2000[Olivier, E., Soury, E., Ruminy, P., Husson, A., Parmentier, F., Daveau, M. & Salier, J.-P. (2000). Biochem. J. 350, 589-597.]) according to the PROSITE database (motif PDOC00966). This supports the hypothesis that the legumain-binding loop of the CY1 domain may play a similar adhesive role at the CY1/CY2 interface in other fetuins. CY1 and CY2 can be superimposed for 78 residues with an r.m.s.d. of 1.6 Å despite just 17% sequence identity, and the respective sheet strands (β6–β10 in CY2) and α-helices (α3 in CY2) coincide [Fig. 1[link](d)].

Akin to CY1 strands β4 and β5, a type-B disulfide (C237–C263) staples β9 and β10 together in the CY2 domain. The largest difference between the domains is found in the legumain-binding loop (Lβ8β9 in CY2), which spans 18 residues and is partially disordered in the CY2 domain, so that the corresponding type-A disulfide (C217–C224) is not resolved in the final Fourier map. Moreover, CY2 is not glycosylated, Lβ7β8 (hairpin I) is two residues shorter than in CY1 and Lβ9β10 (hairpin II) is partially disordered at its tip, as is Lβ8β9. Overall, the CY1 and CY2 domains are topologically equivalent to ovocystatin, except for the legumain-binding loop [Fig. 1[link](d); Bode et al., 1988[Bode, W., Engh, R., Musil, D., Thiele, U., Huber, R., Karshikov, A., Brzin, J., Kos, J. & Turk, V. (1988). EMBO J. 7, 2593-2599.]]: 83 and 97 aligned ovo­cystatin residues can be superposed with r.m.s.d. values of 2.1 and 2.0 Å onto CY2 and CY1 residues, respectively, despite low sequence-identity values (20% and 16%, respectively).

After the C-terminal strand of CY2 (β10), the polypeptide chain of fetuin-B enters the CTR, which is generally irregularly folded, disordered between P273 and A302, and cleaved at S296–S297. After P303, the chain runs parallel to strand β5 of CY1 as strand β11, thus providing a first anchor to CY1 for the downstream part of the domain. The remaining 78 residues of the CTR could be traced but are flexible: they lack regular secondary structure and run irregularly along the convex surface of the CY1 β-sheet. Only two short β-strands (β12 and β13), which coincide with secondary-structure predictions for the CTR, interact with each other in an antiparallel manner. Towards the C-terminus of the domain, C374 is disulfide-linked to C39, in this way providing a second anchor between the CTR and CY1. It is noteworthy that the tripeptide C374PG376 is conserved within fetuins, kininogens and histidine-rich glycoproteins (Olivier et al., 2000[Olivier, E., Soury, E., Ruminy, P., Husson, A., Parmentier, F., Daveau, M. & Salier, J.-P. (2000). Biochem. J. 350, 589-597.]). This suggests a similar link between CY1 and CTR in other fetuins, as experimentally proven for human fetuin-A and human kininogen (Kellermann et al., 1989[Kellermann, J., Haupt, H., Auerswald, E.-A. & Müller-Ester, W. (1989). J. Biol. Chem. 264, 14121-14128.]).

The structure of unbound intact mouse fetuin-B is defined for residues L31–P388 (except for the segments T218–H228, S244–V255, A270–T295 and E316–S325), and its superposition on the structure from the astacin complex reveals no significant conformational rearrangement [Fig. 1[link](e)]. The respective cores comprising CY1+LNK+CY2 are undefined for both the tip of hairpin II and Lβ8β9 of CY2, and they superpose with an r.m.s.d. of 1.7 Å. CY1 and CY2 are just rotated by ∼10° with respect to each other when going from the unbound to the complexed structure. The only functionally relevant differences in the cores are found at the tips of hairpins I and II of CY2 [inset in Fig. 1[link](e)], which however are very likely to result from crystal contacts in the unbound structure. Hairpin I is in a noncompetent conformation for binding, and becomes folded out towards the CPDCP-trunk upon complex formation. In addition, despite being disordered in both structures, hairpin II shows some extra ordered residues in the complex structure, possibly owing to fixing interactions with the target MP (see below). Although the CTR of unbound fetuin-B is intact and that of the complexed form is cleaved (see above), they show similar overall chain traces (except for S348–K357, D314–E328 and the C-terminal tail after N381), as revealed by an r.m.s.d. of 1.3 Å upon superposition.

3.3. Structure-based mechanism of the inhibition of astacin by fetuin-B

Astacin is a 202-residue MP (A1–H202). It consists of an upper subdomain (USD; A1–G99) and a lower subdomain (LSD; F100–H202) separated by an extended, deep, narrow active-site cleft (Bode et al., 1992[Bode, W., Gomis-Rüth, F. X., Huber, R., Zwilling, R. & Stöcker, W. (1992). Nature (London), 358, 164-167.]; Gomis-Rüth et al., 1993[Gomis-Rüth, F. X., Stöcker, W., Huber, R., Zwilling, R. & Bode, W. (1993). J. Mol. Biol. 229, 945-968.]; Guevara et al., 2010[Guevara, T., Yiallouros, I., Kappelhoff, R., Bissdorf, S., Stöcker, W. & Gomis-Rüth, F. X. (2010). J. Biol. Chem. 285, 13958-13965.]). The USD contains a twisted five-stranded β-sheet, the fourth strand of which shapes the upper rim of the cleft and runs antiparallel to the other strands and to substrates bound in the cleft. The USD further contains an active-site helix. This helix includes two of the catalytic zinc ligands and the general base/acid for catalysis (H92, H96 and E93, respectively) embedded in a long zinc-binding consensus sequence (H92EXXHXXGXXH102), which is a hallmark of the astacins (Gomis-Rüth, Trillo-Muyo et al., 2012[Gomis-Rüth, F. X., Trillo-Muyo, S. & Stöcker, W. (2012). Biol. Chem. 393, 1027-1041.]) and other MPs of the metzincin clan (Bode et al., 1993[Bode, W., Gomis-Rüth, F. X. & Stöcker, W. (1993). FEBS Lett. 331, 134-140.]; Gomis-Rüth, 2009[Gomis-Rüth, F. X. (2009). J. Biol. Chem. 284, 15353-15357.]; Cerdà-Costa & Gomis-Rüth, 2014[Cerdà-Costa, N. & Xavier Gomis-Rüth, F. (2014). Protein Sci. 23, 123-144.]). After G99, the chain enters the LSD, which contains little repetitive secondary structure, most notably a kinked C-terminal helix at the end of the chain. Two further zinc ligands come from the LSD (H102 and Y149), as well as another structural hallmark of astacins and other metzincins, the Met-turn (Tallant et al., 2010[Tallant, C., García-Castellanos, R., Baumann, U. & Gomis-Rüth, F. X. (2010). J. Biol. Chem. 285, 13951-13957.]), which contains a conserved methionine (M147). The C-terminus of the protein resides on the molecular surface and two disulfides (C42–C198 and C64–C84) contribute to the overall stability and rigidity of the molecule.

Mouse fetuin-B potently inhibits its physiological target ovastacin, human meprin β and archetypal crayfish astacin in vitro (see Section 3.4[link] and Karmilin et al., 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]). To account for this inhibition, the LNK and CY2 form a bipartite wedge, which slots into the active-site cleft of the target [Figs. 2[link](a), 2[link](b) and 2[link](c)]. In contrast, CY1 makes only a minor contribution to binding and the CTR does not participate at all. Upon inhibitor binding, the LSD of astacin undergoes a slight clamshell motion of maximally ∼2.5 Å (at the Cα atom of Q118), which closes the active-site cleft as in the complex with an inhibitor mimicking a reaction intermediate (PDB entry 1qji; Grams et al., 1996[Grams, F., Dive, V., Yiotakis, A., Yiallouros, I., Vassiliou, S., Zwilling, R., Bode, W. & Stöcker, W. (1996). Nat. Struct. Mol. Biol. 3, 671-675.]). The astacin–fetuin-B complex interface buries an area of 1059 Å2 and yields a theoretical gain of solvation energy on complex formation of −10.8 kcal mol−1. The interface involves 33 residues of fetuin-B and 38 of astacin, which collectively establish ten hydrogen bonds, two ionic interactions and 20 hydrophobic interactions (see Table 2[link]). Astacin segments participating in the interaction include S62–V68 from the upper-rim strand and a preceding bulge above the primed side of the cleft, Q76 from the fifth strand, G83–C84 from the loop connecting the fifth sheet strand to the active-site helix, plus the disulfide C64–C84, the zinc-binding site and Y101, D121–D129, F154–W158 and Y177 from the LSD [Fig. 2[link](c)]. Fetuin-B segments involved include M111–Y113 from the legumain-binding loop of CY1, K147–S159 from the LNK and CY2 strand β6, N198–Y206 from CY2 hairpin I and, to a lesser extent, S245 plus Q253–V255 from CY2 hairpin II, the tip of which is disordered [see above and Figs. 2[link](a) and 2[link](c)].

Table 2
Interactions at the astacin–fetuin-B interface

The first residue/atom belongs to fetuin-B and the second to astacin. The overall resolution of the structure is 3.1 Å, the estimated overall average coordinate error is 0.42 Å based on the Luzzati plot, and the dispersion precision indicator based on the free R factor is 0.49 Å. Thus, the distances are approximate.

Hydrogen bonds
Y113 Oη–S62 Oγ, 3.0 Å
H151 Nδ1–N127 Oδ1, 3.1 Å
T152 O–Y101 Oη, 2.9 Å
T153 O–V68 N, 2.7 Å
D156 Oδ1–Y149 Oη, 2.7 Å
D156 Oδ2–E93 O2, 2.9 Å
Q199 N2–S62 O, 3.1 Å
S202 O–Y177 Oη, 3.9 Å
S202 O–W158 N1, 3.6 Å
Y206 Oη–D121 Oδ1, 3.3 Å
Ionic interactions
D156 Oδ1–Zn999, 2.3 Å
D156 Oδ2–Zn999, 2.3 Å
Hydrophobic interactions (<4 Å)
F112–W65
F112–Y67
C154–W65
P155–V68
P155–H96
P155–H102
D156–W65
C157–W65
P158–W65
N198–M124
Q199–G63
W200–M124
W200–F154
V201–G63
V201–C64
V201–G83
V201–C84
C154–W65
S202–Y177
S245–W158
[Figure 2]
Figure 2
The astacin–fetuin-B complex. (a) Cross-eyed stereo cartoon displaying fetuin-B [colors as in Fig. 1[link](b)] inserted into the active-site cleft of astacin (tan ribbon). The MP is shown after a horizontal ∼90° rotation downwards from the traditional standard orientation of MPs (Gomis-Rüth, Botelho et al., 2012[Gomis-Rüth, F. X., Botelho, T. O. & Bode, W. (2012). Biochim. Biophys. Acta, 1824, 157-163.]), so the cleft runs from left (nonprimed subsites) to right (primed subsites). The catalytic zinc is shown as a magenta sphere and its astacin ligands as stick models, as are P155 and D156 from fetuin-B. (b) Detail of the final (2mFobs − DFcalc)-type Fourier map at 3.1 Å resolution contoured at 1σ above the threshold, depicting the zinc-binding site of astacin (carbons in tan, zinc ion as a magenta sphere) and the CPDCP-trunk of fetuin B (carbons in pink). (c) Close-up view of (a) after a horizontal 35° rotation, displaying selected residues participating in the interaction between fetuin-B (Cα ribbon in plum; from left to right, segments S146–D162, M111–E114, N197–Y206 and T241–S245; carbons in pink and residue numbers according to UniProt Q9QXC1 in blue) and astacin (Cα ribbon in tan, carbons in orange, catalytic zinc as a magenta sphere and residue numbers in black italics according to UniProt P07584). The C154–C157 disulfide from the CPDCP-trunk of fetuin-B is shown with green sulfurs. (d) Scheme of the proposed `raised elephant-trunk model'; the view mirrors that in (a) and (b). The CPDCP-trunk ([\bigcirc \kern -.7em \scriptstyle 1]) would be the raised trunk (in red) and hairpin I ([\bigcirc \kern -.7em \scriptstyle 2]) and II ([\bigcirc \kern -.7em \scriptstyle 3]) the two front and back limbs (in turquoise), respectively (see top-left inset). The trunk contacts the catalytic zinc ion (magenta sphere) of astacin (in brown) by residues nestling into cleft subsites S4, S3, S2, S2 and S3, as shown by purple arrows. Dashed lines represent distantly connected segments or disordered loops. Hydrogen bonds between main-chain atoms, between side-chain and main-chain atoms, and between side-chain atoms are shown as gray, red and blue dashed lines, respectively. The electrostatic interaction between the catalytic zinc and D156 is in light blue. The disulfide connecting the CPDCP-trunk is shown as a green solid line. (e) Superposition of the experimental structure of mouse fetuin-B (white carbons and black labels) and the comparative model of human fetuin-A (cyan carbons and blue labels with regular residue numbers in italics) depicting the regions of the CPDCP warhead and the tips of hairpins I and II of each structure. The view results from (c) after a 90° rotation in the plane. The catalytic zinc ion of astacin is further shown as a magenta sphere for reference.

The main interaction is performed by the rigid CPDCP-trunk of the LNK, which blocks the nonprimed side of the cleft and binds the catalytic zinc through the side chain of D156 in a bidentate manner, thus occupying the position of a scissile carbonyl during catalysis. The D156 carboxylate is also at binding distance from the zinc-binding atom Y149 Oη and, notably, the carboxylate of the general base/acid E93, so that either carboxylate could be protonated. P155 nestles into cleft subsite S2 [active-site cleft subsites are shown in bold; for nomenclature, see Schechter & Berger (1967[Schechter, I. & Berger, A. (1967). Biochem. Biophys. Res. Commun. 27, 157-162.]) and Gomis-Rüth, Botelho et al. (2012[Gomis-Rüth, F. X., Botelho, T. O. & Bode, W. (2012). Biochim. Biophys. Acta, 1824, 157-163.])], C154 into S3 and T153 into S4. The cysteine sulfurs of the trunk make hydrophobic interactions with the ring of W65. In addition, residues M111–Y113 from the legumain-binding loop of CY1 play a scaffolding role for the CPDCP-trunk and further establish mixed hydrophobic and hydrophilic interactions with S62, W65, Y67 and Q76 of astacin [see Table 2[link] and Figs. 2[link](c) and 2[link](d)].

Next in importance to complex formation is CY2 hairpin I, which blocks the primed side of the cleft. The Q199 side chain both positions the CPDCP-trunk through a hydrogen bond to D156 O and binds the main chain of the astacin bulge at S62 O. The downstream residue W200 makes a hydrophobic inter­action through one face of its side chain with F154 and M124; through the other it is kept in place by the side chain of P204 upstream in the hairpin. The V201 side chain occupies S2 and interacts hydrophobically with the sulfurs of the disulfide C64–C84, while S202 enters S3 [Figs. 2[link](c) and 2[link](d)]. Owing to its missing side chain, the downstream residue G203 is essential to prevent a clash with W158 of astacin. Moreover, Y206 establishes a hydrogen bond to D121. Overall, these interactions are key for complex formation and comprise a motif for the tip of CY2 hairpin I (Q199WVSGP204), which is conserved as QWVXGP among mammalian orthologs (see below, Table 3[link] and Supplementary Figs. S2 and S3). Finally, weak interactions are provided by the defined residues of CY2 hairpin II. In particular, S245 and Q253–V255 are close to Y177 and Q157–V160 of astacin.

Table 3
Potential fetuin-B orthologs in vertebrates

All sequences contain the intact CPDCP-trunk motif. UP, UniProt; GB, GenBank. Sequence identity with human fetuin-B was determined only for mammalian sequences, as they share the same overall protein length and architecture. Other vertebrate sequences share domains CY1 and CY2 but have highly variable CTRs.

Organism Database code Sequence identity (%) Hairpin I sequence
Mammals
 Human (Homo sapiens) UP Q9UGM5 100 QWVVGP
 Western lowland gorilla (Gorilla gorilla) UP G3QMB1 98 QWVVGP
 Chimpanzee (Pan troglodytes) UP A0A2J8M5X2 97 QWVVGP
 Sumatran orangutan (Pongo abelii) UP A0A2J8WHT1 97 QWVVGP
 Fruit bat (Pteropus vampyrus) GB XP_011380247 75 QWVFGP
 African elephant (Loxodonta africana) UP G3SVW6 70 QWVVGP
 Dog (Canis lupus familiaris) UP E2R9B6 69 QWVYGP
 Sperm whale (Physeter catodon) GB XP_007116341 69 QWVFGP
 Naked mole rat (Heterocephalus glaber) UP G5BT88 67 QWVVGP
 European hedgehog (Erinaceus europaeus) GB XP_007519827 66 QWVFGP
 Bovine (Bos taurus) UP Q58D62 66 QWVFGP
 Horse (Equus caballus) UP F6RRV1 63 QWVVGP
 Mouse (Mus musculus) UP Q9QXC1 63 QWVSGP
 Rat (Rattus norvegicus) UP Q9QX79 61 QWVVGP
 Pig (Sus scrofa) UP F1SFI6 60 QWVFGP
 Koala (Phascolarctus cinereus) GB XP_020844617 50 QWVFGP
 Opossum (Monodelphis domestica) GB XP_001373317 47 QWVVGP
Birds
 Mallard (Anas platyrhynchos) UP U3IFE7   QMVIGP
 Zebra finch (Taeniopygia guttata) UP H0ZJK5   QWVIGP
 Rock dove (Columba livia) UP F1NHT5   QWVIGP
 Collared flycatcher (Ficedula albicollis) UP U3K0K6   QWVVGP
Reptiles
 American chameleon (Anolis carolinensis) UP H9GD35   QWVVGP
 Bearded dragon (Pogona vitticeps) GB XP_020666227   QWVVGP
 Chinese softshell turtle (Pelodiscus sinensis) UP K7GGY5   QWVIGP
 Chinese alligator (Alligator sinensis) UP A0A1U7RBY2   QWVVGP
 Burmese python (Python bivittatus) GB XP_025031668   QWVVGP
Amphibians
 African clawed frog (Xenopus laevis) UP Q7SYH2   QWVVGP
 Western clawed frog (Xenopus tropicalis) GB XP_002937874   QWVVGP
 High Himalaya frog (Nanorana parkerii) GB XP_018422589   QWMFGQ
 Axolotl (Ambystoma mexicanum)   QWVFGP
Bony fishes
 Channel catfish (Silurus punctatus) UP W5UDH5   QWIVGP
 Rainbow trout (Oncorhynchus mykiss) GB XP_021460414   QWVVGP
 Atlantic salmon (Salmo salar) UP A0A1S3KKB8   QWVVGP
 Turbot (Scophthalmus maximus) UP A0A221J5J5   QWVVGP
 Three-spined stickleback (Gasterosteus aculeatus) UP G3NQ86   QWVVGP
 Nile tilapia (Oreochromis niloticus) UP I3JY83   QWVVGP
 Japanese killifish (Oryzias latipes) UP H2MAI0   QWVVWP
 Zebrafish (Danio rerio) UP A5D6T5   QWMVGA
 Coelacanth (Latimeria chalumnae) UP M3XK68   QWMVGA
Cartilaginous fishes
 Elephant shark (Callorhincus milii) GB XP_007897454   QGMFFE
†With human fetuin-B.
‡http://www.ambystoma.org/index.php?option=com_content&view=article&id=67:ambystoma-get-contig&catid=40:codepages&Itemid=2&species=2&mcid=contig331328.

Overall, the CPDCP-trunk and CY2 hairpin I both bind in the direction of a substrate and cover the nonprimed and primed sides of the active-site cleft from S4 to S3. This matches the preference of the enzyme for elongated substrates (Grams et al., 1996[Grams, F., Dive, V., Yiotakis, A., Yiallouros, I., Vassiliou, S., Zwilling, R., Bode, W. & Stöcker, W. (1996). Nat. Struct. Mol. Biol. 3, 671-675.]) and explains the high inhibitory potency of fetuin-B. However, subsites S1 and S1 are spared by the inhibitor, which is noteworthy as S1 is the main specificity pocket of astacins and most MPs (Gomis-Rüth, Botelho et al., 2012[Gomis-Rüth, F. X., Botelho, T. O. & Bode, W. (2012). Biochim. Biophys. Acta, 1824, 157-163.]; Gomis-Rüth, Trillo-Muyo et al., 2012[Gomis-Rüth, F. X., Trillo-Muyo, S. & Stöcker, W. (2012). Biol. Chem. 393, 1027-1041.]). This enables fetuin-B to potently inhibit astacin, meprin β and ovastacin (see Section 3.4[link]), despite their differences in substrate specificity (Gomis-Rüth, Trillo-Muyo et al., 2012[Gomis-Rüth, F. X., Trillo-Muyo, S. & Stöcker, W. (2012). Biol. Chem. 393, 1027-1041.]), and strongly supports astacin–fetuin-B as a valid model for other complexes with members of the astacin family of MPs. The free S1 pocket may also explain why the zinc ligand Y149, which undergoes a `tyrosine-switch' motion upon substrate or inhibitor binding (Grams et al., 1996[Grams, F., Dive, V., Yiotakis, A., Yiallouros, I., Vassiliou, S., Zwilling, R., Bode, W. & Stöcker, W. (1996). Nat. Struct. Mol. Biol. 3, 671-675.]; Gomis-Rüth, Trillo-Muyo et al., 2012[Gomis-Rüth, F. X., Trillo-Muyo, S. & Stöcker, W. (2012). Biol. Chem. 393, 1027-1041.]), is found in the conformation present in unbound astacin structures and the proenzyme, which likewise evince empty S1 pockets (Gomis-Rüth et al., 1993[Gomis-Rüth, F. X., Stöcker, W., Huber, R., Zwilling, R. & Bode, W. (1993). J. Mol. Biol. 229, 945-968.]; Guevara et al., 2010[Guevara, T., Yiallouros, I., Kappelhoff, R., Bissdorf, S., Stöcker, W. & Gomis-Rüth, F. X. (2010). J. Biol. Chem. 285, 13958-13965.]). Thus, the bipartite blockage of the cleft prevents cleavage of the inhibitor, as occurs for example in standard-mechanism inhibitors of MPs, which typically run entirely along the cleft (Arolas et al., 2011[Arolas, J. L., Botelho, T. O., Vilcinskas, A. & Gomis-Rüth, F. X. (2011). Angew. Chem. Int. Ed. 50, 10357-10360.]). In addition, this mode of inhibition also explains why fetuin-B does not inhibit BMP-1 and its tolloid(-like) relatives (Karmilin et al., 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]): these proteins have astacin residue W65, which is located within the upper-rim strand above the cleft, replaced by the tripeptide CCG, with the two vicinal cysteines connected by a cis-peptide bond and a disulfide bond. This creates a highly strained bulge above the zinc site (see PDB entry 3edh; Mac Sweeney et al., 2008[Mac Sweeney, A., Gil-Parrado, S., Vinzenz, D., Bernardi, A., Hein, A., Bodendorf, U., Erbel, P., Logel, C. & Gerhartz, B. (2008). J. Mol. Biol. 384, 228-239.]), which prevents the fetuin-B CPDCP-trunk from entering the active site. Moreover, the mechanism also explains why CY1 is not inhibitory: despite its overall structural similarity (see Section 3.2[link]), CY1 lacks an upstream CPDCP-trunk and its hairpin I is two residues longer at its tip [Fig. 1[link](d)]. This impairs the functionality of the QWVXGP motif (H76YQEDMGS83 in CY1) and would cause collision with the primed side of the astacin cleft. Finally, CY1 hairpin II is seven residues shorter than in CY2, so that even the minor contribution to inhibition by this structural element is missing. To sum up, the CPDCP-trunk and CY2 hairpin I are the major determinants of inhibition, while CY1 and CTR merely play a scaffolding role.

As to structurally similar inhibitors, human latexin likewise consists of tandem cystatin-like domains connected by a helical linker (PDB entry 2bo9; Pallarès et al., 2005[Pallarès, I., Bonet, R., García-Castellanos, R., Ventura, S., Avilés, F. X., Vendrell, J. & Gomis-Rüth, F. X. (2005). Proc. Natl Acad. Sci. USA, 102, 3978-3983.]). Latexin is also very specific to a particular MP family, here the A/B-type funnelins, which are also known as M14 metallocarboxy­peptidases (Gomis-Rüth, 2008[Gomis-Rüth, F. X. (2008). Crit. Rev. Biochem. Mol. Biol. 43, 319-345.]). Like fetuin-B, it inhibits its targets through the C-terminal cystatin domain. However, in latexin the two cystatin domains are packed against each other through the helices, which are surrounded by the curved β-sheets arranged in a compact open barrel. In addition, latexin inhibition is not exerted by segments that are topologically reminiscent of the CPDCP-trunk plus hairpins I and II, but rather through loops from the opposite edge of the sheet, equivalent to Lβ6α3, Lα3β7 and Lβ8β9 of fetuin-B CY2. In summary, despite the similarity in the building blocks, the overall structure and the working mechanism of fetuin-B are novel for MPs.

3.4. Experimental validation of the structure-derived mechanism

To verify the abovementioned mechanism, we tested the inhibition of several constructs and mutants of human, mouse and bovine fetuin-B against crayfish astacin, human meprin β and mouse ovastacin [Fig. 3[link](a)], which add to previous studies (Karmilin et al., 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]). We also determined the constants of inhibition (Ki) of selected constructs and mutants of mouse fetuin-B to provide quantitative data [Figs. 3[link](b) and 3[link](c)].

[Figure 3]
Figure 3
In vitro validation of the inhibitory mechanism. (a) Activity (%) of astacin (at 1 nM concentration), meprin β (at 1 nM concentration) and ovastacin (at 5.6 nM) in the presence of fetuin-B variants at 300 nM (CPDC and CPRC) or 12.5 nM (other variants). Full astacin, meprin β and ovastacin activities (100%) correspond to substrate-turnover rates of 35.4 ± 5.7, 7.7 ± 0.2 and 10.7 ± 0.8 nM s−1, respectively; w/o, without fetuin; hFB, human fetuin-B; bFB, bovine fetuin-B; mFB, murine fetuin-B; mFBCY2, CY2 domain of mFB; mFBCTR, CTR domain of mFB; mFBC39S,C374S, double point mutant of murine fetuin-B; CPDC and CPRC, cyclic peptides comprising the respective sequences. Experiments were performed in triplicate; error bars indicate standard deviations. (b) Plot of fractional velocity (logarithmic scale) with different fetuin-B variants: mFB, murine fetuin-B; mFBABA, variant with CY1 and CTR from mouse fetuin-A and CY2 from mFB; mFBH1-swap, mutant replacing CY2 hairpin I with CY1 hairpin I; mFBP155A, mFBD156A, mFBP155A,D156A and mFBC154S,C157S are point mutants of murine fetuin-B. Enzyme and substrate concentrations were 1.0 nM and 195 µM for astacin, 0.5 nM and 26 µM for meprin β and 0.7 nM and 26 µM for ovastacin, respectively. Error bars indicate standard deviations. (c) Ki values determined from (b) using Morrison's equation with the respective standard errors (SE).

The three full-length wild-type orthologs completely inhibited all enzymes [Fig. 3[link](a)], in accordance with previous studies (Karmilin et al., 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]). Wild-type mouse fetuin-B inhibited astacin, meprin β and ovastacin with similar Ki values of 100, 81 and 46 pM, respectively [Fig. 3[link](c)]. In contrast, the Ki values of a chimera of mouse fetuin-B, in which CY1 and CTR were replaced by the homonymous domains of non-inhibitory mouse fetuin-A (mutant mFBABA), was decreased by more than three (astacin) and four (meprin β and ovastacin) orders of magnitude, respectively [Figs. 3[link](b) and 3[link](c)]. The mutants mFBBAA and mFBBAA could not be produced, which underpins the requirement for a complementary CY1–CTR interface for the proper folding of fetuin-B.

A cyclic peptide including a sequence CPDC, which mimics the isolated CPDCP-trunk between CY1 and CY2, did not inhibit either enzyme and was indistinguishable from a variant with an arginine replacing the aspartate [Fig. 3[link](a)]. Taken together with the lack of inhibition of isolated domain CY2 [Fig. 3[link](a)], this indicates that these modules are only functional within the context of a multi-domain fetuin. Domain CTR alone did not show inhibition either [Fig. 3[link](a)], which is consistent with recombinant carp fetuin truncated after CY2 inhibiting meprin α/β and carp nephrosin with similar potency as the functional full-length protein purified from fish blood (Hedrich et al., 2010[Hedrich, J., Lottaz, D., Meyer, K., Yiallouros, I., Jahnen-Dechent, W., Stöcker, W. & Becker-Pauly, C. (2010). Biochemistry, 49, 8599-8607.]). Together, these data point to a negligible role of the CTR in inhibition. Consistently, no reduction of activity was observed when the cysteines of disulfide C39–C374 were replaced by serines [Fig. 3[link](a)], which indicates that the minimal functional structure does not require the overarching clamp between CY1 and CTR.

Point mutations affecting the trunk within full-length mouse fetuin-B all had negative effects on inhibition. The P155A mutant caused a decrease in the Ki values of roughly one order of magnitude for astacin and meprin β and two orders of magnitude for ovastacin. For the D156A mutant, we measured an approximate decrease of two orders of magnitude for astacin and meprin β and four orders of magnitude for ovastacin. The corresponding double mutant caused an even greater loss towards ovastacin and meprin β (four orders of magnitude) and astacin (two orders of magnitude) [Figs. 3[link](b) and 3[link](c)]. Thus, similar to the mFBABA mutant, the double mutant is merely a micromolar inhibitor [Fig. 3[link](c)] that is insufficient to block ovastacin under physiological conditions. Likewise, the C154S+C157S mutant led to a substantial decrease in Ki. Finally, grafting the sequence of CY1 hairpin I onto CY2 (H1-swap mutant; QEDMGP instead of QWVSGP) also led to substantially higher Ki values [Figs. 3[link](b) and 3[link](c)]. This indicates that the conserved QXVXGP sequence motif of H1 is required for efficient inhibition of astacins.

Overall, the fact that the mutants and variants have comparable effects on astacin, meprin β and ovastacin supports the crayfish enzyme as a valid structural model for the mammalian orthologs. The results confirm that the major functional interactions between fetuin-B and astacin/ovastacin are made by the CPDCP-trunk and hairpin I of domain CY2 and that any modification leads to a decrease in inhibitory power. Moreover, an overall scaffold encompassing CY1+LNK+CY2 is indispensable to exert full inhibitory potency.

3.5. Comparison with the inhibition mechanism of cysteine peptidases by cystatins

The inhibitory mechanism of cysteine peptidases by monomeric cystatins has been dubbed the `elephant-trunk model' (Turk & Bode, 1991[Turk, V. & Bode, W. (1991). FEBS Lett. 285, 213-219.]) and is mainly exerted by the N-terminal segment of the inhibitor (the `trunk'). This segment precedes the first strand of the β-sheet and occupies the S1 and upstream nonprimed subsites of the cleft (Stubbs et al., 1990[Stubbs, M. T., Laber, B., Bode, W., Huber, R., Jerala, R., Lenarcic, B. & Turk, V. (1990). EMBO J. 9, 1939-1947.]; Turk & Bode, 1991[Turk, V. & Bode, W. (1991). FEBS Lett. 285, 213-219.]). On the primed side, hairpin I contains the conserved sequence QXVXG, which functions similarly to hairpin I of fetuin-B with motif QWVXGP (see above). Finally, hairpin II of cystatin provides a scaffold for hairpin I and performs ancillary interactions with the enzyme through the conserved PW motif. Overall, the inhibitory mechanism of MPs by fetuin-B, in which the extended N-terminal trunk is replaced by a compact CPDCP-trunk that resembles a `raising trunk' [inset in Fig. 2[link](d)], contains elements of similarity to the hydrophobic tripartite wedge of cystatins tackling cysteine peptidases. This suggests that gene duplication and molecular evolution of cysteine-peptidase targeting cystatins eventually led to inhibition of peptidases of a different chemical class. We thus hereby propose the term `raised-elephant-trunk mechanism' to describe the mode of inhibition of astacins by fetuin-B [Fig. 2[link](d)].

3.6. Implications for inhibitory fetuins

The alignment of representative sequences of (potential) mammalian fetuin-B orthologs revealed that they share the same overall length (374–388 residues) and show high sequence identity to the human form, which ranges from 98% (gorilla) to 47% (opossum) (see Table 3[link]). The orthologs contain the sequence of the CPDCP-trunk and the length and sequence of the QWVXGP motif (Table 3[link], Supplementary Figs. S2 and S3), as well as a similar CTR. This suggests that mammalian fetuin-B orthologs probably exhibit equivalent structures and inhibit astacins via the proposed raised-elephant-trunk mechanism, as has been confirmed to date for the bovine, human and murine forms [Fig. 3[link](a)].

Outside mammals, carp fetuin efficiently inhibited nephrosin (Tsai et al., 2004[Tsai, P.-L., Chen, C.-H., Huang, C.-J., Chou, C.-M. & Chang, G.-D. (2004). J. Biol. Chem. 279, 11146-11155.]) and human meprin α and β, but not astacin (Hedrich et al., 2010[Hedrich, J., Lottaz, D., Meyer, K., Yiallouros, I., Jahnen-Dechent, W., Stöcker, W. & Becker-Pauly, C. (2010). Biochemistry, 49, 8599-8607.]). This fetuin occurs in two isoforms, which have a CTR that is either shorter or longer than that in mouse fetuin-B but is in both cases dispensable for inhibition (Hedrich et al., 2010[Hedrich, J., Lottaz, D., Meyer, K., Yiallouros, I., Jahnen-Dechent, W., Stöcker, W. & Becker-Pauly, C. (2010). Biochemistry, 49, 8599-8607.]). Although somewhat more distant in evolution than the mammalian orthologs (24% sequence identity between the first 306 residues of mouse fetuin-B and the entire short variant), carp fetuin also contains the CPDCP-trunk to bind the catalytic zinc, but hairpin I has one extra residue in the sequence: QWMFSGQ. This would entail steric clashes with the primed side of the cleft of astacin, but not of meprin β (PDB entry 1gwn; Arolas et al., 2012[Arolas, J. L., Broder, C., Jefferson, T., Guevara, T., Sterchi, E. E., Bode, W., Stöcker, W., Becker-Pauly, C. & Gomis-Rüth, F. X. (2012). Proc. Natl Acad. Sci. USA, 109, 16131-16136.]), the cleft of which is deeper. This explains the distinct inhibitory behavior towards these enzymes.

Other vertebrates (birds, reptiles, amphibians, bony fishes and cartilaginous fishes) contain potential orthologs with both the CPDCP-trunk and a putative hairpin I with a QWVXGP-derived sequence (Table 3[link]). They should thus indiscriminately inhibit meprins, astacin and ovastacin. However, the CTRs strongly deviate from mammalian sequences and range from short tails to hundreds of residues. In particular, the Chinese softshell turtle sequence only spans 264 residues and is apparently a natural variant containing just CY1+LNK+CY2. Moreover, no sequences were found within cyclostomes, cephalochordates, urochordates, echinoderms or hemichordates within deuterostomes. Thus, fetuin-B sequences are absent from jawless fish and are restricted to jawed vertebrates, for which the raised-elephant-trunk mechanism might be valid.

3.7. Why fetuin-A is not an astacin inhibitor

To investigate the lack of inhibitory activity of fetuin-A (Karmilin et al., 2019[Karmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.]), we calculated a homology model for the human form based on the coordinates of mouse fetuin-B (28% sequence identity) and the experimentally determined disulfide pattern (Kellermann et al., 1989[Kellermann, J., Haupt, H., Auerswald, E.-A. & Müller-Ester, W. (1989). J. Biol. Chem. 264, 14121-14128.]). According to this model, the overall structure would be very similar to that of fetuin-B and likewise would encompass CY1 (G22–S138; human fetuin A residue numbers are shown in italics; see UniProt P02765), LNK (A139–P150), CY2 (L151–Q251) and CTR (T252–V367) domains. It would also include a potential CPDCP-trunk (C146QDCP150), which would be solvent-exposed, as well as the upstream major physiological trypsin cleavage site of human fetuin-A (K143–K144; Kellermann et al., 1989[Kellermann, J., Haupt, H., Auerswald, E.-A. & Müller-Ester, W. (1989). J. Biol. Chem. 264, 14121-14128.]). Moreover, the segments equivalent to Y77–Q78 and H151TT153 of fetuin-B are sensitive to proteolysis in some fetuin-A orthologs (Olivier et al., 2000[Olivier, E., Soury, E., Ruminy, P., Husson, A., Parmentier, F., Daveau, M. & Salier, J.-P. (2000). Biochem. J. 350, 589-597.], Kellermann et al., 1989[Kellermann, J., Haupt, H., Auerswald, E.-A. & Müller-Ester, W. (1989). J. Biol. Chem. 264, 14121-14128.]). These segments map to CY1 hairpin I and helix α3, respectively, which are likewise surface-located.

The second position of the CPDCP-trunk is occupied by a glutamine in human fetuin-A (Q147), which could potentially interfere with active-site residues of the MP and thus impair inhibition. In contrast to the fetuin-B orthologs, the trunk motif shows variability across mammalian fetuin-A orthologs: CPRCP in mouse, CPQCP in cat, CPNCP in pig etc. In the bovine and rabbit orthologs the sequence even corresponds to that of fetuin-B. Collectively, these findings support the above results in vitro that the `warhead' is a necessary but not sufficient element of inhibition. Indeed, the main reason for the lack of inhibition of human fetuin-A is that CY2 hairpin I is one residue shorter, shows a disparate sequence (V191PL­PP195 instead of QWVXGP) and probably shows a distorted geometry owing to the two extra prolines [Fig. 2[link](e)]. In addition, the CY2 hairpin II would be nine residues shorter (K237–G240 instead of S244–E256). Taken together, these differences in crucial and ancillary structural elements of the raised-elephant-trunk model would explain the lack of inhibitory competence of fetuin-A.

4. Conclusion

Discovered over seven decades ago in mammalian serum, the fetuins are a widespread family of vertebrate proteins. Fetuin-B is the only endogenous, specific, potent inhibitor of non-BMP-1-like astacin MPs through a conserved aspartate within a CPDCP-trunk that blocks the active-site cleft. This is reminiscent of the aspartate-switch mechanism that is responsible for latency in astacin zymogens, in which an aspartate likewise blocks the catalytic zinc (Guevara et al., 2010[Guevara, T., Yiallouros, I., Kappelhoff, R., Bissdorf, S., Stöcker, W. & Gomis-Rüth, F. X. (2010). J. Biol. Chem. 285, 13958-13965.]; Arolas et al., 2012[Arolas, J. L., Broder, C., Jefferson, T., Guevara, T., Sterchi, E. E., Bode, W., Stöcker, W., Becker-Pauly, C. & Gomis-Rüth, F. X. (2012). Proc. Natl Acad. Sci. USA, 109, 16131-16136.]).

CY1, LNK and CY2 are required for the inhibitory function of fetuin-B, but not the CTR, which is flexible and depleted of regular secondary structure, i.e. it can potentially adopt several conformations. By lining the exposed surface of CY1 and being covalently linked to it, the CTR provides fetuins with an adaptable surface to potentially bind many proteins and exert disparate signaling functions (Dunker et al., 2000[Dunker, A. K., Obradovic, Z., Romero, P., Garner, E. C. & Brown, C. J. (2000). Genome Inf. 11, 161-171.]), which may be independent of inhibition of proteolysis.

Finally, we found sequences of proven and potential fetuin-B orthologs encompassing the CPDCP and QWVXGP motifs throughout jawed vertebrates down the tree of life as far as cartilaginous fish. In turn, astacins are present across animals, including lower vertebrates and invertebrates such as nematodes, mostly as several paralogs per organism (Gomis-Rüth, Trillo-Muyo et al., 2012[Gomis-Rüth, F. X., Trillo-Muyo, S. & Stöcker, W. (2012). Biol. Chem. 393, 1027-1041.]). This phylogenetic coexistence suggests that the raised-elephant-trunk mechanism described here may represent a general mechanism of endogenous regulation of astacin MPs through dedicated and potent protein inhibitors.

Footnotes

These authors contributed equally and share first authorship.

Acknowledgements

We are grateful to Joan Pous and Xandra Kreplin from the joint IBMB/IRB Automated Crystallography Platform and to Roman Bonet of the Protein Purification Service for assistance during crystallization and purification experiments, respectively. We further acknowledge the help from Nele von Wiegen and Büsra Nur Togru during biochemical experiments and from local contacts from the ALBA and ESRF synchrotrons. Funding for data collection was provided in part by ALBA and ESRF. The authors declare no competing financial interests. Author contributions are as follows. F. X. Gomis-Rüth, W. Jahnen-Dechent, L. Jovine and W. Stöcker designed and coordinated the research. A. Cuppari, T. Guevara, D. de Sanctis, C. Schmitz, H. Körschgen, K. Karmilin, M. Kuske, M. Olf, E. Dietzel, D. Fahrenkamp, J. Floehr, I. Yiallouros, R. Weiskirchen, L. Jovine and F. X. Gomis-Rüth performed experiments and/or data analysis. F. X. Gomis-Rüth wrote the paper with input from all authors.

Funding information

This study was supported in part by grants from German (DFG FA1518/1-1, JA562/16 and FL1033/1; IZKF Aachen Medical Faculty Research Fund; JGU Research Fund), Spanish (BFU2015-64487R, MDM-2014-0435) and Catalan (2017SGR3, Fundació `La Marató de TV3' 201815-10) public and private agencies. Further support was provided by the Swedish Research Council (2016-03999), the Center for Innovative Medicine (Sweden) and an EMBO long-term fellowship (ALTF 143-2017). The Structural Biology Unit of IBMB is a `María de Maeztu' Unit of Excellence of the Spanish Ministry of Science, Innovation and Universities. The funders had no role in the study design, data collection and interpretation, or the decision to submit the work for publication.

References

First citationAfonine, P. V., Grosse-Kunstleve, R. W., Echols, N., Headd, J. J., Moriarty, N. W., Mustyakimov, M., Terwilliger, T. C., Urzhumtsev, A., Zwart, P. H. & Adams, P. D. (2012). Acta Cryst. D68, 352–367.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationAgirre, J., Iglesias-Fernández, J., Rovira, C., Davies, G. J., Wilson, K. S. & Cowtan, K. D. (2015). Nat. Struct. Mol. Biol. 22, 833–834.  CrossRef CAS PubMed Google Scholar
First citationAricescu, A. R., Lu, W. & Jones, E. Y. (2006). Acta Cryst. D62, 1243–1250.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationArolas, J. L., Botelho, T. O., Vilcinskas, A. & Gomis-Rüth, F. X. (2011). Angew. Chem. Int. Ed. 50, 10357–10360.  CrossRef CAS Google Scholar
First citationArolas, J. L., Broder, C., Jefferson, T., Guevara, T., Sterchi, E. E., Bode, W., Stöcker, W., Becker-Pauly, C. & Gomis-Rüth, F. X. (2012). Proc. Natl Acad. Sci. USA, 109, 16131–16136.  Web of Science CrossRef CAS PubMed Google Scholar
First citationBecker, C., Kruse, M. N., Slotty, K. A., Köhler, D., Harris, J. R., Rösmann, S., Sterchi, E. E. & Stöcker, W. (2003). Biol. Chem. 384, 825–831.  CAS Google Scholar
First citationBecker-Pauly, C., Höwel, M., Walker, T., Vlad, A., Aufenvenne, K., Oji, V., Lottaz, D., Sterchi, E. E., Debela, M., Magdolen, V., Traupe, H. & Stöcker, W. (2007). J. Invest. Dermatol. 127, 1115–1125.  CAS Google Scholar
First citationBerman, H., Henrick, K. & Nakamura, H. (2003). Nat. Struct. Biol. 10, 980.  Web of Science CrossRef PubMed Google Scholar
First citationBode, W., Engh, R., Musil, D., Thiele, U., Huber, R., Karshikov, A., Brzin, J., Kos, J. & Turk, V. (1988). EMBO J. 7, 2593–2599.  CrossRef CAS PubMed Web of Science Google Scholar
First citationBode, W., Gomis-Rüth, F. X., Huber, R., Zwilling, R. & Stöcker, W. (1992). Nature (London), 358, 164–167.  CrossRef PubMed CAS Web of Science Google Scholar
First citationBode, W., Gomis-Rüth, F. X. & Stöcker, W. (1993). FEBS Lett. 331, 134–140.  CrossRef CAS PubMed Web of Science Google Scholar
First citationBokhove, M., Sadat Al Hosseini, H., Saito, T., Dioguardi, E., Gegenschatz-Schmid, K., Nishimura, K., Raj, I., de Sanctis, D., Han, L. & Jovine, L. (2016). J. Struct. Biol. 194, 1–7.  CrossRef CAS Google Scholar
First citationBurkart, A. D., Xiong, B., Baibakov, B., Jiménez-Movilla, M. & Dean, J. (2012). J. Cell Biol. 197, 37–44.  CrossRef CAS Google Scholar
First citationCerdà-Costa, N. & Xavier Gomis-Rüth, F. (2014). Protein Sci. 23, 123–144.  Google Scholar
First citationDall, E., Fegg, J. C., Briza, P. & Brandstetter, H. (2015). Angew. Chem. Int. Ed. 54, 2917–2921.  CrossRef CAS Google Scholar
First citationDenecke, B., Gräber, S., Schäfer, C., Heiss, A., Wöltje, M. & Jahnen-Dechent, W. (2003). Biochem. J. 376, 135–145.  CrossRef CAS Google Scholar
First citationDietzel, E., Wessling, J., Floehr, J., Schäfer, C., Ensslen, S., Denecke, B., Rösing, B., Neulen, J., Veitinger, T., Spehr, M., Tropartz, T., Tolba, R., Renné, T., Egert, A., Schorle, H., Gottenbusch, Y., Hildebrand, A., Yiallouros, I., Stöcker, W., Weiskirchen, R. & Jahnen-Dechent, W. (2013). Dev. Cell, 25, 106–112.  CrossRef CAS Google Scholar
First citationDrozdetskiy, A., Cole, C., Procter, J. & Barton, G. J. (2015). Nucleic Acids Res. 43, W389–W394.  Web of Science CrossRef CAS PubMed Google Scholar
First citationDunker, A. K., Obradovic, Z., Romero, P., Garner, E. C. & Brown, C. J. (2000). Genome Inf. 11, 161–171.  CAS Google Scholar
First citationDziegielewska, K. M. & Brown, W. M. (1995). Fetuin. Heidelberg: Springer-Verlag.  Google Scholar
First citationEdgar, R. C. (2004). Nucleic Acids Res. 32, 1792–1797.  Web of Science CrossRef PubMed CAS Google Scholar
First citationEmsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. (2010). Acta Cryst. D66, 486–501.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationEngh, R. A., Dieckmann, T., Bode, W., Auerswald, E. A., Turk, V., Huber, R. & Oschkinat, H. (1993). J. Mol. Biol. 234, 1060–1069.  CrossRef CAS PubMed Web of Science Google Scholar
First citationFiser, A. & Šali, A. (2003). Methods Enzymol. 374, 461–491.  CrossRef CAS Google Scholar
First citationFloehr, J., Dietzel, E., Neulen, J., Rösing, B., Weissenborn, U. & Jahnen-Dechent, W. (2016). Hum. Reprod. 31, 630–637.  CrossRef CAS Google Scholar
First citationFloehr, J., Dietzel, E., Schmitz, C., Chappell, A. & Jahnen-Dechent, W. (2017). Mol. Hum. Reprod. 23, 34–44.  CrossRef CAS Google Scholar
First citationFridrich, S., Karmilin, K. & Stöcker, W. (2016). Curr. Protoc. Protein Sci. 83, 21.16.1–21.16.20.  CrossRef Google Scholar
First citationGalembeck, F. & Cann, J. R. (1974). Arch. Biochem. Biophys. 164, 326–331.  CrossRef CAS Google Scholar
First citationGomis-Rüth, F. X. (2008). Crit. Rev. Biochem. Mol. Biol. 43, 319–345.  Web of Science PubMed Google Scholar
First citationGomis-Rüth, F. X. (2009). J. Biol. Chem. 284, 15353–15357.  Web of Science PubMed Google Scholar
First citationGomis-Rüth, F. X. (2013). Encyclopedia of Metalloproteins., edited by V. N. Uversky, R. H. Kretsinger & E. A. Permyakov, pp. 2473–2479. Heidelberg: Springer-Verlag.  Google Scholar
First citationGomis-Rüth, F. X., Botelho, T. O. & Bode, W. (2012). Biochim. Biophys. Acta, 1824, 157–163.  Web of Science PubMed Google Scholar
First citationGomis-Rüth, F. X., Stöcker, W., Huber, R., Zwilling, R. & Bode, W. (1993). J. Mol. Biol. 229, 945–968.  PubMed Web of Science Google Scholar
First citationGomis-Rüth, F. X., Trillo-Muyo, S. & Stöcker, W. (2012). Biol. Chem. 393, 1027–1041.  Web of Science PubMed Google Scholar
First citationGoulas, T., Garcia-Ferrer, I., Marrero, A., Marino-Puertas, L., Duquerroy, S. & Gomis-Rüth, F. X. (2017). Biol. Chem. 398, 975–994.  CrossRef CAS Google Scholar
First citationGouy, M., Guindon, S. & Gascuel, O. (2010). Mol. Biol. Evol. 27, 221–224.  Web of Science CrossRef PubMed CAS Google Scholar
First citationGrams, F., Dive, V., Yiotakis, A., Yiallouros, I., Vassiliou, S., Zwilling, R., Bode, W. & Stöcker, W. (1996). Nat. Struct. Mol. Biol. 3, 671–675.  CrossRef CAS Google Scholar
First citationGuevara, T., Yiallouros, I., Kappelhoff, R., Bissdorf, S., Stöcker, W. & Gomis-Rüth, F. X. (2010). J. Biol. Chem. 285, 13958–13965.  CrossRef CAS Google Scholar
First citationHedrich, J., Lottaz, D., Meyer, K., Yiallouros, I., Jahnen-Dechent, W., Stöcker, W. & Becker-Pauly, C. (2010). Biochemistry, 49, 8599–8607.  CrossRef CAS Google Scholar
First citationJahnen-Dechent, W., Heiss, A., Schäfer, C. & Ketteler, M. (2011). Circ. Res. 108, 1494–1509.  CAS Google Scholar
First citationJuanhuix, J., Gil-Ortiz, F., Cuní, G., Colldelram, C., Nicolás, J., Lidón, J., Boter, E., Ruget, C., Ferrer, S. & Benach, J. (2014). J. Synchrotron Rad. 21, 679–689.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationKabsch, W. (2010a). Acta Cryst. D66, 125–132.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationKabsch, W. (2010b). Acta Cryst. D66, 133–144.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationKällberg, M., Wang, H., Wang, S., Peng, J., Wang, Z., Lu, H. & Xu, J. (2012). Nat. Protoc. 7, 1511–1522.  Google Scholar
First citationKarmilin, K., Schmitz, C., Kuske, M., Körschgen, H., Olf, M., Meyer, K., Hildebrand, A., Felten, M., Fridrich, S., Yiallouros, I., Becker-Pauly, C., Weiskirchen, R., Jahnen-Dechent, W., Floehr, J. & Stöcker, W. (2019). Sci. Rep. 9, 546.  CrossRef Google Scholar
First citationKellermann, J., Haupt, H., Auerswald, E.-A. & Müller-Ester, W. (1989). J. Biol. Chem. 264, 14121–14128.  CAS Google Scholar
First citationKolodziejczyk, R., Michalska, K., Hernandez-Santoyo, A., Wahlbom, M., Grubb, A. & Jaskolski, M. (2010). FEBS J. 277, 1726–1737.  CrossRef CAS Google Scholar
First citationKörschgen, H., Kuske, M., Karmilin, K., Yiallouros, I., Balbach, M., Floehr, J., Wachten, D., Jahnen-Dechent, W. & Stöcker, W. (2017). Mol. Hum. Reprod. 23, 607–616.  Google Scholar
First citationKrissinel, E. & Henrick, K. (2004). Acta Cryst. D60, 2256–2268.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationKrissinel, E. & Henrick, K. (2007). J. Mol. Biol. 372, 774–797.  Web of Science CrossRef PubMed CAS Google Scholar
First citationKrissinel, E., Uski, V., Lebedev, A., Winn, M. & Ballard, C. (2018). Acta Cryst. D74, 143–151.  Web of Science CrossRef IUCr Journals Google Scholar
First citationLee, C. (2009). Front. Biosci. 14, 2911–2922.  CrossRef CAS Google Scholar
First citationLee, H. X., Ambrosio, A. L., Reversade, B. & De Robertis, E. M. (2006). Cell, 124, 147–159.  CrossRef CAS Google Scholar
First citationLütteke, T. & von der Lieth, C. W. (2004). BMC Bioinformatics, 5, 69.  Google Scholar
First citationMac Sweeney, A., Gil-Parrado, S., Vinzenz, D., Bernardi, A., Hein, A., Bodendorf, U., Erbel, P., Logel, C. & Gerhartz, B. (2008). J. Mol. Biol. 384, 228–239.  Web of Science CrossRef PubMed CAS Google Scholar
First citationMcCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. & Read, R. J. (2007). J. Appl. Cryst. 40, 658–674.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationMorrison, J. F. (1969). Biochim. Biophys. Acta, 185, 269–286.  CrossRef CAS PubMed Web of Science Google Scholar
First citationOlivier, E., Soury, E., Ruminy, P., Husson, A., Parmentier, F., Daveau, M. & Salier, J.-P. (2000). Biochem. J. 350, 589–597.  CrossRef CAS Google Scholar
First citationPallarès, I., Bonet, R., García-Castellanos, R., Ventura, S., Avilés, F. X., Vendrell, J. & Gomis-Rüth, F. X. (2005). Proc. Natl Acad. Sci. USA, 102, 3978–3983.  Web of Science PubMed Google Scholar
First citationPannu, N. S., Waterreus, W.-J., Skubák, P., Sikharulidze, I., Abrahams, J. P. & de Graaff, R. A. G. (2011). Acta Cryst. D67, 331–337.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationPedersen, K. O. (1944). Nature (London), 154, 575.  CrossRef Google Scholar
First citationPedersen, K. O. (1947). J. Phys. Colloid Chem. 51, 164–171.  CrossRef CAS Google Scholar
First citationPettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C. & Ferrin, T. E. (2004). J. Comput. Chem. 25, 1605–1612.  Web of Science CrossRef PubMed CAS Google Scholar
First citationRawlings, N. D. & Barrett, A. J. (1990). J. Mol. Evol. 30, 60–71.  CrossRef CAS Google Scholar
First citationReeves, P. J., Callewaert, N., Contreras, R. & Khorana, H. G. (2002). Proc. Natl Acad. Sci. USA, 99, 13419–13424.  Web of Science CrossRef PubMed CAS Google Scholar
First citationSalvesen, G., Parkes, C., Abrahamson, M., Grubb, A. & Barrett, A. J. (1986). Biochem. J. 234, 429–434.  CrossRef CAS Google Scholar
First citationSanctis, D. de, Beteva, A., Caserotto, H., Dobias, F., Gabadinho, J., Giraud, T., Gobbo, A., Guijarro, M., Lentini, M., Lavault, B., Mairs, T., McSweeney, S., Petitdemange, S., Rey-Bakaikoa, V., Surr, J., Theveneau, P., Leonard, G. A. & Mueller-Dieckmann, C. (2012). J. Synchrotron Rad. 19, 455–461.  Web of Science CrossRef IUCr Journals Google Scholar
First citationSchechter, I. & Berger, A. (1967). Biochem. Biophys. Res. Commun. 27, 157–162.  CrossRef CAS PubMed Web of Science Google Scholar
First citationSmart, O. S., Womack, T. O., Flensburg, C., Keller, P., Paciorek, W., Sharff, A., Vonrhein, C. & Bricogne, G. (2012). Acta Cryst. D68, 368–380.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationStöcker, W., Karmilin, K., Hildebrand, A., Westphal, H., Yiallouros, I., Weiskirchen, R., Dietzel, E., Floehr, J. & Jahnen-Dechent, W. (2014). Biol. Chem. 395, 1195–1199.  Google Scholar
First citationStubbs, M. T., Laber, B., Bode, W., Huber, R., Jerala, R., Lenarcic, B. & Turk, V. (1990). EMBO J. 9, 1939–1947.  CrossRef CAS PubMed Web of Science Google Scholar
First citationTallant, C., García-Castellanos, R., Baumann, U. & Gomis-Rüth, F. X. (2010). J. Biol. Chem. 285, 13951–13957.  CrossRef CAS Google Scholar
First citationTerwilliger, T. C., Grosse-Kunstleve, R. W., Afonine, P. V., Moriarty, N. W., Zwart, P. H., Hung, L.-W., Read, R. J. & Adams, P. D. (2008). Acta Cryst. D64, 61–69.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationTsai, P.-L., Chen, C.-H., Huang, C.-J., Chou, C.-M. & Chang, G.-D. (2004). J. Biol. Chem. 279, 11146–11155.  CrossRef CAS Google Scholar
First citationTurk, V. & Bode, W. (1991). FEBS Lett. 285, 213–219.  CrossRef PubMed CAS Web of Science Google Scholar
First citationWilliams, C. J., Headd, J. J., Moriarty, N. W., Prisant, M. G., Videau, L. L., Deis, L. N., Verma, V., Keedy, D. A., Hintze, B. J., Chen, V. B., Jain, S., Lewis, S. M., Arendall, W. B. III, Snoeyink, J., Adams, P. D., Lovell, S. C., Richardson, J. S. & Richardson, D. C. (2018). Protein Sci. 27, 293–315.  CrossRef CAS Google Scholar
First citationWinn, M. D., Ballard, C. C., Cowtan, K. D., Dodson, E. J., Emsley, P., Evans, P. R., Keegan, R. M., Krissinel, E. B., Leslie, A. G. W., McCoy, A., McNicholas, S. J., Murshudov, G. N., Pannu, N. S., Potterton, E. A., Powell, H. R., Read, R. J., Vagin, A. & Wilson, K. S. (2011). Acta Cryst. D67, 235–242.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationYamamoto, K. & Sinohara, H. (1993). J. Biol. Chem. 268, 17750–17753.  CAS Google Scholar
First citationYoshida, K., Suzuki, Y., Yamamoto, K. & Sinohara, H. (1996). Biochem. Mol. Biol. Int. 39, 1023–1028.  CAS Google Scholar

This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.

IUCrJ
Volume 6| Part 2| March 2019| Pages 317-330
ISSN: 2052-2525