SerpinA12 (vaspin) is thought to be mainly expressed in adipose tissue and has multiple beneficial effects on metabolic, inflammatory and atherogenic processes related to obesity. KLK7 (kallikrein 7) is the only known protease target of vaspin to date and is inhibited with a moderate inhibition rate. In the crystal structure, the cleavage site (P1-P1′) of the vaspin reactive centre loop is fairly rigid compared with the flexible residues before P2, possibly supported by an ionic interaction of P1′ glutamate (Glu379) with an arginine residue (Arg302) of the β-sheet C. A P1′ glutamate seems highly unusual and unfavourable for the protease KLK7. We characterized vaspin mutants to investigate the roles of these two residues in protease inhibition and recognition by vaspin. Reactive centre loop mutations changing the P1′ residue or altering the reactive centre loop conformation significantly increased inhibition parameters, whereas removal of the positive charge within β-sheet C impeded the serpin–protease interaction. Arg302 is a crucial contact to enable vaspin recognition by KLK7 and it supports moderate inhibition of the serpin despite the presence of the detrimental P1′ Glu379, which clearly represents a major limiting factor for vaspin-inhibitory activity. We also show that the vaspin-inhibition rate for KLK7 can be modestly increased by heparin and demonstrate that vaspin is a heparin-binding serpin. Noteworthily, we observed vaspin as a remarkably thermostable serpin and found that Glu379 and Arg302 influence heat-induced polymerization. These structural and functional results reveal the mechanistic basis of how reactive centre loop sequence and exosite interaction in vaspin enable KLK7 recognition and regulate protease inhibition as well as stability of this adipose tissue-derived serpin.

INTRODUCTION

SerpinA12 (vaspin) is an inhibitory serpin first identified in white adipose tissue of the OLETF (Otsuka Long–Evans Tokushima fatty) rat Type 2 diabetes model [1]. In humans, elevated vaspin mRNA and serum levels are associated with parameters of obesity and insulin resistance [2,3]. Vaspin is a promising drug target with various beneficial effects on metabolic and atherogenic processes impaired or induced in overweightness and obesity [4]. It has been suggested to function as a compensatory, anti-diabetic and anti-atherogenic protein in obesity-related disorders such as insulin resistance and inflammation [5]. Under a high fat diet, vaspin transgenic mice gain less weight while exhibiting improved glucose tolerance and reduced adipose tissue inflammation [6]. In addition, central effects have been reported and central administration of vaspin reduced food intake in mice [7]. KLK7 (kallikrein 7) is the only known protease target of vaspin so far and at least the beneficial effects on glucose tolerance in vivo are dependent on the protease inhibitor activity of vaspin [8]. KLK7 seems to be mainly expressed in skin, where it is a key player in the desquamation of cornified skin layers [9]. Both proteins, KLK7 and vaspin, are reported to be involved in skin inflammation as in psoriasis, a chronic skin disease often related to obesity [10,11]. KLK7-overexpressing mice exhibit increased epidermal thickness, hyperkeratosis and dermal inflammation [12,13]. In humans, elevated KLK7 mRNA as well as KLK7 protein overexpression have been reported in various tumours such as ovarian, breast and cervical cancer [1416]. Along these lines, a better understanding of vaspin function and regulation on a molecular level might lead to new strategies for pharmaceutical intervention or modulation of vaspin activity.

Serpins inhibit protease targets using their RCL (reactive centre loop) as substrate bait. After protease attack and acyl-enzyme complex formation, the protease-bound RCL integrates into the central serpin β-sheet, thereby causing active-site distortion and preventing hydrolysis, final cleavage and release of the protease. The primary determinants of serpin specificity are the cleavage site and surrounding residues of the RCL. The crystal structure of human vaspin in its native conformation has been determined previously [8]. Vaspin exhibits interesting molecular aspects. As with many other inhibitory serpins, vaspin exhibits a mostly flexible RCL unresolved in the crystal structure. The protease cleavage site appears more rigid than in other serpin structures, possibly due to ionic interactions of P1′ Glu379 and Arg302 of the β-sheet C in the crystal structure (Figure 1A). First of all, the highly unusual glutamate residue at the P1′ position of the protease cleavage site, which is unique among all human serpins, indicates functional relevance, particularly as the protease target KLK7 has a clear preference for basic residues at P1′ [17]. This might explain the moderate association rate and a comparatively high stoichiometry of inhibition observed for vaspin and KLK7 [8]. Also, Arg302, a well-conserved residue in the serpin family, has been found to be of importance for protease recognition in other serpins [18,19]. In the present study, we used a mutational approach to determine the influence of these two residues on KLK7 inhibition and on vaspin stability. Our results demonstrate the significant, yet opposing, impact of both residues on vaspin activity. Whereas P1′ Glu379 is an obvious down-regulator of vaspin inhibitory activity, Arg302 represents a critical contact for KLK7 and enables a moderate inhibition rate for KLK7, potentially by offsetting the limitations of the unfavourable RCL sequence. On the basis of these results and properties from other serpins, we also investigated the possible activation of vaspin by heparin and found significant heparin binding with a modest activation of vaspin. Furthermore, investigating structural serpin parameters, we registered vaspin as a remarkably thermostable serpin and found that both residues, Arg302 and Glu379, influence serpin polymerization.

Crystal structure of human vaspin and vaspin mutants E379S and D305C/V383C

Figure 1
Crystal structure of human vaspin and vaspin mutants E379S and D305C/V383C

(A) Structure of human vaspin (PDB code 4IF8) with β-strands and loops shown in cyan and α-helices in red. The dotted black line indicates flexible and unresolved RCL residues. The detailed view of the RCL–s2C interactions is from the back left with residues 301–302 (s2C) and 378–384 (RCL) shown as sticks and selected molecules as spheres. The feature-enhanced electron density map calculated with PHENIX is displayed at 1σ contour level in red. (B) Feature-enhanced maps of RCL and selected s2C residues of vaspin E379S (PDB code 4Y3K) contoured at 1 σ. (C) corresponding residues of the D305C/V383C variant (PDB code 4Y40) also demonstrating successful formation of the introduced disulfide bond. To more clearly present electron density of RCL residues and the disulfide bond, note that the viewpoint for (B) and (C) is slightly rotated to the left compared with (A).

Figure 1
Crystal structure of human vaspin and vaspin mutants E379S and D305C/V383C

(A) Structure of human vaspin (PDB code 4IF8) with β-strands and loops shown in cyan and α-helices in red. The dotted black line indicates flexible and unresolved RCL residues. The detailed view of the RCL–s2C interactions is from the back left with residues 301–302 (s2C) and 378–384 (RCL) shown as sticks and selected molecules as spheres. The feature-enhanced electron density map calculated with PHENIX is displayed at 1σ contour level in red. (B) Feature-enhanced maps of RCL and selected s2C residues of vaspin E379S (PDB code 4Y3K) contoured at 1 σ. (C) corresponding residues of the D305C/V383C variant (PDB code 4Y40) also demonstrating successful formation of the introduced disulfide bond. To more clearly present electron density of RCL residues and the disulfide bond, note that the viewpoint for (B) and (C) is slightly rotated to the left compared with (A).

MATERIALS AND METHODS

Materials

Human recombinant KLK7 and thermolysin were from R&D Systems. Fluorogenic peptide substrate Mca-RPKPVE-Nva-WR-K(Dnp)-NH2 [where Mca is (7-methoxycoumarin-4-yl)acetyl, Nva is norvaline, and Dnp is 2,4-dinitrophenyl] was from AnaSpec and the substrate peptide Abz-KLYSSK-Q-EDDnp [where Abz is o-aminobenzoyl and Q-EDDnp is glutaminyl-N-(2,4-dinitrophenyl) ethylenediamine] was a gift from Professor Dr Maria A. Juliano (Department of Biophysics, Escola Paulista de Medicina, Universidade Federal de São Paulo, São Paulo, Brazil). Human AT (antithrombin III) purified from human plasma was obtained from Thermo Fisher Scientific. Ufh (unfractionated heparin) was from Sigma–Aldrich, and heparin oligosaccharide dp4 was from Iduron. All calculations were carried out using Prism 5.03 (GraphPad Software).

Generation of recombinant vaspin and KLK7 proteins

Mutagenesis, recombinant expression and purification of vaspin and mutants was carried out as described previously with minor changes [8]. Disulfide-bond-containing mutants were dialysed using an L-cysteine/-cystamine redox system (5 mM/1 mM). A polishing gel-filtration step was performed after final dialysis using a HiLoad 16/60 Superdex 200 column on an ÄKTA protein purification system (all from GE Healthcare). Purified proteins were analysed by RP (reversed-phase)-HPLC and SDS/PAGE, as well as MALDI–TOF-MS, and were stored at 4°C for up to 1 year without loss of activity. Details of recombinant expression and purification of KLK7 are available on request from J.T.H. Briefly, KLK7 was expressed as a fusion to the SUMO (small ubiquitin-like modifier) protein as inclusion bodies in Escherichia coli. The fusion protein was refolded from guanidine by fast dilution in an arginine/glycerol buffer supplemented with a glutathione redox system. The SUMO protein was cleaved off with SUMO protease and purification of active KLK7 was performed using a HiTrap SP HP column followed by a final gel-filtration step using a HiLoad 16/60 Superdex 200 column. Enzyme activity of recombinant KLK7 was comparable with commercially available KLK7 using the fluorogenic peptide Mca-RPKPVE-Nva-WR-K(Dnp)-NH2.

Crystallization, X-ray data collection and structure determination

For crystallization of vaspin mutants, gel-filtration peak fractions were pooled and concentrated to 12 mg/ml. In a hanging-drop vapour-diffusion set-up at 292 K, 1 μl of protein and 1 μl of crystallization buffer were mixed and equilibrated against 0.5 ml of crystallization buffer as reservoir solution (Table 1). The analysed crystals were transferred stepwise to a buffer containing 20% ethylene glycol in addition to the components in the crystallization buffer and frozen in liquid nitrogen. X-ray data collection was carried out at 100 K on beamline 14.1 of the Berlin Synchrotron (BESSY, Berlin, Germany) equipped with a PILATUS 6M detector (Dectris). The diffraction data were indexed, integrated and scaled with XDS [20] and aimless [21] of the CCP4 suite [22]. Coordinates from wild-type vaspin (PDB code 4IF8) were taken as starting models for rigid body refinement in REFMAC5 [23,24]. Models were further improved by iterative cycles of TLS (Translation–Libration–Screw-rotation) refinement using REFMAC5 [23,24] as well as BUSTER-TNT (version 2.10.1, Global Phasing Ltd) and manual rebuilding with COOT [26]. The artificial disulfide bridge of chain A in the variant D305C/V383C was refined with an occupancy of 25% in the reduced form. Reduction of the disulfide linkage could be unambiguously ascribed to radiation damage by using only the first 40% of the dataset (yielding a data completeness of 97.9%) for refinement. With this dataset, the electron density showed no evidence for an open conformation of the disulfide bridge. In chain B, no alternative conformation was observed for the artificial disulfide bridge. We attribute this to the generally lower quality of the electron density of this chain, which is accompanied by significantly higher B-factors, most likely as a result of static disorder of this protein chain in the crystal. TLS groups (chain A: residues 36–70, 71–242, 243–308 and 309–414; chain B: residues 36–70, 71–176, 177–303 and 304–413) were determined using the TLSMD web server [27,28]. Structures were validated by MOLPROBITY [29]. Final structure building and interpretation was aided by feature enhanced maps calculated with PHENIX [30,31]. For crystallization of the E379S variant, a heparin tetrasaccharide was added to the crystallization buffer, but no binding of the heparin was observed in the electron density maps. Crystals obtained in the absence of heparin showed identical features (results not shown). Final co-ordinates and structure factors have been deposited in the PDB (http://www.rcsb.org [32]) under accession codes 4Y3K (E379S) and 4Y40 (D305C/V383C) and relevant crystallographic data are provided in Table 1. Figures were prepared using PyMOL (http://www.pymol.org).

Table 1
Details of the crystal structure analysis

Values in parentheses correspond to the highest resolution shell.

 Vaspin D305C/V383C Vaspin E379S 
Crystallization and data collection   
 Crystallization buffer 6% (w/v) PEG 6000 and 0.1 M citric acid (pH 3.5–4.5) 2.5 M ammonium sulfate and 0.1 M sodium citrate (pH 4.0–4.6) 
 Crystallization drop 1 μl of crystallization buffer and 1 μl of 6 mg/ml protein in 50 mM Tris/HCl (pH 7.5) and 150 mM NaCl 1 μl of crystallization buffer and 1 μl of 10.2 mg/ml protein in 50 mM Tris/HCl (pH 7.5), 150 mM NaCl and 0.52 mM heparin tetrasaccharide dp4 
 Cryobuffer Crystallization buffer and 20% (v/v) ethylene glycol Crystallization buffer and 20% (v/v) ethylene glycol 
 Wavelength (Å) 0.91841 0.91841 
 Resolution range (Å) 19.69–2.2 (2.26–2.2) 48.37–2.2 (2.26–2.2) 
 Space group CC
 Unit cell parameters a, b, c (Å), α, β, γ (°) 133.5, 152.3, 61.6, 90, 97.1, 90 133.0, 151.4, 61.6, 90, 98.4, 90 
 Total reflections 342549 (19778) 208807 (15788) 
 Unique reflections 61536 (4375) 60841 (4481) 
 Multiplicity 5.6 (4.5) 3.4 (3.5) 
 Completeness (%) 99.6 (96.3) 99.8 (99.8) 
 Mean I/σ(I7.9 (0.9) 7.2 (0.9) 
R-merge 0.145 (1.554) 0.125 (1.413) 
R-pim 0.067 (0.798) 0.080 (0.885) 
 CC1/2 0.996 (0.285) 0.996 (0.206) 
 CC* 0.999 (0.666) 0.999 (0.584) 
 Wilson B-factor (Å246.70 45.62 
 Monomers/asymmetric unit 
Refinement   
R-work 0.1801 (0.2251) 0.1862 (0.2485) 
R-free 0.2058 (0.2406) 0.2193 (0.2574) 
 Protein residues 727 (chain A: 36–366, 380–414; chain B: 34–277, 281–365, 382–413) 728 (chain A: 35–367, 381–414; chain B: 35–277, 280–365, 383–414) 
 Number of non-hydrogen atoms   
  Total 6237 6224 
  Protein 5949 5983 
  Ligands 33 53 
  Water 255 188 
 RMSD   
  Bonds (Å) 0.009 0.010 
  Angles (°) 1.08 1.12 
 Ramachandran statistics   
  Favoured (%) 96.54 95.87 
  Allowed (%) 3.46 4.13 
  Outliers (%) 
B-factor (Å2  
  Average 66.4 71.9 
  Protein 66.9 72.6 
   Chain A 55.4 56.3 
   Chain B 78.6 89.1 
  Ligands 60.858 59.968 
  Water 57.038 53.758 
 Vaspin D305C/V383C Vaspin E379S 
Crystallization and data collection   
 Crystallization buffer 6% (w/v) PEG 6000 and 0.1 M citric acid (pH 3.5–4.5) 2.5 M ammonium sulfate and 0.1 M sodium citrate (pH 4.0–4.6) 
 Crystallization drop 1 μl of crystallization buffer and 1 μl of 6 mg/ml protein in 50 mM Tris/HCl (pH 7.5) and 150 mM NaCl 1 μl of crystallization buffer and 1 μl of 10.2 mg/ml protein in 50 mM Tris/HCl (pH 7.5), 150 mM NaCl and 0.52 mM heparin tetrasaccharide dp4 
 Cryobuffer Crystallization buffer and 20% (v/v) ethylene glycol Crystallization buffer and 20% (v/v) ethylene glycol 
 Wavelength (Å) 0.91841 0.91841 
 Resolution range (Å) 19.69–2.2 (2.26–2.2) 48.37–2.2 (2.26–2.2) 
 Space group CC
 Unit cell parameters a, b, c (Å), α, β, γ (°) 133.5, 152.3, 61.6, 90, 97.1, 90 133.0, 151.4, 61.6, 90, 98.4, 90 
 Total reflections 342549 (19778) 208807 (15788) 
 Unique reflections 61536 (4375) 60841 (4481) 
 Multiplicity 5.6 (4.5) 3.4 (3.5) 
 Completeness (%) 99.6 (96.3) 99.8 (99.8) 
 Mean I/σ(I7.9 (0.9) 7.2 (0.9) 
R-merge 0.145 (1.554) 0.125 (1.413) 
R-pim 0.067 (0.798) 0.080 (0.885) 
 CC1/2 0.996 (0.285) 0.996 (0.206) 
 CC* 0.999 (0.666) 0.999 (0.584) 
 Wilson B-factor (Å246.70 45.62 
 Monomers/asymmetric unit 
Refinement   
R-work 0.1801 (0.2251) 0.1862 (0.2485) 
R-free 0.2058 (0.2406) 0.2193 (0.2574) 
 Protein residues 727 (chain A: 36–366, 380–414; chain B: 34–277, 281–365, 382–413) 728 (chain A: 35–367, 381–414; chain B: 35–277, 280–365, 383–414) 
 Number of non-hydrogen atoms   
  Total 6237 6224 
  Protein 5949 5983 
  Ligands 33 53 
  Water 255 188 
 RMSD   
  Bonds (Å) 0.009 0.010 
  Angles (°) 1.08 1.12 
 Ramachandran statistics   
  Favoured (%) 96.54 95.87 
  Allowed (%) 3.46 4.13 
  Outliers (%) 
B-factor (Å2  
  Average 66.4 71.9 
  Protein 66.9 72.6 
   Chain A 55.4 56.3 
   Chain B 78.6 89.1 
  Ligands 60.858 59.968 
  Water 57.038 53.758 

Inhibition parameters

The SI (stoichiometry of inhibition) was determined as described previously [8]. A discontinuous assay was used to determine the rate of inhibition under pseudo-first-order conditions for vaspin wild-type and mutants D305C/V383C, R302A, R302E and R302E/E379S with at least 10-fold molar excess of inhibitor, as described previously [33]. Briefly, recombinant human KLK7 was activated according to the manufacturer's protocol and the assay concentration of KLK7 was 19.2 nM in TBS (50 mM Tris/HCl and 150 mM NaCl, pH 8.5). Inhibition reactions were stopped by the addition of fluorogenic peptide substrate (30 μM), and residual KLK7 activity was measured on a FlexStation3 Multi-Mode Microplate Reader (Molecular Devices). The pseudo-first-order rate constant (kobs) was obtained from linear regression of semi-logarithmic plots of remaining KLK7 activity against time. Linear regression of kobs plotted against inhibitor concentration [I] resulted in the estimate of the second-order rate constant, ka.

A continuous method was used to determine the rate of inhibition for vaspin mutant E379S with at least 5-fold molar excess of inhibitor, as described previously [34]. The assay concentration of KLK7 was 10 nM, the fluorogenic peptide was used at 25 μM. Here, kobs was determined by non-linear regression fitting of the progress curve to eqn (1):

 
formula
1

where P is the concentration of product at time t and v0 is the initial velocity for each [I]. Resulting kobs values were plotted against [I] and the uncorrected second-order rate constant k′ was determined from the slope of linear fit. Finally, the second-order rate constant, ka, was calculated by correcting k′ for substrate concentration [S], the Km of the protease and SI using eqn (2):

 
formula
2

The Km value for KLK7 was 15 μM. SI values and inhibition rates listed in Table 2 represent the means ± S.D. for at least three experiments from at least two protein batches.

Table 2
Kinetic parameters for the inhibition of KLK7 by vaspin and mutants

Data are presented as means ± S.D. for at least three experiments with the exception of ka for R302E. I, inhibitor; E, enzyme.

Vaspin variant ka (mM−1·s−1SI (I/E, mol/mol) 
Wild-type 12.3 ± 0.3 3.6 ± 0.3 
D305C/V383C 56.9 ± 9.7 2.0 ± 0.2 
E379S 403.0 ± 34.4 2.0 ± 0.1 
R302A 0.3 ± 0.1 2.7 ± 0.6* 
R302E 0.03 3.2 ± 0.2* 
R302E/E379S 20.7 ± 4.2 3.2 ± 0.2 
Vaspin variant ka (mM−1·s−1SI (I/E, mol/mol) 
Wild-type 12.3 ± 0.3 3.6 ± 0.3 
D305C/V383C 56.9 ± 9.7 2.0 ± 0.2 
E379S 403.0 ± 34.4 2.0 ± 0.1 
R302A 0.3 ± 0.1 2.7 ± 0.6* 
R302E 0.03 3.2 ± 0.2* 
R302E/E379S 20.7 ± 4.2 3.2 ± 0.2 

*Densitometric estimation.

The pseudo-first-order rate constant (kobs) was furthermore determined in the presence of various concentrations of ufh (ufh/vaspin ratios of 1.56-, 12.5- and 100-fold) using the discontinuous method with 10-fold molar excess of inhibitor to maintain pseudo-first-order conditions. In these assays, the substrate peptide Abz-KLYSSK-Q-EDDnp [17] was used at a concentration of 7 μM. kobs values indicating reaction speed were normalized to the reaction without ufh to estimate the relative heparin induced increase in activity.

Complex formation analysis by SDS/PAGE

To investigate complex formation, recombinant KLK7 and vaspin, or mutants, were incubated at a ratio of 3:1 (protease/serpin; KLK7 concentration was 3.5 μM) in TBS. At given time points, samples containing 0.5 μg of serpin were taken and the reaction was stopped by immediate addition of reducing SDS sample buffer and 5 min of heating at 95°C. Protein separation by SDS/PAGE was performed using 4–12% Bis-Tris Plus pre-cast gels and NuPAGE MES/SDS running buffer (all from Life Technologies). Gels were stained with Coomassie Blue, and digitization and densitometric analysis were performed using GeneTools analysis software (Syngene). Estimation of SI values were calculated as the sum of band intensities for complexed and cleaved vaspin divided by the band intensity of complexed vaspin [35]. To analyse the effect of heparin on the vaspin–KLK7 reaction, complex formation assays were performed as described above with various concentrations of ufh (molar ratios of heparin to vaspin from 0.1 to 100) and the reactions were stopped after 1 min of incubation. Complex band intensities were then normalized to the reaction without heparin.

Heparin-affinity chromatography

Recombinant protein (200 μg) was used for heparin-affinity chromatography with 1 ml HiTrap Heparin HP columns on the ÄKTA protein purification system (both from GE Healthcare). Elution was performed by an NaCl gradient from 150 mM to 2 M (flow rate 1 ml/min) and monitored at 220 nm.

Thermal stability

CD spectra were recorded using 1-mm pathlength quartz cuvettes (Hellma) on a J-715 spectropolarimeter (Jasco) equipped with a Peltier-type temperature control system at 20°C. Spectra were acquired as an average of three scans (2 nm bandwidth, 4 s response, 50 nm/min scan rate) and baselines were corrected by subtraction of the buffer spectrum. The protein concentration was 1.5 μM in 10 mM phosphate buffer (pH 7.8). Thermal unfolding was performed at a heating rate of 50°C/h and monitored at 208 nm (data pitch 0.1°C, sensitivity 100 mdeg, 4 s response, bandwidth 2 nm). Measured ellipticity θ was converted into mean residue molar ellipticity [θ] in deg·cm2·dmol−1 as described previously [36].

Temperature-induced polymerization

AT, vaspin or mutants at 5 μM were heated for 120 min at 60°C or 70°C in TBS. During heating, samples were taken at indicated time points and non-reducing SDS-sample buffer was added. Vaspin–KLK7 complex formation after heating was analysed as described above, with 2 min of cooling on ice before incubation with KLK7 for 15 min. Polymerization and complex formation were analysed by SDS/PAGE.

Peptide synthesis and RP-HPLC-based analysis of KLK7 cleavage

Peptide synthesis, purification and analytics were performed as described previously [10]. For cleavage analysis, 250 μg of vaspin-(365–388) was incubated with 2 μg of KLK7 (327 nM) in a 250 μl reaction volume for 120 min at 37°C. After incubation, proteolytic activity was terminated by boiling and the peptide mixtures were analysed by RP-HPLC (Phenomenex Jupiter Proteo C18 column, 4.6 mm×250 mm, 90 Å pore size; 1 Å=0.1 nm).

RESULTS

Altering vaspin RCL constraints via two designed variants

Vaspin and its mutants crystallize in space group C2 with two monomers in the asymmetric unit (Table 1). Although no systematic differences are observed between the two molecules, in monomer B, the lack of crystal contacts results in a rigid body disorder of the whole subunit such that the regions close to the interaction interface with monomer A have low B-factors and well-defined density, and the complete subdomain comprising sheet 3 and the RCL exhibit high B-factors and ill-defined density (Supplementary Figures 1A and 1B). As a result, the density of the RCL and nearby regions is less well-defined and fewer residues of the RCL could be modelled, including residues involved in the interactions described below. Since this is a result of a rigid body disorder of the whole subunit, these interactions are likely to be present, but they are not evident from the electron density maps which are affected by the rigid body disorder. Therefore the following observations, Figures and calculations are based on chain A.

In the crystal structure, the P1′ Glu379 forms a weak salt bridge with the side chain of Arg302 within strand 2 of β-sheet C (s2C) and water-mediated hydrogen bonds with the main-chain CO of Arg301. Also P2′ Thr380 interacts with the Arg302 side chain via a water-mediated hydrogen bond (Figure 1A). Collectively, these interactions could limit the overall flexibility of the C-terminal part of the RCL with the cleavage site. Noteworthily, RCL orientations in serpin crystal structures display great variability and do not necessarily represent the conformation of the free serpin in solution.

We aimed to alter the cleavage site by mutating the P1′ glutamate to serine, a very common P1′ residue in the serpin family (E379S mutant) [37]. Also, a recent study investigated the primed subsite (S1′–S3′) preferences of KLK7 and demonstrated a clear P1′ preference for hydrophilic residues, especially arginine and serine [17]. To specifically dissect the effect of RCL flexibility or conformation from P1′ specificity of KLK7, we designed a variant in which we introduced a disulfide linkage at the C-terminal side of the cleavage site via the D305C/V383C double mutant. This disulfide bridge was intended to induce an altered RCL conformation while the native RCL sequence with respect to protease substrate specificity is preserved.

Overall, the crystal structures of both mutants show no conformational differences from wild-type vaspin further away from the mutation sites, as demonstrated by RMSDs of 0.38 Å (E379S) and 0.33 Å (D305C/V383C) calculated for the Cα atoms of chain A. The RCL hinge region of residues P15 to P13/P12 (residues 364–366/367) is well-defined and extends away from the central β-sheet A. In contrast, the electron density for RCL residues P12/P11–P2 (G367/A368–P377) and the N-terminal RCL part is weak or absent, indicating high flexibility in both structures, similar to the situation in wild-type vaspin (Supplementary Figure 1C). The only exception concerns RCL residues P1–P2′ (Met378–Thr380). As mentioned above, in wild-type vaspin, these residues are rather well defined (Figure 1A), but there is weak ambiguous electron density in the E379S mutant structure also indicating a very flexible C-terminal RCL (Figure 1B). In the D305C/V383C mutant, formation of the engineered disulfide bond was verified by the electron density maps (Figure 1C) and resulted in a band with a distinct shift on a non-reducing SDS gel, confirming full formation of the disulfide bond (Supplementary Figure S2). Comparable with the E379S mutant, the P1 and P1′ residues (Met378 and Glu379) also exhibit only weak electron density and thus are much more flexible compared with the wild-type vaspin structure (Figure 1C). Hence the successful introduction of the artificial disulfide linkage was sufficient to alter RCL conformation and both mutant structures exhibit less constrained RCLs.

Change of RCL sequence and conformation result in accelerated and more efficient KLK7 inhibition

Both the introduction of the disulfide bond and P1′ exchange resulted in significant improvement in serpin activity (Table 2). As expected, exchange of the P1′ residue in the E379S variant resulted in a dramatic rate acceleration of ~35-fold, requiring a change to a continuous assay, whereas all other mutants were characterized in a discontinuous assay (Table 2 and Figure 2A). Also the D305C/V383C mutant exhibits a significant ~4.5-fold increase in KLK7 inhibition rate (Table 2 and Figure 2B). This mutant features the native RCL sequence and thus the effect is independent of substrate specificity of KLK7. The SI measured for both vaspin mutants was ~2 (Table 2 and Figure 2C).

Kinetic analysis for the inhibition of KLK7 by vaspin and mutants

Figure 2
Kinetic analysis for the inhibition of KLK7 by vaspin and mutants

(A) KLK7 inhibition by vaspin E379S was determined by using the progress curve method. KLK7 inactivation was monitored for 45 min and kobs was calculated by a non-linear regression fit of each curve. Accounting for Km and SI of KLK7 for the peptide substrate, the slope of kobs against I0 resulted in the ka. (B) Inhibition of KLK7 by vaspin wild-type (wt) and mutants D305C/V383C, R302E/E379S and R302A were measured under pseudo-first-order conditions in a discontinuous assay using at least six different inhibitor concentrations. The ka was determined as the slope of the linear fit after plotting kobs against I0. (C) Stoichiometry of inhibition was determined by linear regression to extrapolate the inhibitor to enzyme (I0/E0) ratio for complete loss of KLK7 activity. The mean ± S.D. values for at least three experiments are reported in Table 2.

Figure 2
Kinetic analysis for the inhibition of KLK7 by vaspin and mutants

(A) KLK7 inhibition by vaspin E379S was determined by using the progress curve method. KLK7 inactivation was monitored for 45 min and kobs was calculated by a non-linear regression fit of each curve. Accounting for Km and SI of KLK7 for the peptide substrate, the slope of kobs against I0 resulted in the ka. (B) Inhibition of KLK7 by vaspin wild-type (wt) and mutants D305C/V383C, R302E/E379S and R302A were measured under pseudo-first-order conditions in a discontinuous assay using at least six different inhibitor concentrations. The ka was determined as the slope of the linear fit after plotting kobs against I0. (C) Stoichiometry of inhibition was determined by linear regression to extrapolate the inhibitor to enzyme (I0/E0) ratio for complete loss of KLK7 activity. The mean ± S.D. values for at least three experiments are reported in Table 2.

These results are also reflected in accelerated complex formation observed in SDS/PAGE analysis for both mutants (Figure 3A). The native RCL sequence seems to prevent a fast protease attack. After 60 min almost all serpin molecules are either cleaved or complexed for both mutants, whereas ~25% of wild-type vaspin remains unaffected after 120 min (Figure 3A). In conclusion, mutation of the P1′ residue Glu379 to the preferred serine demonstrates the explicit rate-reducing effect of the native P1′ residue in this system. Yet, an alteration of the RCL conformation could also moderately improve serpin activity despite the detrimental P1′ Glu379.

Complex formation analyses of vaspin wild-type and mutants with KLK7 by SDS/PAGE

Figure 3
Complex formation analyses of vaspin wild-type and mutants with KLK7 by SDS/PAGE

(A) Coomassie Blue-stained SDS gels of vaspin wild-type (wt) and mutants D305C/V383C and E379S incubated with KLK7 for the indicated times and respective control samples without protease. (B) Respective Coomassie Blue-stained gels for mutants R302A and R302E incubated with KLK7. (C) RP-HPLC monitoring of synthetic vaspin RCL comprising peptide vaspin-(365–388) incubated with or without KLK7 reveals no detectable cleavage (linear gradient of acetonitrile/water from 10 to 50% in 40 min). (D) Coomassie Blue-stained gels for mutants R302A/D305C/V383C, R302E/D305C/V383C and R302E/E379S incubated with KLK7. Notable and indicated bands are: 1, serpin–protease complex; 2, active serpin; 3, active, Tag-cleaved serpin; 4, cleaved serpin; 5, protease.

Figure 3
Complex formation analyses of vaspin wild-type and mutants with KLK7 by SDS/PAGE

(A) Coomassie Blue-stained SDS gels of vaspin wild-type (wt) and mutants D305C/V383C and E379S incubated with KLK7 for the indicated times and respective control samples without protease. (B) Respective Coomassie Blue-stained gels for mutants R302A and R302E incubated with KLK7. (C) RP-HPLC monitoring of synthetic vaspin RCL comprising peptide vaspin-(365–388) incubated with or without KLK7 reveals no detectable cleavage (linear gradient of acetonitrile/water from 10 to 50% in 40 min). (D) Coomassie Blue-stained gels for mutants R302A/D305C/V383C, R302E/D305C/V383C and R302E/E379S incubated with KLK7. Notable and indicated bands are: 1, serpin–protease complex; 2, active serpin; 3, active, Tag-cleaved serpin; 4, cleaved serpin; 5, protease.

Elimination of the positive charge at position 302 reveals a crucial contact for KLK7 recognition

Since conformational alterations as induced by the disulfide mutant increased vaspin inhibition rates, we investigated the existence of further determinants of activity and specificity in addition to the P1′ residue. We analysed the importance of Arg302 in KLK7 inhibition by vaspin. This positively charged residue is well conserved throughout the serpin family and the orientation of the side chain is in good superposition in many serpin structures (Supplementary Figure S3). Furthermore, this arginine residue has been found to be a critical contact point in the Michaelis complex of heparin cofactor 2 (serpinD1) and thrombin [18], and is part of a basic exosite for the inhibition of tissue kallikrein by kallistatin (serpinA4) [19].

For vaspin, the mutation of arginine to alanine (R302A) resulted in dramatic loss of inhibitory activity (~40-fold) compared with wild-type (Table 2 and Figure 2B). This is also evident in the very slow complex formation (Figure 3B). Introduction of a negatively charged glutamate residue (R302E) resulted in almost complete prevention of the serpin–protease interaction and no measurable inhibitory activity (Table 2 and Figures 2B and 3B). After 16 h of incubation, only faint bands of complexed and cleaved serpin were detected, with ~90% of serpin remaining active. These results are also in line with the finding that a synthetic peptide comprising the vaspin RCL sequence, vaspin-(365–388), is not cleaved by KLK7 (Figure 3C). Densitometric estimation of the SI for the R302A and R302E mutants indicated an SI comparable with that of wild-type vaspin (Table 2). Mutation of the Arg302 in the disulfide bond mutant (R302A/D305C/V383C, R302E/D305C/V383C) resulted in a similar loss of activity and serpin–protease interaction as observed in wild-type vaspin (Table 2 and Figure 3D). As expected, loss of Arg302 in the E379S mutant did not completely prevent inhibition of KLK7, as the RCL sequence is adjusted to KLK7 subsite preference, but still slowed down the reaction by >90% (Table 2 and Figure 3D). In conclusion, Arg302 comprises a residue crucial for initial KLK7–vaspin Michaelis complex formation with subsequent cleavage site attack promoting the inhibition reaction and bypassing the repressing effect of P1′ Glu379.

Heparin activates vaspin and accelerates KLK7 inhibition

We investigated further whether heparin, a common activator of serpin activity, is able to bind and activate vaspin. Indeed, vaspin as well as KLK7 bound to heparin–Sepharose and were eluted at ~580 mM and ~420 mM NaCl in heparin-affinity chromatography experiments (Figure 4A). When evaluating the effect of heparin on vaspin activity, we found a concentration-dependent increase of vaspin activity both by analysing vaspin–KLK7 complex formation on SDS/PAGE gels (Figures 4B and 4C) and by measuring the inhibition rate for wild-type vaspin (Figure 4D). At the optimal ratio of heparin to vaspin (12.5-fold ufh in our experiments), we observed a modest ~5-fold increase in vaspin activity towards KLK7. The bell-shaped dose–response curve for heparin activation may indicate activation via the bridging mechanism for vaspin and KLK7 by heparin.

Heparin binding and activation of vaspin

Figure 4
Heparin binding and activation of vaspin

(A) Elution profiles of recombinant wild-type vaspin and KLK7 by heparin-affinity chromatography using a NaCl gradient (black dotted line) monitored at 220 nm. Vaspin is eluted at ~580 mM and KLK7 at 420 mM NaCl. (B) Analysis of concentration-dependent heparin-accelerated complex formation of vaspin and KLK7. Coomassie Blue-stained SDS gel of a fixed molar ratio (1:3) of vaspin wild-type and KLK7 incubated with increasing concentrations of ufh (0.1–100-fold vaspin) for 1 min and a reference sample without heparin (0). The control (con) is vaspin and heat-inactivated KLK7. (C) Densitometric quantification of complex band intensities in relation to the molar ratio of ufh to vaspin of SDS gels as in (B). A bell-shaped curve is obtained suggesting the bridging mechanism as the major contribution to heparin activation of vaspin. (D) Inhibition of KLK7 by wild-type vaspin with different concentrations of unfractionated heparin was measured under pseudo-first-order conditions in a discontinuous assay. Shown is the increase in second-order rate constant as x-fold over control (without heparin). The ka values were determined as described in the materials and methods section.

Figure 4
Heparin binding and activation of vaspin

(A) Elution profiles of recombinant wild-type vaspin and KLK7 by heparin-affinity chromatography using a NaCl gradient (black dotted line) monitored at 220 nm. Vaspin is eluted at ~580 mM and KLK7 at 420 mM NaCl. (B) Analysis of concentration-dependent heparin-accelerated complex formation of vaspin and KLK7. Coomassie Blue-stained SDS gel of a fixed molar ratio (1:3) of vaspin wild-type and KLK7 incubated with increasing concentrations of ufh (0.1–100-fold vaspin) for 1 min and a reference sample without heparin (0). The control (con) is vaspin and heat-inactivated KLK7. (C) Densitometric quantification of complex band intensities in relation to the molar ratio of ufh to vaspin of SDS gels as in (B). A bell-shaped curve is obtained suggesting the bridging mechanism as the major contribution to heparin activation of vaspin. (D) Inhibition of KLK7 by wild-type vaspin with different concentrations of unfractionated heparin was measured under pseudo-first-order conditions in a discontinuous assay. Shown is the increase in second-order rate constant as x-fold over control (without heparin). The ka values were determined as described in the materials and methods section.

Thermostability and vaspin polymerization

Far-UV CD data reflected the structural similarity and demonstrated structural integrity for all mutants tested (Figure 5A). Thermal denaturation revealed a remarkable thermostability, with temperature midpoint (Tm) of 70°C and a well-defined isochromatic point suggesting a two-state unfolding pathway progressing from the native to unfolded state (Figures 5B and 5C). Thermal unfolding seems to be irreversible and heat-induced aggregation and precipitation were observed. Vaspin was found to be considerably resistant to heat-induced inactivation by polymerization. Prolonged incubation of up to 120 min at 60°C did not noticeably alter serpin activity analysed via complex formation with KLK7 (Figure 5D). Also, polymers were almost undetectable in non-reducing non-heated SDS samples, whereas for AT, serving as a control, polymerization was apparent after 15 min at 60°C (Figure 5E). As expected from the Tm values, increasing the temperature to 70°C resulted in distinct vaspin polymer formation (Figure 5F). Polymerization was markedly enhanced in Arg302-deficient mutants and also observed for the disulfide mutant D305C/V383C. Polymerization for the E379S-containing mutants was minor and similar to that of the wild-type. In conclusion, vaspin represents a highly thermostable serpin and Arg302 seems to hinder polymer formation in wild-type variants featuring the native Glu379.

Thermostability and heat-induced polymerization of vaspin wild-type and mutants

Figure 5
Thermostability and heat-induced polymerization of vaspin wild-type and mutants

(A) Far-UV CD spectra of vaspin wild-type (wt) and mutants D305C/V383C, E379S, R302A and R302E at 20°C (collected in 1.5 μM phosphate buffer, pH 7.8). (B) Thermal unfolding of vaspin and mutants from single samples observed by far-UV CD at 208 nm at a heating rate of 50°C/h. (C) CD spectra of vaspin at different stages of thermal denaturation from a single sample. (D) SDS/PAGE analysis of vaspin–KLK7 complex formation or (E) vaspin and AT polymerization after incubation at 60°C for indicated times. (F) Heat-induced polymerization for wild-type vaspin and R302A, R302E, E379S, R302E/E379S and D305C/V383C variants after incubation at 70°C for up to 60 min. Mutation of Arg302 significantly increases polymer formation.

Figure 5
Thermostability and heat-induced polymerization of vaspin wild-type and mutants

(A) Far-UV CD spectra of vaspin wild-type (wt) and mutants D305C/V383C, E379S, R302A and R302E at 20°C (collected in 1.5 μM phosphate buffer, pH 7.8). (B) Thermal unfolding of vaspin and mutants from single samples observed by far-UV CD at 208 nm at a heating rate of 50°C/h. (C) CD spectra of vaspin at different stages of thermal denaturation from a single sample. (D) SDS/PAGE analysis of vaspin–KLK7 complex formation or (E) vaspin and AT polymerization after incubation at 60°C for indicated times. (F) Heat-induced polymerization for wild-type vaspin and R302A, R302E, E379S, R302E/E379S and D305C/V383C variants after incubation at 70°C for up to 60 min. Mutation of Arg302 significantly increases polymer formation.

DISCUSSION

We have previously identified KLK7 as the first target protease of human vaspin inhibited with a moderate inhibition rate [8]. The RCL P1′ Glu379 residue, unique among all human serpins, indicates functional relevance. By changing the RCL sequence or conformation, the first by mutation of Glu379, the latter by an engineered disulfide bond, we observed increasing RCL flexibility in the crystal structures. In addition, the mutants featured increased inhibition rates with a far more pronounced increase in activity for the E379S mutant. This can be attributed to a much more preferred P1′ serine residue over glutamate for KLK7. The primed subsite (S1′–S2′) of KLK7 clearly prefers the amino acids serine and arginine at the substrate's P1′ and P2′ position, and in a soluble peptide library P1′ glutamate-featuring peptide substrates were all found to be resistant to hydrolysis [17]. In accordance with these data, no cleavage of a synthetic peptide comprising the RCL sequence of vaspin by KLK7 was observed. Yet the significant effect on activity of the disulfide bond mutant demonstrates a moderate contribution of RCL conformation or flexibility, as this mutant exhibits the native RCL sequence. Together, these results clearly reveal that the P1′ glutamate is detrimental to vaspin inhibitory activity and it may function as a regulatory element in reducing the inhibition rate.

Related to this, an almost complete loss of inhibition and interaction was observed when mutating the positively charged Arg302 within s2C for all variants featuring the P1′ glutamate. The finding that also the KLK7-adjusted RCL-mutant E379S loses ~90% of activity when lacking Arg302 further demonstrates the importance of this residue for KLK7 recognition. Together with the specificity studies on KLK7, these results highlight the necessity of Arg302 for KLK7 inhibition by vaspin to offset the effect of the impedimental P1′ glutamate and explain the moderate inhibitory activity of native vaspin observed for KLK7. In conclusion, Arg302 seems to represent a crucial contact for the serpin–protease interaction and its presence or access to it regulates whether or not the target protease KLK7 is inhibited at all. Thus vaspin is an example of how serpins generate specificity via exosite interactions with their target proteases, but at the cost of maximal inhibitory activity.

In kallistatin, Arg308, the equivalent to Arg302 in vaspin, is part of an exosite with tissue kallikrein that comprises a cluster of positively charged residues (Lys307/Arg308 and Lys312/Lys313 all in s2C). Mutations of these positively charged residues to alanine (K307A/R308A and K312A/K313A) in kallistatin resulted in 5–20-fold lower activity [19]. In kallistatin, this exosite is supposed to enable specific inhibition of kallikreins despite the kallistatin P1 phenylalanine residue being inappropriate for kallikreins. The exosite mutations in kallistatin were performed as double mutations and there are no data on the effect of individual mutations of single charged residues. Whereas with Arg302 in vaspin, where a single mutation almost completely prevents KLK7 recognition and subsequent inhibition, it is likely that more residues contribute to KLK7 recognition and binding, e.g. the neighbouring Arg301, and also that other exosites are still to be identified.

In addition to exosites promoting binary complex formation of serpin and protease, there are also exosites involved in ternary complex formation with additional cofactors [38]. In the light of the detrimental P1′ Glu379, we investigated whether activity of vaspin could also be increased by cofactors, such as the GAG (glycosaminoglycan) heparin. First, we found strong heparin binding for both vaspin and KLK7. Binding and activation of KLK7 by heparin has been reported previously, with KLK7 eluting at 0.3–0.4 mM NaCl from heparin–Sepharose [17]. This is in line with our findings for the recombinant KLK7 used in our assays. Importantly, we also observed a modest, but distinct and concentration-dependent, heparin-induced increase in KLK7 inhibition by vaspin (~5-fold). It is important to note that the heparin-induced vaspin activation was quantified, accounting and normalizing for any heparin-induced increase in KLK7 activity, as KLK7 is also moderately activated by heparin (~2-fold at ~100-fold excess of GAG [17]). The major contribution of GAG activation of serpins comes via the template mechanism, i.e. bridging serpin and protease and directing a preferential orientation of protease and serpin RCL [38]. Also, the bell-shaped dependence on heparin concentration suggests a bridging effect of heparin for vaspin activation, as very high GAG concentrations counteract the activating effect. Thus vaspin is a heparin-binding, although very modestly heparin-activated, serpin. Interestingly, although there are similarities in this exosite residue for the two serpins, i.e. vaspin and kallistatin, heparin binding has opposing effects for their activity. For kallistatin, heparin binding occurs at the exosite comprising Arg308 and, in consequence, counteracts protease inhibition [39]. As heparin is able to activate vaspin, the binding site has to be located somewhere else, or at least Arg302 is not likely to contribute to GAG binding, and we are currently working on the identification of the GAG-binding site in vaspin.

Investigating the structural integrity of the mutants generated, we found vaspin to be remarkably thermostable, with a Tm of 70°C compared with AT (Tm of 57°C) or antitrypsin (Tm of 58°C) [40,41]. Thermopin, a thermostable serpin from the bacterium Thermofibia fusca, has a Tm of 65°C, enabling functionality at the preferred growth temperature of 55°C [42]. The Tm was identical for all vaspin mutants tested. These data reveal no major thermodynamic importance of the potential interaction of Glu379 and Arg302 observed in the crystal structure and thus further determinants are likely to contribute to the thermal stability of vaspin. Surprised by these findings, we were also interested in heat-induced polymerization for vaspin and variants. There are two pathways of serpin inactivation independent of protease attack, which are the latent transition and serpin polymerization. The transition into the latent conformation occurs via incorporation of the RCL into β-sheet A, as for example observed for PAI-1 (plasminogen-activator inhibitor 1) or AT [4345]. On the basis of structural data, the hinge region of vaspin is not partially incorporated into the central A-sheet as, e.g., in native AT and we also have not observed considerable loss of activity for vaspin stored at 4°C for months. Thus vaspin seems not to be prone to undergo the latent transition. Alternatively, serpins are able to polymerize, and the mechanism thought to most likely represent the in vivo polymerization mechanism features a swap of major C-terminal domains involving strands s1C, s4B and s5B, after RCL incorporation into the central β-sheet A, as for the latent transition [46]. In vivo, serpin polymerization is observed for mutations affecting folding, formation and stability of the native state, e.g. for the Z variant (E342K) of antitrypsin [47]. A recent study concluded that these mutations probably first affect the stability of the metastable native state and ultimately facilitate or hinder the irreversible polymerization step [41]. Serpin polymerization is heat-inducible, and elevated temperatures inactivate most inhibitory serpins due to polymerization [48]. For vaspin, Arg302 and Glu379 seem to affect heat-induced polymerization. Mutants featuring the native Glu379 without Arg302 (R302A and R302E) polymerized more readily than the wild-type. In contrast, E379S-containing mutants as well as mutation of Arg302 did not influence polymerization and were comparable with wild-type vaspin. Together, Glu379 seems to facilitate polymer formation, whereas Arg302 counteracts and appears to stabilize vaspin as a monomer in the native state. On a side note, we were initially surprised to observe polymerization of the D305C/V383C mutant, as this disulfide bond should prevent extended RCL movement probably required for a domain-swap mechanism of polymerization. But recently, Irving et al. [41] have also observed polymerization for a similarly disulfide-stabilized antitrypsin variant (S283C/P361C).

In conclusion, our data reveal Glu379 and Arg302 as important regulatory elements for serpin functionality, such as protease specificity, activity and stability and explain the observed moderate activity for the target protease KLK7. Thereby, RCL P1′ residue Glu379 is the down-regulating element, whereas the exosite at Arg302 enables inhibition of KLK7 despite the obviously inappropriate P1′ residue. The reason underlying this deliberate limitation of vaspin activity by Glu379 has to be further investigated. It probably enhances specificity for KLK7, because inhibition of other proteases preferring similar P1 residues could be even more negatively affected by the presence of this P1′ residue. Vaspin serum concentrations are very low and positively correlated with increasing fat mass, obesity and insulin resistance [4]. An important function of vaspin is its positive effect on insulin action, as application or overexpression in mice improves glucose tolerance [1,6]. We have demonstrated that this effect is dependent on the inhibitory activity of vaspin [8]. Also, vaspin expression follows a circadian rhythm and peak levels of vaspin precede post-prandial insulin secretion peaks [49] and an acute insulin dose in humans rapidly decreases vaspin serum levels [50]. In addition to avoiding potentially adverse influence on non-target protease-mediated processes, a high specificity and prevention of vaspin consumption due to interaction with non-target proteases may thus be of great importance, given its low serum concentrations. Furthermore, heparin was identified as a first activating cofactor of vaspin. A detailed and better understanding of vaspin action and its regulation, the identification of further proteases targeted by vaspin and identification of the substrates of these proteases should increase the understanding of the diverse beneficial effects of vaspin and lead to new pharmacological targets and potentially new treatment strategies for obesity-related metabolic and inflammatory diseases.

AUTHOR CONTRIBUTION

David Ulbricht, Jan Pippel, Stephan Schultz and John Heiker expressed recombinant proteins and performed experiments. Jan Pippel and Norbert Sträter crystallized proteins and carried out crystal structure determinations. René Meier helped with interpretation of data. John Heiker conceived and supervised the project, and wrote the paper with contributions from all authors.

The vaspin expression plasmid was kindly provided by Dr J. Wada (Department of Medicine and Clinical Science, Okayama University Graduate School of Medicine, Okayama, Japan). The substrate peptide used in heparin-activation studies was a generous gift from Professor Dr Maria A. Juliano (Department of Biophysics, Escola Paulista de Medicina, Universidade Federal de São Paulo, São Paulo, Brazil). We thank Antje Keim for helping with crystallization trials. We thank the Joint Berlin MX-Laboratory at BESSY II, Berlin, Germany, for beam time and assistance during synchrotron data collection as well as the Helmholtz Zentrum Berlin for travelling support.

FUNDING

This work was funded by the European Union and the Free State of Saxony (J.T.H.) and supported by grants of the German Research Foundation (DFG) within the Collaborative Research Centre SFB1052 ‘Obesity Mechanisms’ (C4 N.S., C7 J.T.H.).

Abbreviations

     
  • Abz

    o-aminobenzoyl

  •  
  • AT

    antithrombin III

  •  
  • Dnp

    2,4-dinitrophenyl

  •  
  • GAG

    glycosaminoglycan

  •  
  • KLK7

    kallikrein 7

  •  
  • Mca

    (7-methoxycoumarin-4-yl)acetyl

  •  
  • Nva

    norvaline

  •  
  • RCL

    reactive centre loop

  •  
  • RP

    reversed-phase

  •  
  • SI

    stoichiometry of inhibition

  •  
  • SUMO

    small ubiquitin-like modifier

  •  
  • TLS

    Translation–Libration–Screw-rotation

  •  
  • ufh

    unfractionated heparin

References

References
1
Hida
K.
Wada
J.
Eguchi
J.
Zhang
H.
Baba
M.
Seida
A.
Hashimoto
L.
Okada
T.
Yasuhara
A.
Nakatsuka
A.
, et al. 
Visceral adipose tissue-derived serine protease inhibitor: a unique insulin-sensitizing adipocytokine in obesity
Proc. Natl. Acad. Sci. U.S.A.
2005
, vol. 
102
 (pg. 
10610
-
10615
)
[PubMed]
2
Klöting
N.
Berndt
J.
Kralisch
S.
Kovacs
P.
Fasshauer
M.
Schon
M.R.
Stumvoll
M.
Blüher
M.
Vaspin gene expression in human adipose tissue: association with obesity and type 2 diabetes
Biochem. Biophys. Res. Commun.
2006
, vol. 
339
 (pg. 
430
-
436
)
[PubMed]
3
Youn
B.S.
Klöting
N.
Kratzsch
J.
Lee
N.
Park
J.W.
Song
E.S.
Ruschke
K.
Oberbach
A.
Fasshauer
M.
Stumvoll
M.
Blüher
M.
Serum vaspin concentrations in human obesity and type 2 diabetes
Diabetes
2008
, vol. 
57
 (pg. 
372
-
377
)
[PubMed]
4
Blüher
M.
Vaspin in obesity and diabetes: pathophysiological and clinical significance
Endocrine
2012
, vol. 
41
 (pg. 
176
-
182
)
[PubMed]
5
Heiker
J.T.
Vaspin (serpinA12) in obesity, insulin resistance, and inflammation
J. Pept. Sci.
2014
, vol. 
20
 (pg. 
299
-
306
)
[PubMed]
6
Nakatsuka
A.
Wada
J.
Iseda
I.
Teshigawara
S.
Higashio
K.
Murakami
K.
Kanzaki
M.
Inoue
K.
Terami
T.
Katayama
A.
, et al. 
Vaspin is an adipokine ameliorating ER stress in obesity as a ligand for cell-surface GRP78/MTJ-1 complex
Diabetes
2012
, vol. 
61
 (pg. 
2823
-
2832
)
[PubMed]
7
Klöting
N.
Kovacs
P.
Kern
M.
Heiker
J.T.
Fasshauer
M.
Schon
M.R.
Stumvoll
M.
Beck-Sickinger
A.G.
Blüher
M.
Central vaspin administration acutely reduces food intake and has sustained blood glucose-lowering effects
Diabetologia
2011
, vol. 
54
 (pg. 
1819
-
1823
)
[PubMed]
8
Heiker
J.T.
Klöting
N.
Kovacs
P.
Kuettner
E.B.
Sträter
N.
Schultz
S.
Kern
M.
Stumvoll
M.
Blüher
M.
Beck-Sickinger
A.G.
Vaspin inhibits kallikrein 7 by serpin mechanism
Cell. Mol. Life Sci.
2013
, vol. 
70
 (pg. 
2569
-
2583
)
[PubMed]
9
Eissa
A.
Diamandis
E.P.
Human tissue kallikreins as promiscuous modulators of homeostatic skin barrier functions
Biol. Chem.
2008
, vol. 
389
 (pg. 
669
-
680
)
[PubMed]
10
Schultz
S.
Saalbach
A.
Heiker
J.T.
Meier
R.
Zellmann
T.
Simon
J.C.
Beck-Sickinger
A.G.
Proteolytic activation of prochemerin by kallikrein 7 breaks an ionic linkage and results in C-terminal rearrangement
Biochem. J.
2013
, vol. 
452
 (pg. 
271
-
280
)
[PubMed]
11
Saalbach
A.
Vester
K.
Rall
K.
Tremel
J.
Anderegg
U.
Beck-Sickinger
A.G.
Blüher
M.
Simon
J.C.
Vaspin: a link of obesity and psoriasis?
Exp. Dermatol.
2012
, vol. 
21
 (pg. 
309
-
312
)
[PubMed]
12
Hansson
L.
Backman
A.
Ny
A.
Edlund
M.
Ekholm
E.
Ekstrand Hammarstrom
B.
Tornell
J.
Wallbrandt
P.
Wennbo
H.
Egelrud
T.
Epidermal overexpression of stratum corneum chymotryptic enzyme in mice: a model for chronic itchy dermatitis
J. Invest. Dermatol.
2002
, vol. 
118
 (pg. 
444
-
449
)
[PubMed]
13
Ny
A.
Egelrud
T.
Epidermal hyperproliferation and decreased skin barrier function in mice overexpressing stratum corneum chymotryptic enzyme
Acta Derm. Venereol.
2004
, vol. 
84
 (pg. 
18
-
22
)
[PubMed]
14
Psyrri
A.
Kountourakis
P.
Scorilas
A.
Markakis
S.
Camp
R.
Kowalski
D.
Diamandis
E.P.
Dimopoulos
M.A.
Human tissue kallikrein 7, a novel biomarker for advanced ovarian carcinoma using a novel in situ quantitative method of protein expression
Ann. Oncol.
2008
, vol. 
19
 (pg. 
1271
-
1277
)
[PubMed]
15
Talieri
M.
Diamandis
E.P.
Gourgiotis
D.
Mathioudaki
K.
Scorilas
A.
Expression analysis of the human kallikrein 7 (KLK7) in breast tumors: a new potential biomarker for prognosis of breast carcinoma
Thromb. Haemost.
2004
, vol. 
91
 (pg. 
180
-
186
)
[PubMed]
16
Santin
A.D.
Cane
S.
Bellone
S.
Bignotti
E.
Palmieri
M.
De Las Casas
L.E.
Roman
J.J.
Anfossi
S.
O’Brien
T.
Pecorelli
S.
The serine protease stratum corneum chymotryptic enzyme (kallikrein 7) is highly overexpressed in squamous cervical cancer cells
Gynecol. Oncol.
2004
, vol. 
94
 (pg. 
283
-
288
)
[PubMed]
17
Oliveira
J.R.
Bertolin
T.C.
Andrade
D.
Oliveira
L.C.
Kondo
M.Y.
Santos
J.A.
Blaber
M.
Juliano
L.
Severino
B.
Caliendo
G.
, et al. 
Specificity studies on Kallikrein-related peptidase 7 (KLK7) and effects of osmolytes and glycosaminoglycans on its peptidase activity
Biochim. Biophys. Acta
2015
, vol. 
1854
 (pg. 
73
-
83
)
[PubMed]
18
Baglin
T.P.
Carrell
R.W.
Church
F.C.
Esmon
C.T.
Huntington
J.A.
Crystal structures of native and thrombin-complexed heparin cofactor II reveal a multistep allosteric mechanism
Proc. Natl. Acad. Sci. U.S.A.
2002
, vol. 
99
 (pg. 
11079
-
11084
)
[PubMed]
19
Chen
V.C.
Chao
L.
Chao
J.
A positively charged loop on the surface of kallistatin functions to enhance tissue kallikrein inhibition by acting as a secondary binding site for kallikrein
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
40371
-
40377
)
[PubMed]
20
Kabsch
W.
Xds
Acta Crystallogr. D Biol. Crystallogr.
2010
, vol. 
66
 (pg. 
125
-
132
)
[PubMed]
21
Evans
P.R.
Murshudov
G.N.
How good are my data and what is the resolution?
Acta Crystallogr. D Biol. Crystallogr.
2013
, vol. 
69
 (pg. 
1204
-
1214
)
22
Collaborative Computational Project, Number 4
The CCP4 suite: programs for protein crystallography
Acta Crystallogr. D Biol. Crystallogr.
1994
, vol. 
50
 (pg. 
760
-
763
)
[PubMed]
23
Murshudov
G.N.
Skubak
P.
Lebedev
A.A.
Pannu
N.S.
Steiner
R.A.
Nicholls
R.A.
Winn
M.D.
Long
F.
Vagin
A.A.
REFMAC5 for the refinement of macromolecular crystal structures
Acta Crystallogr. D Biol. Crystallogr.
2011
, vol. 
67
 (pg. 
355
-
367
)
[PubMed]
24
Murshudov
G.N.
Vagin
A.A.
Dodson
E.J.
Refinement of macromolecular structures by the maximum-likelihood method
Acta Crystallogr. D Biol. Crystallogr.
1997
, vol. 
53
 (pg. 
240
-
255
)
[PubMed]
25
Reference deleted
[PubMed]
26
Emsley
P.
Lohkamp
B.
Scott
W.G.
Cowtan
K.
Features and development of Coot
Acta Crystallogr. D Biol. Crystallogr.
2010
, vol. 
66
 (pg. 
486
-
501
)
[PubMed]
27
Painter
J.
Merritt
E.A.
Optimal description of a protein structure in terms of multiple groups undergoing TLS motion
Acta Crystallogr. D Biol. Crystallogr.
2006
, vol. 
62
 (pg. 
439
-
450
)
[PubMed]
28
Painter
J.
Merritt
E.A.
TLSMD web server for the generation of multi-group TLS models
J. Appl. Crystallogr.
2006
, vol. 
39
 (pg. 
109
-
111
)
29
Chen
V.B.
Arendall
W.B.
3rd
Headd
J.J.
Keedy
D.A.
Immormino
R.M.
Kapral
G.J.
Murray
L.W.
Richardson
J.S.
Richardson
D.C.
MolProbity: all-atom structure validation for macromolecular crystallography
Acta Crystallogr. D Biol. Crystallogr.
2010
, vol. 
66
 (pg. 
12
-
21
)
[PubMed]
30
Adams
P.D.
Afonine
P.V.
Bunkoczi
G.
Chen
V.B.
Davis
I.W.
Echols
N.
Headd
J.J.
Hung
L.W.
Kapral
G.J.
Grosse-Kunstleve
R.W.
, et al. 
PHENIX: a comprehensive Python-based system for macromolecular structure solution
Acta Crystallogr. D Biol. Crystallogr.
2010
, vol. 
66
 (pg. 
213
-
221
)
[PubMed]
31
Afonine
P.V.
Moriarty
N.W.
Mustyakimov
M.
Sobolev
O.V.
Terwilliger
T.C.
Turk
D.
Urzhumtsev
A.
Adams
P.D.
FEM: feature-enhanced map
Acta Crystallogr. D Biol. Crystallogr.
2015
, vol. 
71
 (pg. 
646
-
666
)
[PubMed]
32
Berman
H.M.
Westbrook
J.
Feng
Z.
Gilliland
G.
Bhat
T.N.
Weissig
H.
Shindyalov
I.N.
Bourne
P.E.
The Protein Data Bank
Nucleic Acids Res.
2000
, vol. 
28
 (pg. 
235
-
242
)
[PubMed]
33
Horvath
A.J.
Irving
J.A.
Rossjohn
J.
Law
R.H.
Bottomley
S.P.
Quinsey
N.S.
Pike
R.N.
Coughlin
P.B.
Whisstock
J.C.
The murine orthologue of human antichymotrypsin: a structural paradigm for clade A3 serpins
J. Biol. Chem.
2005
, vol. 
280
 (pg. 
43168
-
43178
)
[PubMed]
34
Ksiazek
M.
Mizgalska
D.
Enghild
J.J.
Scavenius
C.
Thogersen
I.B.
Potempa
J.
Miropin, a novel bacterial serpin from the periodontopathogen Tannerella forsythia, inhibits a broad range of proteases by using different peptide bonds within the reactive center loop
J. Biol. Chem.
2014
, vol. 
290
 (pg. 
658
-
670
)
[PubMed]
35
Schechter
N.M.
Jordan
L.M.
James
A.M.
Cooperman
B.S.
Wang
Z.M.
Rubin
H.
Reaction of human chymase with reactive site variants of α1-antichymotrypsin: modulation of inhibitor versus substrate properties
J. Biol. Chem.
1993
, vol. 
268
 (pg. 
23626
-
23633
)
[PubMed]
36
Greenfield
N.J.
Methods to estimate the conformation of proteins and polypeptides from circular dichroism data
Anal. Biochem.
1996
, vol. 
235
 (pg. 
1
-
10
)
[PubMed]
37
Gettins
P.G.
Serpin structure, mechanism, and function
Chem. Rev.
2002
, vol. 
102
 (pg. 
4751
-
4804
)
[PubMed]
38
Gettins
P.G.
Olson
S.T.
Exosite determinants of serpin specificity
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
20441
-
20445
)
[PubMed]
39
Chen
V.C.
Chao
L.
Pimenta
D.C.
Bledsoe
G.
Juliano
L.
Chao
J.
Identification of a major heparin-binding site in kallistatin
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
1276
-
1284
)
[PubMed]
40
Huntington
J.A.
Gettins
P.G.
Conformational conversion of antithrombin to a fully activated substrate of factor Xa without need for heparin
Biochemistry
1998
, vol. 
37
 (pg. 
3272
-
3277
)
[PubMed]
41
Irving
J.A.
Haq
I.
Dickens
J.A.
Faull
S.V.
Lomas
D.A.
Altered native stability is the dominant basis for susceptibility of α1-antitrypsin mutants to polymerization
Biochem. J.
2014
, vol. 
460
 (pg. 
103
-
115
)
[PubMed]
42
Irving
J.A.
Cabrita
L.D.
Rossjohn
J.
Pike
R.N.
Bottomley
S.P.
Whisstock
J.C.
The 1.5 Å crystal structure of a prokaryote serpin: controlling conformational change in a heated environment
Structure
2003
, vol. 
11
 (pg. 
387
-
397
)
[PubMed]
43
Bruce
D.
Perry
D.J.
Borg
J.Y.
Carrell
R.W.
Wardell
M.R.
Thromboembolic disease due to thermolabile conformational changes of antithrombin Rouen-VI (187 Asn→Asp)
J. Clin. Invest.
1994
, vol. 
94
 (pg. 
2265
-
2274
)
[PubMed]
44
Mottonen
J.
Strand
A.
Symersky
J.
Sweet
R.M.
Danley
D.E.
Geoghegan
K.F.
Gerard
R.D.
Goldsmith
E.J.
Structural basis of latency in plasminogen activator inhibitor-1
Nature
1992
, vol. 
355
 (pg. 
270
-
273
)
[PubMed]
45
Carrell
R.W.
Stein
P.E.
Fermi
G.
Wardell
M.R.
Biological implications of a 3 Å structure of dimeric antithrombin
Structure
1994
, vol. 
2
 (pg. 
257
-
270
)
[PubMed]
46
Yamasaki
M.
Sendall
T.J.
Pearce
M.C.
Whisstock
J.C.
Huntington
J.A.
Molecular basis of α1-antitrypsin deficiency revealed by the structure of a domain-swapped trimer
EMBO Rep.
2011
, vol. 
12
 (pg. 
1011
-
1017
)
[PubMed]
47
Lomas
D.A.
Evans
D.L.
Finch
J.T.
Carrell
R.W.
The mechanism of Z α1-antitrypsin accumulation in the liver
Nature
1992
, vol. 
357
 (pg. 
605
-
607
)
[PubMed]
48
Patston
P.A.
Hauert
J.
Michaud
M.
Schapira
M.
Formation and properties of C1-inhibitor polymers
FEBS Lett.
1995
, vol. 
368
 (pg. 
401
-
404
)
[PubMed]
49
Jeong
E.
Youn
B.S.
Kim
D.W.
Kim
E.H.
Park
J.W.
Namkoong
C.
Jeong
J.Y.
Yoon
S.Y.
Park
J.Y.
Lee
K.U.
Kim
M.S.
Circadian rhythm of serum vaspin in healthy male volunteers: relation to meals
J. Clin. Endocrinol. Metab.
2010
, vol. 
95
 (pg. 
1869
-
1875
)
[PubMed]
50
Kovacs
P.
Miehle
K.
Sandner
B.
Stumvoll
M.
Blüher
M.
Insulin administration acutely decreases vaspin serum concentrations in humans
Obes. Facts
2013
, vol. 
6
 (pg. 
86
-
88
)
[PubMed]

Supplementary data