An important regulatory mechanism of serine proteases is the proteolytic conversion of the inactive pro-enzyme, or zymogen, into the active enzyme. This activation process is generally considered an irreversible process. In the present study, we demonstrate that an active enzyme can be converted back into its zymogen form. We determined the crystal structure of uPA (urokinase-type plasminogen activator) in complex with an inhibitory antibody, revealing that the antibody ‘rezymogenizes’ already activated uPA. The present study demonstrates a new regulatory mechanism of protease activity, which is also an extreme case of protein allostery. Mechanistically, the antibody binds a single surface-exposed loop, named the autolysis loop, thereby preventing the stabilization of uPA in its active conformation. We argue that this autolysis loop is a key structural element for rezymogenation of other proteases, and will be a new target site for pharmacological intervention with serine protease activity.
- autolysis loop
- crystal structure
Serine proteases cleave peptide bonds and play central roles in a wide range of physiological processes, including food digestion, blood clotting, immune defence, hormone activation and development . Serine proteases are regulated at the transcriptional level by differential expression and at the post-translational level by binding of endogeneous inhibitors and cofactors, and by activation of inactive zymogens into active enzymes. The zymogen to protease transition is governed by endoproteolytic cleavage of a single peptide bond between residues 15 and 16 (chymotrypsinogen numbering), leading to insertion of the newly formed N-terminus into a hydrophobic pocket termed the activation pocket, and formation of a salt bridge between residues 16 and 194 [2,3]. The pioneering structural works by Huber and Bode  revealed that zymogen activation was accompanied by the stabilization of four flexible loop regions, i.e. the activation loop (residues 16–21), the autolysis loop (residues 142–154), the oxyanion-stabilizing loop (residues 184–194) and the S1-entrance frame (residues 216–223), collectively referred to as the activation domain, converting the meta-stable zymogen into a structurally compact and active enzyme. The activation process is generally referred to as an irreversible process [3,5]. In the present paper, we describe a surprising finding that an activated enzyme can be converted back into the inactive zymogen conformation.
Proteins and reagents
The human uPA (urokinase-type plasminogen activator) protease domain (residues Ile16–Glu244, uPA16–244) was expressed in Pichia pastoris yeast strain X-33 and purified as described previously . Human two-chain uPA was purchased from ProSpec–Tany TechnoGene. The monoclonal antibody mAb-112 and the Fab fragment of mAb-112, Fab-112, were produced and purified as described previously . In brief, mAb-112 was isolated from hybridoma cell clones generated from fusion of NSO myeloma cells with spleen cells obtained from mice immunized with recombinant human single-chain pro-uPA. The antibody was purified from conditioned hybridoma cell medium on a Protein G–Sepharose 4FF column. Fab-112 was generated by treating mAb-112 with papain in a 1:100 (w/w) ratio of enzyme to antibody in 15 mM cysteine and 2 mM EDTA for 6 h at 37°C. Fab fragments were purified by Protein A-affinity chromatography in PBS followed by MonoSTM 5/50 GL cation exchange in 25 mM acetate (pH 5.5). PAB (p-aminobenzamidine) was purchased from Sigma–Aldrich. The chromogenic substrate BIOPHEN CS-61(44) (Glu-Gly-Arg-p-nitroanilide) was from Aniara Diagnostica.
The sequences of the variable domains of mAb-112 (Supplementary Figure S1 at http://www.BiochemJ.org/bj/449/bj4490161add.htm) were determined by the following procedure. Total RNA was extracted from a hybridoma cell line expressing mAb-112 using RNeasy minikit (Qiagen). The mRNA was purified from total RNA with beads coated with oligo-dT, using the Oligotex mRNA minikit (Qiagen). cDNA was produced from the purified mRNA using random hexamer primers and First Strand cDNA Synthesis Kit for RT (reverse transcription)–PCR [AMV (avian myeloblastosis virus)] (Roche). The cDNA was purified using a PCR purification kit (Invitrogen), and DNA encoding the variable domains of heavy and light chains was amplified by PCR. The redundant PCR primers were designed to hybridize to consensus sequences found at both ends of the DNA coding for the variable domains. For the light chain variable region, primer sequences 5′-CCGTTTGATTTCCAGCTTGGTGCC-3′, 5′-CCGTTTTATTTCCAGCTTGGTCCC-3′, 5′-CCGTTTTATTTCCAACTTTGTCCC-3′, 5′-CCGTTTCAGCTCACCCAGTCTC-3′ and 5′-GACATTGAGCTCACCCAGTCTC-3′ were used. For the heavy chain variable region, primer sequences 5′-TGAGGAGACGGTGACCGTGGTCCCTTGGCCCC-3′, 5′GAGACGGTGACCGTGGTCCC-3′ and 5′-AGGTSMARCTGCAGSAGTCWGG-3′ were used. The amplified DNA was purified using the QIAquick gel extraction kit (Qiagen), ligated into an SrfI-digested pCR-Script SK(+) vector, and transformed into XL10-gold Escherichia coli cells using the pCR-Script® Amp Cloning Kit (Stratagene). Plasmids from positive clones were sequenced using the ABI PRISM® 310 Genetic Analyser and evaluated with Vector NTI software (Applied Biosystems).
The Fab-112–uPA16–244 complex was prepared in a 5-fold molar ratio excess of uPA. The complex was purified by gel-filtration chromatography in a buffer containing 50 mM Tris/HCl (pH 7.4) and 150 mM NaCl. The presence of both Fab-112 and uPA16–244, and thus their complex, was verified by SDS/PAGE. The fractions were pooled and concentrated to a final concentration of 10 mg/ml.
Crystallization and data collection
The crystallization screens of Fab-112 and Fab-112–uPA16–244 were performed using the sitting-drop vapour-diffusion method in 96-well plates. Typically, 0.75 μl of protein solution (10 mg/ml) was mixed with an equal volume of precipitant solution and equilibrated against 100 μl of the latter at room temperature (25°C). The Fab-112 was crystallized in 2.3 M ammonium sulfate, 100 mM Tris/HCl (pH 7.4) and 5% PEG [poly(ethylene glycol)] 400. The Fab-112–uPA16–244 complex was crystallized in 25% (w/v) PEG2000 MME (monomethyl ether), 100 mM Tris/HCl (pH 8.0) and 0.21 M ammonium sulfate. The crystals of the complex appeared in approximately 2 days. It is important to note that no other precipitation was observed in the crystallization drop, suggesting that the complex has a homogeneous conformation. For X-ray data collection, a single crystal was cryoprotected in the mother liquor supplemented with 20% (v/v) glycerol and flash-frozen in a liquid nitrogen stream at beamline.
The crystallization procedure for unbound uPA16–244 was essentially as described previously for uPA in complex with a peptide inhibitor . In brief, crystals of unbound uPA16–244 were obtained using the sitting-drop method in 2.0 M ammonium sulfate, 50 mM sodium citrate (pH 4.6) and 5% PEG400. For X-ray data collection, a single crystal was cryoprotected in a solution containing 20% (v/v) glycerol, 2.1 M ammonium sulfate, 50 mM sodium citrate (pH 4.6) and 5% PEG400 and flash-frozen in a liquid nitrogen stream.
All data collection was performed at 100 K and a wavelength of 1.0 Å (1 Å=0.1 nm), at the beamline BL17U of the SSRF (Shanghai Synchrotron Radiation Facility). The data were processed and scaled using the HKL2000 suite . The crystal of Fab-112 belongs to the space group P212121 with unit cell dimension of a=71.18 Å, b=86.63 Å and c=155.33 Å. The crystal of Fab-112–uPA16–244 belongs to the same space group as Fab-112 with unit cell dimension of a=38.96 Å, b=94.96 Å and c=154.76 Å. The crystal of unbound uPA belongs to space group R3 with cell dimensions a=121.28 Å, b=121.28 Å and c=42.81 Å. Data collection statistics are reported in Table 1.
Structure determination and refinement
The structures of free uPA16–244, Fab-112 and Fab-112–uPA16–244 complex were all solved by molecular replacement . To obtain the crystal structures of free uPA16–244 and Fab-112, the upain-1–uPA16–244 structure (PDB code 2NWN)  and the structures of Fab light chain of MN20B9.34 anti-P1.4 antibody (PDB code 2BRR)  and Fab heavy chain of anti-β2-adrenoceptor antibody (PDB code 2R4R)  were used as search models. The 2Fo−Fc and Fo−Fc electron-density maps were examined, and the protein model was manually adjusted after each refinement cycle using the molecular graphics program COOT . To determine the structure of the Fab-112–uPA16–244 complex, the crystal structure of Fab-112 was used as search model. The 2Fo−Fc density was weak for several regions of the uPA catalytic domain, notably residues 16–20, 187–190 and 218–223. Therefore these residues were removed entirely from the model or their occupancies were set to zero. Through several cycles of refinement and map inspection, the region 142–154 was rebuilt. The structure was refined with iterative cycles of manual model building using COOT  and restrained refinement with Refmac5 . Solvent molecules were added using an Fo−Fc Fourier difference map contoured at 2.5σ in the final refinement steps. The structures were validated with PROCHECK  and analysed by PyMol (http://www.pymol.org).
Carbamylation of the N-terminal Ile16 of uPA in the absence or presence of mAb-112
To assess the susceptibility of the N-terminal Ile16 to chemical modification by KCNO, two-chain uPA (1 μM) alone or in the presence of mAb-112 (1 μM) or mAb-101 (1 μM) were incubated in assay buffer containing 30 mM Hepes (pH 7.4), 135 mM NaCl, 1 mM EDTA, 0.1% PEG8000 and 0.2 M KCNO. The reactions were started at different time points and subsequently stopped by 1:100 dilution into assay buffer without KCNO and supplemented with 0.1% BSA [15,16]. The diluted samples containing uPA and mAb-112 at concentrations of 10 nM were incubated for 2 h at 37°C to allow dissociation of the uPA–mAb-112 complex. Control experiments without KCNO were conducted in parallel. Residual uPA activity was determined from the hydrolysis of a small chromogenic substrate, BIOPHEN CS-61(44) (1 mM). The experiments were conducted with both full-length uPA and uPA16–244.
Binding of PAB to uPA in the absence or presence of mAb-112
The binding of PAB was determined essentially as described previously . In brief, fluorescence emission spectra were recorded at 25°C on a PTI Quantamaster spectrofluorimeter, equipped with a single monocromator on both the excitation and emission side. The experiments were carried out in a 0.2 cm×1.0 cm quartz cuvette in a buffer containing 30 mM Hepes (pH 7.4), 135 mM NaCl and 1 mM EDTA supplemented with 0.1% PEG8000 to avoid protein adsorption to the quartz cuvette. The total reaction volume was 300 μl. Excitation was at 335 nm and the emission was scanned from 340 to 400 nm. The experiments were conducted over a range of mAb-112 concentrations (0.03–2 μM) together with two-chain uPA (0.4 μM) in the presence of 20 μM PAB. The emission spectrum in the presence of a single concentration of mAb-101 (2 μM) served as a negative control . Emission spectra were collected using an integration of 1–2 s over a 1.0-nm step resolution. The experiments were conducted with both full-length uPA and uPA16–244.
RESULTS AND DISCUSSION
Active two-chain uPA is converted into a zymogen-like conformation in the presence of mAb-112
We developed an inhibitory antibody (mAb-112) against human uPA  by immunization of mice with the single-chain zymogen pro-uPA. The antibody binds pro-uPA with a subnanomolar Kd value. In addition, it also binds to activated two-chain uPA, albeit with much lower affinity (300-fold lower), and is a non-competitive inhibitor of two-chain uPA. The antibody strongly delays activation of single-chain pro-uPA into two-chain active uPA either by stabilizing pro-uPA in its zymogen conformation and/or by sterically hindering access of proteases to the activation bond at Lys15–Ile16. The antibody epitope was mapped by alanine-scanning mutagenesis to the autolysis loop of uPA, and the surface-exposed residues Glu144, Tyr149 and Tyr151 were pinpointed as crucial residues for antibody binding . The in vivo applicability of this antibody for intervention with uPA activity was demonstrated by efficient inhibition of tumour cell intravasation in a chicken embryo chorioallantoric membrane model and in an orthotopic prostate carcinoma mouse model [7,19].
To provide structural information on the mechanism by which the antibody inhibits uPA, we crystallized the Fab fragment of mAb-112, Fab-112 and the protease domain of active uPA (residues 16–244, uPA16–244). It should be pointed out that we were unable to form crystals of the pro-uPA–Fab-112 complex. Instead, we determined the crystal structures of Fab-112, active uPA16–244 and the Fab-112–uPA16–244 complex to a resolution of 1.93, 1.94 and 2.97 Å respectively (Figure 1 and Table 1 and see Supplementary Figure S2 at http://www.BiochemJ.org/bj/449/bj4490161add.htm). Surprisingly, the uPA16–244 in the Fab-112–uPA16–244 complex adopts a conformation very different from the free active uPA16–244 (Figure 2A). The structure of free active uPA16–244 displays an empty specificity pocket (S1 pocket) for substrate binding. In the Fab-112–uPA16–244 structure, this pocket is occluded owing to a 9.7 Å displacement of the peptide segment Trp215–Arg217 and burial of the Arg217 side chain in the pocket (Figure 2B). In addition, the N-terminus is disordered, and the oxyanion hole formed by the backbone amides of Gly193 and Ser195 is not properly organized and shifted by 3.2 Å compared with active uPA16–244 (Figure 2B). These are hallmark features of many zymogen structures, e.g. trypsinogen  and Factor XI . We superimposed further the structures of free and antibody-complexed uPA and showed that the differences were mostly confined to four regions (Figure 2C): the activation loop (residues 16–21), the autolysis loop (residues 142–154), the oxyanion-stabilizing loop (residues 184–194) and the S1 entrance frame (residues 216–223). These regions are ordered in the structure of active free uPA16–244, but are disordered or have undergone major conformational changes in the Fab-112–uPA16–244 complex (Figure 2C). These regions are exactly the regions originally defined as the activation domains . In addition, Asp194 in the Fab-112–uPA16–244 structure is stabilized by hydrogen bonds to Tyr40, Tyr140 and Phe141 (Supplementary Figure S3 at http://www.BiochemJ.org/bj/449/bj4490161add.htm). This hydrogen-bonding network is reminiscent of the interactions involving residues 194, 143, 40 and 32 proposed to stabilize chymotrypsinogen . Together, these structural results demonstrate that uPA in the Fab-112–uPA16–244 complex is in a zymogen conformation and suggest that Fab-112 induces complete conversion of activated uPA into its zymogen.
To confirm further that Fab-112-complexed uPA16–244 is in a zymogen conformation, we measured the effect on uPA activity of KCNO treatment, which is a probe for an exposed N-terminal amino group of Ile16. The rate of activity loss in the presence of KCNO depends on the frequency of solvent exposure of the N-terminal Ile16 because KCNO chemically modifies Ile16 when it is solvent-exposed and prevents reinsertion and the formation of the salt bridge of Ile16 to Asp194. Free active uPA16–244 alone and active uPA16–244 in the presence of a control antibody (mAb-101)  lost negligible activities in response to 2 h of incubation with KCNO. However, in the presence of Fab-112, activated uPA16–244 lost its activity more than 4-fold faster than active uPA16–244 alone (Figure 3A). This observation is in agreement with the X-ray crystal structure analysis, showing that the N-terminus of uPA16–244 is exposed to the solvent in the presence of Fab-112.
We also analysed the effect of Fab-112 on the integrity of the uPA16–244 S1 pocket by measuring the binding of PAB to uPA16–244 in the presence of increasing concentrations of the Fab fragment. PAB is a common probe for analysing the conformational state of the S1 pocket of serine proteases. The fluorescence of PAB increases strongly upon binding in the pocket. As a consequence, any disturbance of the structural integrity of the S1 pocket and an ensuing decreased binding of PAB can be measured as a decrease in fluorescence. We found that Fab-112 displaced PAB from uPA16–244. No displacement of PAB was observed in the presence of a control anti-uPA antibody, mAb-101 , at concentrations up to 2 μM (Figure 3B). This result supports the observations from the Fab-112–uPA16–244 structure, in which the S1 pocket is deformed.
Structural mechanism of rezymogenation of active uPA
The Fab-112–uPA16–244 structure revealed no direct interactions between the uPA active site and Fab-112. Instead, Fab-112 binds solely to the uPA16–244 autolysis loop (residues 142–154) (Figure 4) far away from the active site through a range of hydrogen bonds (Supplementary Table S1 at http://www.BiochemJ.org/bj/449/bj4490161add.htm). This is in agreement with previous alanine-scanning mutagenesis which also identified the autolysis loop as the primary binding site for mAb-112 . The autolysis loop is buried in a binding pocket formed by the antibody CDR (complementarity-determining region) H1, H3 and L2 loops (Figure 4A). Tyr149 of the autolysis loop is stacked between Asp101 of the CDR H3 loop and Tyr49 of the CDR L2 loop (Figure 4B). As a consequence, the autolysis loop is therefore structurally well ordered, as opposed to its disordered appearance in other zymogen structures.
The interactions between Fab-112 and activated uPA16–244 result in the translocation of the autolysis loop away from the core of the protease by a distance of 12.8 Å, as compared with the active unbound uPA16–244 structure (Figure 2A). In the active enzyme, the autolysis loop stabilizes the N-terminus through direct interactions between Ile16 and Phe142 and Gly143 (Figure 4C). Thus the translocation of the autolysis loop induced by the antibody destabilizes the N-terminus, leading to its displacement from the hydrophobic pocket (Supplementary Figure S4 at http://www.BiochemJ.org/bj/449/bj4490161add.htm). Notably, it is the main-chain atoms that mediate these stabilizing interactions, and thus these interactions are conserved in most active serine proteases. In addition, the autolysis loop stabilizes the oxyanion-stabilizing loop in the active uPA16–244 through Gln192 interactions with Gly142, Lys143 and Tyr151. This effect is perturbed by Fab-112, leading to the movement of the oxyanion-stabilizing loop (average 5.8 Å), the occlusion of the activation pocket, the distortion of the oxyanion hole and the dislocation of Asp189 of the S1 pocket. A schematic representation of the inhibitory mechanism of Fab-112 is shown in Figure 4(D).
The movement of the autolysis loop induced by Fab-112 binding pivots on the hinge residues Gly142 and Pro152 on each side of the loop. The dihedral angle ψ of Pro152 undergoes a nearly 180° flip (152.4° to −35.6°) between free and bound uPA16–244. Gly142 is part of a segment (G140XG142, where X=F, Y or W) that is highly conserved in most serine proteases. Pro152 is also conserved in many serine proteases. These results suggest the propensity of flexibility in the autolysis loop for most serine proteases.
Targeting the autolysis loop may be a new general strategy for pharmacological intervention of protease activity
The present study shows that the autolysis loop of uPA is the only important region for binding of Fab-112. The binding rezymogenizes already activated uPA. These results indicate that the autolysis loop is a hotspot region important for the regulation of proteolytic activity of uPA. This conclusion may be generalized to other serine proteases since a flexible autolysis loop and an identical activation mechanism are common features . Thus monoclonal antibodies raised against peptides corresponding to the autolysis loop of serine proteases may be inhibitors acting similarly to mAb-112. The sequence of the autolysis loop among different serine proteases are quite variable ; thus antibodies against the autolysis loop are likely to be specific for the targeted protease.
The autolysis loop may also be involved in natural regulation of protease activity. In the co-crystal structure of zymogen prethrombin-2 in complex with staphylocoagulase, a non-proteolytical zymogen activator, staphylocoagulase binds to the autolysis loop of prethrombin-2 and insert its N-terminal end into the activation pocket of prethrombin-2, thereby triggering an allosteric rearrangement of thrombin zymogen into the ordered and proteolytically active thrombin conformation .
The inactivation of uPA by Fab-112 represents a clear-cut example of allosteric regulation of a serine protease, which is a means of modulating protease activity in a manner which is more specific than targeting directly the active site . Allosteric regulation of proteolytic activities through direct interaction on non-active sites are widely observed . For example, in caspases, two specific compounds bound to caspase 7 at an allosteric site and inhibited caspase 7 by trapping a zymogen-like conformation . However, the complete conversion into zymogen of serine proteases has not been seen before.
Rezymogenation may be a natural mechanism for inhibition of serine protease activity. Thus it has been argued that ‘slow’ thrombin is zymogen-like . Also, the reaction of serine proteases with inhibitors of the serpin family implies some aspects of rezymogenization, for instance the change of the oxyanion hole [28,29].
Longguang Jiang, Kenneth Botkjaer, Lisbeth Andersen and Cai Yuan performed experiments and analysed the results. Peter Andreasen designed the experiments, provided reagents and revised the paper. Mingdong Huang designed the experiments and wrote the paper.
This work was supported by Natural Science Foundation of China [grant numbers 31161130356 and 31170707], Fujian Province [grant number 2012J05071] and the Danish National Research Foundation [grant number 26-331-6].
We thank the Shanghai Synchrotron Radiation Facility beamline BL17U for X-ray data collection. Anni Christensen is thanked for excellent technical assistance.
↵1 All authors are members of the Danish–Chinese Centre for Proteases and Cancer (http://www.proteasesandcancer.org), a collaboration between the State Key Laboratory of Structural Chemistry, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, and the Department of Molecular Biology, Aarhus University.
The atomic co-ordinates of the three kinds of crystal structures have been deposited in the PDB under accession codes 4DVA (uPA16–244), 4DVB (Fab-112) and 4DW2 (Fab-112–uPA16–244).
Abbreviations: CDR, complementarity-determining region; PAB, p-aminobenzamidine; PEG, poly(ethylene glycol); uPA, urokinase-type plasminogen activator
- © The Authors Journal compilation © 2013 Biochemical Society