PRSS3/mesotrypsin is an atypical isoform of trypsin, the up-regulation of which has been implicated in promoting tumour progression. Mesotrypsin inhibitors could potentially provide valuable research tools and novel therapeutics, but small-molecule trypsin inhibitors have low affinity and little selectivity, whereas protein trypsin inhibitors bind poorly and are rapidly degraded by mesotrypsin. In the present study, we use mutagenesis of a mesotrypsin substrate, APPI (amyloid precursor protein Kunitz protease inhibitor domain), and of a poor mesotrypsin inhibitor, BPTI (bovine pancreatic trypsin inhibitor), to dissect mesotrypsin specificity at the key P2′ position. We find that bulky and charged residues strongly disfavour binding, whereas acidic residues facilitate catalysis. Crystal structures of mesotrypsin complexes with BPTI variants provide structural insights into mesotrypsin specificity and inhibition. Through optimization of the P1 and P2′ residues of BPTI, we generate a stable high-affinity mesotrypsin inhibitor with an equilibrium binding constant Ki of 5.9 nM, a >2000-fold improvement in affinity over native BPTI. Using this engineered inhibitor, we demonstrate the efficacy of pharmacological inhibition of mesotrypsin in assays of breast cancer cell malignant growth and pancreatic cancer cell invasion. Although further improvements in inhibitor selectivity will be important before clinical potential can be realized, the results of the present study support the feasibility of engineering protein protease inhibitors of mesotrypsin and highlight their therapeutic potential.

INTRODUCTION

Proteases that are aberrantly expressed in the tumour microenvironment represent key contributors to tumour growth and progression [1], as well as a potentially promising category of drug targets [2,3]. However, a major challenge in the development of therapeutic protease inhibitors has been the identification of sufficiently selective inhibitors, since individual oncogenic proteases typically belong to large families of closely structurally related enzymes, such as the matrix metalloproteinases or the trypsin-like serine proteases [3]. Insufficiently selective drugs are associated with toxicity caused by interference with diverse physiological processes, and may lack efficacy as a result of simultaneously targeting pro- and anti-tumorigenic proteases [4].

Recently, evidence has begun to accumulate suggesting that the trypsin isoform PRSS3/mesotrypsin promotes cancer growth, invasion and progression. Mesotrypsin, encoded by the PRSS3 gene, is produced and secreted as a digestive zymogen by the pancreas [5]; a splice isoform with an alternative exon 1, lacking a classical secretion signal, is expressed in the brain [6,7]. PRSS3 expression has also been reported in a variety of tumours and cancer cell lines, and has been associated with cancer progression [810]. PRSS3/mesotrypsin expression is associated with metastasis and poor survival in NSCLC (non-small cell lung cancer) patients [8] and in pancreatic cancer patients [10]. Mesotrypsin enhanced transendothelial migration of NSCLC cells, suggesting a functional role in metastasis [8], and mesotrypsin promoted pancreatic cancer cell proliferation and invasion in culture models, and tumour progression and metastasis in animal models [10]. We have found PRSS3 expression to be up-regulated with advancing malignancy in a culture model of breast cancer progression, where it contributes to malignant growth [9]. Thus mesotrypsin represents a potential target for cancer therapy, if potent and selective inhibitors can be developed; this effort depends upon defining aspects of binding specificity which differ between mesotrypsin and other trypsin-like serine proteases.

Although mesotrypsin shows high sequence homology with the major digestive trypsins, we and others have found that unique sequence and structural features, most notably Arg193 (in most other serine proteases a highly conserved glycine residue), contribute to distinct specificity and functional properties [1117]. Structural studies reveal that Arg193 contributes to a positively charged electrostatic surface potential in the vicinity of the S2′ subsite [12], and must undergo substantial conformational rearrangements to permit binding of substrates or inhibitors with bulky P2′ residues [15,17]. Consequences include minimal or markedly reduced activity towards specific protein substrates of other trypsins [13], unusual resistance to many polypeptide trypsin inhibitors [11,12,18], the ability to degrade some trypsin inhibitors as substrates [13,16,17], and unusual substrate specificity for small polar residues at the P1′ position [14] and for polypeptide substrates constrained in a canonical conformation [16]. Importantly, the unique structural features of the mesotrypsin active site may also offer the opportunity to develop tailored inhibitors that can selectively target mesotrypsin, for use as probes to study the role of mesotrypsin in cancer models, and perhaps ultimately as cancer therapeutics.

In the present study, we use mutagenesis of a model mesotrypsin substrate to define the unusual P2′ substrate specificity of mesotrypsin. We demonstrate the relevance of optimizing interactions at this site by engineering the poor mesotrypsin inhibitor BPTI (bovine pancreatic trypsin inhibitor) [15] into a high-affinity mesotrypsin inhibitor, and gain structural insights into mesotrypsin substrate specificity and inhibition by solving high-resolution crystal structures of mesotrypsin–inhibitor complexes. We further show that our optimized mesotrypsin inhibitor functions in the pharmacological inhibition of mesotrypsin in physiologically relevant assays of breast cancer cell malignant growth and pancreatic cancer cell invasion.

EXPERIMENTAL

Production of recombinant proteins

Recombinant human mesotrypsinogen, human cationic trypsinogen and human anionic trypsinogen, as well as a catalytically inactive S195A mutant of mesotrypsinogen, were expressed in Escherichia coli, isolated from inclusion bodies, refolded, purified and activated with bovine enteropeptidase as described previously [15,19,20]. Kunitz domain inhibitors were expressed in the methylotrophic yeast Pichia pastoris under control of the AOX1 (alcohol oxidase) promoter using the expression vector pPICZαA (Invitrogen); constructs, expression and purification of APPI (amyloid precursor protein Kunitz protease inhibitor domain)-WT (wild-type), BPTI-WT and several mutant inhibitors has been described previously [16,17,21,22]. Additional mutations were introduced using the QuikChange® kit (Stratagene), and sequence verification and expression screening were conducted as described previously [17].

Inhibition studies

Mesotrypsin, cationic trypsin and anionic trypsin concentrations were quantified by active-site titration using 4-nitrophenyl 4-guanidinobenzoate (Sigma), and APPI and BPTI variant concentrations were determined by titration with bovine trypsin (Sigma), as described previously [15]. Concentrations of the chromogenic substrate Z-GPR-pNA (benzyloxycarbonyl-Gly-Pro-Arg-p-nitroanalide; Sigma) were determined by end-point assay. For determination of mesotrypsin inhibition constants, enzyme assays performed at 37°C in the presence of various concentrations of substrate and inhibitor were followed spectroscopically for 3–5 min, and initial rates were determined from the absorbance increase caused by the release of p-nitroaniline (ϵ410=8480 M−1·cm−1), as described previously [15]. Data were globally fitted by multiple regression to the classic competitive inhibition equation (eqn 1), using GraphPad Prism (GraphPad Software, San Diego, CA, U.S.A.). Reported inhibition constants are average values obtained from multiple independent experiments.

 
formula
(1)

For measurement of inhibition constants with cationic and anionic trypsins, the observation of slow tight-binding behaviour required an alternative kinetic treatment, using methods that we have described previously [15]. Reactions were run at 25°C, and were followed spectroscopically for 16 h so that reliable steady-state rates could be obtained. Inhibition constants were calculated using eqn (2) as described previously [15], where vi and v0 are the steady-state rates in the presence and absence of inhibitor, KM is the Michaelis constant for substrate cleavage, and [S]0 and [I]0 are the initial concentrations of substrate and inhibitor. Calculations were performed using KM values of 36.5 μM for cationic trypsin and 22.6 μM for anionic trypsin, determined from Michaelis–Menten kinetic studies.

 
formula
(2)

Impact of P2′ residues on free energies of association, catalysis and transition-state stabilization

Using the absence of a P2′ sidechain in APPI-M17G as the baseline for comparison of all other variants, and using 1/Ki as an approximation of Ka, the equilibrium association constant, we calculated the change in free energy of association ΔΔGa°(Gly17X) from eqn (3). Changes in the free energy of catalysis ΔΔGcat(Gly17X) and in the transition-state stabilization energy ΔΔGT(Gly17X) were similarly calculated from eqns (4) and (5) respectively.

 
formula
(3)
 
formula
(4)
 
formula
(5)

Inhibitor hydrolysis studies

The depletion of intact APPI and BPTI variants in time-course incubations with active mesotrypsin was monitored by SDS/PAGE and HPLC as described previously [1517]; SDS/PAGE was used to obtain initial qualitative estimates of reaction rates, whereas HPLC was used to quantitatively determine catalysis rates (kcat). Incubations of mesotrypsin with BPTI mutants were carried out in 0.1 M Tris/HCl (pH 8.0) and 1 mM CaCl2 at 37°C; the BPTI variant concentration was 50 μM and mesotrypsin concentration was 1–5 μM. Aliquots for HPLC analysis were withdrawn from BPTI hydrolysis reactions at periodic intervals, adjusted to 6 M urea and 2 mM DTT (dithiothreitol), incubated for 10 min at 37°C, quenched by acidification to pH 1 and then frozen at −20°C until analysed as described previously [15]. Incubations of mesotrypsin with APPI mutants were carried out similarly, except that the APPI concentration was 50 μM and the mesotrypsin concentration was in the range of 10–500 nM. APPI hydrolysis time-point samples were not denatured or reduced; instead samples were quenched immediately by acidification to pH≤1 and then frozen at −20°C until analysed as described previously [16,17]. Enzyme, Kunitz inhibitors and hydrolysis products were resolved by HPLC and the disappearance of intact Kunitz inhibitors over time was quantified by peak integration as described previously [1517]; initial rates were obtained by linear regression using a minimum of five data points within the initial linear phase of the reaction, and not exceeding 50% conversion of intact inhibitor into hydrolysis products. Hydrolysis rates reported for each inhibitor represent the average of two to three independent experiments.

Crystalization of mesotrypsin–BPTI-K15R/R17G and mesotrypsin–BPTI-K15R/R17D variant complexes

The catalytically inactive mesotrypsin-S195A mutant was mixed with BPTI-K15R/R17G or BPTI-K15R/R17D at a 1:1 stoichiometric molar ratio. The heterodimeric complexes were further purified by HiLoad Superdex 75 gel-filtration chromatography (GE Healthcare), exchanged into 10 mM sodium acetate (pH 6.5) and concentrated to ~3 mg/ml. Complexes were crystallized at 22°C in hanging drops, over a reservoir of 25% PEG [poly(ethylene) glycol] 4000, 0.2 M sodium acetate and 100 mM Tris/HCl (pH 8.0); drops (4 μl) were prepared by mixing equal volumes of protein and reservoir solutions. Crystals (0.2 mm×0.4 mm×0.1 mm) appeared within 2 days and grew over the course of 6 days. Crystals were harvested, soaked in a cryoprotectant solution [30% PEG4000, 0.2 M sodium acetate, 100 mM Tris/HCl (pH 8.0) and 15% glycerol] and flash-frozen in liquid nitrogen.

X-ray data collection, structure solution and model refinement

Synchrotron X-ray data were collected from crystals at 100 K using ADSC CCD (charge-coupled device) detectors at beam lines X12-B, X12-C and X25 at the National Synchrotron Light Source, Brookhaven National Laboratory, NY, U.S.A. The software package HKL2000 [23] was used for integrating, scaling and merging the reflection data. The structures were solved by molecular replacement using the program Phaser [24] operated by PHENIX [25], using the mesotrypsin–BPTI-WT complex structure (PDB code 2R9P) [15] as the search model. The successful solution contained one copy of the heterodimeric complex in the asymmetric unit. Cycles of manual rebuilding in COOT [26] were alternated with automated refinement using the refinement module of the PHENIX software suite [27]. A test set comprising 10% of the total reflections was excluded from refinement to allow calculation of the free R factor. Waters, ions and alternative conformations of protein residues were added using COOT [26]. At early stages, TLS refinement was employed, with each protein chain assigned to a separate TLS group. At later stages of refinement, full anisotropic treatment of atomic displacement parameters was employed, and hydrogen atoms were added in the riding positions. All superpositions and structure figures were created using the graphics software PyMOL (http://pymol.sourceforge.net).

Breast cancer cell three-dimensional malignant growth assays

HMT-3522 T4-2 human mammary epithelial cells were cultured as described previously [9]. Three-dimensional cultures in Matrigel™ were established using the ‘on-top’ protocol [28], in which different concentrations of BPTI-WT or BPTI-K15R/R17G inhibitors were added to the upper layer of Matrigel™ and to the culture medium replaced every 2 days, as described previously [9]. After 7 days in culture, cells were photographed and assessed for colony size.

Pancreatic cancer cell invasion assays

Capan-1 human pancreatic cancer cells (a gift from Dr P. Storz, Mayo Clinic, Jacksonville, FL, U.S.A.) were cultured in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% FBS (fetal bovine serum). Cells were transduced with lentiviral short hairpin RNA construct NM_002771.2-454s1c1 targeting human PRSS3, from the MISSION TRC-Hs1.0 library (Sigma), or with a negative control vector containing a short hairpin that does not recognize any human genes, as described previously [9]. After selection with 1 μg/ml puromycin, pooled transductants were split into three parallel cultures for (i) analysis of PRSS3 transcript levels by qRT-PCR (quantitative real-time PCR) as described previously [9], (ii) analysis of mesotrypsin protein levels, and (iii) assessment of cellular invasiveness. Knockdown was assessed at the protein level in cell lysates by Western blotting using a custom rabbit polyclonal antiserum (Cocalico Biologicals) raised against the mesotrypsin peptide acetyl-TQAECKASYPGKITNS-NH2 conjugated to KLH (Keyhole limpet haemocyanin; EZ Biolab). For the transwell invasion assays, cells were suspended in DMEM+0.1% BSA, mixed with BPTI-K15R/R17G as noted, and 1×105 cells/well were used to seed 24-well 8.0 μm transwell inserts (BD Biosciences) coated with 50 μg of Matrigel™ (BD Biosciences). Cells were allowed to invade for 24 h at 37°C and 5% CO2 toward a chemoattractant in the lower chamber comprised of NIH 3T3-conditioned serum-free medium containing 50 μg/ml ascorbic acid and 50 ng/ml SDF-1β (R&D Systems) before methanol fixation, Crystal Violet staining, photographing of filters and counting of cells on the underside of the filter using Image Pro Plus 6.3 software (MediaCybernetics).

RESULTS

The P2′ position plays a critical role in mesotrypsin substrate-binding affinity

Previous structural studies have found that the distinctive Arg193 of mesotrypsin influences the steric contour and electrostatic potential of the substrate-binding site, and interacts closely with the P2′ residue of a bound substrate or inhibitor molecule [12,15,17]. Thus we hypothesized that the identity of the P2′ residue might be a major determinant of mesotrypsin substrate specificity. To investigate this hypothesis, we selected the recombinant Kunitz domain of the amyloid precursor protein (APPI) as a model protein substrate, since we have recently shown that mesotrypsin rapidly and efficiently hydrolyses a specific Arg–Ala peptide bond in APPI [16], and we employed site-directed mutagenesis to replace the native P2′ methionine residue of APPI with a variety of alternative amino acid residues.

We measured binding affinity in competitive inhibition experiments in which the APPI variants were studied as inhibitors of the hydrolysis by mesotrypsin of a competing substrate, the chromogenic tripeptide Z-GPR-pNA. All of the inhibition data were well described by the competitive model, as we have reported previously for the inhibition of mesotrypsin by APPI-WT [16]; values of the equilibrium inhibition constants Ki for all APPI P2′ variants are summarized in Table 1. Because the competitive inhibition equation also describes the enzymatic reaction of one substrate in the presence of a competing substrate, with the condition that Ki must be equivalent to the Michaelis constant Km for the competing substrate [29], the Ki values in Table 1 also represent Km values for cleavage of the APPI variants as mesotrypsin substrates.

Table 1
Strength of association of mesotrypsin with APPI variants

Ki values are means±S.D.

Inhibitor P2′ Ki (M) Relative Ki (fold difference) ΔΔGa°(Gly17X) (kcal/mol) 
APPI-M17G Glycine (4.71±0.13)×10−9 1.0 0.00 
APPI-M17A Alanine (1.71±0.27)×10−8 3.6 0.80 
APPI-WT* Methionine (1.36±0.19)×10−7 28.9 2.07 
APPI-M17F Phenylalanine (1.26±0.11)×10−7 26.8 2.02 
APPI-M17Y Tyrosine (5.33±0.32)×10−8 11.3 1.49 
APPI-M17R** Arginine (9.20±0.99)×10−7 195.3 3.25 
APPI-M17D Aspartate (3.46±0.18)×10−7 73.5 2.65 
APPI-M17E Glutamate (3.09±0.18)×10−7 65.5 2.58 
Inhibitor P2′ Ki (M) Relative Ki (fold difference) ΔΔGa°(Gly17X) (kcal/mol) 
APPI-M17G Glycine (4.71±0.13)×10−9 1.0 0.00 
APPI-M17A Alanine (1.71±0.27)×10−8 3.6 0.80 
APPI-WT* Methionine (1.36±0.19)×10−7 28.9 2.07 
APPI-M17F Phenylalanine (1.26±0.11)×10−7 26.8 2.02 
APPI-M17Y Tyrosine (5.33±0.32)×10−8 11.3 1.49 
APPI-M17R** Arginine (9.20±0.99)×10−7 195.3 3.25 
APPI-M17D Aspartate (3.46±0.18)×10−7 73.5 2.65 
APPI-M17E Glutamate (3.09±0.18)×10−7 65.5 2.58 
*

Kinetic constants reported previously in [16].

**

Kinetic constants reported previously in [17].

We found that replacement of the bulky hydrophobic methionine residue, which occupies the P2′ position of APPI-WT, with small (glycine, alanine), aromatic (phenylalanine, tyrosine), basic (arginine) or acidic (aspartate, glutamate) amino acids produced a spectrum of Ki values spanning over two orders of magnitude, illustrating the importance of this position in mesotrypsin-binding discrimination. The tightest binding of the APPI variants tested was APPI-M17G, which completely lacks a side chain at the P2′ position and possesses a Ki value toward mesotrypsin of 4.7 nM. We then considered APPI-M17G as a baseline for assessing the impact of the various other P2′ side chains on mesotrypsin association. Using 1/Ki as an approximation of Ka, the equilibrium association constant, we calculated the change in free energy of association ΔΔGa°(Gly17X); these ΔΔGa° values are listed in Table 1.

It is apparent that all P2′ side chains are deleterious for binding by comparison with glycine. When ΔΔGa°(Gly17X) values are plotted against the number of non-hydrogen atoms in each side chain [30,31], there is a clear trend towards weaker binding (larger ΔΔGa°) with increasing steric bulk of the P2′ residue (Figure 1A). Side-chain charge also appears to contribute to a deleterious effect on binding, since variants possessing acidic (aspartate, glutamate) or basic (arginine) side chains each display higher ΔΔGa° than those featuring uncharged residues of similar size.

Impact of P2′ side-chain size on free energies of association, catalysis and transition-state stabilization for the mesotrypsin interaction with APPI

Figure 1
Impact of P2′ side-chain size on free energies of association, catalysis and transition-state stabilization for the mesotrypsin interaction with APPI

The number of non-hydrogen atoms in the P2′ side chain is plotted against the change in free energy of association relative to P2′ glycine ΔΔGa°(Gly17X) (A), the change in free energy of catalysis catalysis ΔΔGcat(Gly17X) (B) and the change in the transition-state stabilization energy ΔΔGT(Gly17X) (C); P2′ residues associated with each data point are indicated using the one-letter amino acid code.

Figure 1
Impact of P2′ side-chain size on free energies of association, catalysis and transition-state stabilization for the mesotrypsin interaction with APPI

The number of non-hydrogen atoms in the P2′ side chain is plotted against the change in free energy of association relative to P2′ glycine ΔΔGa°(Gly17X) (A), the change in free energy of catalysis catalysis ΔΔGcat(Gly17X) (B) and the change in the transition-state stabilization energy ΔΔGT(Gly17X) (C); P2′ residues associated with each data point are indicated using the one-letter amino acid code.

Acidic residues at the P2′ position accelerate mesotrypsin-catalysed hydrolysis

We next used HPLC assays to directly quantify cleavage of APPI variants by mesotrypsin in time-course incubations [16,17]. Because hydrolysis studies used APPI concentrations >50-fold higher than Km values, mesotrypsin-binding capacity was saturated and observed rates of hydrolysis approximate true catalytic rate constants (kcat). Values of kcat and the specificity constant kcat/Km for each APPI variant considered as a substrate for mesotrypsin are summarized in Table 2. We again used APPI-M17G as a baseline for comparison, and calculated changes in the free energy of catalysis ΔΔGcat(Gly17X) and in transition-state stabilization energy ΔΔGT(Gly17X) reflecting contributions to catalysis of each non-glycine P2′ residue tested.

Table 2
Catalytic cleavage of APPI variants by mesotrypsin

kcat and kcat/Km values are means±S.D.

Inhibitor P2′ kcat (s−1Relative kcat (fold difference) ΔΔGcat (kcal/mol) kcat/Km (s−1·M−1kcat/Km (fold difference) ΔΔGT (kcal/mol) 
APPI-M17G Glycine (2.22±0.13)×10−2 1.0 0.00 (4.71±0.31)×106 1.00 0.00 
APPI-M17A Alanine (2.45±0.14)×10−2 1.1 −0.06 (1.43±0.24)×106 0.30 0.73 
APPI-WT* Methionine (4.32±0.40)×10−2 1.9 −0.41 (3.18±0.53)×105 0.067 1.66 
APPI-M17F Phenylalanine (3.44±0.25)×10−2 1.6 −0.27 (2.73±0.31)×105 0.058 1.75 
APPI-M17Y Tyrosine (1.57±0.07)×10−2 0.71 0.21 (2.95±0.22)×105 0.063 1.71 
APPI-M17R** Arginine (5.36±0.16)×10−2 2.4 −0.54 (5.83±0.65)×104 0.012 2.71 
APPI-M17D Aspartate (7.64±0.76)×10−1 34.4 −2.18 (2.21±0.25)×106 0.47 0.47 
APPI-M17E Glutamate (6.72±0.70)×10−1 30.3 −2.10 (2.18±0.26)×106 0.46 0.48 
Inhibitor P2′ kcat (s−1Relative kcat (fold difference) ΔΔGcat (kcal/mol) kcat/Km (s−1·M−1kcat/Km (fold difference) ΔΔGT (kcal/mol) 
APPI-M17G Glycine (2.22±0.13)×10−2 1.0 0.00 (4.71±0.31)×106 1.00 0.00 
APPI-M17A Alanine (2.45±0.14)×10−2 1.1 −0.06 (1.43±0.24)×106 0.30 0.73 
APPI-WT* Methionine (4.32±0.40)×10−2 1.9 −0.41 (3.18±0.53)×105 0.067 1.66 
APPI-M17F Phenylalanine (3.44±0.25)×10−2 1.6 −0.27 (2.73±0.31)×105 0.058 1.75 
APPI-M17Y Tyrosine (1.57±0.07)×10−2 0.71 0.21 (2.95±0.22)×105 0.063 1.71 
APPI-M17R** Arginine (5.36±0.16)×10−2 2.4 −0.54 (5.83±0.65)×104 0.012 2.71 
APPI-M17D Aspartate (7.64±0.76)×10−1 34.4 −2.18 (2.21±0.25)×106 0.47 0.47 
APPI-M17E Glutamate (6.72±0.70)×10−1 30.3 −2.10 (2.18±0.26)×106 0.46 0.48 
*

Kinetic constants reported previously in [16].

**

Kinetic constants reported previously in [17].

Mutations at the P2′ position generally appear to have very little impact on mesotrypsin catalytic rate constants, irrespective of residue size (Figure 1B); the only exceptions noted were for replacement of the P2′ amino acid with the acidic residues aspartate or glutamate, which both accelerated kcat by ~30-fold, contributing ~−2 kcal/mol (1 kcal=4.184 kJ) to the free energy of catalysis. Since most P2′ residues had a substantial impact on Km, but a negligible effect on kcat, the impact on the specificity constant kcat/Km is dominated by Km effects, and a plot of ΔΔGT(Gly17X) shows positive (deleterious) changes in transition-state stabilization energy trending upward with residue size (Figure 1C). Aspartate and glutamate diverge from this pattern; for these residues, the deleterious effect of the acidic side chain on binding is substantially offset by the favourable impact on catalytic rate.

Optimization of the P2′ residue converts BPTI into a potent mesotrypsin inhibitor

We have found previously that although APPI is a rapidly cleaved substrate of mesotrypsin [16], the structurally homologous trypsin inhibitor BPTI is a slowly hydrolysed weak inhibitor of mesotrypsin (Ki=14 μM) [15,17]. We hypothesized that P2′ mutations characterized in the context of APPI would have parallel effects on the binding affinity and vulnerability to mesotrypsin hydrolysis of BPTI, and that this information might be used to engineer BPTI as a more potent inhibitor of mesotrypsin. Starting with the BPTI-K15R variant, which possesses a P1 arginine residue and binds 6-fold more tightly to mesotrypsin than BPTI-WT [17], we introduced either glycine or aspartate at the P2′ position. These new mutants, along with the previously reported BPTI-K15R and BPTI-K15R/R17M variants [17], allow us to compare the impact of an absent (glycine), hydrophobic (methionine), basic (arginine) or acidic (aspartate) side chain at the BPTI P2′ position. For each variant, Ki was obtained from competitive inhibition experiments and kcat was derived from HPLC-based hydrolysis studies as described previously [15,17]. Similarly to our studies described above with APPI, we have used the BPTI variant with P2′ glycine as a baseline for comparison, allowing evaluation of the energetic contributions of methionine, arginine or aspartate side chains to mesotrypsin binding and catalysis.

Comparing the Ki values for mesotrypsin obtained with the BPTI variants (Table 3), it is apparent that as in the case of the substrate APPI, all non-glycine residues at the P2′ position tested proved deleterious for binding. The rank order of BPTI variants ranked according to Ki matched that observed for the APPI variants: glycine showed the tightest association, followed by methionine, aspartate and then arginine, suggesting that for BPTI as for APPI, the bulk and charge of the P2′ residue were important determinants of binding specificity. Mutation of the P2′ residue to glycine in BPTI-K15R/R17G produced the tightest-binding inhibitor of mesotrypsin reported to date, with a Ki value of 5.9 nM, over 2000-fold lower than that of BPTI-WT, and over 400-fold lower than that of BPTI-K15R mutated only at the P1 position, another strong determinant of mesotrypsin-binding specificity [17].

Table 3
Strength of association of mesotrypsin with BPTI variants

Ki values are means±S.D.

Inhibitor P2′ Ki (M) Relative Ki (fold difference) ΔΔGa°(Gly17X) (kcal/mol) 
BPTI-K15R/R17G Glycine (5.9±1.7)×10−9 1.0 0.00 
BPTI-K15R/R17M* Methionine (2.18±0.01)×10−7 36.7 2.22 
BPTI-K15R* Arginine (2.38±0.20)×10−6 401.6 3.69 
BPTI-K15R/R17D Aspartate (1.50±0.09)×10−6 252.4 3.41 
Inhibitor P2′ Ki (M) Relative Ki (fold difference) ΔΔGa°(Gly17X) (kcal/mol) 
BPTI-K15R/R17G Glycine (5.9±1.7)×10−9 1.0 0.00 
BPTI-K15R/R17M* Methionine (2.18±0.01)×10−7 36.7 2.22 
BPTI-K15R* Arginine (2.38±0.20)×10−6 401.6 3.69 
BPTI-K15R/R17D Aspartate (1.50±0.09)×10−6 252.4 3.41 
*

Kinetic constants reported previously in [17].

An effective polypeptide inhibitor must resist inactivation and degradation through proteolytic cleavage as well as compete effectively with substrates for binding to the active site of the protease. The kcat values for mesotrypsin cleavage of BPTI variants (Table 4) show that the identity of the P2′ residue has a significant impact on the proteolytic stability or vulnerability to mesotrypsin of BPTI. As in the Ki comparisons, the rank orders of corresponding BPTI and APPI variants ranked according to kcat remain the same, with P2′ glycine the most resistant to proteolysis followed by methionine, arginine and then aspartate, but for the BPTI variants the impact of P2′ substitutions on kcat is magnified. In particular, the P2′ aspartate variant BPTI-K15R/R17D displays a 90-fold enhanced catalytic rate relative to the P2′ glycine variant. Notably, the P2′ glycine variant is the best inhibitor both from the viewpoint of binding affinity and from the viewpoint of proteolytic stability to mesotrypsin. Table 4 also reports kcat/Km substrate-specificity constants for each BPTI variant considered as a substrate rather than an inhibitor of mesotrypsin; all of the BPTI variants feature specificity constants two to three orders of magnitude lower than those seen for the corresponding APPI variants in Table 2 as a consequence of their much slower rates of cleavage by mesotrypsin; this resistance to proteolysis is an attribute conferred by the BPTI scaffold [17].

Table 4
Catalytic cleavage of BPTI variants by mesotrypsin

kcat and kcat/Km values are means±S.D.

Inhibitor P2′ kcat (s−1Relative kcat (fold difference) ΔΔGcat (kcal/mol) kcat/Km (s−1M−1kcat/Km (fold difference) ΔΔGT (kcal/mol) 
BPTI-K15R/R17G Glycine (5.09±0.88)×10−5 1.0 0.00 (8.6±2.9)×103 1.00 0.00 
BPTI-K15R/R17M* Methionine (5.14±0.23)×10−4 10.1 −1.42 (2.36±0.11)×103 0.28 0.79 
BPTI-K15R* Arginine (8.7±2.7)×10−4 17.0 −1.75 (3.6±1.2)×102 0.042 1.95 
BPTI-K15R/R17D Aspartate (4.62±0.17)×10−3 90.8 −2.78 (3.09±0.21)×103 0.36 0.63 
Inhibitor P2′ kcat (s−1Relative kcat (fold difference) ΔΔGcat (kcal/mol) kcat/Km (s−1M−1kcat/Km (fold difference) ΔΔGT (kcal/mol) 
BPTI-K15R/R17G Glycine (5.09±0.88)×10−5 1.0 0.00 (8.6±2.9)×103 1.00 0.00 
BPTI-K15R/R17M* Methionine (5.14±0.23)×10−4 10.1 −1.42 (2.36±0.11)×103 0.28 0.79 
BPTI-K15R* Arginine (8.7±2.7)×10−4 17.0 −1.75 (3.6±1.2)×102 0.042 1.95 
BPTI-K15R/R17D Aspartate (4.62±0.17)×10−3 90.8 −2.78 (3.09±0.21)×103 0.36 0.63 
*

Kinetic constants reported previously in [17].

Optimized BPTI variant retains a strong affinity towards other trypsins

To assess the potential binding selectivity of BPTI-K15R/R17G toward mesotrypsin relative to other trypsins, we also measured the Ki values for BPTI-WT and BPTI-K15R/R17G against recombinant human cationic trypsin and human anionic trypsin. Because the association of BPTI with these trypsins follows a slow tight-binding model [15], it was necessary to use an alternative approach to the measurement of Ki values as described in the Experimental section. As shown in Table 5, both inhibitors bind substantially more tightly to cationic and anionic trypsin than to mesotrypsin. However, although the dual mutations introduced in BPTI-K15R/R17G have improved the Ki value towards mesotrypsin by a factor of 2277, they have weakened binding toward cationic and anionic trypsin by a factor of 1.6–1.8. Thus, although BPTI-WT shows strong selectivity towards both cationic and anionic trypsins in preference to mesotrypsin by a gap of nearly six orders of magnitude, for BPTI-K15R/R17G this gap has been narrowed to a factor of 180–200. BPTI-K15R/R17G appears to be a relatively non-selective inhibitor with an unusual capacity to potently inhibit mesotrypsin in addition to other trypsins.

Table 5
Inhibition of human trypsin isoforms by BPTI and BPTI-K15R/R17G

Ki values are means±S.D.

 BPTI-WT BPTI-K15R/R17G  
Trypsin isoform Ki (M) Selectivity index* Ki (M) Selectivity index* Fold change 
Mesotrypsin (1.4±0.2)×10−5 (5.9±1.7)×10−9 2277 
Cationic trypsin (2.0±0.1)×10−11** 1.5×10−6 (3.3±0.1)×10−11 5.6×10−3 0.61 
Anionic trypsin (1.7±0.2)×10−11 1.2×10−6 (3.0±0.3)×10−11 5.1×10−3 0.55 
 BPTI-WT BPTI-K15R/R17G  
Trypsin isoform Ki (M) Selectivity index* Ki (M) Selectivity index* Fold change 
Mesotrypsin (1.4±0.2)×10−5 (5.9±1.7)×10−9 2277 
Cationic trypsin (2.0±0.1)×10−11** 1.5×10−6 (3.3±0.1)×10−11 5.6×10−3 0.61 
Anionic trypsin (1.7±0.2)×10−11 1.2×10−6 (3.0±0.3)×10−11 5.1×10−3 0.55 
*

Selectivity Index (SI)=Ki/Ki(mesotrypsin)

**

Ki reported previously in [15].

Structures of mesotrypsin in complex with BPTI P2′ variants reveal multiple conformations of Arg193

To determine why the P2′ residue plays such a critical role in inhibitor binding and hydrolysis by mesotrypsin, we crystallized and solved X-ray structures for the mesotrypsin–BPTI-K15R/R17G and mesotrypsin–BPTI-K15R/R17D complexes, and compared these with the previously published crystal structure of mesotrypsin bound to BPTI-WT (PDB code 2R9P) [15]. To avoid heterogeneity associated with BPTI proteolysis, we used an inactive mesotrypsin-S195A mutant. Diffraction data were measured to 1.6 and 1.3 Å (1 Å=0.1 nm) resolution for mesotrypsin–BPTI-K15R/R17G and mesotrypsin–BPTI-K15R/R17D complexes respectively. Both crystals belong to the space group P21 and exhibit very similar unit cell parameters, with a single heterodimer in the asymmetric unit. Both structures were solved by molecular replacement; Table 6 summarizes the crystal, data collection and refinement statistics.

Table 6
Data collection and refinement statistics for mesotrypsin–BPTI complexes

ASU, asymmetric unit; R.m.s.d., root mean square deviation.

Parameter Mesotrypsin–BPTI-K15R/R17G Mesotrypsin–BPTI-K15R/R17D 
PDB code 3P92 3P95 
Complexes per ASU 
Space group P21 P21 
Unit cell (Å) 44.3, 39.2, 68.8 43.9, 39.1, 68.5 
 90°, 100.3°, 90° 90°, 100.1°, 90° 
Resolution range (Å) 33.85–1.60 33.82–1.30 
Unique reflections 27579 53441 
Completeness (%) 88.9 (46.7*) 94.4 (65.0**) 
Multiplicity 6.5 (3.8) 6.3 (3.8) 
I/S.D. 39.8 (18.4) 17.1 (6.5) 
Rsym 0.032 (0.081) 0.068 (0.227) 
Rcryst/Rfree 11.13/15.86 11.05/13.19 
R.m.s.d. bonds (Å) 0.007 0.016 
R.m.s.d. angles (°) 0.907 1.335 
Parameter Mesotrypsin–BPTI-K15R/R17G Mesotrypsin–BPTI-K15R/R17D 
PDB code 3P92 3P95 
Complexes per ASU 
Space group P21 P21 
Unit cell (Å) 44.3, 39.2, 68.8 43.9, 39.1, 68.5 
 90°, 100.3°, 90° 90°, 100.1°, 90° 
Resolution range (Å) 33.85–1.60 33.82–1.30 
Unique reflections 27579 53441 
Completeness (%) 88.9 (46.7*) 94.4 (65.0**) 
Multiplicity 6.5 (3.8) 6.3 (3.8) 
I/S.D. 39.8 (18.4) 17.1 (6.5) 
Rsym 0.032 (0.081) 0.068 (0.227) 
Rcryst/Rfree 11.13/15.86 11.05/13.19 
R.m.s.d. bonds (Å) 0.007 0.016 
R.m.s.d. angles (°) 0.907 1.335 
*

Completeness at 1.85 Å is 95.1%.

**

Completeness at 1.43 Å is 97.1%.

In comparing the structures of the BPTI-K15R/R17G, BPTI-K15R/R17D and BPTI-WT complexes, we find that although each model features BPTI similarly bound in the canonical fashion at the mesotrypsin active site, the different P2′ residues show significant differences in interface topology (Figures 2A–2C). In the mesotrypsin–BPTI-K15R/R17G structure, Arg193 protrudes downward, enveloping the BPTI Gly17 backbone (Figure 2A). In this conformation, Arg193 has 96.4 Å2 of ASA (accessible surface area), 65.5 Å2 (68%) of which becomes buried by contact with BPTI-K15R/R17G (calculations from the PDBePISA server [32]). By contrast, in the mesotrypsin–BPTI-WT complex, the presence of the BPTI P2′ arginine residue displaces Arg193 to a position buried between the two β-barrel domains of the enzyme; in this conformation, Arg193 has only 48.9 Å2 ASA, 30.2 Å2 (62%) of which becomes buried by contact with BPTI (Figure 2B). The mesotrypsin–BPTI-K15R/R17D complex is intermediate between these two extremes; Arg193 adopts yet a third distinct conformation, with 62.3 Å2 ASA, 30.8 Å2 (49%) of which becomes buried by contact with BPTI (Figure 2C).

Distinct conformations of mesotrypsin Arg193 shaped by alternative P2′ residues of bound BPTI

Figure 2
Distinct conformations of mesotrypsin Arg193 shaped by alternative P2′ residues of bound BPTI

Crystal structures of mesotrypsin in complex with BPTI-K15R/R17G (A), BPTI-WT (B) and BPTI-K15R/R17D (C) show globally similar complexes (left-hand panels) with significant differences in interface topology in the vicinity of Arg193 (surface shown in colour). (A) In the BPTI-K15R/R17G complex, Arg193 extends downwards, enveloping BPTI Gly17. In the BPTI-WT complex, Arg193 recedes into a crevice on the surface of the enzyme (B), whereas a distinct intermediate conformation of Arg193 is found in the BPTI-K15R/R17D complex (C). 2FoFc density maps (contoured at 1.5 σ) reveal well-ordered side chains for Arg193 in the BPTI-K15R/R17G complex (D) and for both Arg193 and the P2′ arginine in the BPTI-WT complex (E); Arg193 and the P2′ aspartate side chain are less well defined in the BPTI-K15R/R17D complex despite the higher resolution of this structure.

Figure 2
Distinct conformations of mesotrypsin Arg193 shaped by alternative P2′ residues of bound BPTI

Crystal structures of mesotrypsin in complex with BPTI-K15R/R17G (A), BPTI-WT (B) and BPTI-K15R/R17D (C) show globally similar complexes (left-hand panels) with significant differences in interface topology in the vicinity of Arg193 (surface shown in colour). (A) In the BPTI-K15R/R17G complex, Arg193 extends downwards, enveloping BPTI Gly17. In the BPTI-WT complex, Arg193 recedes into a crevice on the surface of the enzyme (B), whereas a distinct intermediate conformation of Arg193 is found in the BPTI-K15R/R17D complex (C). 2FoFc density maps (contoured at 1.5 σ) reveal well-ordered side chains for Arg193 in the BPTI-K15R/R17G complex (D) and for both Arg193 and the P2′ arginine in the BPTI-WT complex (E); Arg193 and the P2′ aspartate side chain are less well defined in the BPTI-K15R/R17D complex despite the higher resolution of this structure.

The discrete conformations of Arg193 found in the mesotrypsin–BPTI-K15R/R17G structure and in the mesotrypsin–BPTI-WT structure are well-ordered, as shown by strong electron density for the entire side chain (Figures 2D and 2E). The P2′ arginine residue of BPTI-WT is also well-ordered (Figure 2E). By contrast, the intermediate position adopted by Arg193 in the mesotrypsin–BPTI-K15R/R17D structure, as well as the position of the P2′ Asp17 residue of BPTI, appear to be somewhat less stable and less well-ordered, as both Arg193 and Asp17 show weaker electron density despite the higher resolution of this 1.3 Å structure (Figure 2F). While Nϵ of Arg193 and Oδ2 of Asp17 are separated by only 2.75 Å in the model, the weak electron density of these side chains, and the presence of multiple weak water peaks as alternative hydrogen-bonding partners to Asp17 Oδ2 (Figure 2F), suggest that the interaction between these residues may be weak and transient in nature. This structural interpretation is consistent with our biochemical data as we found that BPTI-K15R/R17D exhibited only marginally better mesotrypsin affinity than BPTI-K15R, which features positively charged Arg17 at the P2′ position (Table 3). Thus the Arg193–Asp17 electrostatic interaction appears to contribute minimally to binding affinity.

Further insights into the differential mesotrypsin affinities of BPTI P2′ variants come from superimposing the structures on to the previously reported structure of mesotrypsin bound to benzamidine (PDB code 1H4W) [12]. Benzamidine is a small molecule that fills only the trypsin-specificity pocket occupied by the P1 arginine or lysine side chain of a substrate or polypeptide inhibitor; the mesotrypsin–benzamidine complex is expected to closely approximate the free enzyme. The superimpositions reveal that very little adjustment is required of the mesotrypsin active site in order to accommodate BPTI-K15R/R17G binding; thus it appears that enzyme and inhibitor are preconfigured for optimal complementarity, and that their interaction resembles a lock-and-key-type molecular recognition (Figure 3A). By contrast, conformational changes involving displacement of Arg193 in order to avoid steric clash with P2′ arginine or aspartate residues are required for binding to BPTI-WT (Figure 3B) or BPTI-K15R/R17D (Figure 3C). The binding data, showing 250–400-fold weaker binding for BPTI variants substituting the P2′ aspartate or arginine residue for glycine (Table 3), suggest that the reorganization of the enzyme active site required to accommodate the conformational shift in Arg193 confers an energetic penalty.

Conformational changes of mesotrypsin upon inhibitor binding

Figure 3
Conformational changes of mesotrypsin upon inhibitor binding

Superimposition of mesotrypsin–BPTI structures with the mesotrypsin–benzamidine structure, PDB code 1H4W (mesotrypsin, beige; benzamidine, red), in which the primed-side subsites are unfilled, reveals the conformational rearrangements required of mesotrypsin Arg193 upon BPTI binding. Only minor adjustments of Arg193 are observed in the mesotrypsin–BPTI-K15R/R17G complex (A), whereas Arg193 is shifted upward by ~6 Å in the mesotrypsin–BPTI-WT structure (B) and by ~3.5 Å in the mesotrypsin–BPTI-K15R/R17D structure (C).

Figure 3
Conformational changes of mesotrypsin upon inhibitor binding

Superimposition of mesotrypsin–BPTI structures with the mesotrypsin–benzamidine structure, PDB code 1H4W (mesotrypsin, beige; benzamidine, red), in which the primed-side subsites are unfilled, reveals the conformational rearrangements required of mesotrypsin Arg193 upon BPTI binding. Only minor adjustments of Arg193 are observed in the mesotrypsin–BPTI-K15R/R17G complex (A), whereas Arg193 is shifted upward by ~6 Å in the mesotrypsin–BPTI-WT structure (B) and by ~3.5 Å in the mesotrypsin–BPTI-K15R/R17D structure (C).

BPTI-K15R/R17G inhibits cancer cell malignant growth and invasion

Since mesotrypsin has been functionally implicated in cancer cell malignant growth and invasion [810], it may offer a target for cancer therapy. Although a clinically useful mesotrypsin inhibitor would probably require greater selectivity towards mesotrypsin than is possessed by BPTI-K15R/R17G, the high mesotrypsin affinity of this inhibitor (Ki=5.9 nM) may allow for its use in studies evaluating the impact of pharmacological inhibition on malignant cancer cell phenotypes conferred by mesotrypsin activity. We have shown previously that in the HMT-3522 breast cancer progression cell series, up-regulation of mesotrypsin contributes to loss of basal polarity and increased proliferation, and that suppression of PRSS3 expression in malignant HMT-3522 T4-2 cells inhibited disorganized proliferation and restored basal polarity and acinar morphology in a physiologically relevant three-dimensional culture model [9]. Using this model to evaluate the effect of BPTI-WT compared with BPTI-K15R/R17G on malignant growth, we found a concentration-dependent reduction in colony size with both inhibitors; notably, the BPTI-K15R/R17G mutant was at least 10-fold more potent, since colony growth in the presence of 100 nM BPTI-K15R/R17G was similar to that observed in cultures treated with 1000 nM BPTI-WT (Figure 4).

Effect of BPTI-K15R/R17G on malignant growth of T4-2 breast cancer cells in three-dimensional culture

Figure 4
Effect of BPTI-K15R/R17G on malignant growth of T4-2 breast cancer cells in three-dimensional culture

Cells treated with 100–1000 nM BPTI-WT, BPTI-K15R/R17G or with buffer only (control) were grown in Matrigel™ for 7 days, then photographed and assessed for colony size. (A) Inhibitor concentration-dependent reduction in colony size relative to control cultures is more pronounced for the BPTI-K15R/R17G variant. Scale bar=100 μm. (B) Quantification of colony size revealed significant inhibition of colony growth with BPTI-WT treatment (◇) at 500 nM (P<0.001, unpaired Student's t test) and 1000 nM concentration (P<0.0001, unpaired t test), and highly significant growth inhibition with BPTI-K15R/R17G treatment (●) at all concentrations tested (P<0.0001, unpaired Student's t test), relative to control cells. Values are means±S.E.M. for morphometry of 30 colonies per condition.

Figure 4
Effect of BPTI-K15R/R17G on malignant growth of T4-2 breast cancer cells in three-dimensional culture

Cells treated with 100–1000 nM BPTI-WT, BPTI-K15R/R17G or with buffer only (control) were grown in Matrigel™ for 7 days, then photographed and assessed for colony size. (A) Inhibitor concentration-dependent reduction in colony size relative to control cultures is more pronounced for the BPTI-K15R/R17G variant. Scale bar=100 μm. (B) Quantification of colony size revealed significant inhibition of colony growth with BPTI-WT treatment (◇) at 500 nM (P<0.001, unpaired Student's t test) and 1000 nM concentration (P<0.0001, unpaired t test), and highly significant growth inhibition with BPTI-K15R/R17G treatment (●) at all concentrations tested (P<0.0001, unpaired Student's t test), relative to control cells. Values are means±S.E.M. for morphometry of 30 colonies per condition.

PRSS3/mesotrypsin has also been found to promote pancreatic cancer cell proliferation and invasion [10]. Using Capan-1 pancreatic cancer cells, which express high levels of PRSS3/mesotrypsin [10], we measured the impact of PRSS3 knockdown by RNA interference and of mesotrypsin inhibition by BPTI-K15R/R17G treatment on cellular invasion in Matrigel™ transwell invasion assays (Figure 5). Transduction of Capan-1 cells with a lentiviral shRNA (short hairpin RNA) construct specifically targeting PRSS3 led to efficient suppression of PRSS3/mesotrypsin expression both at the transcript level (Figure 5A) and at the protein level (Figure 5B). In Matrigel™ transwell invasion assays, either PRSS3 knockdown or treatment with 100 nM BPTI-K15R/R17G led to equivalent inhibition of invasion by 70–75% relative to control cells (Figure 5C). Taken together, these results confirm that BPTI-K15R/R17G inhibits malignant growth and invasion in cancer models in which PRSS3/mesotrypsin has been found to play a role in promoting these hallmarks of cancer.

Effect of BPTI-K15R/R17G on invasion of pancreatic cancer cells

Figure 5
Effect of BPTI-K15R/R17G on invasion of pancreatic cancer cells

Transduction of Capan-1 cells with an shRNA construct specifically targeting PRSS3 resulted in suppression of expression at the transcript level as assessed by qRT-PCR (A), and at the protein level as assessed by Western blot (B). (C) In Matrigel™ transwell invasion assays, Capan-1 cells assayed in the presence of 100 nM BPTI-K15R/R17G (NT+I) showed reduced invasion relative to control cells (NT); 100 nM inhibitor gave results similar to shRNA knockdown of PRSS3 (KD). The histogram shows means and S.E.M. for quadruplicate membranes; representative fields from invasion filters are shown above the histogram.. *P<0.05; ***P<0.0005 (unpaired Student's t test). KD, knockdown; NT, transduced with non-target control virus.

Figure 5
Effect of BPTI-K15R/R17G on invasion of pancreatic cancer cells

Transduction of Capan-1 cells with an shRNA construct specifically targeting PRSS3 resulted in suppression of expression at the transcript level as assessed by qRT-PCR (A), and at the protein level as assessed by Western blot (B). (C) In Matrigel™ transwell invasion assays, Capan-1 cells assayed in the presence of 100 nM BPTI-K15R/R17G (NT+I) showed reduced invasion relative to control cells (NT); 100 nM inhibitor gave results similar to shRNA knockdown of PRSS3 (KD). The histogram shows means and S.E.M. for quadruplicate membranes; representative fields from invasion filters are shown above the histogram.. *P<0.05; ***P<0.0005 (unpaired Student's t test). KD, knockdown; NT, transduced with non-target control virus.

DISCUSSION

In the present study, we have used site-directed mutagenesis of the mesotrypsin substrate APPI and inhibitor BPTI to dissect specificity at the key S2′ subsite, finding that bulky and charged residues strongly disfavour binding, whereas acidic residues facilitate catalysis. We have solved crystal structures to determine the structural underpinnings of these findings, and we have used the results to identify the mesotrypsin inhibitor BPTI-K15R/R17G, which is greatly improved in both mesotrypsin affinity and proteolytic stability. Because mesotrypsin is an enzyme that can contribute to tumour progression, we have tested BPTI-K15R/R17G in assays of breast cancer cell malignant growth and pancreatic cancer cell invasion, and have found significant anticancer activity.

The broad range of equilibrium association constants for APPI variants with dissimilar P2′ residues suggests that specificity at the S2′ subsite is likely to play a major role in shaping mesotrypsin substrate discrimination in vivo, creating specificity distinct from that of the less-selective trypsins. The best mesotrypsin substrates as evaluated by specificity constants kcat/Km possessed glycine or alanine at P2′, with specificity dominated by strong binding interactions, or alternatively possessed aspartate or glutamate residues at P2′, with specificity driven by rapid catalytic rates. These results contrast with previous studies of other trypsins: bovine trypsin displays little specificity at the P2′ position, but strongly disfavours binding and cleavage of a substrate possessing a P2′ aspartate residue [33], whereas rat anionic trypsin shows limited specificity favouring P2′ hydrophobic residues [34]. The mesotrypsin P2′ specificity defined through studies with the substrate APPI was recapitulated in studies with the inhibitor BPTI, where glycine at P2′ maximized inhibitor affinity, whereas aspartate at P2′ seriously compromised inhibitor resistance to mesotrypsin proteolysis. The crystal structure of mesotrypsin in complex with BPTI-K15R/R17G suggests that a major factor contributing to improved affinity and complex stability is the precise steric complementarity of the P2′ glycine variant to the pre-existing conformation of the mesotrypsin active site (Figures 2A and 3A).

It is less obvious why acidic P2′ residues should enhance catalysis, but our crystal structure of the mesotrypsin–BPTI-K15R/R17D complex does offer several clues. The relatively weak electron densities for mesotrypsin Arg193 and BPTI Asp17 (Figure 2F) are suggestive of conformational dynamics at the interface that may facilitate catalysis; our recent study comparing structures of mesotrypsin bound to BPTI compared with APPI also suggested that mobility at the primed-side interface could be a major determinant of hydrolytic rates [17]. This factor may be particularly important for mesotrypsin cleavage within the constrained binding loops of canonical protease inhibitors such as APPI and BPTI, which are typically highly resistant to cleavage by target proteases. Studies of BPTI variants by the Goldenberg laboratory [35,36] and our previous studies with chymotrypsin inhibitor 2 [37,38] have indicated that the rate-determining step in proteolysis occurs after acyl-enzyme formation for some canonical inhibitors, when retention of the primed-side amine leaving group in the active site blocks access to water and favours religation of the scissile peptide bond. Enhanced mobility of the primed-side residues in the acyl-enzyme could accelerate dissociation of the amine leaving group, allowing access to water and progress of deacylation. The presence of an acidic residue at the P2′ position may further assist in chaperoning solvent into the active site, as we see evidence for multiple water peaks surrounding the aspartate side chain carboxylate in our structure (Figure 2F).

The stability and experimental tractability of the Kunitz domain has led to its exploration as a scaffold for development of protein therapeutics targeting serine proteases. BPTI, also known as aprotinin, was for decades used clinically under the trade name Trasylol as an antifibrinolytic agent [39], and a number of drugs based on human Kunitz scaffolds are under development targeting a variety of proteases [4043]. Bikunin, a fragment of inter-α-trypsin inhibitor comprised of two Kunitz domains, has been studied as an antimetastatic agent for treatment of ovarian carcinoma [40,44]; its anticancer activity is derived in part through inhibition of cell-surface-associated plasmin [45]. The first Kunitz domain of TFPI-1 (tissue factor pathway inhibitor-1) has been engineered by phage display as a potent inhibitor of plasma kallikrein [46], for use in treatment of HAE (hereditary angiodoema) [41]. Using a similar approach, the same scaffold has been optimized for plasmin specificity [47] and PEGylated for improved pharmacokinetics, for potential use as an antimetastatic agent [42]. The first Kunitz domain of HAI (hepatocyte growth factor activator inhibitor)-1 has also been recombinantly produced and PEGylated as an antimetastatic agent targeting hepsin, a transmembrane serine protease that promotes prostate cancer progression [43]. These studies lend support to the idea that a Kunitz-based inhibitor could theoretically be developed as a mesotrypsin-targeted therapeutic. However, while our prototype BPTI-K15R/R17G offers impressive mesotrypsin affinity and stability, it offers little selectivity among trypsin isoforms, and may also inhibit other serine proteases featuring trypsin-like specificity. As P2′ engineering has had limited impact on affinity toward the major trypsins, future efforts to incorporate greater mesotrypsin selectivity will need to identify and exploit additional molecular determinants of binding specificity.

PRSS3 expression has been associated with cancer progression in several tumour types and model systems, suggesting that mesotrypsin may offer a target for cancer therapy. We have found that PRSS3 expression is upregulated with advancing malignancy in the HMT-3522 breast cancer progression series, where it promotes loss of cellular polarity in three-dimensional culture and enhances proliferation [9]. Treatment with recombinant mesotrypsin stimulated malignant growth, whereas PRSS3 knockdown suppressed malignant growth [9]. In the present study, we find that treatment of HMT-3522 T4-2 cells with BPTI-K15R/R17G also suppresses malignant growth in this assay, and does so an order of magnitude more potently than BPTI-WT (Figure 4). It is perhaps surprising that biological potency was not enhanced to an even greater extent, given that mesotrypsin affinity was improved by more than three orders of magnitude in the mutant inhibitor. However, both BPTI and BPTI-K15R/R17G inhibit multiple trypsin isoforms (Table 5), and presumably other serine proteases of tryptic specificity. Since multiple serine proteases can contribute to malignancy, the net biological effects of these relatively non-specific inhibitors represent a composite view of activity against a spectrum of proteases.

Ultimately, mesotrypsin inhibitors may be most likely to offer therapeutic benefit for cancers in which the enzyme contributes to metastasis, and for which existing therapies are of limited efficacy. In lung cancer patients, PRSS3 expression was associated with increased metastasis and poorer survival; in functional studies using NSCLC cells, overexpression of PRSS3 enhanced transendothelial migration, suggesting a mechanism by which mesotrypsin may stimulate metastasis [8]. As lung cancer is frequently diagnosed at advanced stages and has an overall 5 year survival rate of only 10–15% due primarily to metastasis [48], there is a need for better intervention strategies, one component of which might be the use of mesotrypsin inhibitors to suppress invasion. PRSS3/mesotrypsin expression has also been found to contribute to pancreatic cancer progression, correlating with metastasis and poor survival [10]. Mesotrypsin promoted pancreatic cancer cell proliferation and invasion in culture models, and stimulated tumour progression and metastasis in animal models [10]. Mesotrypsin may offer a promising therapeutic target in pancreatic cancer, a disease with a median survival of less than 6 months due to the aggressive metastasis of pancreatic cancers and their poor response to established chemotherapies [49]; this idea is supported by the efficacy of BPTI-K15R/R17G in suppressing the invasion of pancreatic cancer cells (Figure 5). Although further strategies to improve inhibitor selectivity are needed, our studies support the feasibility of protein engineering for development of mesotrypsin inhibitors, and highlight the potential for eventual therapeutic applications.

Abbreviations

     
  • APPI

    amyloid precursor protein Kunitz protease inhibitor domain

  •  
  • ASA

    accessible surface area

  •  
  • BPTI

    bovine pancreatic trypsin inhibitor

  •  
  • DMEM

    Dulbecco's modified Eagle's medium

  •  
  • NSCLC

    non-small cell lung cancer

  •  
  • PEG

    poly(ethylene) glycol

  •  
  • qRT-PCR

    quantitative real-time PCR

  •  
  • shRNA

    short hairpin RNA

  •  
  • WT

    wild-type

  •  
  • Z-GPR-pNA

    benzyloxycarbonyl-Gly-Pro-Arg-p-nitroanalide

AUTHOR CONTRIBUTION

Moh'd Salameh, Alexandra Hockla, Derek Radisky and Evette Radisky designed the research; Moh'd Salameh, Alexei Soares, Alexandra Hockla and Derek Radisky performed the research; Moh'd Salameh, Alexei Soares, Alexandra Hockla, Derek Radisky and Evette Radisky analysed the data; and Moh'd Salameh, Derek Radisky and Evette Radisky wrote the paper.

We thank Dr Peter Storz and Heike Döppler for the gift of Capan-1 cells. We thank Dr Nicole Murray and Amanda Butler for the suggestion of using SDF-1 as a chemoattractant in the pancreatic cancer cell-invasion assays.

FUNDING

This work was supported by the Bankhead–Coley Florida Biomedical Research Program [grant number 07BN-07 (to E.S.R.)]; the US Department of Defense [grant number PC094054 (to E.S.R.)]; the US National Cancer Institute [grant number CA091956 (to E.S.R., primary investigator Don Tindall), CA122086 (to D.C.R.), CA116201 (to D.C.R., primary investigator James Ingle)]; and the Susan B. Komen Foundation [grant number FAS0703855 (to D.C.R.)]. Diffraction data were measured at beamlines X12-B, X12-C and X25 of the National Synchrotron Light Source, which is supported by the Offices of Biological and Environmental Research and of Basic Energy Sciences of the US Department of Energy, and the National Center for Research Resources of the National Institutes of Health.

References

References
1
Kessenbrock
K.
Plaks
V.
Werb
Z.
Matrix metalloproteinases: regulators of the tumor microenvironment
Cell
2010
, vol. 
141
 (pg. 
52
-
67
)
2
Turk
B.
Targeting proteases: successes, failures and future prospects
Nat. Rev. Drug Discovery
2006
, vol. 
5
 (pg. 
785
-
799
)
3
Cudic
M.
Fields
G. B.
Extracellular proteases as targets for drug development
Curr. Protein Pept. Sci.
2009
, vol. 
10
 (pg. 
297
-
307
)
4
Lopez-Otin
C.
Matrisian
L. M.
Emerging roles of proteases in tumour suppression
Nat. Rev. Cancer
2007
, vol. 
7
 (pg. 
800
-
808
)
5
Rinderknecht
H.
Activation of pancreatic zymogens. Normal activation, premature intrapancreatic activation, protective mechanisms against inappropriate activation
Dig. Dis. Sci.
1986
, vol. 
31
 (pg. 
314
-
321
)
6
Wiegand
U.
Corbach
S.
Minn
A.
Kang
J.
Muller-Hill
B.
Cloning of the cDNA encoding human brain trypsinogen and characterization of its product
Gene
1993
, vol. 
136
 (pg. 
167
-
175
)
7
Toth
J.
Siklodi
E.
Medveczky
P.
Gallatz
K.
Nemeth
P.
Szilagyi
L.
Graf
L.
Palkovits
M.
Regional distribution of human trypsinogen 4 in human brain at mRNA and protein level
Neurochem. Res.
2007
, vol. 
32
 (pg. 
1423
-
1433
)
8
Diederichs
S.
Bulk
E.
Steffen
B.
Ji
P.
Tickenbrock
L.
Lang
K.
Zanker
K. S.
Metzger
R.
Schneider
P. M.
Gerke
V.
, et al. 
S100 family members and trypsinogens are predictors of distant metastasis and survival in early-stage non-small cell lung cancer
Cancer Res.
2004
, vol. 
64
 (pg. 
5564
-
5569
)
9
Hockla
A.
Radisky
D. C.
Radisky
E. S.
Mesotrypsin promotes malignant growth of breast cancer cells through shedding of CD109
Breast Cancer Res. Treat.
2010
, vol. 
124
 (pg. 
27
-
38
)
10
Jiang
G.
Cao
F.
Ren
G.
Gao
D.
Bhakta
V.
Zhang
Y.
Cao
H.
Dong
Z.
Zang
W.
Zhang
S.
, et al. 
PRSS3 promotes tumour growth and metastasis of human pancreatic cancer
Gut
2010
, vol. 
59
 (pg. 
1535
-
1544
)
11
Nyaruhucha
C. N.
Kito
M.
Fukuoka
S. I.
Identification and expression of the cDNA-encoding human mesotrypsin(ogen), an isoform of trypsin with inhibitor resistance
J. Biol. Chem.
1997
, vol. 
272
 (pg. 
10573
-
10578
)
12
Katona
G.
Berglund
G. I.
Hajdu
J.
Graf
L.
Szilagyi
L.
Crystal structure reveals basis for the inhibitor resistance of human brain trypsin
J. Mol. Biol.
2002
, vol. 
315
 (pg. 
1209
-
1218
)
13
Szmola
R.
Kukor
Z.
Sahin-Toth
M.
Human mesotrypsin is a unique digestive protease specialized for the degradation of trypsin inhibitors
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
48580
-
48589
)
14
Szepessy
E.
Sahin-Toth
M.
Human mesotrypsin exhibits restricted S1′ subsite specificity with a strong preference for small polar side chains
FEBS J.
2006
, vol. 
273
 (pg. 
2942
-
2954
)
15
Salameh
M. A.
Soares
A. S.
Hockla
A.
Radisky
E. S.
Structural basis for accelerated cleavage of bovine pancreatic trypsin inhibitor (BPTI) by human mesotrypsin
J. Biol. Chem.
2008
, vol. 
283
 (pg. 
4115
-
4123
)
16
Salameh
M. A.
Robinson
J. L.
Navaneetham
D.
Sinha
D.
Madden
B. J.
Walsh
P. N.
Radisky
E. S.
The amyloid precursor protein/protease nexin 2 Kunitz inhibitor domain is a highly specific substrate of mesotrypsin
J. Biol. Chem.
2010
, vol. 
285
 (pg. 
1939
-
1949
)
17
Salameh
M. A.
Soares
A. S.
Navaneetham
D.
Sinha
D.
Walsh
P. N.
Radisky
E. S.
Determinants of affinity and proteolytic stability in interactions of Kunitz family protease inhibitors with mesotrypsin
J. Biol. Chem.
2010
, vol. 
285
 (pg. 
36884
-
36896
)
18
Rinderknecht
H.
Renner
I. G.
Abramson
S. B.
Carmack
C.
Mesotrypsin: a new inhibitor-resistant protease from a zymogen in human pancreatic tissue and fluid
Gastroenterology
1984
, vol. 
86
 (pg. 
681
-
692
)
19
Sahin-Toth
M.
Human cationic trypsinogen. Role of Asn-21 in zymogen activation and implications in hereditary pancreatitis
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
22750
-
22755
)
20
Kukor
Z.
Toth
M.
Sahin-Toth
M.
Human anionic trypsinogen: properties of autocatalytic activation and degradation and implications in pancreatic diseases
Eur. J. Biochem.
2003
, vol. 
270
 (pg. 
2047
-
2058
)
21
Navaneetham
D.
Jin
L.
Pandey
P.
Strickler
J. E.
Babine
R. E.
Abdel-Meguid
S. S.
Walsh
P. N.
Structural and mutational analyses of the molecular interactions between the catalytic domain of factor XIa and the Kunitz protease inhibitor domain of protease nexin 2
J. Biol. Chem.
2005
, vol. 
280
 (pg. 
36165
-
36175
)
22
Navaneetham
D.
Sinha
D.
Walsh
P. N.
Mechanisms and specificity of factor XIa and trypsin inhibition by protease nexin 2 and basic pancreatic trypsin inhibitor
J. Biochem.
2010
, vol. 
148
 (pg. 
467
-
479
)
23
Otwinowski
Z.
Minor
W.
Carter
C. W.
Jr
Sweet
R. M.
Processing of X-ray diffraction data collected in oscillation mode
Macromolecular Crystallography, Part A
1997
New York
Academic Press
(pg. 
307
-
326
)
24
McCoy
A. J.
Grosse-Kunstleve
R. W.
Storoni
L. C.
Read
R. J.
Likelihoodenhanced fast translation functions
Acta Crystallogr. D Biol. Crystallogr.
2005
, vol. 
61
 (pg. 
458
-
464
)
25
Adams
P. D.
Grosse-Kunstleve
R. W.
Hung
L. W.
Ioerger
T. R.
McCoy
A. J.
Moriarty
N. W.
Read
R. J.
Sacchettini
J. C.
Sauter
N. K.
Terwilliger
T. C.
PHENIX: building new software for automated crystallographic structure determination
Acta Crystallogr. D Biol. Crystallogr.
2002
, vol. 
58
 (pg. 
1948
-
1954
)
26
Emsley
P.
Cowtan
K.
Coot: model-building tools for molecular graphics
Acta Crystallogr. D Biol. Crystallogr.
2004
, vol. 
60
 (pg. 
2126
-
2132
)
27
Afonine
P. V.
Grosse-Kunstleve
R. W.
Adams
P. D.
The Phenix refinement framework
CCP4 Newsletter
2005
, vol. 
42
 
28
Lee
G. Y.
Kenny
P. A.
Lee
E. H.
Bissell
M. J.
Three-dimensional culture models of normal and malignant breast epithelial cells
Nat. Methods
2007
, vol. 
4
 (pg. 
359
-
365
)
29
Cornish-Bowden
A.
Fundamentals of Enzyme Kinetics
1995
London
Portland Press
30
Bigler
T. L.
Lu
W.
Park
S. J.
Tashiro
M.
Wieczorek
M.
Wynn
R.
Laskowski
M.
Jr
Binding of amino acid side chains to preformed cavities: interaction of serine proteinases with turkey ovomucoid third domains with coded and noncoded P1 residues
Protein Sci.
1993
, vol. 
2
 (pg. 
786
-
799
)
31
Lu
W.
Apostol
I.
Qasim
M. A.
Warne
N.
Wynn
R.
Zhang
W. L.
Anderson
S.
Chiang
Y. W.
Ogin
E.
Rothberg
I.
, et al. 
Binding of amino acid side-chains to S1 cavities of serine proteinases
J. Mol. Biol.
1997
, vol. 
266
 (pg. 
441
-
461
)
32
Krissinel
E.
Henrick
K.
Inference of macromolecular assemblies from crystalline state
J. Mol. Biol.
2007
, vol. 
372
 (pg. 
774
-
797
)
33
Coombs
G. S.
Dang
A. T.
Madison
E. L.
Corey
D. R.
Distinct mechanisms contribute to stringent substrate specificity of tissue-type plasminogen activator
J. Biol. Chem.
1996
, vol. 
271
 (pg. 
4461
-
4467
)
34
Schellenberger
V.
Turck
C. W.
Rutter
W. J.
Role of the S′ subsites in serine protease catalysis. Active-site mapping of rat chymotrypsin, rat trypsin, α-lytic protease, and cercarial protease from Schistosoma mansoni
Biochemistry
1994
, vol. 
33
 (pg. 
4251
-
4257
)
35
Zakharova
E.
Horvath
M. P.
Goldenberg
D. P.
Functional and structural roles of the Cys14-Cys38 disulfide of bovine pancreatic trypsin inhibitor
J. Mol. Biol.
2008
, vol. 
382
 (pg. 
998
-
1013
)
36
Zakharova
E.
Horvath
M. P.
Goldenberg
D. P.
Structure of a serine protease poised to resynthesize a peptide bond
Proc. Natl. Acad. Sci. U.S.A.
2009
, vol. 
106
 (pg. 
11034
-
11039
)
37
Radisky
E. S.
Koshland
D. E.
Jr
A clogged gutter mechanism for protease inhibitors
Proc. Natl. Acad. Sci. U.S.A.
2002
, vol. 
99
 (pg. 
10316
-
10321
)
38
Radisky
E. S.
Kwan
G.
Lu
C. J.
Koshland
D. E.
Jr
Binding, proteolytic, and crystallographic analyses of mutations at the protease-inhibitor interface of the subtilisin BPN'/chymotrypsin inhibitor 2 complex
Biochemistry
2004
, vol. 
43
 (pg. 
13648
-
13656
)
39
Fritz
H.
Wunderer
G.
Biochemistry and applications of aprotinin, the kallikrein inhibitor from bovine organs
Arzneimittelforschung
1983
, vol. 
33
 (pg. 
479
-
494
)
40
Kobayashi
H.
Yagyu
T.
Inagaki
K.
Kondo
T.
Suzuki
M.
Kanayama
N.
Terao
T.
Therapeutic efficacy of once-daily oral administration of a Kunitz-type protease inhibitor, bikunin, in a mouse model and in human cancer
Cancer
2004
, vol. 
100
 (pg. 
869
-
877
)
41
Williams
A.
Baird
L. G.
DX-88 and HAE: a developmental perspective
Transfus. Apher. Sci.
2003
, vol. 
29
 (pg. 
255
-
258
)
42
Devy
L.
Rabbani
S. A.
Stochl
M.
Ruskowski
M.
Mackie
I.
Naa
L.
Toews
M.
van Gool
R.
Chen
J.
Ley
A.
, et al. 
PEGylated DX-1000: pharmacokinetics and antineoplastic activity of a specific plasmin inhibitor
Neoplasia
2007
, vol. 
9
 (pg. 
927
-
937
)
43
Li
W.
Wang
B. E.
Moran
P.
Lipari
T.
Ganesan
R.
Corpuz
R.
Ludlam
M. J.
Gogineni
A.
Koeppen
H.
Bunting
S.
, et al. 
PEGylated kunitz domain inhibitor suppresses hepsin-mediated invasive tumor growth and metastasis
Cancer Res.
2009
, vol. 
69
 (pg. 
8395
-
8402
)
44
Kobayashi
H.
Yagyu
T.
Inagaki
K.
Kondo
T.
Suzuki
M.
Kanayama
N.
Terao
T.
Bikunin plus paclitaxel markedly reduces tumor burden and ascites in mouse model of ovarian cancer
Int. J. Cancer
2004
, vol. 
110
 (pg. 
134
-
139
)
45
Kobayashi
H.
Shinohara
H.
Takeuchi
K.
Itoh
M.
Fujie
M.
Saitoh
M.
Terao
T.
Inhibition of the soluble and the tumor cell receptor-bound plasmin by urinary trypsin inhibitor and subsequent effects on tumor cell invasion and metastasis
Cancer Res.
1994
, vol. 
54
 (pg. 
844
-
849
)
46
Markland
W.
Ley
A. C.
Ladner
R. C.
Iterative optimization of high-affinity protease inhibitors using phage display. 2. Plasma kallikrein and thrombin
Biochemistry
1996
, vol. 
35
 (pg. 
8058
-
8067
)
47
Markland
W.
Ley
A. C.
Lee
S. W.
Ladner
R. C.
Iterative optimization of high-affinity proteases inhibitors using phage display. 1
Plasmin. Biochemistry
1996
, vol. 
35
 (pg. 
8045
-
8057
)
48
Wang
T.
Nelson
R. A.
Bogardus
A.
Grannis
F. W.
Jr
Five-year lung cancer survival: which advanced stage nonsmall cell lung cancer patients attain long-term survival?
Cancer
2010
, vol. 
116
 (pg. 
1518
-
1525
)
49
Li
D.
Xie
K.
Wolff
R.
Abbruzzese
J. L.
Pancreatic cancer
Lancet
2004
, vol. 
363
 (pg. 
1049
-
1057
)

Author notes

The atomic co-ordinates and structure factors (PDB codes 3P92 and 3P95) have been deposited in the Protein Data Bank.