A hybrid, broad-spectrum inhibitor of Colorado potato beetle aspartate and cysteine digestive proteinases

Protein engineering approaches are currently being devised to improve the inhibitory properties of plant proteinase inhibitors against digestive proteinases of herbivorous insects. Here we engineered a potent hybrid inhibitor of aspartate and cysteine digestive proteinases found in the Colorado potato beetle, Leptinotarsa decemlineata Say. Three cathepsin D inhibitors (CDIs) from stressed potato and tomato were first compared in their potency to inhibit digestive cathepsin D-like activity of the insect. After showing the high inhibitory potency of tomato CDI (M(r) approximately 21 kDa), an approximately 33-kDa hybrid inhibitor was generated by fusing this inhibitor to the N terminus of corn cystatin II (CCII), a potent inhibitor of cysteine proteinases. Inhibitory assays with recombinant forms of CDI, CCII, and CDI-CCII expressed in Escherichia coli showed the CDI-CCII fusion to exhibit a dual inhibitory effect against cystatin-sensitive and cathepsin D-like enzymes of the potato beetle, resulting in detrimental effects against 3rd-instar larvae fed the hybrid inhibitor. The inhibitory potency of CDI and CCII was not altered after their fusion, as suggested by IC(50) values for the interaction of CDI-CCII with target proteinases similar to those measured for each inhibitor. These observations suggest the potential of plant CDIs and cystatins as functional inhibitory modules for the design of effective broad-spectrum, hybrid inhibitors of herbivorous insect cysteine and aspartate digestive proteinases.     

Response of digestive cysteine proteinases from the colorado potato beetle (Leptinotarsa decemlineata) and the black vine weevil (Otiorynchus sulcatus) to a recombinant form of human stefin A

The effects of the cystatins, human stefin A (HSA) and oryzacystatin I (OCI) on digestive cysteine proteinases of the Colorado potato beetle (CPB), Leptinotarsa decemlineata, and the black vine weevil (BVW), Otiorynchus sulcatus, were assessed using complementary inhibition assays, cystatin-affinity chromatography, and recombinant forms of the two inhibitors. For both insects, either HSA and OCI used in excess (10 or 20 μM) caused partial and stable inhibition of total proteolytic (azocaseinase) activity, but unlike for OCI the HSA-mediated inhibitions were significantly increased when the inhibitor was used in large excess (100 μM). As demonstrated by complementary inhibition assays, this two-step inhibition of the insect proteases by HSA was due to the differential inactivation of two distinct cysteine proteinase populations in either insect extracts, the rapidly (strongly) inhibited population corresponding to the OCI-sensitive fraction. After removing the cystatin-sensitive proteinases from CPB and BVW midgut extracts using OCI- (or HSA-) affinity chromatography, the effects of the insect “non-target” proteases on the structural integrity of the two cystatins were assessed. While OCI remained essentially stable, HSA was subjected to hydrolysis without the accumulation of detectable stable intermediates, suggesting the presence of multiple exposed cleavage sites sensitive to the action of the insect proteases on this cystatin. This apparent susceptibility of HSA to proteolytic cleavage may partially explain its low efficiency to inactivate the insect OCI-insensitive cysteine proteinases when not used in large excess. It could also have major implications when planning the use of cystatin-expressing transgenic plants for the control of coleopteran pests. © 1996 Wiley-Liss, Inc.     

Contributions of individual residues in the N-terminal region of cystatin B (stefin B) to inhibition of cysteine proteinases

The importance of individual residues in the N-terminal region of cystatin B for proteinase inhibition was elucidated by measurements of the affinity and kinetics of binding of N-terminally truncated, recombinant variants of the bovine inhibitor to cysteine proteinases. Removal of Met-1 caused an 8- to 10-fold lower affinity for papain and cathepsin B, decreased the affinity also for cathepsin L but only minimally affected cathepsin H affinity. Additional truncation of Met-2 further weakened the binding to papain and cathepsin B by 40-70-fold, whereas the affinity for cathepsins L and H was essentially unaffected. Removal of Cys-3 had the most drastic effects on the interactions, resulting in a further affinity decrease of approximately 1500-fold for papain, approximately 700-fold for cathepsin L and approximately 15-fold for cathepsin H; the binding to cathepsin B could not be assessed. The binding kinetics could only be evaluated for papain and cathepsin H and showed that the reduced affinities for these enzymes were predominantly due to increased dissociation rate constants. These results demonstrate that the N-terminal region of cystatin B contributes appreciably to proteinase inhibition, in contrast to previous proposals. It is responsible for 12-40% of the total binding energy of the inhibitor to the proteinases investigated, being of least importance for cathepsin H binding. Cys-3 is the most important residue of the N-terminal region for inhibition of papain, cathepsin L and cathepsin H, the role of the other residues of this region varying with the target proteinase.     

Phage display selection of hairpin loop soyacystatin variants that mediate high affinity inhibition of a cysteine proteinase

Two hairpin-loop domains in cystatin family proteinase inhibitors form an interface surface region that slots into the active site cleft of papain-like cysteine proteinases, and determine binding affinity. The slot region surface architecture of the soybean cysteine proteinase inhibitor (soyacystatin N, scN) was engineered using techniques of in vitro molecular evolution to define residues that facilitate interaction with the proteinase cleft and modulate inhibitor affinity and function. Combinatorial phage display libraries of scN variants that contain mutations in the essential motifs of the first (QVVAG) and second (EW) hairpin-loop regions were constructed. Approximately 1010-1011 phages expressing recombinant scN proteins were subjected to biopanning selection based on binding affinity to immobilized papain. The QVVAG motif in the first hairpin loop was invariant in all functional scN proteins. All selected variants (30) had W79 in the second hairpin-loop motif, but there was diversity for hydrophobic and basic amino acids in residue 78. Kinetic analysis of isolated scN variants identified a novel scN isoform scN(LW) with higher papain affinity than the wild-type molecule. The variant contained an E78L substitution and had a twofold lower Ki (2.1 pM) than parental scN, due to its increased association rate constant (2.6 +/- 0.09 x 107 M-1sec-1). These results define residues in the first and second hairpin-loop regions which are essential for optimal interaction between phytocystatins and papain, a prototypical cysteine proteinase. Furthermore, the isolated variants are a biochemical platform for further integration of mutations to optimize cystatin affinity for specific biological targets.     

Conserved cystatin segments as models for designing specific substrates and inhibitors of cysteine proteinases

Peptide segments derived from consensus sequences of the inhibitory site of cystatins, the natural inhibitors of cysteine proteinases, were used to develop new substrates and inhibitors of papain and rat liver cathepsins B, H, and L. Papain hydrolyzedAbz-QVVAGA-EDDnp andAbz-LVGGA-EDDnp at about the same rate, with specificity constants in the 10⁷M^–1 sec^–1 range; cathepsin L also hydrolyzes both substrates with specificity constants in the 10⁵ M^–1 sec^–1 range due to lowerk _cat values, with theK _m 's being identical to those with papain. OnlyAbz-LVGGA-EDDnp was rapidly hydrolyzed by cathepsin B, and to a lesser extent by cathepsin H. Peptide substrates that alternate these two building blocks (LVGGQVVAGAPWK and QVVAGALVGGAPWK) discriminate the activities of cathepsins B and L and papain. Cathepsin L was highly selective for cleavage at the G-G bond of the LVGG fragment in both peptides. Papain and cathepsin B cleaved either the LVGG fragment or the QVVAG fragment, depending on their position within the peptide. While papain was more specific for the segment located C-terminally, cathepsin B was specific for that in N-terminal position. Peptidyl diazomethylketone inhibitors based on these two sequences also reacted differently with papain and cathepsins. GlcA-QVVA-CHN₂ was a potent inhibitor of papain and reacted with papain 60 times more rapidly (k ₊₀= 1,100,000 M^–1 sec^–1) than with cathepsin L, and 220 times more rapidly than with cathepsin B. Cathepsins B and L were preferentially inhibited by Z-RLVG-CHN₂. Thus cystatin-derived peptides provide a valuable framework for designing sensitive, selective substrates and inhibitors of cysteine proteinases.     

Increase of cyclin B by overexpression of cystatin α

Degradation of cyclin B was effectively suppressed when cells were treated with ALLN (N-acetylleucylleucylnorleucinal) which inhibits proteasome, calpain and cysteine proteinase cathepsins. In order to examine which protease degrades cyclin B, the effect of a cathepsin inhibitor, cystatin α, was investigated. The cystatin α gene was inserted into an inducible expression vector, pMSG, and transfected into NIH3T3 mouse fibroblasts. The expression of cystatin α was induced effectively in the transfected cells after treatment with dexamethasone. Overexpression of cystatin α resulted in an increase of the amount of cyclin B, suggesting that cysteine proteinase cathepsins might be involved in the degradation of cyclin B.     

Importance of the second binding loop and the C-terminal end of cystatin B (stefin B) for inhibition of cysteine proteinases.

The importance of residues in the second hairpin loop and the C-terminal end of mammalian cystatin B for binding of proteinases was elucidated by mutagenesis of the bovine inhibitor. Bovine cystatin B was modeled onto the crystal structure of the human inhibitor in complex with papain with minimal structural changes. Substitution of the two deduced contact residues in the second hairpin loop, Leu-73 and His-75, with Gly resulted in appreciably reduced affinities for papain and cathepsins H and B. These losses indicated that the two residues together contribute 20-30% of the free energy of binding of cystatin B to these enzymes and that Leu-73 is responsible for most of this contribution. In contrast, the small decrease in the affinity for cathepsin L suggested that the second hairpin loop is less important for inhibition of this proteinase. Replacement of the contact residue in the C-terminal end, Tyr-97, with Ala resulted in losses in affinity for papain and cathepsins L and H that were consistent with Tyr-97 contributing 6-12% of the energy of binding of cystatin B to these enzymes. However, this substitution minimally affected the affinity for cathepsin B, indicating that the C-terminal end is of limited importance for binding of this proteinase. All affinity decreases were due predominantly to increased dissociation rate constants. These results show that both the second hairpin loop and the C-terminal end of cystatin B contribute to anchoring the inhibitor to target proteinases, each of the two regions interacting with a different domain of the enzyme. However, the relative contributions of these two interactions vary with the proteinase.     

On the mechanism of human stefin B folding: I. Comparison to homologous stefin A. Influence of pH and trifluoroethanol on the fast and slow folding phases

The folding of human stefin B has been studied by several spectroscopic probes. Stopped-flow traces obtained by circular dichroism in the near and far UV, by tyrosine fluorescence, and by extrinsic probe ANS fluorescence are compared. Most (60 ± 5%) of the native signal in the far UV circular dichroism (CD) appeared within 10 ms in an unresolved “burst” phase, which was followed by a fast phase (t = 83 ms) and a slow phase (t = 25 s) with amplitudes of 30% and 10%, respectively. Similar fast and slow phases were also evident in the near UV CD, ANS fluorescence, and tyrosine fluorescence. By contrast, human stefin A, which has a very similar structure, exhibited only one kinetic phase of folding (t = 6 s) detected by all the spectroscopic probes, which occurred subsequent to an initial “burst” phase observed by far UV CD. It is interesting that despite close structural similarity of both homologues they fold differently, and that the less stable human stefin B folds faster by an order of magnitude (comparing the non-proline limited phase). To gain more information on the stefin B folding mechanism, effects of pH and trifluoroethanol (TFE) on the fast and slow phases were investigated by several spectroscopic probes. If folding was performed in the presence of 7% of TFE, rate acceleration and difference in the mechanism were observed. Protein 32:296–303, 1998. © 1998 Wiley-Liss, Inc.     

Cathepsins J and K: high molecular weight cysteine proteinases from human tissues

Human tissue extracts contained two high Mr proteinases active in hydrolyzing the fluorogenic substrate Cbz-phe-arg-aminomethylcoumarin. By gel filtration chromatography, cathepsins J and K had apparent molecular weights of 230,000 and 650,000, respectively. Both enzymes were cysteine proteinases with optimum activity at pH 6.2-6.8; neither had aminopeptidase activity. Human kidney, lung and spleen were rich sources of these enzymes, while liver contained moderate amounts. Cathepsins J and K were partially characterized and appeared to differ from the mammalian high Mr cysteine proteinases described in the literature. In rat liver and kidney and in mouse liver, cathepsin J was localized in the particulate fraction, whereas cathepsin K was not detected in these tissues.     

Purification and characterization of a low molecular weight cysteine proteinase inhibitor from bovine muscle

A low-M_r tight binding proteinase inhibitor was purified from bovine muscle by alkaline denaturation of cysteine proteinases, gel filtration on Sexphadex G-75 and affinity chromatography on carboxymethyl-papain-Sepharose. Chromatofocusing separated three isoforms which are similar in their M_r of about 14 000, their stability with heating at 80°C and their inhibitory activity towards cathepsin H, cathepsin B and papain. The equilibrium constants (K_i) were determined for these three cysteine proteinases but for cathepsin H, association (k_ass) and dissociation (k_diss) rate constants were also evaluated. K_i values of 56 nM and 8.4 nM were found for cathepsin B and cathepsin H, respectively. For papain, K_i was in the range of 0.1–1 nM. The kinetic features of enzyme-inhibitor binding suggest a possible role for this low-M_r protein inhibitor in controlling ‘in vivo’ cathepsin H proteolytic activity. With regard to cathepsin B, such a physiological role was less evident.     

The N-terminal region of cystatin A (stefin A) binds to papain subsequent to the two hairpin loops of the inhibitor. Demonstration of two-step binding by rapid-kinetic studies of cystatin A labeled at the N-terminus with a fluorescent reporter group

The three-dimensional structures of cystatins, and other evidence, suggest that the flexible N-terminal region of these inhibitors may bind to target proteinases independent of the two rigid hairpin loops forming the remainder of the inhibitory surface. In an attempt to demonstrate such two-step binding, which could not be identified in previous kinetics studies, we introduced a cysteine residue before the N-terminus of cystatin A and labeled this residue with fluorescent probes. Binding of AANS- and AEDANS-labeled cystatin A to papain resulted in approximately 4-fold and 1.2-fold increases of probe fluorescence, respectively, reflecting the interaction of the N-terminal region with the enzyme. Observed pseudo-first-order rate constants, measured by the loss of papain activity in the presence of a fluorogenic substrate, for the reaction of the enzyme with excess AANS-cystatin A increased linearly with the concentration of the latter. In contrast, pseudo-first-order rate constants, obtained from measurements of the change of probe fluorescence with either excess enzyme or labeled inhibitor, showed an identical hyperbolic dependence on the concentration of the reactant in excess. This dependence demonstrates that the binding occurs in two steps, and implies that the labeled N-terminal region of cystatin A interacts with the proteinase in the second step, subsequent to the hairpin loops. The comparable affinities and dissociation rate constants for the binding of labeled and unlabeled cystatin A to papain indicate that the label did not appreciably perturb the interaction, and that unlabeled cystatin therefore also binds in a similar two-step manner. Such independent binding of the N-terminal regions of cystatins to target proteinases after the hairpin loops may be characteristic of most cystatin-proteinase reactions.     

Differential tissue-specific expression of cysteine proteinases forms the basis for the fine-tuned mobilization of storage globulin during and after germination in legume seeds

The temporal and spatial distribution of cysteine proteinases (CPRs) was analyzed immunologically and by in situ hybridization to identify the CPRs involved in the initiation of storage-globulin degradation in embryonic axes and cotyledons of germinating vetch (Vicia sativa L.). At the start of germination several CPRs were found in protein bodies in which they might have been stored in the mature seeds. Cysteine proteinase 1 was predominantly found in organs like the radicle, which first start to grow during germination. Cysteine proteinase 2 was also present at the start of germination but displayed a less-specific histological pattern. Proteinase B was involved in the globulin degradation of vetch cotyledons as well. The histological pattern of CPRs followed the distribution of their corresponding mRNAs. The latter were usually detected earlier than the CPRs but the in situ hybridization signals were histologically not as restricted as the immunosignals. Proteolytic activity started in the radicle of the embryonic axis early during germination. Within 24 h after imbibition it had also spread throughout the whole shoot. At the end of germination, newly synthesized CPRs might have supplemented the early detectable CPRs in the axis. In the cotyledons, only the abaxial epidermis and the procambial strands showed proteinase localization during germination. Both CPR1 and CPR2, as well as the less common proteinase B, might have been present as stored proteinases. Three days after imbibition, proteolytic activity had proceeded from the cotyledonary epidermis towards the vascular strands deeper inside the cotyledons. The histochemical detection of the CPRs was in accordance with the previously described histological pattern of globulin mobilization in germinating vetch [Tiedemann J, et al. (2000)]. A similar link between the distribution of CPRs and globulin degradation was found in germinating seeds of Phaseolus vulgaris L. The coincidence of the histological patterns of globulin breakdown with that of the CPRs indicates that at least CPR1, CPR2 and proteinase B are responsible for bulk globulin mobilization in the seeds of the two legumes. Received: 14 February 2000 / Accepted: 16 August 2000     

Grafting of features of cystatins C or B into the N-terminal region or second binding loop of cystatin A (stefin A) substantially enhances inhibition of cysteine proteinases

Replacement of the three N-terminal residues preceding the conserved Gly of cystatin A by the corresponding 10-residue long segment of cystatin C increased the affinity of the inhibitor for the major lysosomal cysteine proteinase, cathepsin B, by approximately 15-fold. This tighter binding was predominantly due to a higher overall association rate constant. Characterization of the interaction with an inactive Cys29 to Ala variant of cathepsin B indicated that the higher rate constant was a result of an increased ability of the N-terminal region of the chimeric inhibitor to promote displacement of the cathepsin B occluding loop in the second binding step. The low dissociation rate constant for the binding of cystatin A to cathepsin B was retained by the chimeric inhibitor, which therefore had a higher affinity for this enzyme than any natural cystatin identified so far. In contrast, the N-terminal substitution negligibly affected the ability of cystatin A to inhibit papain. However, substitutions of Gly75 in the second binding loop of cystatin A by Trp or His, making the loop similar to those of cystatins C or B, respectively, increased the affinity for papain by approximately 10-fold. This enhanced affinity was due to both a higher association rate constant and a lower dissociation rate constant. Modeling of complexes between the two variants and papain indicated the possibility of favorable interactions being established between the substituting residues and the enzyme. The second-loop substitutions negligibly affected or moderately reduced the affinity for cathepsin B. Together, these results show that the inhibitory ability of cystatins can be substantially improved by protein engineering.     

In vitro study of stability and amyloid-fibril formation of two mutants of human stefin B (cystatin B) occurring in patients with EPM1

Myoclonus epilepsy of type 1 (EPM1) is a rare monogenic progressive and degenerative epilepsy, also known under the name Unverricht-Lundborg disease. With the aim of comparing their behavior in vitro, wild-type (wt) human stefin B (cystatin B) and the G4R and the R68X mutants observed in EPM1 were expressed and isolated from the Escherichia coli lysate. The R68X mutant (Arg68Stop) is a peptide of 67 amino acids from the N terminus of stefin B. CD spectra have shown that the R68X peptide is not folded, in contrast to the G4R mutant, which folds like wild type. The wild type and the G4R mutant were unfolded by urea and by trifluoroethanol (TFE). It has been shown that both proteins have closely similar stability and that at pH 4.8, where a native-like intermediate was demonstrated, TFE induces unfolding intermediates prior to the major transition to the all-alpha-helical state. Kinetics of fibril formation were followed by Thioflavin T fluorescence while the accompanying changes of morphology were followed by the transmission electron microscopy (TEM). For the two folded proteins the optimal concentration of TFE producing extensive lag phases and high fibril yields was predenaturational, 9% (v/v). The unfolded R68X peptide, which is highly prone to aggregate, formed amyloid fibrils in aqueous solution and in predenaturing 3% TFE. The G4R mutant exhibited a much longer lag phase than the wild type, with the accumulation of prefibrillar aggregates. Implications for pathology in view of the higher toxicity of prefibrillar aggregates to cells are discussed.     

Differential localization of two acid proteinases in germinating barley (Hordeum vulgare) seed

A cathepsin D-like aspartic proteinase (EC 3.4.23) is abundant in ungerminated barley ( Hordeum vulgare ) seed while a 30 kDa cysteine endoproteinase (EC 3.4.22) is one of the proteinases synthesized de novo in the germinating seed. In this work, the localization of these two acid proteinases was studied at both the tissue and subcellular levels by immunomicroscopy. The results confirm that they have completely different functions. The aspartic proteinase was present in the ungerminated seed and, during germination, it appeared in all the living tissues of the grain, including the shoot and root. Contrary to previous suggestions, it was not observed in the starchy endosperm. By immunoblotting, the high molecular mass form of the enzyme (32 + 16 kDa) was found in all the living tissues, whereas the low molecular mass form (29 + 11 kDa) was not present in the shoot or root, indicating that the two enzyme forms have different physiological roles. The aspartic proteinase was localized first in the scutellar protein bodies of germinating seed, and later in the vacuoles which are formed by fusion of the protein bodies. In contrast to the aspartic proteinase, the expression of the 30 kDa cysteine proteinase began during the first germination day, and it was secreted into the starchy endosperm; first from the scutellum and later from the aleurone layer. It was not found in either shoots or roots. The 30 kDa cysteine proteinase was detected in the Golgi apparatus and in the putative secretory vesicles of the scutellar epithelium. These results suggest that the aspartic proteinase functions only in the living tissues of the grain, as opposed to the 30 kDa cysteine proteinase which is apparently one of the proteases initiating the hydrolysis of storage proteins in the starchy endosperm.     

Toxicity toChrysomela tremulae (Coleoptera: Chrysomelidae) of transgenic poplars expressing a cysteine proteinase inhibitor

The aim of this study was to test the potential of proteinase inhibitors to controlChrysomela tremulae, a beetle that causes severe damage in young plantations and in short-rotation intensive culture (SRIC) of poplar. As a first step, cysteine proteinases were determined to be the major digestive proteinases ofC. tremulae and oryzacystatin OCI, a cysteine proteinase inhibitor, was shown to inhibit this activityin vitro. The gene encoding OCI was introduced into poplar (Populus tremula ×P. tremuloides) and transgenic plants expressing OCI at a high level were selected. Feeding tests on these transgenic plants demonstrate the toxicity of OCI-producing poplar leaves againstC. tremulae larvae.J.C. Leplé and M. Bonadé-Bottino contributed equally to the research presented in this paper.     

High affinity copper binding by stefin B (cystatin B) and its role in the inhibition of amyloid fibrillation

We show that human stefin B, a protease inhibitor from the family of cystatins, is a copper binding protein, unlike stefin A. We have used isothermal titration calorimetry to directly monitor the binding event at pH 7 and pH 5. At pH 7 stefin B shows a picomolar affinity for copper but at pH 5 the affinity is in the nanomolar range. There is no difference in the affinity of copper between the wildtype stefin B (E31 isoform) and a variant (Y31 isoform), whereas the mutant (P79S), which is tetrameric, does not bind copper. The conformation of stefin B remains unaltered by copper binding. It is known that below pH 5 stefin B undergoes a conformational change and amyloid fibril formation. We show that copper binding inhibits the amyloid fibril formation and, to a lesser degree, the initial aggregation. Similarities to and differences from other copper binding amyloidogenic proteins are discussed.     

Nuclear localization of cystatin B, the cathepsin inhibitor implicated in myoclonus epilepsy (EPM1)

Cystatin B is an anti-protease implicated in myoclonus epilepsy, a degenerative disease of the central nervous system. In vitro, cystatin B interacts with and inhibits proteases of the cathepsin family. Confocal microscopy analysis of the subcellular localization of cystatin B and cathepsin B shows that, in vivo, the two proteins are concentrated in different cell compartments. In fact, cystatin B is found mainly in the nucleus of proliferating cells and both in the nucleus and in the cytoplasm of differentiated cells, while cathepsin B, in either case, is essentially cytoplasmic. However, colocalization of cystatin and cathepsin B is observed in the isolated cell matrix and in the nuclear scaffold of differentiated neuroblastoma cells but not of proliferating cells. This suggests that at least a fraction of cystatin B is bound to the protease in differentiated cells. The electron microscopy analysis of the cell matrix confirms the observation made with confocal microscopy. The cellular activity of cathepsin B was analyzed with a fluorogenic cytochemical assay. A fluorescent signal is observed in the cytoplasm of proliferating cells but is undetectable in the cytoplasm of differentiated cells, suggesting that cathepsin B is active mainly during the cell cycle. This result is consistent with the separate compartimentalization of cystatin B and cathepsin B that we have observed in growing cells.     

Inhibition of the cysteine proteinases cathepsins K and L by the serpin headpin (SERPINB13): a kinetic analysis

Headpin (SERPINB13) is a novel member of the serine proteinase inhibitor (Serpin) gene family that was originally cloned from a keratinocyte cDNA library. Western blot analysis using a headpin-specific antiserum recognized a protein with the predicted M(r) of 44kDa in lysates derived from a transformed keratinocyte cell line known to express headpin mRNA. Similarity of the reactive-site loop (RSL) domain of headpin, notably at the P1-P1(') residues, with other serpins that inhibit cysteine and serine proteinases suggests that headpin may inhibit similar proteinases. This study demonstrates that recombinant headpin indeed inhibits cathepsins K and L, but not chymotrypsin, elastase, trypsin, subtilisin A, urokinase-type plasminogen activator, plasmin, or thrombin. The second-order rate constants (k(a)) for the inhibitory reactions of rHeadpin with cathepsins K and L were 5.1+/-0.6x10(4) and 4.1+/-0.8x10(4)M(-1)s(-1), respectively. Headpin formed SDS-stable complexes with cathepsins K and L, a characteristic property of inhibitory serpins. Interactions of the RSL domain of headpin with cathepsins K and L were indicated by cleavage of headpin near the predicted P1-P1(') residues by these proteinases. These results demonstrate that the serpin headpin possesses specificity for inhibiting lysosomal cysteine proteinases.