![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Vol. 18, Issue 9, 3607-3619, September 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,
,
,,
*Institut National de la Santé et de la Recherche Médicale, U563, Toulouse, F-31300 France; Université Toulouse III Paul-Sabatier, Unité Mixte de Recherche-S563, Toulouse, F-31400 France; Centre Hospitalier Universitaire de Toulouse, Hopital Purpan, Departement de Génétique Médicale, Toulouse, F-31000 France; ||Université Toulouse III Paul-Sabatier, Faculté de Médecine Toulouse-Rangueil, Institut Louis Bugnard (IFR31), Toulouse, F-31400 France; ¶Wellcome Trust Centre for Human Genetics, Oxford OX3 7BN, United Kingdom; #Department of Head and Neck Surgery, M. D. Anderson Cancer Center, Houston, TX 77030; @Arexis AB/Biovitrum, 413 46 Gothenburg, Sweden; **Department of Public Health and Clinical Medicine, Section for Dermatology and Venereology, Umeå University, SE-901 87 Umeå, Sweden; and Department of Dermatology, Vrije Universiteit Brussels, 1090 Brussels, Belgium
Submitted February 14, 2007;
Revised June 11, 2007;
Accepted June 18, 2007
Monitoring Editor: M. Bishr Omary
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). LEKTI, belongs to the Kazal serine proteinase inhibitor family whose numerous members generally bear 3–7 tandem kazal domains. Interestingly, LEKTI contains a signal peptide and exhibits as many as 15 potential serine proteinase inhibitory domains (D1–D15) separated by 14 spacing segments. These linker regions are characterized by the presence of several dibasic residues, potentially sensitive to the cleavage by the subtilisin-like proprotein convertases. Two of these domains (D2 and D15) resemble typical Kazal-type serine proteinase inhibitors, as deduced from their primary structure and characteristic pattern of six cysteine residues. The other 13 domains share high homology with this class of inhibitors but lack one of the three conserved typical disulfide bridges. Despite this difference, these sequences adopt a hairpin structure, creating an inhibitory binding loop (Lauber et al., 2003). Several authors have studied the inhibitory capacity of different forms of LEKTI. The full-length LEKTI recombinant protein has been shown to inhibit trypsin, subtilisin A, plasmin, cathepsin G, and neutrophil elastase, but not chymotrypsin (Mitsudo et al., 2003). A partial recombinant form of LEKTI containing domains 6–9 (rLEKTI6–9) has been shown to inhibit trypsin, subtilisin A, chymotrypsin, kallikrein 5 (KLK5, stratum corneum tryptic enzyme [SCTE]), and kallikrein 7 (KLK7, stratum corneum chymotryptic enzyme [SCCE]) but not plasmin, cathepsin G, or elastase (Jayakumar et al., 2004; Schechter et al., 2005). In addition, the single domain D6 was shown to be a potent inhibitor of trypsin, KLK5, and KLK7, whereas D15 was not effective against these two kallikreins (Egelrud et al., 2005). Altogether, these results highlight the fact that each form of LEKTI exhibits its own inhibitory specificity.
LEKTI is expressed specifically in the most differentiated viable layers of stratified epithelial tissues and in the Hassal corpuscules of the thymus (Bitoun et al., 2003). In the epidermis, it is mainly restricted to the granular layer (GR), where critical biochemical and morphological changes that occur during terminal differentiation lead to cornification (stratum corneum [SC] formation). LEKTI is transported by specific intracellular lamellar granule cargoes until its secretion in the extracellular space, between granular cells and cornified cells (Ishida-Yamamoto et al., 2005).
In normal human keratinocytes (NHK), LEKTI is expressed as three precursors derived from alternative pre-mRNA processing (Tartaglia-Polcini et al., 2006). In addition to the previously described, full-length isoform of 15 domains (145 kDa), SPINK5 encodes a shorter LEKTI isoform (125 kDa) composed of the first 13 domains generated from the use of an alternative polyadenylation signal, as well as a longer isoform (148 kDa) carrying a 30-amino acid residue insertion between the 13th and the 14th inhibitory domains, generated from the activation of cryptic splice junction sequences. RNase protection assay experiments have identified the full-length transcript as the most abundant isoform in NHK (Tartaglia-Polcini et al., 2006). LEKTI precursors are rapidly processed into proteolytic fragments in a postendoplasmic reticulum compartment (Bitoun et al., 2003; Jayakumar et al., 2005). C-terminal LEKTI fragments have been detected in the conditioned medium of NHK (Tartaglia-Polcini et al., 2006). In addition, a 30-kDa polypeptide with an N-terminal extremity, corresponding to D8, has also been purified from NHK-conditioned medium (Ahmed et al., 2001).
SPINK5 mutations lead to Netherton syndrome (NS; Chavanas et al., 2000), a severe autosomal recessive skin disorder characterized by congenital ichthyosiform erythroderma, a specific hair shaft defect (trichorrhexis invaginata) and atopic manifestations (Traupe, 1989). To decipher the biological functions of LEKTI, we have genetically engineered mice with a targeted disruption of Spink5. Spink5-null mice faithfully replicate key features of Netherton syndrome, including abnormal desquamation, impaired keratinization, hair malformation, and a severe skin barrier defect. LEKTI deficiency causes abnormal desmosome cleavage in the upper GL through desmoglein 1 degradation due to the hyperactivity of KLK5 and KLK7. This leads to accelerated SC shedding and consequent loss of skin barrier function (Yang et al., 2004; Descargues et al., 2005; Hewett et al., 2005). This work identified LEKTI as a key regulator of epidermal protease activity. In addition, the presence of LEKTI domains (D1, D5, and D6) in the blood circulation (Magert et al., 1999, 2002) suggests that LEKTI could also have biological effects at a distance from the skin. The extent of atopic manifestations in NS predicts a role for LEKTI as an inhibitor of proteases involved in the inflammation process.
To gain further insight into LEKTI functions, we studied the inhibitory properties of physiological LEKTI fragments. The different LEKTI forms present in the epidermis, in NHK, as well as in a mammalian heterologous expression system were detected using three antibodies directed against the N-terminal, the internal and the C-terminal part of full-length LEKTI. We produced several of these physiological LEKTI proteolytic fragments and characterized their inhibitory properties against a large panel of serine proteinases involved in skin homeostasis and inflammation. Kinetic parameters of the interaction between LEKTI fragments and their proteinase targets were studied by surface plasmon resonance (SPR). Using the same technology, we showed that the pH gradient occurring through the SC controls the interaction strength between LEKTI and KLK5, thus allowing the controlled release of active proteinase in the most superficial layers of SC.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
), a rabbit polyclonal antibody raised against D8–D11 LEKTI domains (Ishida-Yamamoto et al., 2005) and a rabbit polyclonal antibody raised against D13–D15 LEKTI domains (Bitoun et al., 2003). Trypsin, tryptase, chymotrypsin, plasmin, thrombin, neutrophile elastase, cathepsin G, and kallikreins 1 and 3 were purchased from Sigma-Aldrich (St. Louis, MO). KLK8 was purchased from R&D Systems (Minneapolis, MN). Recombinant kallikreins KLK5, KLK7, and KLK14 were obtained as previously described (Brattsand et al., 2005). All chromogenic substrates were commercially obtained (Sigma-Aldrich).
Human Epidermis and Cell Culture
Normal and NS human primary keratinocytes were isolated from skin biopsies and cultured in Green medium containing 1.2 mM calcium as previously described (Bitoun et al., 2003). Chinese hamster ovary (CHO) and furin-deficient CHO cells were a generous gift of Dr. Leppla (Laboratory of Microbial Ecology, National Institute of Dental Research, NIH, Bethesda, MD 20892) (Gordon et al., 1995). Cells were grown in F12 medium (Invitrogen, Carlsbad, CA) supplemented with 10% fetal calf serum and 100 U/ml penicillin/streptomycin. Human foreskin was obtained after medical surgery at Purpan Hospital (Toulouse, France). The epidermis was mechanically separated from the dermis after heating at 55°C for 15 min. The medical ethical committee CCPPRB (Comité consultatif de personnes se prêtant à des recherches biomédicales) of Toulouse hospitals approved all described studies (research project no. 0102908). The study was conducted according to the Declaration of Helsinki principles.
Cloning, Expression, and Purification of LEKTI Domains
Partial LEKTI cDNA fragments encoding the following human LEKTI fragments: D1 (residues 23-77), D5 (residues 292-353), D6 (residues 356-423), and D8–D11 (residues 490-759) were amplified by PCR from a vector containing the full-length cDNA as a template; amino acid numbering is according to Magert et al. (1999). PCR products of D5 and D6 were cloned into pGEX in order to use glutathione S-transferase (GST) as a C-terminal tag. D1 and D8–D11 PCR products were cloned into pET22b, to introduce an in-frame N-terminal His6 tag. All the constructs were transformed into the Origami bacterial strain (Novagen, Madison, WI) in order to produce recombinant LEKTI fragments. Previous studies have demonstrated that similar prokaryotic expression systems are suitable to obtain functional LEKTI fragments (Kreutzmann et al., 2004; Egelrud et al., 2005).
Production of recombinant LEKTI fragments was induced by adding 1 mM isopropyl-1-thio-
-D-galactopyranoside (IPTG) to Origami cultures (Novagen). After 3 h at 30°C, cells were harvested, lysed in PBS containing 0.1 mM EGTA, 0.25% Tween 20, 100 µg/ml lysosyme, and subjected to sonication for optimal solubilization. For GST fusion proteins, the soluble fraction was loaded onto a glutathione Sepharose 4B column according to the recommendations of the MicroSpin GST purification module (GE Healthcare, Waukesha, WI). For His-fusion proteins, the soluble fraction was subjected to nickel affinity (Chelating Sepharose Fast Flow, GE Healthcare, Waukesha, WI). Elution was performed using a gradient of imidazole (0–500 mM in PBS) on fast protein liquid chromatography (FPLC). Nickel-eluted fractions were subsequently loaded onto a cation exchange column (SP Sepharose High performance, GE Healthcare), for which a gradient of NaCl (0–2 M) was carried out to elute proteins. Finally, size exclusion chromatography (HiLoad 16/60 Superdex 75 pq, GE Healthcare) was used to obtain pure protein fractions, which were pooled for further experiments. Purified proteins were analyzed by SDS-PAGE after Coomassie Blue staining. HisD9–D15 was purified following the experimental procedures as previously described (Jayakumar et al., 2004). Proteins were dialyzed against a solution of HEPES 10 mM, pH 7.4, for inhibitory activity assay and Biacore analysis.
Cloning of LEKTI cDNAs into pEF-DEST51 Expression Vector
LEKTI full-length cDNA (GenBank NM_006846) or LEKTI short-length (GenBank DQ149929) were amplified by long-range PCR (PfU Turbo, Stratagene, La Jolla, CA) from a vector containing the full-length cDNA as a template. The PCR products were subcloned into pDEST8 vector by homologous recombination and transferred into the mammalian expression vector pEF-DEST51 using the Gateway technology, according to the manufacturer's instructions (Invitrogen). pEF-DEST51-SPINK5f-l and pEF-DEST51-SPINK5sh constructs were fully sequenced using the Big Dye Terminator Sequencing Kit and an ABI 3100 automated sequencer (Applied Biosystems, Foster City, CA).
CHO Cell Transfections
One day before transfection, CHO cells were plated at 5 x 104 cells per well of a six-well plate. The day of the transfection, cells at 70% confluence were transiently transfected with 1 µg of pEF-DEST51, pEF-DEST51-SPINK5f-l, or pEF-DEST51-SPINK5sh plasmid DNAs, using the FuGENE 6 Transfection Reagent, according to the manufacturer's recommendations (Roche Applied Science, Indianapolis, IN). Twenty-four hours after transfection, the medium was replaced with serum-free medium, and cells were maintained in culture for an additional 24 h. Both intracellular and extracellular protein extracts were prepared for Western blot analysis.
Western Blotting
Epidermis was crushed in a protein extraction buffer (PEB) containing Tris-Cl 50 mM, pH 8, NaCl 150 mM, EDTA 5 mM, pH 8, 1% NP40, 1 mM PMSF, 10 µg/ml leupeptin, 10 mg/ml pepstatin A, and 1 mg/ml antipain with an Ultra-Turrax. Cultured cells were lysed in PEB. Lysates were clarified from insoluble material by centrifugation at 13000 g, 4°C for 5 min. The conditioned medium was concentrated by overnight acetone precipitation. Proteins were recovered by centrifugation at 13000 g, 4°C for 30 min, and resuspended in lysis buffer. Proteins were quantified by Bradford protein assay kit (Bio-Rad Laboratories, Hercules, CA). Protein fractions were mixed with Laemmli buffer (Bio-Rad Laboratories), incubated for 5 min at 65°C, and then separated by SDS-PAGE. After migration, proteins were transferred to Hybond-C extra membranes (GE Healthcare). After incubation with primary and secondary antibodies, enhanced chemiluminescence detection was performed as recommended by the manufacturer. PNGase F (New England Biolabs, Beverly, MA), O-glycosidase, and 
-(2
3,6,8,9)-neuraminidase (Sigma-Aldrich) treatments were performed at 37°C, according to the manufacturer's instructions.
Proteinase Activity Assay
Varying concentrations of substrates (Table 1) were incubated with a fixed amount of proteinase in a suitable buffer activity, and initial velocities were measured by monitoring the absorbance at 405 nm. Double reciprocal Lineweaver-Burke plots of 1/[V] versus 1/[S] were used to determine the Km of each substrate for its partner enzyme. Affinity constant (Km) between an enzyme and a substrate is defined as the substrate concentrate at 1/2 maximum velocity. Six separate mixtures of enzymes and inhibitors in various ratios were incubated for 5 min. The proteinase activity was initiated by adding the appropriate synthetic substrate (Table 1), and the activity of free enzyme was determined spectrophotometrically at 405 nm by monitoring the release of p-nitrophenyl acetate (pNA). All time courses were performed at 25°C, during 15 min, in duplicate. Reaction velocities were linear over the course of the reaction. Initial velocities were measured by monitoring absorbance at 405 nm, and IC50 was calculated by plotting [V0/Vi]-1 versus [I]. To account for the effect of substrate Km on the inhibition constant, IC50 were converted to Ki using the formula (Morris et al., 2002):
 |
|
Surface Plasmon Resonance Analysis
Principle.
Binding events between two molecules are monitored in real time, without the use of any label, using an optical phenomenon called SPR. Biomolecular binding events cause changes in the refractive index close to the surface layer of a chip, which are detected as changes in the SPR signal. During a binding analysis SPR changes occur as a solution is passed over the surface of a sensorchip. To perform an analysis, one interactant (ligand) is immobilized over a carboxymethylated dextran matrix of a sensorchip. The sensor surface forms one wall of a flow cell. Sample containing the other interactant (analyte) is injected over this surface in a precisely controlled flow. The progress of an interaction is monitored as a sensorgram that expresses resonance units (RU) as a function of time. Analyte binds to the surface-attached ligand during sample injection, resulting in an increase in signal. At the end of the injection, the sample is replaced by a continuous flow of buffer, and the decrease in signal reflects dissociation of interactant from the surface-bound complex.
Materials. All binding studies based on SPR phenomenon were performed on a four-channel BIACORE 3000 optical biosensor instrument (BIAcore AB, Uppsala, Sweden). All experiments were performed on sensorchips CM5 obtained from Biacore AB.
Immobilization of Recombinant LEKTI Domains. Both flow cells of a CM5 sensor chip were coated with recombinant proteins by amine coupling, allowing immobilization of the proteins in the same orientation, independent of the tag fused to the protein. Various levels of RU were immobilized to take into account the differences in molecular weights: His-D1 (500 RU), GST-D5 (2100 RU), GST-D6 (2100 RU), His-D8–D11 (2000 RU), and His-D9-D15 (3800 RU).
BIA Analysis. Binding analyses were performed with multiple injections of different protein concentrations over the immobilized surfaces at 15°C. All samples were diluted in HBS-EP buffer (HEPES 10 mM, NaCl 150 mM, EDTA 3 mM, and polysorbate 0.005%) and were injected over the sensor surface for 3 min at a flow rate of 30 µl/min. All diluted samples were injected at the same time over the four channels (flow cells). A gradient of 0.05% to 0.5% SDS in HBS-EP buffer was used to regenerate the chip. D1 was considered as a negative control according to its inability to affect proteinase activity. D1 sensorgrams were subtracted from sensorgrams obtained with immobilized fusion proteins to yield true binding responses. Kinetics constants (ka, kd, KD = kd/ka) were calculated using BIAevaluation 4.0.1 software and the 1/1 Langmuir binding model was chosen. This model determines the association constant (ka) and takes into account the dissociation occurring during the association phase. Therefore, the calculated values do not necessary correlate with apparent slope of the sensorgram.
Casein Gel Zymography
Epidermis from wild-type (WT) and knockout (KO) animals was crushed in 1 M acetic acid solution with an Ultra-Turrax. After overnight extraction at 4°C, soluble proteins were lyophilized and resuspended in PBS. After acetone precipitation, proteins were assayed (Bradford, Bio-Rad), and 5 µg of soluble fractions were mixed in a nondenaturing loading buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, and 0.1% bromophenol blue) were loaded onto casein copolymerized with acrylamide gels (15% acrylamide, 0.05% -casein, Sigma-Aldrich) for electrophoresis. Gels were washed with 2.5% Triton X-100 for 1 h to remove SDS and incubated 24 h at 37°C in a reaction buffer containing 50 mM Tris, pH 8. Gels were stained with 1% Coomassie Brilliant blue for 30 min. Areas of caseinolytic activity appeared as clear zones against a dark blue background. To assess inhibitory capacity of D8–D11 LEKTI domain, a solution of D8–D11 fragment (5 µM) was added to the sample before electrophoresis (15 min on ice), as well as in the reaction buffer.
In Situ Zymography
Frozen sections of WT or Spink5–/– mouse skin (5-mm thickness) were rinsed with a washing solution (2% Tween 20 in deionized water) and incubated at 37°C overnight with 100 µl of BODIPY FL casein using the EnzChek Ultra Protease Assay kit (Invitrogen) in 50 mM Tris-Cl, pH 8, in order to visualize global protease activity. Cryostat sections were incubated in the same conditions with 100 µl of Boc-Val-ProArg-AMC or Suc-Leu-Leu-Val-Tyr-AMC (Sigma-Aldrich) at 100 mM in Tris 50 mM, CaCl2 10 mM for the detection of trypsin- and chymotrypsin-like activity, respectively. For some sections, LEKTI D8–D11 fragment (5 µM) was added to the substrate in order to assess its inhibitory capacity. All sections were rinsed with PBS solution and visualized with the inverted high-end microscope Axiovert 200 (Zeiss, Thornwood, NY) at an excitation wavelength of 485 and 400 nm and an emission wavelength of 530 and 460 nm for BODIPY FL and AMC fluorescent dyes, respectively. Frozen sections from WT and KO skin were photographed at equal time points and exposure time. Images were captured and analyzed with Metamorph Imaging system software, version 3.6 (Universal Imaging, West Chester, PA). The intensity of the fluorescence signals was coded as color gradient, ranging from 0 (dark) to 255 (white).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
D1–D6, D8–D11, and D13–D15; Figure 1, lane 1). The same analysis was carried out using intracellular and extracellular fractions of differentiated normal human keratinocytes in culture (NHK; Figure 1, lane 2). Signal specificity was confirmed using extracts of NS keratinocytes, which do not express LEKTI (Bitoun et al., 2003; data not shown).
|
), respectively, and were not detected in human epidermal extracts.
LEKTI fragments of 10, 11, 13, 15, and 20 kDa were detected in human epidermis with D1–D6 antibodies (Figure 1, lane 1). The same bands were detected only in the extracellular fraction of NHK, which indicated that these LEKTI N-terminal fragments are secreted (Figure 1, lane 2). Using D8–D11 antibodies, 20-, 31-, 37-, 42-, and 65-kDa proteolytic fragments were detected in human epidermis and in the conditioned medium of NHK (Figure 1, lanes 1 and 2).
In the epidermis and in the extracellular fraction of NHK, D13–D15 antibodies detected two C-terminal fragments at 
65 and 42 kDa (Figure 1, lanes 1 and 2). Altogether, these results highlight a similar proteolytic processing of LEKTI in human epidermis in vivo and in NHK in vitro. They also demonstrate the heterogeneity in molecular weights of LEKTI proteolytic fragments. In the C-terminal part of LEKTI, high-molecular-weight proteolytic fragments were produced (65 or 42 kDa) and were not cleaved in a larger extent, in contrast to N-terminal extremity, the proteolytic processing of which produced several small LEKTI fragments.
No LEKTI precursor could be detected in the epidermis with any of the LEKTI antibodies used, suggesting that they are rapidly processed into proteolytic fragments. The fact that LEKTI precursors were detected only in the intracellular fraction of NHK and that LEKTI proteolytic fragments were observed exclusively in the extracellular fraction confirms that LEKTI processing takes place intracellularly (Bitoun et al., 2003) and suggests a rapid secretion of fragments upon cleavage of the precursors. These observations support the notion that secreted LEKTI fragments are the relevant LEKTI biologically active forms.
Heterologous Expression of LEKTI in CHO Cells Reproduces Physiological LEKTI Processing
To discriminate LEKTI fragments deriving from the full-length LEKTI precursor from those deriving from the shorter LEKTI precursor, we developed an heterologous system for LEKTI expression. Transient transfection of CHO cells was performed with mammalian expression vectors carrying the full-length (pEF-DEST51-LEKTIf-l) or short-length (pEF-DEST51-LEKTIsh) LEKTI cDNAs under the control of the Elongation Factor 1 promoter. Intracellular and extracellular fractions of transfected CHO cells were analyzed by Western blotting using the three anti-LEKTI antibodies (D1-D6, D8–D11, and D13–D15 antibodies; Figure 1, lanes 3–5). Bands of 145 and 125 kDa were detected in the intracellular fraction of CHO cells transfected with pEF-DEST51-LEKTIf-l and pEF-DEST51-LEKTIsh, respectively. These high-molecular-weight signals correspond to LEKTI precursors. They were not detected in the conditioned medium, indicating that proLEKTI is processed intracellularly, as observed in NHK. Using D1–D6 antibodies, 10-, 13-, and 20-kDa fragments were detected in both the extracellular fraction of CHO transfected with pEF-DEST51-LEKTIf-l and pEF-DEST51-LEKTIsh. These fragments are present in the NHK extracellular fraction (Figure 1, lane 2), and their size is consistent with LEKTI physiological cleavage. However, the 11- and 15-kDa fragments detected in NHK were not visualized in CHO extracts.
Using D8–D11 antibodies, 20-, 31-, and 37-kDa bands were detected in the medium of CHO transfected with pEF-DEST51-LEKTIf-l and pEF-DEST51-LEKTIsh (Figure 1, lanes 3–5).
In addition, D8–D11 and D13–D15 antibodies (Figure 1, lanes 3–5) detected a 65-kDa fragment in the medium of CHO transfected with pEF-DEST51-LEKTIf-l and a 42-kDa fragment in the extracellular fraction of CHO transfected with pEF-DEST51-LEKTIsh, as shown by Tartaglia et al. (2006).
As a result, heterologous expression of LEKTIf-l and LEKTIsh in CHO cells overall reproduces the proteolytic processing of LEKTI precursors seen in NHK and human epidermis. In addition, it provides evidence that both precursors are submitted to similar proteolytic processing, generating fragments of similar molecular weight. The only difference observed with the C-terminal fragments is due to the lack of D14 and D15 domains in the shortest precursor LEKTIsh.
Furin Is a Key Enzyme for LEKTI Intracellular Proteolytic Processing
Subtilisin-like proprotein convertases (SPCs) are a family of endoproteinases involved in the processing of a variety of proproteins. Among them, furin has been proposed as a good candidate for the proteolysis of LEKTI (Bitoun et al., 2003). To demonstrate the involvement of furin in LEKTI processing, pEF-DEST51-LEKTIf-l and pEF-DEST51-LEKTIsh vectors were used to transfect furin-deficient CHO cells. Intracellular and extracellular fractions were analyzed by Western blot using the three anti-LEKTI antibodies (Figure 2). In contrast to the various LEKTI proteolytic fragments visualized in the extracellular fraction of transfected CHO (Figure 2, lanes 1–3), no proteolytic fragment could be observed in the medium of transfected furin-deficient CHO cells (Figure 2, lanes 4–6). Instead, the 145- and 125-kDa LEKTI precursors were detected in the extracellular fractions of furin-deficient CHO cells transfected with pEF-DEST51-LEKTIf-l and pEF-DEST51-LEKTIsh, respectively, with each of the three antibodies used. These results showed that furin-deficient CHO cells are unable to process LEKTI, and that unprocessed precursors are secreted even so. Nevertheless, the furin-deficient CHO cells that we used were generated by ethyl methane sulfonate mutagenesis, and it is indeed possible that mutations elsewhere than in the furin gene could have occurred. To eliminate the possibility that such mutations were responsible for the lack of LEKTI processing, pEF-DEST51-LEKTIf-l was transfected in furin-deficient CHO cells stably retransfected with a murine cDNA of furin (Gu et al., 1995). Extracellular fractions were analyzed by Western blot and revealed LEKTI processing rescue (Supplementary Figure 1). All together, these results prove evidence that furin plays a major role in LEKTI physiological processing.
|
|
D1–D6 antibodies were used (data not shown). On the basis of their molecular weights, some of these unglycosylated N-terminal fragments could correspond to single LEKTI domains D1, D5, or D6, previously identified (Magert et al., 1999; Figure 1, right panel).
In contrast, when D8–D11 antibodies were used, a 2-kDa shift of the 31- and 37-kDa bands was detected after N-deglycosylation. This treatment had no effect on the 20-kDa band (Figure 3B). No shift was observed for any of these three bands after O-deglycosylation (data not shown). The N-glycosylation result suggests that the 31-kDa fragment detected in epidermis and NHK correspond to LEKTI D8–D11 fragment (predicted molecular weight of 30 kDa). This is consistent with the description of the 30-kDa fragment reported in NHK conditioned medium, the N-terminal of which corresponds to D8 (Ahmed et al., 2001).
Finally, N-deglycosylation treatment resulted in a reduction of 4 kDa of the 65- and 42-kDa fragments detected with D13–D15 antibodies, demonstrating that these C-terminal fragments are N-glycosylated (Figure 3B). O-deglycosylation reduced the molecular weight of the 65-kDa C-terminal LEKTI fragment signal to 61 kDa, whereas the 42-kDa band was not affected (Figure 3C). This result is concordant with the O-glycosylation predicted sites on D15, which are absent from the shortest LEKTI isoform.
Each of the N- and O-deglycosylation experiment showed a 4-kDa shift of the 65-kDa fragment, revealing that this fragment carries several glycosylated residues accounting for 8 kDa. The molecular weight of the unglycosylated fragment (57 kDa) is concordant with the one calculated for the primary sequence of the last seven C-terminal domains of LEKTI (D9–D15), which is 57.4 kDa. To summarize, deglycosylation experiments lead us to propose that D1, D5, D6, D8–D11, and D9–D15 are physiological LEKTI domains derived from proteolytic processing of the full-length precursor.
LEKTI Domains, Except D1, Inhibit Epidermal Kallikreins
Based on the homology with other Kazal family members, LEKTI domains are predicted to inhibit serine proteases. In the Kazal family, the cognate protease is dictated by an amino acid occupying the P1 position.1 Among all LEKTI domains, only D1, D2, and D15 do not possess an arginine at this position. Kazal inhibitor bearing an arginine at P1 position are known to inhibit trypsin. However, prediction is not simple because interactions between the inhibitory loop of the inhibitor and the active site of the inhibited protease also depend on the microenvironment of the complex. Moreover, the behavior of multidomain inhibitors toward their targets is not easily predictable.
To study the anti-protease function of physiological LEKTI fragments, we expressed recombinant LEKTI domains D1, D5, D6, D8–D11, and D9–D15. To provide evidence for the formation of disulfide bridges in the LEKTI fragments produced in prokaryotic expression system, these fragments were submitted to SDS-PAGE under reducing or nonreducing conditions (Supplementary Figure 2). After Coomassie staining, a single band was observed at the expected molecular weight with a slight difference between reducing and nonreducing conditions. This is consistent with the fact that the recombinant LEKTI fragments expressed in the Origami bacteria contained intramolecular disulfide bonds.
The capacity of purified recombinant LEKTI fragments to function in vitro as a serine protease inhibitor was then assessed against a large set of 13 serine proteinases involved in skin desquamation and inflammation: trypsin, tryptase, chymotrypsin, plasmin, thrombin, neutrophile elastase, cathepsin G, and kallikreins 1, 3, 5, 7, 8, and 14 (Table 1). Inhibition tests showed that the different LEKTI fragments were not equally effective against target proteinases: D1 was unable to inhibit any proteinase of the set, despite evidence for the formation of disulfide bridges (Supplementary Figure 2). In contrast, the other fragments inhibited trypsin, KLK5, KLK7, and KLK14 to various extents. The other proteinases tested in the panel were not inhibited by LEKTI fragments. The strongest inhibitory activity was observed with D8–D11 against KLK5 and KLK14 with Ki values of 3.7 and 3.1 nM, respectively (Figure 4). Despite a 68% sequence identity from the first to the fourth cysteine residues and an identical cysteine connectivity pattern, D5 and D6 differentially inhibited target proteinases. D6 appeared as a weaker inhibitor of trypsin, KLK5, and KLK7 than D5 and was not effective against KLK14. Specificity and capacity of inhibition of D6 are concordant with previously published data (Kreutzmann et al., 2004; Egelrud et al., 2005).
|
|
D1, D5, D6, D8–D11, and D9–D15 fragments were immobilized onto a sensorchip, and their binding capacity toward KLK5, KLK7, and KLK14 was tested in real time. The D1 LEKTI fragment, which was devoid of inhibitory capacity, was considered first. As expected, D1 was unable to bind any proteinase and was then considered to be a negative control in each BIA experiment.
Similarly to specific inhibition profiles of each LEKTI fragment, LEKTI fragments were not equally effective in their interaction capacities toward the tested proteinases (Figure 5). As illustrated by sensorgrams, kinetics profiles and amounts of bound proteinase molecules were highly variable from one LEKTI fragment to another for each proteinase.
|
The interaction between single LEKTI domains D5 and D6 with KLK7 is weaker than the one determined with KLK5, as shown by their higher affinity constant (KD) for KLK7. This difference is not due to their dissociation constants (kd) toward KLK5 and KLK7, which are similar, but instead, to their weaker association constant (ka) for KLK7 (KD = kdx 1/ka).
Sensorgrams of KLK14 interacting with different LEKTI fragments showed very particular profiles, different from those previously described for KLK5 and KLK7. Specifically for KLK14/D9–D15 interaction, the SPR signal abnormally decreased during the association phase before the end of injection, indicating a proteolytic activity of KLK14 against D9–D15. D9–D15 degradation was confirmed when KLK14 was injected at 20 nM and showed a loss of ligand immobilized during the dissociation phase, with a level of the SPR signal below the baseline. This phenomenon was less visible on the other sensorgrams but certainly occurred and prevented kinetic constant measurement.
pH Dependent–binding of KLK5 and LEKTI Is a Key Factor for the Regulation of the Desquamation Process
LEKTI and KLK5 are transported in different cargo vesicles before they are released into the extracellular space, at the stratum granulosum–SC interface. Immunoelectron microscopy experiments showed localization of the two molecules at this interface near corneodesmosomes in human skin (Ishida-Yamamoto et al., 2005). KLK5 is a major proteinase involved in the desquamation process, through the cleavage of the corneodesmosomal components (Caubet et al., 2004). To allow the detachment of the superficial layers of SC, KLK5 activity must be tightly controlled, in a spatially and temporally manner. Recent studies have suggested that the ultimate desquamation of corneocytes from the SC surface may be orchestrated by localized changes in pH (Elias, 2004; Hachem et al., 2004). In this context, we analyzed the influence of pH on the stability of KLK5 and D8–D11 interaction using the BIAcore technology (Figure 6). Injection of KLK5 at 2.5 nM was performed at pH 7.5, a pH occurring at the GR-SC interface. At the end of the injection, a buffer adjusted at different pH was injected, and the dissociation phase was followed for various pH conditions (7.5–4.5), miming the pH gradient in the SC layers from the depth to the surface. The sensorgrams presented in Figure 6A clearly show that the interaction between the two partners is affected by pH variable conditions, with an acceleration of the dissociation phase concomitantly with pH decrease. On release, both partners are potentially able to reform a complex at lower pH values. Therefore, the interaction (association and dissociation) was evaluated for each pH value (Figure 6B). The calculated values showed that the association decreases, whereas dissociation increases with acidification (pH 7.4 to pH 4.5). The effect is much stronger on dissociation (10–8 to 10–4 s–1) than on association (104 to 103 M–1 s–1). Therefore, acidification decreases the strength of binding between LEKTI D8–D11 and KLK5, by favoring dissociation of the complex. To test whether the binding between LEKTI D8–D11 and KLK7 was also pH-sensitive, we performed a binding experiment between these two partners at the same pH values (Figure 6C). Similarly, a decreased interaction was observed with acidification, with a high effect from pH 5.5.
|
The highest inhibitory capacity for LEKTI D8–D11 was obtained at pH 7.5 and declined at inferior pH values, to become very weak at pH 4.5. This is concordant with the binding profile determined by BIA experiment and confirmed that the interaction and the inhibition capacity of LEKTI D8–D11 are optimal at neutral pH, which corresponds to the pH of the GR–SC interface. SC acidification allows active proteinases to be released from their inhibitor as a result of increased complex dissociation.
LEKTI Fragments Inhibit Native Epidermal Proteinases of the Stratrum Corneum
To confirm the inhibitory capacity of D8–D11 toward disregulated proteinase activities in NS, in situ zymography was carried out on cryosections of Spink5–/– mouse skin. In situ zymography using fluorescein isothiocyanate (FITC)-conjugated casein as a substrate showed that proteinase activity was remarkably increased in the epidermis of Spink5–/– mice in comparison with WT epidermis (Figure 7B). This enzymatic activity mainly localized to the SC. Trypsin and chymotrypsin activities were then assessed with a synthetic substrate conjugated with AMC. Activities were markedly increased in the SC of KO mice compared with WT (Figure 7E,H). Addition of 5 µM D8–D11 LEKTI fragment on KO cryosection resulted in an important decrease in signal intensity (Figure 7, C, F, and I). The substrates (for trypsin and chymotrypsin) used in this study are preferentially cleaved by KLK5 and KLK7 (Debela et al., 2006). However, we cannot exclude the possibility that other proteinase activities may degrade these substrates on skin cryosections.
|
). The addition of LEKTI D8–D11 completely abolished KLK5 activity, whereas a residual activity of KLK7 persisted. A 28-kDa protease also overactivated in KO epidermis was not affected by D8–D11. These results confirmed the involvement of D8–D11 LEKTI domain in the control of native KLK5 and KLK7 activities, showing a stronger inhibition of KLK5, concordant with the inhibition data.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). It is specifically expressed in the GR of the epidermis (Bitoun et al., 2003). Absence of LEKTI expression is responsible for the severe skin disease NS. The study of NS patients as well as the analysis of a mouse model of NS provided evidence that LEKTI plays a key role in the control of the desquamation process (Descargues et al., 2005, 2006). To gain further insights into the role of LEKTI in skin homeostasis, we first focused on the characterization of LEKTI active forms and carried out functional analysis of LEKTI fragments. In NHKs, we showed that LEKTI precursors are rapidly processed intracellularly into several proteolytic fragments, the molecular weights of which are concordant with single domains or several domains linked together. We showed that furin plays a major role in the generation of these proteolytic LEKTI fragments, which is concordant with immunolocalization of furin in the GR (Pearton et al., 2001). In addition, the presence of 11- and 15-kDa N-terminal LEKTI fragments in the epidermis and in NHK that are not detected in CHO cells suggests that (an)other unidentified endoprotease(s) specifically expressed in keratinocytes could account for their production.
Once processed, all LEKTI proteolytic fragments are secreted in the conditioned medium of NHK. In human foreskin epidermis, LEKTI precursors are not detectable, in contrast to the numerous proteolytic fragments. These results indicate that LEKTI fragments, rather than LEKTI precursors, are the biologically relevant LEKTI forms, which are secreted at the GR-SC interface, as shown by immunoelectron microscopy (Ishida-Yamamoto et al., 2005). Our study provides evidence for N- and O-glycosylation of LEKTI fragments. Protein glycans can play several roles (Lee et al., 2001), including promoting protein folding into proper structure or protecting against proteolytic enzymes. The glycans present on LEKTI could then prevent the molecule from furin proteolytic action at sensitive sites in domain-linking regions. This could explain why LEKTI precursors are not processed into 15 domains as proposed (Komatsu et al., 2002), but rather into several multidomain and single domain fragments.
Combination of Western blot analyses, molecular weight data, and glycosylation status lead us to propose the identity of some physiological LEKTI fragments. D1, D5, and D6 have been isolated from human blood; however, no cellular blood compartment has been shown to express LEKTI so far (Tartaglia-Polcini et al., 2006). Therefore, D1, D5, and D6 circulating LEKTI fragments could originate from LEKTI-expressing tissues such as the epidermis and correspond to the smallest signals detected with D1–D6 in the epidermis and NHK extracts. In addition to these single LEKTI domains, using internal and C-terminal antibodies, other LEKTI signals detected in the epidermis and in NHK were concordant with being new physiological LEKTI fragments corresponding to D8–D11 (31 kDa) and D9–D15 (65 kDa).
The physiological inhibition of proteinases depends on several parameters, including temporal and spatial colocalization of the protease and its inhibitor and binding kinetics between the partners involved. The goal of our study was to analyze the inhibitory capacity of the proposed physiological LEKTI fragments and to determine the associated kinetics parameters. In addition to pancreatic trypsin, among all proteases included in the panel studied, KLK5, KLK14, and KLK7 were the only proteinases inhibited by LEKTI fragments. Interestingly, these three proteinases are specifically detected in the GR of the skin and colocalize with LEKTI (Brattsand et al., 2005; Ishida-Yamamoto et al., 2005; Komatsu et al., 2005). Each LEKTI fragment presents a specific inhibitory profile toward these three epidermal proteinases. D1 is devoid of inhibitory capacity as anticipated by its particular 3D structure (Lauber et al., 2003). Except for this domain, the other fragments appear to have a higher inhibitory capacity toward trypsin-like proteases (KLK5, KLK14) compared with chymotrypsin-like proteases (KLK7). Surprisingly, whereas D6 inhibits KLK5, it is not active against KLK14, although KLK5 and KLK14 belong to the same family and share 65% similarity. This discloses a very fine specificity of interaction between Kazal domains and kallikreins.
Our results show that all LEKTI fragments studied, except D1, inhibit KLK5. D8–D11 demonstrated the highest inhibitory capacity with a Ki as low as 3 nM. This result correlates with the rapid and irreversible interaction occurring between the two partners. Although SPR technology did not allow determining the kinetics parameters of the interaction between KLK14 and LEKTI fragments, D5, D8–D11, and D9–D15 LEKTI domains displayed a high inhibitory capacity toward this proteinase. Taken together, these results identify KLK5 and KLK14 as the major targets of LEKTI fragments and KLK7 to a lesser extent. Using zymography analyses, we confirmed the inhibitory capacity of D8–D11 toward the native form of these epidermal proteinases.
Despite a high sequence homology (68%) between D5 and D6, D5 inhibits KLK14, whereas D6 does not. Structural studies of Kazal domain complexes reveal that there are 12 contact positions (P6, P5, P4, P3, P2, P1, P1', P2', P3', P14', P15', and P18') responsible for interactions between Kazal domains and their cognate serine proteinases (Lu et al., 1997). Among these 12 contact positions, P4 and P6 are the only positions where the nature of residues differs between D5 and D6. At P6 position, a lysine is present in D5, whereas an arginine is found in D6. These two amino acids are structurally close and are not likely to explain the functional difference between the two domains. In contrast, the P4 position is occupied by a phenylalanine in D5 and by an alanine in D6. Interestingly, Empie and Laskowski (1982) found that substitution of a voluminous amino acid (Asp) in a small, and uncharged residue (Ala) at position P4 of Kazal ovomucoid third domain had a dramatic consequence on its inhibitory capacity toward trypsin-like enzyme subtilisin. Therefore, the difference of only 1 amino acid at P4 position between D5 and D6 LEKTI domains could account for their selectivity toward different serine proteinases.
This study highlights the specialization of LEKTI in the inhibition of epidermal proteinases KLK5, KLK7, and KLK14 and is consistent with an increased desquamation in NS patients. NS skin is also characterized by chronic inflammation, but the observation that inflammatory proteinases are not the direct targets of the studied LEKTI fragments suggests that KLK5 and KLK7 may have a proinflammatory role by activating PLA2 and IL1, as proposed by Egelrud et al. (2005). Alternatively, it is also possible that additional physiological LEKTI fragments could have a direct activity against inflammatory proteinases.
The Spink5–/– mice revealed that LEKTI is a key regulator of the desquamation process through the control of KLK5 and KLK7 activities. In normal epidermis, LEKTI, KLK5, and KLK7 colocalize in the neighborhood of corneodesmosomes (Ishida-Yamamoto et al., 2005). This suggests a finely regulated interaction between these partners to allow the detachment of the SC superficial layers only. Previous studies have demonstrated that pH is important for maintaining skin homeostasis and that transient increase in SC pH induces abnormality in permeability barrier (Hachem et al., 2006). We mimicked the pH gradient occurring in SC layers during binding studies between KLK5 and D8–D11. At pH 7.5, the complex is very stable, but BIAcore analysis demonstrated that dissociation increases with acidification. The same effect of acidification could be observed on the dissociation rate of LEKTI–KLK7 complexes. This result is consistent with the possibility that during the passage of deep (pH 7.5) to superficial SC (pH 4.5) (Elias, 2004), KLK5 and KLK7 gradually dissociate from LEKTI. At acidic pH, KLK5 and KLK7 retain sufficient activity to degrade corneodesmosomal components, desmoglein-1, desmocollin-,1 and corneodesmosin (Caubet et al., 2004). These results support the role of SC acidification in the control of the detachment of the most superficial corneocytes (Figure 9). The severity of the NS skin phenotype demonstrates the crucial need for a tight control of epidermal proteolytic activity. KLK5 and KLK7, as serine proteinases, display optimal activities at neutral pH, which is precisely the pH at the GR–SC interface. To prevent premature desquamation at the GR–SC interface, KLK5 and KLK7 activities must be strongly inhibited. This is consistent with the observation that the interaction between LEKTI and epidermal kallikreins is very strong at neutral pH. This highlights the importance of skin pH balance in the control of desquamation, acting at two levels: the control of protease activity, and the control of the interaction between proteinases and their inhibitors. All together, the resultant ensures an apparent proteolytic activity in a restricted environment.
|
|
ACKNOWLEDGMENTS |
|---|
|
Footnotes |
|---|
The online version of this article contains supplemental material at MBC Online (http://www.molbiolcell.org). 
1 The amino acid residues involved in the reactive site loop are numbered following the Schechter and Berger (1967) nomenclature.
These authors contributed equally to this work.
Address correspondence to: Alain Hovnanian (alain.hovnanian{at}toulouse.inserm.fr).
Abbreviations used: BIA, biomolecular interaction analysis; CHO, Chinese hamster ovary; GR, granular layer; HBS, HEPES-buffered saline; KLK, kallikrein; LEKTI, Lympho-epithelial Kazal type inhibitor; LEKTIf-l, full-length LEKTI; LEKTIsh, short-length LEKTI; NHK, normal human keratinocytes; NS, Netherton syndrome; SC, stratum corneum; SPINK5, serine proteinase inhibitor Kazal type 5.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Bitoun, E. et al. (2003). LEKTI proteolytic processing in human primary keratinocytes, tissue distribution and defective expression in Netherton syndrome. Hum. Mol. Genet 12, 2417–2430.
Brattsand, M., Stefansson, K., Lundh, C., Haasum, Y., and Egelrud, T. (2005). A proteolytic cascade of kallikreins in the stratum corneum. J. Invest. Dermatol 124, 198–203.[CrossRef][Medline]
Caubet, C., Jonca, N., Brattsand, M., Guerrin, M., Bernard, D., Schmidt, R., Egelrud, T., Simon, M., and Serre, G. (2004). Degradation of corneodesmosome proteins by two serine proteases of the kallikrein family, SCTE/KLK5/hK5 and SCCE/KLK7/hK7. J. Invest. Dermatol 122, 1235–1244.[CrossRef][Medline]
Chavanas, S. et al. (2000). Mutations in SPINK5, encoding a serine protease inhibitor, cause netherton syndrome. Nat. Genet 25, 141–142.[CrossRef][Medline]
Debela, M., Magdolen, V., Schechter, N., Valachova, M., Lottspeich, F., Craik, C. S., Choe, Y., Bode, W., and Goettig, P. (2006). Specificity profiling of seven human tissue kallikreins reveals individual subsite preferences. J. Biol. Chem 281, 25678–25688.
Descargues, P. et al. (2005). Spink5-deficient mice mimic Netherton syndrome through degradation of desmoglein 1 by epidermal protease hyperactivity. Nat. Genet 37, 56–65.[CrossRef][Medline]
, stop of proteinase injection and replacement with buffer; RU, resonance units; Diff. Resp., differential response.
, stop of proteinase injection and replacement with buffer; RU, resonance units; Norm. Diff. Resp., normalized differential response.
