J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 3 9 2 5 – 3 9 3 7
Ava i l ab l e on l i ne a t www.sc i enced i r ec t . com
www.e l sev i e r . com/ loca te / j p ro t
The identification of a new role for LEKTI in the skin: The use ofprotein ‘bait’ arrays to detect defective trafficking of dermcidinin the skin of patients with Netherton syndrome
Kate Bennetta, Wendy Heywooda, Wei-Li Dia, John Harpera, Gary L. Claymanb,Arumugam Jayakumarb, Robin Callarda, Kevin Millsa,⁎aInstitute of Child Health & Great Ormond Street Hospital for Sick Children, University College London, 30 Guilford Street, London, WC1N 1EH,United KingdombDepartment of Head and Neck Surgery, UTMD Anderson Cancer centre, 1515 Holcombe Blvd, Houston, TX, 77030, USA
A R T I C L E I N F O
⁎ Corresponding author. Tel.: +44 207 905 287E-mail address: [email protected] (K. M
1874-3919/$ – see front matter. Crown Copyrdoi:10.1016/j.jprot.2012.04.045
A B S T R A C T
Article history:Received 27 February 2012Accepted 29 April 2012Available online 12 May 2012
Lympho-Epithelial Kazal-Type-related Inhibitor (LEKTI) has been demonstrated to be aninhibitor of various kallikreins and is thought to play a role in the regulation of skindesquamation. In order to identify and investigate the potential of LEKTI to interact withother proteins, a method was developed using immobilised proteins onto arrays andnanoUPLC/MALDI‐TOF MS. Using various domains of LEKTI, we demonstrated that thesedomains bound a number of kallikreins (5, 13 and 14) to varied extents on the array surface.Inhibitory assays confirmed that binding on the protein array surface corresponded directlyto levels of inhibition. The method was then tested using skin epidermal extracts. All formsof rLEKTI with the exception of rLEKTI 12–15, demonstrated the binding of several potentialcandidate proteins. Surprisingly, the major binding partners of LEKTI were found to be theantimicrobial peptide dermcidin and the serine protease cathepsin G and no kallikreins.Using confocal microscopy and Netherton syndrome skin sections, we confirmed the co-localisation of LEKTI with dermcidin and demonstrated altered trafficking of dermcidin inthese patients. This potential new role for LEKTI as a multifunctional protein in theprotection and transport of proteins in the epidermis and its role in disease are discussed.
Keywords:LEKTIKallikreinsDermcidinCathepsin GMass spectrometryQTOF MS
1. Introduction
LEKTI is a Kazal-type protease inhibitor expressed in the mostdifferentiated layers of stratified epithelial tissues including theskin, where expression has been localised to the granular layerand stratum corneum [1,2]. LEKTI is synthesised in its full-lengthform (15 domains) and cleaved rapidly into a range of single ormultidomain fragments that are secreted from the cell [1–4].Recessive mutations in SPINK5, the gene that encodes LEKTI,have been identified as the cause of Netherton syndrome [5], achronic skin disease characterised by accelerated stratum
corneum shedding and loss of skin barrier function [6–8]. LEKTIhas a recognised role as a serine protease inhibitor through itsinhibitory activity against a range of serine proteases in vitro,including members of the kallikrein (KLK) family [4,9–12]. InNetherton syndrome, where LEKTI expression is either absent orreduced, KLK5- and KLK7-like activities are increased, whichleads to premature degradation of corneodesmosomes, over-desquamation and stratum corneum thinning [2,13,14]. In ourprevious study, we showed that LEKTI was able to inhibit thecysteine protease caspase 14 in vitro [15], which indicated thatLEKTI is a multifunctional protease inhibitor. Caspase 14 has
3926 J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 3 9 2 5 – 3 9 3 7
been implicated in both epidermal differentiation and hydrationthroughprocessing of the barrier-related protein, filaggrin [16,17].This was the first study to report the inhibitory activity of LEKTIagainst a cysteine protease, which presents the possibility thatLEKTI targets multiple protease families in the skin and suggestsan involvement of LEKTI in novel biological pathways other thanthose associated with KLKs. In order to identify further potentialproteases or targets for LEKTI, an immobilised protein arraytechnique was developed. This technique uses protein arraysand immobilisation of LEKTI domains as ‘baits’. Binding condi-tions on the array were optimised and confirmed using matrix-assisted laser desorption/ionisation quadrupole time-of-flightmass spectrometry (MALDI TOF MS) analysis of protein stan-dards. Identificationof unknownbindingproteins in the skinwasachieved by scaling up the procedure, extraction off the arraysurface, proteolytic digestion and Q-TOF MS sequencing. Themethod developed in this study has shown that the levels ofKLKs binding to rLEKTI on the array were in an amount thatcorresponded to the levels of inhibitory activity against the KLKs.This technique was then used in vivo to probe potential bindingpartners of LEKTI in the skin using epidermal extracts and thetwo proteins dermcidin and cathepsin G were identified asbinding partners of LEKTI. Confirmation of this discovery wassubstantiated by immunohistochemical staining of control andNetherton syndrome skin sections. This work describes thedevelopment of the technique, the identification of targets ofLEKTI and a potential new function for LEKTI in a protection/transport role of antimicrobial peptides to the skin surface.
2. Materials and methods
2.1. Materials
All standard reagents were of analar grade or equivalent andobtained from Sigma-Aldrich Company (Gillingham, Dorset, UK).RS100 arrays were obtained from Bio-Rad Laboratories (HemelHempstead, UK). Trypsin was of sequencing grade and wasobtained from Promega UK Ltd. (Southampton, Hants, UK). Thefluorogenic peptide substrate (Boc-V-P-R-AMC) was obtainedfrom R&D Systems Europe Ltd (Abingdon, UK). EndopeptidaseLys-C was obtained from Sigma-Aldrich Company (Gillingham,Dorset, UK).
2.2. Proteins
The rLEKTI forms containing intact domains 1–6 (rLEKTI 1–6),domains 6–8 and partial domain 9 (rLEKTI 6–9′), domains 9–12(rLEKTI 9–12), domains 12–15 (rLEKTI 12–15) and domains 1–15(full-length rLEKTI)were created by Dr. A. Jayakumar, Universityof Texas,M.D. AndersonCancerCentre. Eachproteinwas clonedand expressed in Sf9 cells using the baculovirus expressionvector system as reported previously [9,18]. Recombinanthuman KLK standard proteins 1, 5, 13 and 14 were obtainedfrom R&D Systems Europe Ltd (Abingdon, UK).
2.3. Patients and skin biopsies
Skin biopsies were obtained from three patients with a clinicaldiagnosis of Netherton syndrome and from eight children
with normal skin who attended the plastic surgery unit atGreat Ormond Street Hospital (London, UK). Local ethicscommittee approval and consent from the parents wereobtained prior to the procedure. Skin samples were eithersnap frozen or paraffin embedded and stored at −80 °C.
2.4. Protein extraction from skin
Control skin was obtained from age- and anatomy-matchedbiopsies from patients admitted to the hospital for routinesurgery. The dermis was removed from all skin biopsies andthe remaining epidermis washed with 3×1 ml of ice‐coldphosphate buffered saline solution (pH 7.4) to remove allcontaminating blood-derived proteins. The epidermis wassectioned into 10 μm pieces using a cryostat. The sampleswere then resuspended in lysis buffer (50 mM Tris pH 8.0,150 mM NaCl, 5 mM EDTA, 1% Triton) and left on ice for 1 hand disruption of the tissue was achieved using ultrasonication (Soniprep 150, 3 cycles of 10 amplitudes for 10 s).After this time the sample was left on ice for a further 30 minand the suspension was spun down using a bench-topcentrifuge at 16,100 g for 10 min at 4 °C. The protein superna-tant was aliquoted into separate Eppendorf tubes and storedat −80 °C prior to analysis.
2.5. Immobilisation of proteins to create ‘bait’ arrays
2.5.1. Coupling rLEKTI to the array surfaceRecombinant LEKTI (rLEKTI) was diluted in phosphate buff-ered saline (PBS, pH 7.4) to a final concentration of ~1 pmol/μl.A total of 4.3 pmol was added to each spot on RS100 arrays(BioRad, UK). The array was incubated overnight at 4 °C in ahumidity chamber. Excess buffers were removed using a cleantissue. Any unreacted carbonyldiimidazole sites present onthe surface of the array were blocked by the addition of 5 μl of0.5 M ethanolamine/50 mM Tris, pH 8.0, followed by incuba-tion at room temperature (20 °C), for 30 min in a humiditychamber. After this time, the excess blocking solution wasremoved and the array washed for 15 minwith 8 ml of PBS, pH7.4, 0.1% Triton X-100. The array was washed for a further15 min in 8 ml PBS containing no Triton X-100, followed by afinal 10 minute wash with 10 ml of PBS, pH 7.4 containing noTriton. Excess PBS wash buffer was removed from the areasaround the spots using a tissue. Care was taken not to touchthe surface of the spots containing bound rLEKTI. All proteinarrays were used immediately.
2.5.2. Optimisation of the affinity capture of serine proteasesusing immobilised LEKTI arraysAll proteases were diluted to a final concentration of 2.2 pmol/μl using 50 mM ammonium bicarbonate, pH 7.8 and 5 μl of thediluted sample (10.8 pmol) was added to the array surfacecontaining the immobilised rLEKTI. The array was incubatedfor 2 h at room temperature (20 °C) in a humidity chamber.Excess sample solution was removed from the spots and thearray was washed twice in 8 ml of PBS, pH 7.4 containing 0.1%Triton X-100 for 15 min. The array was transferred to a freshtube and washed in 8 ml PBS, containing no Triton, for 5 min.The buffer was poured off and the array was rinsed briefly(10 s) 3 times in 8 ml of 5 mM ammonium acetate, pH 7 to
3927J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 3 9 2 5 – 3 9 3 7
wash away any salts that may otherwise interfere with MSanalysis. The arrays were allowed to air-dry prior to MS.
2.5.3. MALDI-TOF MS of proteins captured using theimmobilised LEKTI arrayMALDI-TOF MS matrices were prepared by dissolving 5 mg ofsinapinic acid in 200 μl of 50% acetonitrile containing 0.1%trifluoroacetic acid (TFA) and 1 μl was added to each spot inpreparation for MALDI-TOF MS. After drying a further 1 μl ofmatrix preparation was added. The array was allowed to air-dry at room temperature for 30 min before being placed intothe mass spectrometer. Mass spectral analyses were analysedin positive ionmode and carried out using a PBS-IIc linear TOFMS (Bio-Rad) with a flight path of 0.8 m, capable of massresolution (m/Δm) of 1000, mass accuracy of ±1000 ppm andfitted with a 337 nmUV laser. Spots were air dried and two 1 μlof Sinnapinic acid (Sigma Aldrich, UK) in 50% ACN, 0.05% TFAwere added to each spot and air dried before being analysedby a SELDI TOF PBSIIc instrument. Twomass spectral analyseswere performed on each array. The first involved a lowermassscan optimised to analyse proteins of mass 1.5–30 kDa. Asecond scan was performed to analyse proteins of highermolecular weight at up to 180 kDa with settings optimisedbetween 50 and 110 kDa. After baseline subtraction spectrawere normalised by means of the total ion current method.Peak detection was done by Biomarker Wizard software(Ciphergen) using a signal to noise ratio of three. All spectrawere normalised from 1.5 kDa for the low mass scans andfrom 10 kDa for high mass scans.
2.6. Inhibitory assays
2.6.1. The inhibition of kallikreins by different rLEKTI formsAll experiments were carried out in triplicate in 0.1 M sodiumphosphate, 0.01% Tween buffer, pH 8. Proteases (final con-centration 3.7 nM) were pre-incubated in black microtiterplate wells with various concentrations (0–30 nM) of rLEKTI,at room temperature for different time periods (0–60 min). Afluorogenic peptide substrate (Boc-V-P-R-AMC) containing thehighly fluorescent group AMC was added in a total volume of100 μl (final concentration of 100 μM). Each KLK cleaved thesubstrate at the carboxyl side of the arginine residue, whichresulted in the release of AMC. Free AMC fluorescence wasmeasured as arbitrary fluorescence units (AFU) at varioustime intervals on a TECAN GENios microplate fluorescencereader (Salzburg, Austria), with excitation and emission filtersof 360 nm and 465 nm, respectively. Protease-free reactions,with the fluorogenic substrate only, were used as negativecontrols and the background counts were subtracted fromeach value. The amount of KLK activity that was inhibited byeach rLEKTI form was calculated using the following equa-tion: % inhibition=100×[1− (velocity in the presence of inhib-itor/velocity of uninhibited control)]. Velocities (AFU/time)were calculated using GraphPad Prism 4 from the plots of timeversus AFU.
2.6.2. The extraction of proteins bound to the protein ‘bait’arraysRecombinant LEKTI (4.3 pmol) was bound covalently to anRS100 array surface (all 8 sites on the array), followed by the
addition of 10.8 pmol protein sample. After the final wash inammonium acetate, 5 μl of FAPH (50% formic acid, 25%acetonitrile, 15% isopropanol) was added to 7 sites, followedby repeated aspiration of the solution to extract highermolecular weight proteins (>10 kDa) from the chip surface.Each wash was combined and collected in Eppendorf tubes.To extract lower molecular weight proteins (<10 kDa) theprocedure was repeated with 70% acetonitrile, containing0.2% TFA. The protein solutions were combined, concentratedand residual acetonitrile and TFA removed by centrifugalevaporation, prior to in-solution digestion [15].
2.6.3. In-solution digestion of proteinsThe lyophilised protein was reconstituted in 20 μl of 100 mMTris, pH 7.8, containing 6 M urea to denature the protein.Disulphide bridges were reduced by the addition of 1.5 μl of100 mM Tris–HCl buffer, pH 7.8, containing 5 M DTE andfollowed by incubation at room temperature for 60 min. Freethiol groups were carboamidomethylated followed by incuba-tion with 6 μl of 100 mM Tris–HCl, pH 7.8, containing 5 Miodoacetamide for 30 min at room temperature. The solutionwas then diluted with H2O to a final volume of 200 μl, vortexedand 1 μg of sequence grade trypsin added to the solution.Samples were incubated overnight at 37 °C in water bath.
2.6.4. NanoUPLC–MS/MS (ESI-QTOF MS)Proteins extracted from the array surface were identified andquantitated by direct analysis of the reaction mixture de-scribed above. All analyses were performed using a nanoAc-quity UPLC and QTOF Premier mass spectrometer (WatersCorporation, Manchester, UK). Peptides were trapped anddesalted prior to reverse phase separation using a SymmetryC18 5 μm, 5 mm×300 μm pre-column. Peptides were thenseparated prior to mass spectral analysis using a25 cm×75 μm C18 reverse phase analytical column. Peptideswere loaded onto the pre-column at a flow rate of 4 μl/min in0.1% formic acid for a total time of 4 min. Peptides were elutedoff the pre-column and separated on the analytical columnusing a gradient of 3–40% acetonitrile [0.1% formic acid] over aperiod of 90 min and at a flow rate of 250 nl/min. The columnwas washed and re-generated at 250 nl/min for 10 min using a99% acetonitrile [0.1%] rinse. After all non-polar and non-peptide materials were removed the column was re-equilibrated at the initial starting conditions for 20 min. Allcolumn temperatures were maintained at 35 °C. Mass accu-racy was maintained during the run using a lock spray of thepeptide [glu1]-fibrinopeptide B delivered through the auxiliarypump of the nanoAcquity at a concentration of 300 fmol/l andat a flow rate of 300 nl/min. Peptides were analysed in positiveion mode using a Q-Tof Premier mass spectrometer (WatersCorp., Manchester, UK) and was operated in v-mode with atypical resolving power of 10,000 FWHM. Prior to analyses, theTOF analyser was calibrated using [glu1]-fibrinopeptide Bfragments obtained using a collision energy of 25 eV andover the mass range 50–2000 m/z. Post calibration of data fileswas corrected using the doubly charged precursor ion of[glu1]-fibrinopeptide B (785.8426 m/z) with a sampling fre-quency of 30 s. Accurate mass LC–MS data were collected in adata independent and alternating, low and high collisionenergy mode. Each low/high acquisition was 1.5 s with 0.1 s
3928 J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 3 9 2 5 – 3 9 3 7
interscan delay. Low energy data collections were performedat a constant collision energy of 4 eV, high collision energyacquisitions were performed using a 15–40 eV ramp over a1.5 s time period and a complete low/high energy acquisitionwas achieved every 3.2 s.
2.6.5. Data analysis of skin samples analysed by QTOF MSProteinLynx GlobalServer version 2.4 was used to process alldata acquired. Protein identifications were obtained bysearching UniProt/Swiss-Prot databases to which the se-quence of P00924 yeast enolase was added manually. Proteinidentification from the low/high collision spectra for eachsample was processed using a hierarchical approach wheremore than three fragment ions per peptide, seven fragmentions per protein and more than two peptides per protein hadto be matched. Lock mass window was set at 0.1 Da, low andelevated energy thresholds set at 150 and 50 counts, respec-tively. Elution start and end times were 15 and 110 min,respectively and with chromatographic peak width set toautomatic. Protein identification parameters used in thedatabase searching included a <10 ppm mass accuracytolerance, fixed modifications of carboamidomethylation ofcysteines, dynamic modifications of deamidation of aspara-gine/glutamine and oxidation of methionine, up to 3 missedcleavage sites andmaximumproteinmass 400 kDa. Data weresearched against human canonical species of the SwissProtdatabase (May 2011, www.expasy.org).
2.6.6. Immunofluorescence stainingImmunofluorescence staining was performed on paraffin skinsections from normal donors (n=8) and patients with Nether-ton syndrome (n=3) using antibodies against Dermcidin G-81(Novus Biologicals, NBP1-49620, Cambridge, UK). Paraffin-embedded tissues were cut into 6 μm sections, which werede-waxed, antigen retrieved and blocked with 3% serum inPBS and avidin/biotin blocking solution (Vector Laboratories,Burlingame, CA, USA). The sections were then incubated withmouse anti-dermcidin antibody in 1:50 dilutions over night at4 °C. The following day, after rinsing with PBS, the sectionswere further incubated with anti-LEKTI antibody [19] overnight at 4 °C. Sections were further incubated with biotiny-lated anti-mouse IgM antibody (Vector Laboratories, Burlin-game, CA, USA) and Alexa568 goat anti-rabbit IgG antibody(Invitrogen, Paisley, UK) for 1 h and FITC-streptavidin conju-gates for 30 min. Stained tissues were recorded by Zeiss laserconfocal microscopy 710 (Carl Zeiss, Welwyn Garden City, UK)and images were processed using Adobe Photoshop CS (AdobeSystems Inc, San Jose, Ca, USA).
3. Results
3.1. The evaluation of the LEKTI ‘bait’ array and itsinteractions with kallikrein protein standards
A novel method based on immobilised protein array technol-ogy was developed to investigate the interaction of differentrLEKTI forms with various KLKs (KLK1, 5, 13 and 14). TherLEKTI fragments rLEKTI 1–6, rLEKTI 6–9′, rLEKTI 9–12 andrLEKTI 12–15 used in this study, span all fifteen domains of
LEKTI and are numbered according to the domains theycontain. Each rLEKTI form (4 pmol) was bound covalently to aprotein array surface, followed by the addition of 10 pmol ofthe kallikreins 1, 5, 13 and 14 and analysed by MALDI TOF MS.Separate experiments were performed for each KLK.
KLK5 has previously been shown to be inhibited by all fiverLEKTI forms to various extents in vitro [11,12]. This presentedKLK5 as an ideal protein target to assess the efficiency of ourmethod against all five rLEKTI forms. Both KLK5 (~33,392 m/z)and KLK14 (~25,203 m/z) bound to all forms of rLEKTI tovarious extents (Fig. 1A and B, respectively). The peak areaswere used to determine relative amounts of protein thatbound to rLEKTI on the array surface. rLEKTI 6–9′ demon-strated the highest level of binding to rKLK5 (designated100%), followed by rLEKTI 1–6, rLEKTI 9–12, rLEKTI 12–15 andfull-length rLEKTI (Fig. 1A(iii), (ii), (iv), (v) and (vi), respectively;Table 1). Similarly, rLEKTI 1–6 was also observed to bind highamounts of rKLK14, followed by rLEKTI 6–9′, full-lengthrLEKTI, rLEKTI 9–12 and rLEKTI 12–15 (Fig. 1B(ii), (iii), (vi), (iv)and (v), respectively; Table 1). Finally, rLEKTI 1–6 also boundthe highest amounts of rKLK13 (~30,934 m/z) followed byrLEKTI 6–9′ and full-length rLEKTI (Fig. 1C(ii), (iii) and (vi),respectively; Table 1). However, both rLEKTI 9–12 and rLEKTI12–15 were unable to bind rKLK13 (Fig. 1C(iv) and (v)).
Previous studies have shown that KLK1 is not inhibited byany of the four rLEKTI fragments in vitro [12]. If binding ofrLEKTI on the chip surface is a representation of inhibition,KLK1 should not bind to any of the four rLEKTI fragments onthe array surface and could be used as a negative control. Noprotein of mass 33,142 m/z corresponding to KLK1 wasobserved following incubation with any bound rLEKTI frag-ment (data not shown). For this experiment rKLK5 was usedas a positive control.
To eliminate the possibility of non-specific binding to thehydrophobic chip surface, KLKs were also incubated onprotein-free array surfaces, in which all the reactive siteshad been blocked with ethanolamine (Fig. 1A(i)–C(i)). Further-more, to eliminate the possibility of non-specific binding tonon-related proteins, the unrelated plasma protein transfer-rin was also bound to the chip surface, followed by incubationwith each KLK (data not shown). In each case, no KLKs weredetected in the negative controls, which indicated thatbinding to rLEKTI forms was specific and not an artefact orconsequence of non-specific binding to the array.
3.2. The relationships between the binding and inhibitionof kallikreins 5, 13 and 14 by different forms of rLEKTI
In order to determine whether binding to LEKTI on the arraysurface corresponded to levels of inhibition by LEKTI, inhib-itory assays were performed in triplicate against each KLK.The effect of each rLEKTI form on the activity of KLKs wasinvestigated by assaying each KLK separately in the presenceof equimolar amounts of rLEKTI. The amount of KLK activitythat was inhibited by each rLEKTI formwasmeasured using %decreases in AFU relative to the uninhibited control, whichwas assumed to represent 0% inhibition.
Almost exclusively, the order of inhibitory potencies for allfour rLEKTI fragments (with the exception of full-length rLEKTI)against KLK5, 13 and 14 corresponded to the general order of
Fig. 1 – MALDI-TOF mass spectra of (A) KLK5, (B) KLK14, and (C) KLK13, (i) no bound protein (ii) rLEKTI 1–6, (iii) rLEKTI 6–9′, (iv)rLEKTI 9–12, (v) rLEKTI 12–15 and (vi) full-length rLEKTI. The spectra cover the mass range 20,000–60,000 m/z.
3929J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 3 9 2 5 – 3 9 3 7
binding efficiencies on the array surface (Fig. 2A–C). In thepresence of equimolar amounts of LEKTI, rLEKTI 1–6 was themost potent KLK inhibitor of all the four rLEKTI fragmentstested, followed by rLEKTI 6–9′, rLEKTI 9–12 and rLEKTI 12–15(Fig. 2(a–c) and Table 2). This correspondedwith the observation
that rLEKTI 1–6 bound the highest levels of KLK14 and 13 on thearray surface (Fig. 2B(ii) andC(ii), respectively). Also, the absenceof KLK13 binding to either rLEKTI 9–12 or rLEKTI 12–15 agreedwith the finding that neither rLEKTI form was able to inhibitrKLK13 activity. In addition, rLEKTI 12–15, which demonstrated
Table 1 – Varied binding efficiencies of different rLEKTIforms towards rKLK5, rKLK14 and rKLK13.
a Percent inhibition=100×[1− (velocity in the presence of inhibitor/velocity of uninhibited control)].
3930 J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 3 9 2 5 – 3 9 3 7
extremely low binding efficiency towards rKLK14, was not ableto inhibit rKLK14 activity (Fig. 2B). In contrast, although rLEKTI1–6 bound high amounts of KLK5, rLEKTI 6–9′ bound slightlyhigher amounts of KLK5 compared to rLEKTI 1–6, despite thefact that rLEKTI 1–6 exhibited much higher levels of inhibitionagainst KLK5 (Fig. 2A).
(A)
(B)
(C)
(i)
(i)
(i) (i
2.5 5.0 7.5 10.00
5000
10000klk13 control1-66-9'9-1212-15full-length
time (min)
AF
U (
em. 4
65n
m)
2.5
0 10 20 30 40 50 60 70
5.0 7.5 10.00
10000
20000
30000
40000klk14 control
klk5 control
1-66-9'9-1212-15full-length
1-6
6-9'
9-1212-15
full-length
time (min)
time (min)
AF
U (
em. 4
65n
m)
0
10000
20000
30000
40000
AF
U (
em. 4
65n
m)
%%
%
Fig. 2 – Inhibition of (A) KLK5, (B) KLK14 and (C) KLK13 by rLEKTIrLEKTI. (i) The plots of time versus arbitrary fluorescence units (A(Boc-V-P-R-AMC) by rKLK in the presence of equimolar amountsaccording to the domains they contained. The uninhibited rKLKefficiencies and relative inhibitory levels of different rLEKTI formrelative to the rLEKTI form that demonstrated the highest level ofthe presence of inhibitor/velocity of uninhibited control)].
Out of all the rLEKTI forms tested, full-length rLEKTI exhibitedthe highest level of inhibition against each KLK tested (Fig. 2A–C).Unlike the four rLEKTI fragments, this did not correspond to the
(ii)
(ii)
i)
0
50
100Binding efficiencyInhibition
1-6 6-9’ 9-12 12-15 Full-length
1-6 6-9’ 9-12 12-15 Full-length 0
50
100Binding efficiencyInhibition
0
50
100Binding efficiencyInhibition
1-6 6-9’ 9-12 12-15 Full-length
1–6, rLEKTI 6–9′, rLEKTI 9–12, rLEKTI 12–15 and full-lengthFU), represent the hydrolysis of an AMC substrateof KLK and LEKTI (mean±SEM). rLEKTI forms are numberedwas used as a control. (ii) Comparison between the bindings towards each KLK. Inhibition values (%) were recalculatedinhibition for each KLK. The % inhibition=100×[1− (velocity in
3931J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 3 9 2 5 – 3 9 3 7
trend of binding efficiencies observed on the array surface. Full-length rLEKTI was only able to bind extremely small amounts ofKLK5 and KLK13 on the array surface, despite demonstratinghigh levels of inhibition (Fig. 2A(ii) and C(ii), respectively). BothrLEKTI 1–6 and rLEKTI 6–9′were found to bind higher amounts ofKLK14 compared to full-length rLEKTI, despite exhibiting muchlower levels of inhibition (Fig. 2B(ii)).
3.3. The detection of LEKTI targets from human skin usingprotein ‘bait’ arrays
Having established the bait technique, the method was thenused to investigate potential targets of LEKTI in the humanskin. Proteins were extracted from the epidermis (3.9 μg/μl)and then incubated with rLEKTI 1–6, rLEKTI 6–9′, rLEKTI 9–12,rLEKTI 12–15 and full-length rLEKTI all coupled separately tothe array surface. MALDI TOF MS analyses showed thatmultiple proteins bound to rLEKTI 1–6, rLEKTI 6–9′, rLEKTI9–12 and full-length rLEKTI (Fig. 3A(iii), (iv), (v) and (vii),respectively). In contrast, no proteins were observed to bind torLEKTI 12–15 (Fig. 3A(vi)). Nine proteins in the mass range of9000 m/z–18,000 m/z, bound to rLEKTI 1–6, rLEKTI 6–9′, rLEKTI9–12 and full-length rLEKTI on the array surface (Fig. 3A(iii),(iv), (v) and (vii), respectively). Overall, rLEKTI 6–9′ appeared tobind the highest levels of each protein, followed by rLEKTI 1–6,rLEKTI 9–12 and full-length rLEKTI. For each individual rLEKTIform, with the exception of rLEKTI 12–15, the protein ofmass~11,300 m/z was found to bind in highest amounts. Inthe mass range 20,000 m/z–150,000 m/z a protein ofmass~23,400m/z bound to all rLEKTI forms, with the exceptionof rLEKTI 12–15 (Fig. 3B). rLEKTI 1–6 and rLEKTI 6–9′ also boundextremely low amounts of proteins of mass~41,900m/z and~70,000m/z (Fig. 3B(iii), (iv)). Analysis of the array positionswhere no protein and a control protein transferrin were added(Fig. 3A and B), demonstrated that no proteins from theepidermal extract bound to the array surface or an unrelatedcontrol protein, in a non-specific manner. In order to investigatewhether the finalised protocol yielded reproducible results,proteins were extracted from six normal epidermal samples,taken from six different human subjects, using the optimisedextraction protocol. MALDI TOFMS analysis showed that severalproteins in the mass range 9000m/z–15,000m/z, all of whichwere detected in the previous experiment were found to bindconsistently to full-length rLEKTI from each skin sample (Fig. 3C(ii–vii)). However, the proteins of molecular mass~17,900m/zand ~23,400m/z, which were detected previously, no longerbound to full-length rLEKTI on the array surface. In contrast,proteins of mass~25,900m/z and ~27,100m/z bound reproduc-ibly to full-length rLEKTI fromall six skin samples (Fig. 3D(ii–vii)).Two additional binding proteins of mass 37,466m/z and42,019 m/z were also detected in one skin sample (Fig. 3D(v)).Again the protein of mass~11,300 m/z was found to bind inhighest amounts to LEKTI from each skin sample.
3.4. The identification of skin proteins that bound to theLEKTI ‘bait’ array
It was found that multiple proteins bound to LEKTI on theprotein array surface from skin extracts. However, to identifythese proteins LEKTI targets had to be extracted from the
protein array surface and identified by MS-sequencing. Asuccessful method was developed for the identification ofsmall quantities of protein mixtures by in-solution digestionand sequencing using nanoUPLC Q-TOF MS. The next stagewas to determine whether this method could be used toidentify the proteins that bound to LEKTI on the array surfacefrom skin extracts. In an attempt to obtain sufficient amountsof protein for successful digestion and sequencing, themethod was ‘scaled up’ using rLEKTI 6–9′ (4 pmol) bound tofour protein arrays (32 spots) and incubation with theepidermal protein extract. For each array 7 out of 8 positionswere treated with two reagents, 70% acetonitrile, 0.2% TFAand 50% formic acid, 25% acetonitrile, 15% isopropanol, todenature and extract low and high molecular weight proteins,respectively. The unextracted sample and 7 pooled/extractedsamples were analysed by MALDI TOF MS to determine thatLEKTI was efficient in acting as a ‘bait’ to isolate bindingpartners and that the extraction procedure was successful.Multiple proteins in the mass range 10,000 m/z–50,000 m/zwere detected in the untreated sample (Fig. 4A) and indicatedthat epidermal proteins had bound to rLEKTI 6–9′ on the arraysurface. The absence of proteins in the samples treated withextraction reagents (Fig. 4B) demonstrated that all proteinshad been extracted from the array surface.
The extracted samples were combined, digested withtrypsin and the peptides were sequenced using nanoUPLC Q-TOF MS. The data was analysed using Proteinlynx 2.4software. The proteins dermcidin and cathepsin G wereidentified with 40.9% and 42.4% sequence coverage, respec-tively (Fig. 5A and B, respectively). Full-length dermcidin has amolecular mass of ~11.3 kDa, which corresponded to that ofthe protein that bound in high amounts to LEKTI frommultiple epidermal samples on the protein array surface(Fig. 3A and C). The coverage of amino acids that weresequenced also suggests that full-length dermcidin wasidentified as opposed to processed peptides (DCD-1/DCD-1L)(Fig. 5A). In humans, several isoforms of cathepsin G existranging from ~23 to 29 kDa (Twining et al. [20] and Edwards[21]). This range corresponded to the molecular masses of 3proteins that bound to LEKTI on the array surface (Fig. 3B andD). The results indicated that both dermcidin and cathepsin Gwere binding partners of LEKTI on the protein array surface.No kallikreins were detected.
3.5. The expression of dermcidin in the epidermis
To date, dermcidin expression has only been described ineccrine sweat glands [22]. In order to confirm that theinteraction between LEKTI and dermcidin on the array surfacemay occur in vivo, the expression of both proteins wasinvestigated in normal skin sections and from patients witha LEKTI deficiency (Netherton syndrome) using immunofluo-rescence and confocal microscopy. Immunofluorescencestaining showed that dermcidin was expressed throughoutthe epidermis with the highest level of expression in theupper granular layer (Fig. 6A). LEKTI expression was found tobe restricted to the granular layer where it co-localised withdermcidin (Fig. 6A). It should also be noted that the expressionof dermcidin in the epidermis was lower than that observed ineccrine sweat glands (data not shown).
50000 150000
00.5
11.5
22.5
50000
00.5
11.5
22.5
50000
00.5
11.5
22.5
50000
15426.6+H17929.5+H
23394.4+H 41901.5+H 70057.6+H
00.5
11.5
22.5
50000
17925.8+H
23397.5+H
25907.6+H 41881.4+H69994.8+H
00.5
11.5
22.5
50000
15769.7+H
17920.8+H
00.5
11.5
22.5
50000
00.5
11.5
22.5
15774.4+H
17965.4+H
23395.7+H
(i)
(ii)
(iii)
(iv)
(v)
(vi)
(vii)
negativecontrol
negativecontrol
1-6
6-9’
9-12
12-15
Full-length
m/z
Inte
nsi
ty
(B)
15769.7+H
23396.2+H
-
(i)
(ii)
(iii)
(iv)
(v)
(vi)
(vii)
10000 15000 20000 25000
05
101520
10000 15000 20000 25000
05
101520
10000 15000 20000 25000
05
101520
10000 15000 20000 25000
11396.0+H
11858.1+H12873.4+H
14102.9+H
15426.6+H 17929.5+H 23394.4+H
05
101520
10000 15000 20000 25000
9367.0+H10303.8+H
11392.2+H
11852.7+H15426.4+H 17925.8+H
19061.9+H23397.5+H
25907.6+H
05
101520
10000 15000 20000 25000
9368.5+H11392.3+H
12535.3+H12738.1+H
14096.7+H15769.7+H 17920.8+H 23396.2+H
05
101520
10000 15000 20000 25000
05
101520
9370.1+H11395.2+H 12536.8+H
14096.6+H 15774.4+H 17965.4+H 23395.7+H
negativecontrol
negativecontrol
1-6
6-9’
9-12
12-15
Full-length
m/z
(A)
11396.0+H
11858.1+H12873.4+H
14102.9+H
15426.6+H 17929.5+H 23394.4+H
10303.8+H 14104.3+H 17925.8+H19061.9+H
23397.5+H25907.6+H
11392.3+H12535.3+H
12738.1+H14096.7+H
15769.7+H 17920.8+H 23396.2+H
15774.4+H 17965.4+H 23395.7+H
Inte
nsi
ty
10000 17500
00.250.5
0.751
10000
0
0.51
1.5
10000
9326.7+H11302.6+H
11765.8+H12441.1+H
14001.1+H
0
1
2
10000
9330.7+H
11305.7+H11776.8+H 12447.4+H
00.250.5
0.751
10000
9329.9+H
11304.8+H11769.0+H
12445.8+H 14003.0+H
0
0.5
1
10000
10236.1+H
11305.0+H12444.5+H 14004.1+H
00.5
11.5
2
10000
9329.5+H
11306.3+H12445.9+H 14007.5+H
15667.8+H
0
0.2
0.4
0.6 11300.9+H11955.3+H 12446.2+H 13988.5+H
(i)
(ii)
(iii)
(iv)
(v)
(vi)
(vii)
negativecontrol
1
2
3
4
5
6
m/z
Inte
nsi
ty
00.250.5
0.751
0
0.51
1.5
0
1
2
00.250.5
0.751
14003.0+H
0
0.5
112444.5+H
00.5
11.5
2
15667.8+H
0
0.2
0.4
0.6
negativecontrol
1
2
3
4
5
6
14004.6+H
14003.0+H
(C)
20000 50000
00.0250.05
0.075
20000
0
0.1
0.20.3
20000
27168.3+H
0
0.2
0.4
20000
15441.8+H25803.2+H
-0.10
0.10.20.3
20000
15721.8+H 25922.1+H
0
0.2
0.4
20000
15396.6+H
25858.8+H
37466.1+H 42019.1+H
0
0.2
0.4
0.6
20000
15788.2+H 27157.1+H
-0.050
0.050.1
0.150.2
15733.2+H25906.3+H
27024.0+H
(i)
(ii)
(iii)
(iv)
(v)
(vi)
(vii)
negativecontrol
1
2
4
5
6
3
m/z
Inte
nsi
ty
25938.7+H
27072.3+H
27076.8+H
27029.1+H
25852.7+H
-
-
negativecontrol
1
2
4
5
6
3
(D)
Fig. 3 – (A and B) MALDI-TOF mass spectra of proteins from epidermal extracts captured on arrays with (i) no bound protein or bound covalently with (ii) transferrin, (iii) rLEKTI1–6, (iv) rLEKTI 6–9′, (v) rLEKTI 9–12, (vi) rLEKTI 12–15 or (vii) full-length rLEKTI. The spectra cover the mass range (A) 9000–30,000 m/z and (B) 15,000–150,000 m/z. (C and D) Theexperiment was repeated with skin samples from six different subjects on arrays with (i) no bound protein or (ii–vii) bound covalently with full-length rLEKTI. The spectra coverthe mass range (C) 8000–17,500 m/z and (D) 15,000–60,000 m/z.
3932JO
UR
NA
LO
FPR
OT
EO
MIC
S75
(2012)
3925–3937
11411.2+H
5
10
15
12765.9+H
15797.4+H
17063.3+H 23579.0+H 27445.5+H 42019.7+H
0
0
5
10
15
20000 30000 40000 50000
Inte
nsi
ty
20000
20000
40000
40000 (A)
(B)
m/z
Fig. 4 – The extraction of LEKTI targets from the protein array surface. MALDI-TOF mass spectra of (A) proteins from skinextracts captured on protein arrays with bound rLEKTI 6–9′. (B) Representative and typical spectra of (A), after it has beentreated with extraction reagents, 70% acetonitrile, 0.2% TFA and 50% formic acid, 25% acetonitrile and 15% isopropanol.
3933J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 3 9 2 5 – 3 9 3 7
3.6. The expression of dermcidin in Netherton syndrome
As a negative control and in order to investigate the effect of anabsence of LEKTI on the levels and localisation of dermcidin,immunofluorescence staining for both LEKTI and dermcidinwere performed on a skin sections taken from patients withNetherton syndrome. Each patient had a compound heterozy-gous mutation in the SPINK5 gene (c.1732 C>T, c.2274insT),leading to the formation of premature stop codons in exon 19
(A)
(B)
Fig. 5 – Sequence coverage obtained for (A) dermcidin and (B) catpeptides sequenced by Q-TOF MS are highlighted in purple. Regsequenced peptides that had been modified by carbamidomethyC-terminal peptides DCD-1 (63–109) and DCD-1L (63–110). Cathep(1–18), propeptide (19–20) and peptidase S1 domain (21–243).
and exon 26, respectively. This mutation leads to a completedeficiency of LEKTI. Immunofluorescence staining demon-strates an absence of LEKTI expression in the epidermis of thispatient (Fig. 6B). The LEKTI-deficient epidermis showed a morediffuse distribution of the protein dermcidin with a markedincrease in the expression of dermcidin throughout the lowerepidermal layers and a reduction in the upper granular layercompared to the control (Fig. 6B). Negative controls performed inthe absence of primary antibodies for both LEKTI and dermcidin
hepsin G. Regions of the protein sequence that match theions highlighted in dark purple or green correspond tolation. Full-length dermcidin (110 aa) is processed intosin G is 255 amino acids long and consists of a signal peptide
Fig. 6 – Immunofluorescence staining for LEKTI and dermcidin in skin sections from a (A) control and (B) a patient withNetherton syndrome. LEKTI expression is shown in red and dermcidin expression is shown in green. Co-localisation of LEKTIand dermcidin (yellow colour) was observed in the upper granular layer of the normal skin section. Areas with intense nucleistaining (blue colour) highlight the viable layers of the epidermis (basal and granular layers). Negative controls were performedin the absence of primary antibodies for the skin section from the (C) control and (D) patient with Netherton syndrome. Scalebars=50 μm.
3934 J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 3 9 2 5 – 3 9 3 7
showed non-specific staining in the cornified layer (uppermostlayer) of both skin sections (Fig. 6C and D, respectively).
4. Discussion
The aim of this study was two-fold, firstly to create andoptimise a method to find potential binding/interactingpartners of LEKTI in vitro and secondly, to apply thistechnique in patient samples to study potential diseasemechanisms in Netherton syndrome. LEKTI has a recognisedrole as a serine protease inhibitor through its action onvarious kallikreins and regulation of skin desquamation. Morerecently, LEKTI was also found to inhibit the cysteine proteasecaspase 14 in vitro [15], an enzyme that is thought to play amajor role in terminal keratinocyte differentiation andepidermal hydration through the processing of filaggrin[16,17]. This finding indicated for the first time that LEKTI isa multi-functional protease inhibitor in the skin.
We have shown previously that protein arrays can be usedin immunocapture type analyses [23] but with modificationswe have demonstrated that protein arrays could be used tostudy other protein:protein (‘bait’) interactions and in aquantitative manner that correlate with enzyme substrateinteractions. In this study, the method was developed toinvestigate the interactions of full length LEKTI and itsrespective domains (rLEKTI 1–6, rLEKTI 6–9′, rLEKTI 9–12,rLEKTI 12–15) against known targets of LEKTI before beingused to identify novel targets of LEKTI. Themethod was firstly
optimised and validated using a number of kallikrein stan-dards. KLK1, a protease that has been shown not to beinhibited by any rLEKTI form in vitro [12] was used as anegative control and did not bind to any rLEKTI form on thearray surface. In addition, KLK 5, 13 and 14 which have allbeen shown previously to be inhibited by rLEKTI 1–6, rLEKTI 6–9′, rLEKTI 9–12, rLEKTI 12–15 or full-length rLEKTI in vitro[9,12,18], bound to the corresponding rLEKTI form(s) on thearray surface. Evidence suggested that these proteases boundto rLEKTI in a specific manner as none of these proteasesbound in a non-specific manner to either the hydrophobiccoating of the chip surface or to the control protein,transferrin. These observations confirmed that the protein‘bait’ array method developed could be used to detectpotential targets of LEKTI or any other specific protein:proteininteraction.
The method was further validated using inhibition studiesto investigate whether binding to LEKTI on the protein arraysurface corresponded to actual levels of inhibition. The overallrank order of inhibitory potencies for each rLEKTI fragmentfrom highest to lowest was rLEKTI 1–6, rLEKTI 6–9′, rLEKTI9–12, rLEKTI 12–15. This remained consistent for KLK5, 13 and14 and in the majority of cases was found to correspond wellwith the order of binding efficiencies for all four rLEKTIfragments. It was therefore concluded that the extent ofbinding on the array surface for rLEKTI 1–6, rLEKTI 6–9′, rLEKTI9–12 and rLEKTI 12–15 had good correlations with the generaltrends in inhibition. Although the inhibitory activities of allfour rLEKTI fragments against KLK1, 5, 13 and 14 have been
3935J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 3 9 2 5 – 3 9 3 7
investigated previously [12], our study is the first to show aclear relationship between the binding and inhibitory activityof these LEKTI fragments (including full-length LEKTI) againstKLK5, 13 and 14.
In contrast, the binding efficiencies of full-length rLEKTI onthe array surface did not correlate with inhibition. Full-lengthrLEKTI was found to bind consistently lower amounts ofKLK13 and 14, compared to rLEKTI 1–6 and rLEKTI 6–9′ andwas also found to bind the lowest amount of KLK5 out of allthe rLEKTI forms tested. This did not correspond to inhibitorydata that showed full-length rLEKTI as the most potentinhibitor of all the KLKs tested. Previous studies have shownthat several KLKs are able to completely hydrolyse specificrLEKTI fragments. Full-length rLEKTI is a large protein (1064amino acids) therefore only a small proportion of the proteinwould have made direct contact with the array surface,exposing multiple putative cleavage sites for hydrolysis byKLKs [12,24]. Although full-length LEKTI has been detected inhuman stratum corneum extracts [2], it is often postulated asan inactive precursor that is processed proteolytically intobiologically active fragments. The fact that full-length rLEKTIwas found to be the most potent inhibitor of all KLKs tested,may suggest a more active role for full-length LEKTI in vivo.
After validation and optimisation of the method, the LEKTIprotein array was used a ‘bait’ to detect potential bindingpartners of LEKTI from epidermal protein extracts and toidentify a novel target(s) of LEKTI in the skin. Previous studieshave focused on the use of inhibitory assays to indicatepotential targets of LEKTI [4,12]. With these types of assaysonly one protease can be tested at a time, the wide range ofproteins in the skin and the varied affinities LEKTI may havefor each protein in vivo are not taken into account. The novelapproach used in this study had an advantage over standardinhibitory assays because it allowed the use of whole proteinextracts from the epidermis in an attempt to detect targets ofLEKTI. As a result, no prior knowledge of the potential targetprotein was required.
Both the high lipid content of the skin and the presence ofsignificant amounts of hydrophobic proteins presented apotential challenge for analysis. By using the detergent TritonX-100 at a concentration of 1% enabled the disruption of thebipolar lipid membranes and the extracellular lipids of cells,allowing release and solubilisation of lipid-bound proteins forefficient protein extraction, but did not compromise thenative conformation of the proteins. The latter were essentialfor subsequent interactions with LEKTI on the array surface.Multiple binding partners (9000 m/z–27,000 m/z) were foundto bind to rLEKTI 1–6, rLEKTI 6–9′, rLEKTI 9–12 and full-lengthrLEKTI on the array surface. Many of these bound reproduc-ibly to full-length rLEKTI from the epidermal samples from sixdifferent human subjects. A protein of molecularmass~11 kDa bound to all forms of rLEKTI (except rLEKTI12–15) in the highest amounts. To identify these proteins amethod was developed to extract LEKTI targets from theprotein array surface and identify them by sequencing usingnanoUPLC Q-TOF MS. In order to obtain sufficient amounts ofLEKTI target for protein sequencing the method was scaled upand the arrays used as preparative devices. The proteins weredigested tryptically and the peptides were sequenced bynanoUPLC Q-TOF MS. Surprisingly, no kallikreins or caspase
14 were detected and the identity of the binding partners wasfound to be the proteins dermcidin and cathepsin G. The factthat LEKTI binds preferentially to these proteins from wholeepidermal extracts suggests that LEKTI has a high affinity forthese proteins in vivo. Full-length dermcidin has a molecularmass of ~11.3 kDa, which corresponded to the molecularmass of the protein that bound consistently to rLEKTI 1–6,rLEKTI 6–9′, rLEKTI 9–12 and full-length rLEKTI on the arraysurface in high amounts. This is the first study to report aninteraction of LEKTI with a protein other than a protease.Dermcidin is an antimicrobial peptide expressed constitutive-ly in human sweat glands and transported via sweat to theepidermal surface [25]. Full-length dermcidin (110 aa) isprocessed into C-terminal peptides DCD-1 (47 aa) and DCD-1L (48 aa) [26]. Both DCD-1 and DCD-1L are highly effectiveagainst numerous bacteria including Staphylococcus aureus,Staphylococcus epidermidis, Escherichia coli, Enterococcus faecalisand Candida albicans preventing/limiting skin infection. Thiswas the first study to report the expression of dermcidin inthe epidermis and also the co-localisation of LEKTI anddermcidin in the upper granular layer of the epidermis. Thefinding that dermcidin is a novel target of LEKTI in the skinwas confirmed using immunohistochemical analysis of theskin from patients with Netherton syndrome where LEKTIexpression was absent. Increased amounts of dermcidin weredetected throughout the lower epidermal layers, with lowerlevels present in the uppermost epidermal layers compared tothe control. These results indicate LEKTI may have a role inthe transport and protection of dermcidin from proteolyticactivation until it reaches the surface of the skin whereactivity is required. This would also explain why LEKTI boundto the full-length form of dermcidin (~11.3 kDa) on the arraysurface.
It may also be possible to postulate that reduced levels ofdermcidin in the uppermost layers of the epidermis inpatients with Netherton syndrome may contribute to thehigh susceptibility of these patients to recurrent staphylococ-cal infections. A similar observation has been shown in atopicdermatitis where reduced amounts of dermcidin-derivedpeptides have been found in the sweat of patients with atopicdermatitis [25]. It has been suggested that reduced expressionof dermcidin in sweat may contribute to the high susceptibil-ity of these patients to S. aureus infection and colonisation.Interestingly, the expression of LEKTI was found to besignificantly decreased in keratinocyte cell cultures of threeAD patients with the Glu420Lys polymorphism [27] that hasbeen associated previously with AD [28]. In addition todermcidin, the serine protease cathepsin G was also identifiedas a binding partner of LEKTI on the array surface. Interest-ingly, cathepsin G has been shown previously to be inhibitedby full-length rLEKTI in vitro [9], which would explain why itbinds to LEKTI on the array surface. In normal skin cathepsinG expression was restricted to monocytes and was undetect-able in the epidermis (data not shown). This corresponds wellto previous studies that have reported the expression ofcathepsin G in primary human monocytes [29]. Preliminarydata has also shown that LEKTI is expressed in CD14+ cells,which strengthens the finding that cathepsin G is a target ofLEKTI in the skin. Interestingly, previous studies have shownthat cathepsin D is involved in the post-secretory processing
3936 J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 3 9 2 5 – 3 9 3 7
of dermcidin in eccrine sweat glands [30]. With this in mind itmay be possible to postulate that dermcidin is also a substrateof cathepsin G and that LEKTI, cathepsin G and dermcidin areall involved in the same pathway. LEKTI may regulate theprocessing of dermcidin by inhibiting cathepsin G untilactivity is required. Patients with Netherton syndrome sufferrecurrent infections caused by a compromised skin barrier. Inthe event of infection, immune cells infiltrate the epidermis. Itmay be possible that upon infection cathepsin G is releasedfrom the monocytes (and LEKTI) to the uppermost layers ofthe epidermis where it is able to process dermcidin into activeantimicrobial peptides. This would explain why both dermci-din and cathepsin G were identified on the array surface.Further functional studies are required to help understand thepathways involving these proteins.
In summary, a novel method involving protein arrays andnanoUPLC Q-TOF MS has been developed to investigate LEKTItarget interactions in the skin. Using this technique we haveshowed that the anti-microbial protein dermcidin and ca-thepsin G are novel targets of LEKTI in the skin. This is thefirst study to report the interaction of LEKTI with a proteinother than a protease. LEKTI and dermcidin were found to co-localise in the uppermost layers of the epidermis. In Nether-ton syndrome where LEKTI expression was absent, increasedlevels of dermcidin were detected in the lower epidermallayers, suggesting that LEKTI is required for the transport ofdermcidin to the epidermal surface. The protein arraymethoddeveloped in this study has proved to be a valuable tool for theinvestigation of protein:protein interactions and the identifi-cation of potential drug targets/disease mechanisms involvedin the skin disease associated with Netherton syndrome.
Acknowledgements
The authors thank the National Eczema Society and the PetoFoundation for the funding of this research, Professor BryanWinchester (UCL) for his guidance with the enzyme kineticanalyses and Great Ormond Street Hospital for Sick Children.
R E F E R E N C E S
[1] Bitoun E, Micheloni A, Lamant L, Bonnart C, Tartaglia-PolciniA, Cobbold C, et al. LEKTI proteolytic processing in humanprimary keratinocytes, tissue distribution and defectiveexpression in Netherton syndrome. Hum Mol Genet2003;12(19):2417–30.
[2] Hachem JP, Wagberg F, Schmuth M, Crumrine D, Lissens W,Jayakumar A, et al. Serine protease activity and residualLEKTI expression determine phenotype in Nethertonsyndrome. J Invest Dermatol 2006;126(7):1609–21.
[3] Tartaglia-Polcini A, Bonnart C, Micheloni A, Cianfarani F,Andre A, Zambruno G, et al. SPINK5, the defective gene inNetherton syndrome, encodes multiple LEKTI isoformsderived from alternative pre-mRNA processing. J InvestDermatol 2006;126(2):315–24.
[4] Deraison C, Bonnart C, Lopez F, Besson C, Robinson R,Jayakumar A, et al. LEKTI fragments specifically inhibit KLK5,KLK7, and KLK14 and control desquamation through apH-dependent interaction. Mol Biol Cell 2007;18(9):3607–19.
[5] Chavanas S, Bodemer C, Rochat A, Hamel-Teillac D, Ali M,Irvine AD, et al. Mutations in SPINK5, encoding a serineprotease inhibitor, cause Netherton syndrome. Nat Genet2000;25(2):141–2.
[6] Yang T, Liang D, Koch PJ, Hohl D, Kheradmand F, OverbeekPA. Epidermal detachment, desmosomal dissociation, anddestabilization of corneodesmosin in Spink5−/− mice. GenesDev 2004;18(19):2354–8.
[7] Descargues P, Deraison C, Bonnart C, Kreft M, Kishibe M,Ishida-Yamamoto A, et al. Spink5-deficient mice mimicNetherton syndrome through degradation of desmoglein 1 byepidermal protease hyperactivity. Nat Genet 2005;37(1):56–65.
[8] Hewett DR, Simons AL, Mangan NE, Jolin HE, Green SM, FallonPG, et al. Lethal, neonatal ichthyosis with increasedproteolytic processing of filaggrin in a mouse model ofNetherton syndrome. Hum Mol Genet 2005;14(2):335–46.
[9] Mitsudo K, Jayakumar A, Henderson Y, Frederick M, Kang Y,Wang M, et al. Inhibition of serine proteinases plasmin,trypsin, subtilisin A, cathepsin G, and elastase by LEKTI: akinetic analysis. Biochemistry 2003;42(13):3874–81.
[10] Egelrud T, Brattsand M, Kreutzmann P, Walden M, VitzithumK, Marx UC, et al. hK5 and hK7, two serine proteinasesabundant in human skin, are inhibited by LEKTI domain 6. BrJ Dermatol 2005;153(6):1200–3.
[11] Schechter NM, Choi EJ, Wang ZM, Hanakawa Y, Stanley JR,Kang Y, et al. Inhibition of human kallikreins 5 and 7 by theserine protease inhibitor lympho-epithelial Kazal-typeinhibitor (LEKTI). Biol Chem 2005;386(11):1173–84.
[12] Borgono CA, Michael IP, Komatsu N, Jayakumar A, Kapadia R,Clayman GL, et al. A potential role for multiple tissuekallikrein serine proteases in epidermal desquamation. J BiolChem 2007;282(6):3640–52.
[13] Komatsu N, Takata M, Otsuki N, Ohka R, Amano O, TakeharaK, et al. Elevated stratum corneum hydrolytic activity inNetherton syndrome suggests an inhibitory regulation ofdesquamation by SPINK5-derived peptides. J Invest Dermatol2002;118(3):436–43.
[14] Descargues P, Deraison C, Prost C, Fraitag S, Mazereeuw-Hautier J, D'Alessio M, et al. Corneodesmosomal cadherinsare preferential targets of stratum corneum trypsin- andchymotrypsin-like hyperactivity in Netherton syndrome. JInvest Dermatol 2006;126(7):1622–32.
[15] Bennett K, Callard R, Heywood W, Harper J, Jayakumar A,Clayman GL, et al. New role for LEKTI in skin barrierformation: label-free quantitative proteomic identification ofcaspase 14 as a novel target for the protease inhibitor LEKTI. JProteome Res 2010;9(8):4289–94.
[16] Denecker G, Hoste E, Gilbert B, Hochepied T, Ovaere P,Lippens S, et al. Caspase-14 protects against epidermal UVBphotodamage and water loss. Nat Cell Biol 2007;9(6):666–74.
[18] Jayakumar A, Kang Y, Mitsudo K, Henderson Y, Frederick MJ,Wang M, et al. Expression of LEKTI domains 6–9′ in thebaculovirus expression system: recombinant LEKTI domains6–9′ inhibit trypsin and subtilisin A. Protein Expr Purif2004;35(1):93–101.
[19] Di WL, Larcher F, Semenova E, Talbot GE, Harper JI, Del RM,et al. Ex-vivo gene therapy restores LEKTI activity andcorrects the architecture of Netherton syndrome-derivedskin grafts. Mol Ther 2011;19(2):408–16.
[20] Twining SS, Schulte DP, Wilson PM, Zhou X, Fish BL, Moulder JE.Neutrophil cathepsin G is specifically decreased under vitamin Adeficiency. Biochim Biophys Acta 1996 Nov 15;1317(2):112–8.
[21] Edwards SW. Biochemistry and physiology of the neutrophil.Cambridge University Press; 2005. p. 70.
[22] Schittek B, Hipfel R, Sauer B, Bauer J, Kalbacher H, StevanovicS, et al. Dermcidin: a novel human antibiotic peptide secretedby sweat glands. Nat Immunol 2001;2(12):1133–7.
3937J O U R N A L O F P R O T E O M I C S 7 5 ( 2 0 1 2 ) 3 9 2 5 – 3 9 3 7
[23] Mills K, Mills P, Jackson M,Worthington V, Beesley C, Mann A,et al. Diagnosis of congenital disorders of glycosylation type-Iusing protein chip technology. Proteomics 2006;6(7):2295–304.
[24] Felber LM, Borgono CA, Cloutier SM, Kundig C, Kishi T, RibeiroCJ, et al. Enzymatic profiling of human kallikrein 14 usingphage-display substrate technology. Biol Chem 2005;386(3):291–8.
[25] Rieg S, Steffen H, Seeber S, Humeny A, Kalbacher H, Dietz K,et al. Deficiency of dermcidin-derived antimicrobial peptidesin sweat of patients with atopic dermatitis correlates with animpaired innate defense of human skin in vivo. J Immunol2005;174(12):8003–10.
[26] Flad T, Bogumil R, Tolson J, Schittek B, Garbe C, Deeg M, et al.Detection of dermcidin-derived peptides in sweat byProteinChip technology. J Immunol Methods 2002;270(1):53–62.
[27] Roedl D, Traidl-Hoffmann C, Ring J, Behrendt H, Braun-FalcoM. Serine protease inhibitor lymphoepithelial Kazaltype-related inhibitor tends to be decreased in atopicdermatitis. J Eur Acad Dermatol Venereol 2009;23(11):1263–6.
[29] Burster T, Macmillan H, Hou T, Boehm BO, Mellins ED.Cathepsin G: roles in antigen presentation and beyond. MolImmunol 2010;47(4):658–65.
[30] Baechle D, Flad T, Cansier A, Steffen H, Schittek B, Tolson J,et al. Cathepsin D is present in human eccrine sweat andinvolved in the postsecretory processing of the antimicrobialpeptide DCD-1L. J Biol Chem 2006;281(9):5406–15.
