Identification and Sequence Analysis of Two New Members of the SKALP/elafin and SPAI-2 Gene Family BIOCHEMICAL PROPERTIES OF THE TRANSGLUTAMINASE SUBSTRATE MOTIF AND SUGGESTIONS FOR A NEW NOMENCLATURE*

The human epithelial proteinase inhibitor SKALP/elafin and the porcine sodium-potassium ATPase inhibitor SPAI-2 are two highly homologous proteins that share an NH2-terminal transglutaminase substrate domain and a COOH-terminal whey acidic protein (WAP) domain. Here we describe the bovine and simian orthologs of SKALP/elafin as well as two new bovine family members that are designated Trappin-4 and Trappin-5 on the basis of a new nomenclature that we propose (Trappin = TRansglutaminase substrate and WAP motif-containing ProteIN). Sequence analysis of Trappin-4 and Trappin-5 revealed a domain structure that is very similar to SPAI-2 (Trappin-1) and SKALP/elafin (Trappin-2). The transglutaminase substrate motifs are conserved although the number of repeats varies among species and among family members. The sequence of Trappin-4 and Trappin-5 diverges from Trappin-1 and Trappin-2 at the putative reactive site in the WAP domain. The bovine ortholog of Trappin-2 is expressed in tongue and snout epidermis; Trappin-4 is expressed in trachea, ileum, and tongue; and Trappin-5 is expressed at low levels in trachea, as determined by RNase protection and Northern blot analysis. Based on the analysis of 67 transglutaminase substrate repeats as present in all known Trappin gene family members from four different mammalian species a consensus sequence could be established: Gly-Gln-Asp-Pro-Val-Lys (GQDPVK). Using biotinylated hexapeptide probes we found that the GQDPVK sequence is a very efficient transglutaminase substrate both for guinea pig liver transglutaminase and for epidermal transglutaminase, and it acts as acyl donor as well as acceptor. We propose that the Trappin protein family forms a new group of enzyme inhibitors with various specificities of the WAP domain, which share transglutaminase substrate motifs that can act as an anchoring sequence.

glutaminase (TGase) substrate, and a COOH-terminal whey acidic protein (WAP) domain that harbors an inhibitory activ ity toward at least two distinct enzymes* Porcine SPAI-2 was the first of these molecules to be described and is expressed mainly in the intestine (3). Human SKALP, otherwise known as elafin (4), or elastase-specific inhibitor (5), is a potent inhib itor of the leukocytic proteinases elastase and proteinase-3 (6 , 7). We found that SKALP/elafin is expressed in several human stratifying squamous epithelia, except for epidermis where it is only expressed in the context of inflammation, such as psoriasis or wound healing (8)(9)(10), We mapped the genomic localization of human SKALP/elafin to chromosome 20q l2-13 (11). Inter estingly, this region contains various other genes involved in TGase-mediated cross-linking processes such as the tissue TGase gene (12), epidermal TGase (13,14), and the genes coding for semenogelin I and semenogelin II (15), which are also epithelial TGase substrates. SKALP/elafin is distinct from SPAI-2 on the basis of its inhibitory activity, the amino acid sequence of the putative active site, and the epithelial expres sion pattern. The cDNAs and genes for porcine SPAI-2 (2, 3), porcine SKALP/elafin (16), human SKALP/elafin. (7,17), and a new porcine family member (16) that is more similar to SPAI-2 than to SKALP/elafin, have been cloned and revealed a high degree of conservation in the gene structure and the intronic sequences but a strong sequence divergence in the second exon.
The SKALP/SPAI-2 gene family members are composed of two evolutionary building blocks that are found in other pro teins as well (Fig. 1), The COOH-terminal WAP domain is homologous to the second domain of secretory leukocyte pro teinase inhibitor (SLPI), which inhibits elastase and cathepsin G (18). The NH2-terminal domain, containing the TGase sub strate motifs, is homologous to the guinea pig seminal vesicle protein-1 (19) and the human semenogelins (15). We showed that the NHa-terminal TGase substrate domain, for which the name "cementoin" was coined by others (20), is actually used in vivo and in vitro for cross-linking to stratum corneum proteins. Recently, it was shown by direct sequencing of cross-linked peptides from human foreskin epidermis that SKALP/elafin is cross-linked in vivo to loricrin and cytokeratin-1 (21), which are structural proteins of the terminally differentiating keratinocyte. The cross-linked SKALP/elafin is proteolytically pro cessed further by unidentified proteinases to yield low molec ular weight COOH-terminal fragments containing the antiproteinase activity, starting at amino acid positions 149, 151 and 156 (numbering according to Fig, 2), as shown by NH2-terminal sequencing of purified SKALP/elafin from epi dermal scale extracts (7,10). We have found that the COOHterminal part of SKALP/elafin is cleared via the plasma and can be recovered from the urine (22,23). This mechanism provides the epidermis with an anchored proteinase inhibitor,

20472
The Trappin Gene Family Fig. 1. Schem atic re p re se n ta tio n of p ro tein s w ith a WAP do m ain andVor TGase su b stra te motifs. The boxed S indicates the signal peptide. Panel A , protein structure of SPAI-2 and SKALP/elafin that share an NHa-terminal TGase (TG ) substrate domain and a COOH-terminal WAP domain. Panel B , protein structure of SLPI, which consists of two WAP domains. The COOH-terminal WAP domain of SLPI harbors proteinase inhibitory activity and is homologous to SKALP/elafin. Panel C, schematic structure of the seminal vesicle pro tein-1 and the human semenogelins. These proteins contain a domain that consists of TGase substrate motifs. which could protect the structural proteins of the stratum comeurn against unwanted proteolysis; in addition it gener ates, after cleavage, a gradient of low molecular weight inhib itors from the epidermis to the dermis, thereby possibly inter fering with polymorphonuclear leukocyte chemotaxis and polymorphonuclear leukocyte-induced degradation of extracel lular matrix proteins such as elastin and basal membrane components.
To study the evolutionary divergence of SKALP/elafin in various species and to identify potential new members of this gene family we performed reverse transcription-polymerase chain reaction (PCR) on mRNA of epithelial tissues from var ious mammals, using degenerate primers that encompass the TGase substrate domain and the WAP domain. In this way we identified the putative simian and bovine orthologs of SKALP/ elafin, and we have identified two new family members from bovine tissue. On the basis of the family members thus far identified we propose a new unifying nomenclature. Prom all known SKALP/elafin gene family members in various species we derived a consensus hexapeptide sequence for the TGase substrate motif. Using synthetic peptides we show that this hexapeptide motif is an extremely efficient substrate for vari ous TGases. 4

EXPERIMENTAL PROCEDURES
Tissues-Bovine and simian (rhesus monkey) tissues were obtained from the central animal laboratory, University of Nijmegen, The Neth erlands. Spontaneously shed scales from psoriatic, eczema, and lamel lar ichthyosis (LI) patients were collected and stored at -2 0 *C. The LI patient was homozygous for a splice site mutation in intron 5 of the TGase 1 gene as determined by single strand conformation polymor phism analysis and sequence analysis. This mutation causes a defi ciency for TGase type 1 as recently described (24).
R N A Isolation and cDNA Synthesis -Total RNA from bovine tongue and rhesus monkey skin were extracted with RNAzol B as suggested by the supplier (Cinna/Biotex). First strand cDNA was generated from total RNA with Moloney murine leukemia virus range H" reverse transcriptase (Expand Reverse Transcriptase, Boehringer Mannheim) under conditions as suggested by the supplier, using excess oligo(dT) primers. The reverse transcriptase reaction products were used for PCR amplification to obtain the partial cDNA of bovine and simian orthologs of SKALP/elafin. PCR; Cloning and Sequencing o f PCR Products -Based on conserved amino acid stretches within SKALP/elafin and porcine SPAI-2 we de signed two oligonucleotide forward primers (positions are shown in Fig.  2): primer SKAl, 5'-AGGGCCAGCAGCTTCTTGATC-3'; and primer SKA2, 5 '-CAAGA(G/T)CCAGTCAA(A/G)GGT-3'. As reverse oligonu cleotide primers we used primer SKA3, 5'-CAGCACTTCTTGA(C/T)(C/ T)CCTGG-3' and the oligo(dT) primer (Fig. 2). PCRs were carried out using a DNA thermal cycler (PTC-200, Biozym) in 25-jnl mixtures. The following buffer conditions were used; 10 mM Tris-HCl, pH 8.8, 1.5 mM magnesium chloride, 50 mM potassium chloride, 0.1% Triton X-100, all four dNTPs (each at 200 /xM), 2 units of PrimeZyme DNA polymerase (Biometra), and 20 pmol of each primer. After an initial incubation of 6 min at 94 °C amplification was conducted for 35 cycles as follows: 1 min at 94 °C, 1 min at annealing temperature, and 2 min at 72 Q C. An additional 10 min at 72 °C was used for the last cycle. Annealing temperatures were 57 °C when using primers SKAl and SKA3, 55 °C for primers SKA2 and SKA3, 51 °C for primer SKÀ2 and the oligo(dT) primer. The PCR products were purified from agarose gel and cloned into the pGEM-T vector (Promega) according to the protocol provided by the manufacturer. Blunted PCR fragments were cloned into a Sm aldigested pBluescript KS(+) vector (Stratagene). Both strands were sequenced using the dideoxy chain termination procedure (25).
DNA sequence gel readings were recorded, edited, and assembled using the GCG package (26) provided by the Dutch CAOS/CAMM center. Deduced protein sequences and alignments were analyzed with the same software.
RNase Protection A ssay-Total RNA from different bovine tissues was extracted with RNAzol B as suggested by the supplier. The follow ing bovine tissues were studied; tongue, kidney, liver, trachea, lung, ileum, epidermal skin from snout, brain, and esophagus. The 360-base pair blunt fragment encoding the possible bovine ortholog of human SKALP/elafin (Trappin-2) was cloned into the S m a I site of the pBlue script KS(+) vector, which contains the T3 promoter for transcription of antisense RNA. In vitro transcription was performed using a transcrip tion kit (Boehringer Mannheim) in the presence of [a-a2P]UTP (Amersham). Antisense RNA probe was transcribed from the Xbal-linearized SKALP/elafin (bovine Trappin-2) template. DNA template was de graded with RNase-free DNase I, and the labeled RNA probe was purified using Chroma spin+TE-100 columns (CLONTECH), acidic phenol/chloroform extraction, and ethanol precipitation. Using the RNase protection kit (Boehringer Mannheim), the radioactively labeled probe (1 X 105 cpm) was hybridized to 10 fxg of total RNA from different bovine tissues or to 10 jug of total RNA from human psoriatic skin biopsies as a negative control. Samples were denatured for 10 min at 80 °C followed by an overnight incubation at 55 °C. Partial RNA'RNA hybrids were treated with DNase-free RNase A and RNase 'I\ according to the manufacturer's instructions. The protected RNA fragments were subjected to electrophoresis on a 6% denaturing polyacrylamide gel containing 7 M urea and were sized using the sequencing reaction of the PCR product encoding bovine Trappin-2. Autoradiography was per formed on Biomax-MR films (Kodak) a t -8 0 °C with intensifying screens. Processing of the autoradiographs was performed using the Imagemaster™ data image system (Pharmacia Biotech Inc.).
Northern Blot Analysis -For Northern blot analysis, 10 /ig of total RNA from different bovine tissues was electrophoretic ally separated in a 1% agarose gel (dissolved in 10 mM sodium phosphate buffer, pH 7.0), as described before (27). The gel was blotted by capillary transfer onto positively charged nylon membrane (Boehringer Mannheim) using 10 X SSC (1.5 M sodium chloride, 0.15 M sodium citrate). After transfer, RNA was fixed to the membrane by ultraviolet irradiation (312 nm, 0.2 J/cm2), The membrane was incubated with ethidium bromide (1 /ig/ml) for 15 min prior to photography. Hybridization was performed in 250 mM phosphate buffer at 60 °C according to Church and Gilbert (28), using 32P-labeled random primed probes encoding bovine Trappin-4 and -5 obtained by reverse transcription-PCR. The blot was washed twice for 15 min at 60 °C using 125 mM phosphate buffer. Autoradiog raphy was performed on Biomax-MR films at -8 0 °C with intensifying screens. Processing of the autoradiographs was performed using the Imagemaster™ data image system.
Cross-linking of the Biotinylated Peptides to Stratum Corneum Pro teins by Endogenous TGase -Epidermal scales from a psoriatic patient (200 mg) were homogenized in 4 m l of buffer containing 50 mM Tris-HCl, pH 7.8, 100 mM sodium chloride, and 1 mM phenylmethylsulfonyl fluoride (Sigma), and centrifuged for 30 min at 25,000 X g. The same protocol was used for scales from patients with eczema and LI. The supernatants were stored at -20 °C until further use. For cross-linking experiments 10 /jl of scale extract was used with 5 jul (100 jig/ml) of biotinylated peptide. The following buffer conditions were used: 50 mM Tris-HCl, pH 7.8, 100 mM sodium chloride, and calcium chloride at a concentration of 2 mM, in a final reaction volume of 50 \id. In some experiments 2 jul (0.0313 unit/ml) of guinea pig liver TGase (Sigma) was added. After 60 min at 37 °C, the reaction was stopped by the addition of 5 fx1 of 100 mM EDTA, pH 7.8. Reaction mixtures containing 10 mM ED TA, pH 7.8, and heat-inactivated scale extracts were used as con trols. The reaction mixtures were blotted directly onto polyvinylidene h T r a p p i n -2 s T R A P P I N -2 p T r a p p i n -2 b T R A F P I N -2 b T R A P P I N -4 b T R A P P I N -5 p T r a p p i n -l pTrappin.-3 10 20 h T r a p p i n -2 s T R A P P I N -2 p T r a p p i n -2 b T R A P P I N -2 b T R A P P I N -4 b T R A P P I N -5 p T r a p p i n -l p T r a p p i n -3 h T r a p p i n -2 s T R A P P I N -2 p T r a p p i n -2 b T R A P P I N -2 b T R A P P I N -4 b T R A P P I N -5 p T r a p p i n -l p T r a p p i n -3 Alignm ent of am ino acid seq u en ces of m em bers from th e T ra p p in p ro te in family. The amino acid sequences (single letter code) of the novel bovine Trappins (bTRAPPIN-4 and -5) and the bovine and simian orthologs of SKALP/elafin (bTRAPPIN-2 and sTRAPPIN-2) are deduced from the partial cDNA sequences obtained, by reverse transcription-PCR (primer sequences are excluded). Comparisons with human SKALP/elafin (hTrappin-2), porcine SKALP/elafin (pTrappin-2), the porcine sodium-potassium ATPase inhibitor SPAI-2 (pTrappin-l), and a new porcine family member (pTrappin-3) are shown. The italicized amino acid residues (from position 176 to 182) correspond to the known active site region of SLPI which is a protein th at consists of two WAP motifs (18,57), The binding site of SKALP/elafin with elastase is found at the amino acid sequence Leu17ö(P5)-Leulö2(P2'), which contains the scissile peptide bond AlaI8fl-Met181 (35), which is marked by a vertical arrow (4 ). The hydrophobic signal sequences of hTrappin-2 (7) and pTrappin-l (2) are indicated in bold. A phase-1 intron ( §> <g) (between the first and second nucleotides of the codon for the amino acid at position 42 separates the exon coding for the signal sequence from the TGase substrate domain. The hexapeptide repeats in the Trappins are grouped in the second and third blocks of the alignment. As pointed out by Tamechika et al. (16), expansion and contraction of the number of hexapep tides in this gene family occur readily, presumably as a result of gene conversions and DNA polymerase slippage. Some sequence similarities suggestive of conversions among the three porcine Trappin genes are underlined. Eight conserved cysteine residues involved in the formation of the WAP motif or four-disulfide core structure of the Trappin gene family are indicated (v). Amino acid residues with 100% conservation are indicated with asterisks (*). The positions of the oligonucleotide forward primers (sk a ï and ska2) and reverse primers (ska3 and oligo-dT) used in the PCR are indicated. difluoride membranes (Millipore) by a slot-blot manifold or were sub jected to SDS-PAGE (12% Ready gel, Bio-Rad) and blotted onto polyvinylidene difluoride membrane. Biotinylated proteins were detected with the Western Light kit (Tropix) according to the manufacturer's instructions. This assay uses avidin-conjugated alkaline phosphatase and chemiluminescence of a sensitive alkaline phosphatase substrate. Positive bands were recorded on X-Omat S i films (Kodak) and were processed with a maximum scanning OD range of 1.5 and quantitated using the Imagemaster™ data image system.
Keratinocyte Secondary C ulture-U nm an primary kératinocytes» cultured according to the Rheinwald-Green system (33), were seeded at 10s cells in keratinocyte growth medium in 60-mm culture dishes as 20474 sonification on ice, followed by centrifugation for 10 min at 25,000 X g.

The Trappin Gene Family
The supernatants were stored at -2 0 °C until further use.

Cloning and Sequence Analysis of Bovine and Simian SKALP fElafin Orthologs and Two New Bovine Members of the SKALP IElafin Gene Family; Suggestions for a New
Nomencla ture-To isolate partial cDNAs of bovine and simian SKALP/ elafin orthologs, three oligonucleotide primers (Fig. 2) were designed based on conserved sequences within human SKALP/ elafin and porcine SPAI-2. Using total RNA derived from bo vine tongue and rhesus monkey skin, first strand cDNA was generated in a reverse transcriptase reaction and amplified by PCR with the designed primers. PCR products were cloned and sequenced. Four clones were identified as members of the SKALP/elafm gene family based on the presence of sequences encoding an NHz-terminal TGase substrate domain and a COOH-terminal WAP-domain. Computer-assisted comparison with published sequences of this gene family was performed to reveal the possible identity of the clones. Fig. 2 shows the alignment of the deduced amino acid sequences of the four clones with the currently known family members. Two se quences can be regarded as the simian and bovine orthologs of SKALP/elafin. Overall, simian SKALP/elafin is 93% identical to human SKALP/elafin, and bovine SKALP/elafm is 71% iden tical to porcine SKALP/elafin. Within a part of the WAP do main (amino acid residues at positions 163-194 in Fig. 2) these percentages are 97% and 81%, respectively (Table I). The se quences of the putative active sites (residues italicized in Fig.  2) of both simian and bovine SKALP/elafin closely correspond to the known protease binding site of SKALP/elafin (35) and the pro tease binding site of SLPI (36) which also belongs to the WAP protein superfamily. The homology in the putative active site of both simian and bovine SKALP/elafm suggests that these molecules could be elastase inhibitors, although we have no data from functional studies to substantiate this contention. It was, however, shown recently that the porcine SKALP/elafin ortholog is indeed an inhibitor of at least porcine pancreatic elastase (16).
At present, the nomenclature of the proteins containing an NH2-terminal TGase substrate domain and a COOH-terminal WAP-domain is very confusing. Until now porcine SPAI-2 (2) and the proteinase inhibitor SKALP/elafin/elastase-specific in hibitor were described in the literature (1, 7, 17, 37). For the NH2-terminal TGase substrate domain a separate name, ce-55  52  47  44  55  52  47  44  68  61  53  47  74  68  53  47  93  48  42  93  48  39  68  68  59  68  65  69 mentoin, was coined by Nara et aL (20). Here we propose to give these proteins the acronym Trappin (Trappin = TRansglutaminase substrate and WAP motif-containing ProteIN) as a new nomenclature for this protein family. Since SPAI-2 and SKALP/elafin/elastase-specific inhibitor were the first mem bers to be described, these are designated Trappin-1 and Trappin-2, respectively. Trappin-3 is a new porcine member of this gene family which was called pWAP-3 by Tamechika et al. (16). The two other sequences from bovine tongue we identified (see Fig. 2) are new members of the SKALP/elafin gene family and are designated Trappin-4 and Trappin-5 on the basis of the new nomenclature that we proposed. Sequence analysis of these new members of the Trappin gene family revealed a domain structure that is very similar to porcine SPAI-2 (Trappin-1) and human SKALP/elafin/elastase-specific inhibitor (Trappin-2). The sequence of Trappin-4 and Trappin-5 diverges from SKALP/elafm and SLPI at the putative reactive site; however, the cysteine residues of the four-disulfide core structure ob tained from the partial cDNA sequence are conserved. The biological function of these new bovine Trappin gene family members is not known. The putative reactive site region in Trappin-4 and Trappin-5 is distinct from SLPI and SKALP/ elafin at the Met181 and Leu182 residues (numbering according to Fig. 2) and could possibly lead to dissimilar functions. Di versity in biological functions among SKALP/elafin, SLPI, and SPAI-2 could be the result of the differences in residues around the protease cleavage site within the WAP domain (see Fig. 2).
(Amino acid residues of substrates numbered PI, P2, etc., are toward the NH2-terminal direction, and PI', P2', etc., are to ward the COOH-terminal direction from the scissile bond, as in the nomenclature of Schechter and Berger (38)). The preferable residues at PI' and P2' in a-lytic protease, which demonstrates an elastase-like primary specificity, are, respectively, methio nine and leucine (39). As the scissile peptide bond in SKALP/ elafin and SLPI is Ala180(Pl)-Met181(Pr), these proteins hold the preferable residues at Met181 (P1') and Leu182(P2 '), sug gesting a specificity like a-lytic protease (35). Although porcine SKALP/elafin varies from human SKALP/elafin and SLPI at the PI and P2' residue, this protein shows the ability to inhibit porcine pancreatic elastase (16), suggesting that the conserved Met181 residue is necessary for elastase specificity. An intact methionine is required for human SKALP/elafm to act as a high affinity inhibitor of leukocyte elastase (35). The putative reactive site in the new bovine Trappin gene family members Trappin-4 and Trappin-5 lacks this methionine residue, which makes it unlikely that these proteins are elastase inhibitors. Recently, Trappin-3 was described (16) which is 59% identical to Trappin-1 within a part of the WAP domain (amino acid residues 163-194 according to Fig. 2 ). Comparison of our new family members Trappin-4 and -5 with porcine Trappin-1 and Trappin-3 revealed less than 48% identity. Trappin-4 and -5 share a 93% identity. On the basis of sequence identity in the active site, the Trappin family members can be divided in three subgroups: Trappin-1 and -3, Trappin-4 and -5, and Trappin-2 (Table I), Tissue Distribution of Trappin Family Members-The. Trap pin gene family members exhibit distinct tissue distributions. Porcine Trappin-1 (SPAI-2) was previously found to be abun dantly expressed in the intestine, whereas porcine Trappin-3 is found in the intestine at relatively low levels (16). The mRNA for porcine Trappin-2 (SKALP/elafm) was mainly found in tra chea and large intestine. This is different from human Trap pin-2 which is expressed in several human stratifying squa mous epithelia (9), except for epidermis where it is only expressed in the context of inflammation, such as psoriasis and wound healing (8 , 10). The observed differences could be the result of different sensitivity of the detection methods or could represent real species differences. To define tissue distribution patterns of bovine Trappin-2, expression of the Trappin-2 mRNA was studied by RNase protection analysis. Protected bands with the expected length (360 base pairs) were detected in the tongue and in the epidermis of snout (Fig. 3A). As a negative control we used RNA from psoriatic skin, which con tains human Trappin-2 mRNA. Because human Trappin-2 has several mismatches compared with the bovine ortholog it will be degraded by RNase and is hence not visible on the gel. The expression of Trappin-4 and -5 mRNA in various bovine tissues was studied by conventional Northern blot analysis. Trappin-4 is abundantly expressed in trachea, and a faint signal was found in ileum and tongue (Fig. 35), whereas Trappin-5 was expressed at relatively low levels in trachea (Fig. 3C). Trap pin-4 an -5 are very similar at the DNA level (88% overall identity), and therefore it cannot be totally excluded that the signal of the Trappin-5 probe on the Northern blot is the result of cross-hybridization.
The high level of sequence divergence in the short amino acid stretch encoding the inhibitor domain of this gene family can be viewed as an example of accelerated evolution as described by Tamechika et al. (16). This is a process assumed to take place in genes after a duplication event (40) and has been interpreted as an effective mechanism to create new reactive site sequences with different substrate specificities. These amino acid changes are established by positive Darwinian selection as first re ported by Hill and co-workers (41, 42), who described this phenomenon in the serine proteinase inhibitor (serpin) family.
Since the Trappin family members are all expressed in tissues that are exposed continuously to microbial stimuli (oral cavity, trachea, intestine) it raises the possibility that some of them are directed against bacterial proteinases rather than exclu sively against self-proteases (as Trappin-2, which is directed against leukocyte elastase and proteinase-3). This contention is supported by the recent finding that the distant family member SLPI, which is homologous to Trappins in its COOH-terminal domain, possesses antiviral and antibacterial activity (43,44). The high substitution rates in the reactive center region of the Trappins could conceivably provide the host with a defense system against pathogens and parasites and give them the capacity to deal with an increasing number of attacking proteinases.
The TGase substrate motifs in the NH2-terminal part of the Trappins are conserved, although the number of amino acid repeats varies among species and among the Trappin gene family members. Porcine and bovine Trappin-2 display 11 and 12 repeats, respectively, whereas six repeats were found in human and simian Trappin-2. The deduced amino acid se quences of the novel members derived from bovine tongue, Trappin-4 and Trappin-5, both contained five repeats. Com bined with 14 repeats found in the published porcine Trappin-1 protein and eight repeats in the new porcine family member Trappin-3, the frequency of amino acids at each position was calculated by comparing the total of 67 repeats in the putative TGase substrate domains of the currently known Trappin pro tein family members from four different mammalian species (Table II). In this way a consensus hexapeptide sequence of GQDPVK could be deduced, in agreement with earlier studies (2, 7).

Conservation of a TGase Substrate Motif in Members o f the Trappin Gene
Family -■ To characterize further the biochemical properties of the above mentioned TGase consensus substrate motif, the biotinylated hexapeptide GQDPVK was synthesized, and enzyme kinetic experiments were performed using a TGase assay (29) to determine optimal reaction conditions. The TGase cross-linking reaction is based on a Ca2 ' -dependent exchange of primary amines for ammonia at the y-carboxamide group of glutamine residues. Peptide-bound lysine residues or poly amines serve as the primary amines to form either e-(y-glu-tamyDlysine or (y-glutamyl)poly amine bonds between proteins (45), which are highly resistant to chemical and enzymatic degradation (46). As human epidermis is known to contain both TGase activity (47,48) and various substrate proteins (e.g. involucrin, loricrin, small proline-rich proteins) (21), an extract of psoriatic scales was used to study the incorporation of the biotinylated hexapeptide. A time course incubation at 37 °C of the biotinylated hexapeptide with scale extract showed that the reaction rate as measured by chemiluminescence detection was linear up to 3 h (data not shown), For further experiments a reaction time of 60 min was used for practical convenience. The effect of the peptide concentration on the rate of cross-linking to epidermal proteins by endogenous TGase is shown in Fig. 4. An apparent Km of 0.46 ju,m was found.
The protective callus layer resulting from terminal differen tiation of the squamous epithelium is thought to be crosslinked by different TGase activities present in mammalian epidermis. These T Gas es are probably involved in the forma tion of the cornified cell envelope of terminally differentiating epidermis and of other stratified squamous epithelia. Three TGases are expressed in the epidermis, a ubiquitous tissue type TGase (TGC or TGase 2), a membrane-associated kerati nocyte TGase (TGK or TGase 1) present in cultured epidermal keratinocytes and in many epithelial and nonepithelial tissues, and the zymogen epidermal TGase (TGE or TGase 3) known to 20476

Consensus amino acid sequence o f the TGase substrate m otif in various Trappin fa m ily members
The established hexapeptide is derived from the frequency of amino acids at each position by comparing 67 repeats in the putative TGase substrate domain within eight members of the Trappin family from four different species. Hexapeptide sequences th a t contain fewer than three residues identical to GQDPVK are excluded from the frequency calculation.  Fig. 4, Effect of peptide co n cen tratio n on the ra te of crosslinking to epiderm al p ro tein s by endogenous TGase. The bioti nylated GQDPVK hexapeptide was incubated with scale extract from a psoriatic patient for 60 min at 37 °C. The reaction mixture was blotted directly onto polyvinylidene difluoride membrane, and the biotinylated proteins were detected by chemiluminescence that was recorded on x-ray film and quantified by image analysis. An apparent Km of 0.46 pM was found. be expressed in differentiated epidermal cells and hair follicles, but not in cultured epidermal keratinocytes (49)(50)(51)(52)(53). Incuba tion of the biotinylated GQDPVK hexapeptide with psoriatic scale extract leads to its cross-linking to stratum corneum proteins by Ca2+-activated endogenous TGase (Fig. 5A, sample 1). Human epidermis and probably also stratum corneum are known to harbor TGases 1 , 2 , and 3 ( 2 1 , 47, 48). From our data, however, it cannot be concluded which type is responsible for cross-linking the biotinylated GQDPVK hexapeptide to sub strate proteins. Since TGase 1 and 3 are not available in puri fied form we only tested guinea pig liver TGase (type 2 TGase), and this was found to cause incorporation of the biotinylated hexapeptide into stratum corneum proteins of a native scale extract (sample 2) or scale extract that had been heat-inacti vated to eliminate endogenous TGase activity (sample 3); so both the TGases present in scale extract and type 2 TGase catalyze the incorporation of biotinylated peptide into stratum corneum proteins. Specificity of the reaction was checked by the omission of guinea pig liver TGase (sample 4) or the addi tion of excess EDTA (not shown).
The biotinylated GQDPVK hexapeptide appeared to be an extremely efficient TGase substrate that acts both as an acyl donor and as an acyl acceptor probe (Fig. 55). Substitution of the acyl acceptor residue lysine (K) for arginine (R) (sample 2) or substitution of the acyl donor residue glutamine (Q) for asparagine (N) (sample 3) showed no influence on cross-linking to stratum corneum proteins. Substitution of both the lysine and the glutamine residue for, respectively, arginine and as paragine, totally abolished cross-linking of the biotinylated hexapeptide (sample 4). The efficiency of cross-linking is se quence dependent as a hexapeptide in the reverse order (KVP-DQG, sample 5) or the exchange of lysine and glutamine (GK-DPVQ, sample 6) virtually eliminated cross-linking to stratum  (29) showed that some residues directly preceding the substrate lysines have a nega tive effect on TGase activity. These residues, like Asp, Gly, Pro, His, and Trp, appeared to be largely avoided in a total of 30 characterized acyl acceptor (lysine) substrates. We would spec ulate that degenerate hexapeptide motifs could also be used in vivo. We have shown previously that a synthetic peptide com prising the NH2-terminal 14 amino acids of processed SKALP/ elafin, which contains the degenerate motif GQDTVK, is also incorporated efficiently by TGase, suggesting that slight vari ations in the surrounding amino acids are tolerated. In the same study we used full-length SKALP/elafin purified from human keratinocytes and showed that the protein is crosslinked to an acyl acceptor probe by the action of type 2 TGase. A recent study by Steinert and Mare ko v (21) showed that the degenerate AQEPVK and GQDKVK sequences were used for cross-linking to loricrin and cytokeratin 1 in vivo, as deter mined by amino acid sequencing of purified peptides from human foreskin.
In addition to cross-linking of the GQDPVK hexapeptide to its natural substrate proteins in stratum corneum, we also used purified control proteins to investigate the substrate spec ificity of GQDPVK for these proteins. The biotinylated GQD PVK hexapeptide was found to be cross-linked efficiently by exogenous type 2 TGase to cd3-crystallin and ß-low-crystallin, structural proteins of the vertebrate eye lens (Fig. 5C). These proteins are known acyl acceptor substrates for type 2 TGase, as the COOH-terminal lysine residue of aB-crystallin was iden tified as the site of linkage (30), and a lysine residue in the NH2-terminal extension acts as the sole acyl acceptor substrate in /3A3-crystallin, which is a component of a /3-low-crystallin preparation (31). Two glutamine residues acting as acyl donor were characterized in the NH2-termmal region of ßA3-crystallin by Berbers et al. (54). No cross-linking capacity was found using oA-crystallin and y-crystallin as a substrate, which is in accordance with experiments described previously by Groenen et al. (30,31). Recombinant SKALP/elafin, which only contains the 57 COOH-terminal amino acids of the full-length molecule (and thus lacking the TGase substrate domain), and bovine serum albumin did not show appreciable cross-linking to GQD PVK. To determine the specificity of the GQDPVK hexapeptide for cross-linking to stratum corneum proteins by endogenous TGase in psoriatic scale extracts, we introduced two other acyl donor probes: a biotinylated TVQQEL hexapeptide that is pat terned on the NHg-terminal extension of bovine /3A3-crystaIlin (30), and a biotinylated PGGQQIV heptapeptide, patterned on the amine acceptor sequence in fibronectin (55). Fig. 5D shows that the GQDPVK hexapeptide reacts to a greater extent with stratum corneum proteins than the other acyl donor probes (samples [1][2][3][4][5], whereas they all show the same reactivity to ward /3-low-crystallin by the action of exogenous type 2 TGase (samples [4][5][6]. This suggests that the GQDPVK motif is a preferred substrate for cross-linking to stratum corneum pro teins by epidermal TGases* TGase activity in stratum corneum extracts could be derived from TGase types 1> 2, and 3» Previous studies, however, have suggested that the bulk of the soluble TGase activity comes from TGase type 3 (49). To investigate the relative contribution of the various TGase types in stratum corneum extracts we used scale extracts from three psoriatic patients (Fig. 5E> sam ples IS), a patient with LI (sarnple 4)> and a patient with eczema (sample 5). Surprisingly, the TGase activity measured in these five scale extracts was very similar. Assuming that TGase 1 would contribute significantly, a decreased TGase activity was expected in scale extract from the patient with LI, since this patient was homozygous for a splice mutation in intron 5 of the TGase type 1 gene as determined by single strand conformation polymorphism analysis and sequence analysis.2 This mutation causes a deficiency for TGase type 1 as described recently (24) and is probably the origin for dis turbed formation of the cornified envelopes which may explain the phenotype of LI. Despite a defect TGase 1 gene in LI, our experiments showed that the GQDPVK hexapeptide is still cross-linked to stratum corneum proteins by endogenous TGases in scale extract from a LI patient. It is most likely that TGase type 2 or 3 is responsible for this phenomenon. In a recent publication TGases 1, 2 , and 3 were shown to utilize loricrin in vitro as a complete substrate, but the types of crosslinking were different (56). TGase 1 mostly formed oligomeric complexes by interchain cross-links, whereas TGase 3 reactiv-  ity is involved in the formation of intrachain cross-links. The participation of TGase 2 in loricrin cross-linking was quite weak. It is therefore likely that TGase 3 is the active enzyme in the scale extract of the LI patient. Evidence that the GQDPVK motif can be used for cross-linking by TGase 1 was obtained by using cultured normal epidermal keratinocytes, which do not express TGase 3 (49), as a source of TGase type 1 and substrate (Fig, 6). Subsequent reaction with the biotinylated GQDPVK hexapeptide and analysis by SDS-PAGE show incorporation in proteins predominantly between 30 and 80 kDa (Fig. 6, lane 2). For comparison the incorporation pattern in stratum corneum proteins from psoriatic epidermis is shown (Fig. 6 , lane 4). Cross-linking of GQDPVK to aB-crystallin by TGase 2 is dem onstrated in lanes 5 and 6.
To conclude, we have identified novel Trappin family mem bers, and we have characterized some of the biochemical prop erties of the GQDPVK motif with respect to TGase cross-link ing in vitro. In addition to the TGase substrate domain, Trappin family members possess a COOH-terminal WAP motif that harbors putative proteinase inhibitory activity. The con stitutive expression of Trappin gene family members in a num ber of normal epithelia which are subjected to continuous me chanical and microbial stress or inflammatory stimuli (e.g. oral epithelia, esophagus, trachea, ileum) is in line with a role for these molecules as proteinase inhibitors with different sub strate specificities for self and (possibly) non-self proteases. Future research will be directed at investigation of the role of other Trappin family members in epithelial homeostasis and human diseases.