Identification and characterization of a novel rat ov-serpin family member, trespin.

Serpins are responsible for regulating a variety of proteolytic processes through a unique irreversible suicide substrate mechanism. To discover novel genes regulated by transforming growth factor-beta1 (TGF-beta 1), we performed differential display reverse transcriptase-PCR analysis of NRP-152 rat prostatic epithelial cells and cloned a novel rat serpin that is transcriptionally down-regulated by TGF-beta and hence named trespin (TGF-beta-repressible serine proteinase inhibitor (trespin). Trespin is a 397-amino acid member of the ov-serpin clade with a calculated molecular mass of 45.2 kDa and 72% amino acid sequence homology to human bomapin; however, trespin exhibits different tissue expression, cellular localization, and proteinase specificity compared with bomapin. Trespin mRNA is expressed in many tissues, including brain, heart, kidney, liver, lung, prostate, skin, spleen, and stomach. FLAG-trespin expressed in HEK293 cells is localized predominantly in the cytoplasm and is not constitutively secreted. The presence of an arginine at the P1 position of trespin's reactive site loop suggests that trespin inhibits trypsin-like proteinases. Accordingly, in vitro transcribed and translated trespin forms detergent-stable and thermostable complexes with plasmin and elastase but not subtilisin A, trypsin, chymotrypsin, thrombin, or papain. Trespin interacts with plasmin at a near 1:1 stoichiometry, and immunopurified mammal-expressed trespin inhibits plasmin in a dose-dependent manner. These data suggest that trespin is a novel and functional member of the rat ov-serpin family.

The serpins are an expanding superfamily of proteins present in the genomes of viruses, plants, and metazoans (reviewed in Refs. 1 and 2). Serpins are identified by their unique conformation (3,4), namely a conserved secondary structure containing three ␤-sheets (designated A, B, and C), ␣-helices (usually nine), and a reactive site loop (RSL), 1 which confers specificity to proteinase recognition. The RSL is composed of ϳ17 amino acids, which form a flexible region capable of distorting a target proteinase upon entry into the enzyme's active site. Inhibition occurs after the target proteinase cleaves the serpin RSL (5)(6)(7)(8), generating a covalent acyl-enzyme intermediate, which parallels a suicide substrate mechanism (9 -11).
The serpin superfamily is divided into 16 different clades based on phylogenic relationships (1). Clade B members, the ov-serpins, were originally identified by their significant sequence homology to chicken ovalbumin (12). They are competitive inhibitors of serine or cysteine proteinases and can target more than one proteinase through the use of several P1 residues (13,14). Ov-serpins also share several properties: 1) the absence of an N-terminal signal sequence, 2) the beginning of their amino acid sequences relative to ␣ 1 -antitrypsin at amino acid position ϳ23, and 3) the lack of a carboxyl-terminal extension. Human ov-serpins are further characterized by gene structure, namely an exon encoding a polypeptide loop between ␣ helices C and D (i.e. CD loop). The CD loop confers unique characteristics to serpins, such as nuclear localization, and provides a binding motif for ancillary proteins (15)(16)(17)(18).
The physiological functions of ov-serpins are not well understood; however, evidence suggests that their roles are diverse. MENT is a nuclear ov-serpin containing a lamin-like chromatin binding domain that induces higher order chromatin assembly when overexpressed (19). Other ov-serpins such as plasminogen inhibitor-9 (PI-9) can inhibit cells from undergoing granzyme B-mediated apoptosis and may regulate host defense against microbial or viral proteinases (20,21). Bomapin (proteinase inhibitor-10) as well as plasminogen activator inhibitor-2 (PAI-2) confers apoptotic resistance to tumor necrosis factor ␣ in several cell lines (22)(23)(24). Maspin (protease inhibitor-5) has been identified as an inhibitor of cell motility and metastasis along with demonstrating anti-angiogenesis properties (25)(26)(27)(28). The variety of physiological roles serpins regu-late foreshadows their potential as novel therapeutic targets for many disease states.
To identify genes whose products may serve as downstream mediators or regulators of transforming growth factor-␤1 signal transduction (reviewed in Refs. 29 and 30), we performed differential display reverse transcription-polymerase chain reaction using NRP-152 rat prostatic cells treated with or without TGF-␤1, either alone or in the presence of insulin, which blocks several of TGF-␤1's effects. 2 NRP-152 is a unique androgen receptor-positive basal prostatic epithelial cell line highly sensitive to TGF-␤1 (31)(32)(33)(34). From these differential display analyses, we have identified and characterized a widely expressed novel rat ov-serpin, trespin, which directly binds and inhibits plasmin with a stoichiometry of 1:1.
Differential Display-Following the method of Zhao et al. (35), differential display RT-PCR was performed using total RNA prepared from either untreated NRP-152 cells (control) or cells treated with TGF-␤1 (10 ng/ml) for 24 h in the absence or presence of insulin (5 g/ml). The RNA for differential display was isolated by a modified RNeasy method (36). The [ 33 P]dATP-labeled PCR products were then resolved through a denaturing polyacrylamide gel, and the selected bands were reamplified and cloned into pCR-TRAP (GenHunter).
Northern Blot Analysis-Total RNA was extracted from cells using an RNeasy Total RNA kit (Qiagen) and resolved through a 1% agarose, 0.66 M formaldehyde gel. Equal loading of the gel was confirmed through ethidium bromide staining of 28 S ribosomal RNA. The RNA was then transferred to a Nytran membrane (Schleicher & Schuell). Nytran membranes were cross-linked using ultraviolet radiation and prehybridized, hybridized, and washed following the Church and Gilbert method (37). The presence of indicated mRNAs was detected with [ 32 P]dCTP random primer-labeled cDNA probes.
Nytran membranes used contained cDNA probes for trespin and ␤-actin. The 1.17-kb trespin cDNA probe was designed to hybridize to bases 79 -1251 of trespin mRNA, whereas the 0.35-kb ␤-actin probe was complementary to exons 2 and 3 of the rat ␤-actin gene. A 1.4-kb fragment of the pcDNA3 expression vector, prepared by digestion with AvaII, was also loaded for nonspecific DNA binding.
RT-PCR-Reverse transcription was performed using murine leukemia virus reverse transcriptase and random hexamer primers (Gene-Amp; PerkinElmer Life Sciences) with 0.2 g of total RNA from each tissue. The cDNA was then PCR-amplified (40 cycles of 94°C for 30 s, 55°C for 30 s, 72°C for 60 s) using AmpliTaq DNA polymerase (Gene-Amp). The sense (5Ј-ACCCTCACTGCAAAAATCCTC-3Ј) and antisense (5Ј-CTGACTGGAGGTCATAACTCTCC-3Ј) primers used in these reactions were designed to synthesize a 629-bp product. The resulting products were resolved through a 1.2% agarose gel containing 0.72 g/ml ethidium bromide.
Development of pcDNA3-FLAG-trespin, pCEP4-FLAG-trespin, and pcDNA3-FLAG-bomapin Vectors-The hemagglutinin tag and translational start site was excised from hemagglutinin-pcDNA3 by digestion with HindIII and BamHI and replaced with 5Ј-HindIII-Kozak-FLAG-BamHI-3Ј cDNA made by annealing complementary oligonucleotides (Integrated DNA Technologies, Inc.). Trespin (codons 2 to stop) were amplified with primers containing BamHI and EcoRI ends and ligated to BamHI-and EcoRI-digested pcDNA3-FLAG to generate pcDNA3-FLAG-trespin. pCEP4-FLAG-trespin was generated by excision of FLAG-trespin from pcDNA3-FLAG-trespin using complete NotI digestion, followed by limited digestion with HindIII and ligation to HindIIIand NotI-digested pCEP4. The bomapin coding sequence (codons 2 to stop) was PCR-amplified from THP-1 total RNA with primers containing BamHI and ClaI ends and ligated into BamHI and ClaI-digested pcDNA-FLAG to generate pcDNA3-FLAG-bomapin. Constructs were sequenced for confirmation.
Subcellular Fractionation and Conditioned Medium-One 150-mm dish of stable transfected pCEP4-FLAG-trespin HEK293 cells were trypsinized, resuspended, and quantified. 25% of the cell suspension was used for whole lysate, and 75% was used for nuclear extract. For preparation of nuclear extract, cells were centrifuged at 4000 rpm for 10 min and resuspended in 2 ml of lysis buffer (10 mM HEPES, pH 7.5, 2 mM EDTA, 5 mM MgCl 2 , 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride, 2 g/ml each leupeptin, aprotinin, antipain, and 4-(2-aminoethyl)benzenesulfonyl fluoride). Cells were incubated on ice for 20 min, homogenized with 10 passes through a 25-gauge needle, and centrifuged at 4000 rpm for 10 min at 4°C. The pellet was washed twice with 1 ml of lysis buffer and resuspended with 300 l of lysis buffer. Samples were incubated for 1 h at Ϫ80°C, thawed, and centrifuged at 14,500 rpm for 30 min at 4°C; supernatant is the nuclear extract. For whole cell lysate, cells were harvested, resuspended in lysis buffer, and immediately incubated at Ϫ80°C for 1 h. The thawed lysate was then centrifuged at 14,500 rpm for 30 min at 4°C; supernatant is the whole cell lysate. The same volume of nuclear extract and whole cell lysate (equal number of nuclei and whole cells per ml) was subjected to Western blot analysis, and FLAG-trespin was detected as mentioned above. Proliferating cell nuclear antigen expression (clone NA03, 1:250; Oncogene Research) was used as loading control.
The secretion of trespin from HEK293-FLAG-trespin cells was compared with the content of total cellular FLAG-trespin as follows. Conditioned medium (treated with 1% Triton X-100 and protease inhibitor mixture) and whole cell lysate (treated similarly) from 10 7 cells were each mixed overnight with 50 l of anti-FLAG-agarose. Resin was then washed extensively with PBS plus 1% Triton-X-100 and eluted with a total of 0.2 ml of 0.1 mg/ml FLAG peptide, as described by the manufacturer. Equal volumes of each eluted fraction were analyzed by Western blot.
Thermostability of Trespin-0.5 g of purified FLAG-trespin was suspended in 20 l of PBS and heated to the indicated temperatures for 5 min (39). Samples were then centrifuged at 14,000 rpm for 30 min at 4°C. The supernatant was combined with SDS loading buffer and resolved in a 4 -12% NuPAGE gel with 1ϫ MOPS buffer, transferred to nitrocellulose, and subjected to Western blot analysis with anti-FLAG (clone M1; Sigma) as described (40).
Binding Stoichiometry-Active plasmin concentration was determined by titrating the plasmin substrate D-Val-Leu-Lys-pNA (VLK-pNA; Sigma) to measure the turnover rate, using a spectrophotometric method similar to Chase and Shaw (41). Plasmin (100 nM) was incubated with the indicated concentrations of FLAG-trespin (0 -88. Plasmin activity proceeded at 37°C, and the rate of product formation was recorded (A 405 ) as described above.

RESULTS
Cloning of Trespin-Differential display RT-PCR was used to identify and isolate genes regulated by TGF-␤1 in NRP-152 cells. This was done with RNA isolated from NRP-152 cells treated with or without 10 ng/ml TGF-␤1 for 24 h either alone or in the presence of 5 g/ml insulin, which blocks several effects of TGF-␤1. 2 Following differential display screening with 20 primer sets, we identified the 3Ј-end of a novel gene, which we initially named TGF-␤-regulated gene 13 (TRG-13). Expression of TRG-13 is decreased by TGF-␤1 in a manner that is blocked in the presence of insulin (Fig. 1A, doublet is observed due to denaturing electrophoresis of double-stranded DNA). We confirmed these results by Northern blot analysis of total RNA from NRP-152 cells, using a [ 32 P]dCTP-labeled cDNA probe prepared by reamplification of the original 300-bp fragment eluted from the differential display gel. TRG-13 migrates as a single 1.4-kb transcript and is down-regulated as early as 5 h after TGF-␤1 exposure (Fig. 1B).
To obtain full-length TRG-13 cDNA, we performed 5Ј-RACE RT-PCR using oligonucleotide primers complementary to the 300-bp differential display product. From these reactions, we amplified a PCR product that contains the entire coding region of TRG-13. This product, totaling 1345 nucleotides, has 1.2 kb of coding region with 78 and 73 bp of 5Ј-and 3Ј-flanking untranslated regions, respectively. Analysis of this sequence with a nucleotide BLAST search revealed that TRG-13 shares high sequence similarity with serpins, with greatest similarity to the human bone marrow serpin bomapin (ϳ80%) (42). cDNA alignment of TRG-13 to bomapin allowed us to identify the open reading frame and the ATG codon representing the translation start site (43). Although TRG-13 shares significant nucleotide and amino acid sequence similarity with other serpins, it is down-regulated by TGF-␤1, which contrasts with the dogmatic view of TGF-␤ influence on proteolysis (44,45).
TGF-␤1 has been shown to regulate mRNA expression through both transcriptional and post-transcriptional mechanisms. To determine whether the decrease in trespin mRNA produced by TGF-␤1 occurred through loss of gene transcription, we measured trespin mRNA production in NRP-152 cells by nuclear run-on assay. The amount of trespin mRNA produced in NRP-152 cells treated with 10 ng/ml TGF-␤1 for 24 h was ϳ20% that of untreated control cells (Fig. 1C), demonstrating that TGF-␤1 lowers trespin mRNA levels by decreasing transcription of the trespin gene. For this reason, we have renamed this serpin trespin (for TGF-␤-repressible serine proteinase inhibitor. Trespin Sequence Analysis-Based on its deduced sequence, trespin consists of 397 amino acids ( Fig. 2A) with a predicted molecular mass of 45.2 kDa. Whereas trespin protein has 72% sequence identity with human bomapin, it also shows about 40% identity with other serpins, including human PAI-2 (46), chicken MENT (19), and human PI-9 (47, 48) (Fig. 2B). Amino acids are shaded according to similarities (red and yellow are most homologous to trespin) by ESPript 2.0. Trespin's predicted fold (Fig. 3) is based on the alignment of structures with Protein Data Bank identification numbers 1BY7 (human PAI-2), 1JRR (human PAI-2 complexed with RSL peptide), 1HLE (equine leukocyte elastase inhibitor), and 1OVA (chicken ovalbumin). The two most divergent regions are 1) reactive loop 350 -364, which is built based on secondary structure prediction (Swiss model) and steric considerations, and 2) loop 64 -75, which cannot be built without experimental data. However, loop 64 -75 is modeled after a similar 1OVA loop where possible. Even so, the in silico structure determination suggests that trespin exhibits a classical serpin fold with nine ␣ helices, three ␤ sheets, and an exposed RSL.
Expression of Trespin in Rat Tissues and Cellular Localization-Consistent with its expression in the NRP-152 prostatic epithelial cell line, we found trespin to be expressed in rat ventral prostate, dorso-lateral prostate, and seminal vesicle as shown by RT-PCR of total RNA (Fig. 4A). RT-PCR analysis also showed expression of trespin mRNA in several other rat tissues, with highest levels in the lung (Fig. 4B). PCR without reverse transcriptase, which was done in parallel for all samples, failed to amplify any product (data not shown).
The above RT-PCR results indicated lung, skin, and stomach tissues have the greatest trespin expression. Since lung has the highest trespin mRNA expression, we probed rat (Fig. 4C, left  panel) and human (Fig. 4C, right panel) lung poly(A) ϩ RNA (2 g/lane) Northern blots to compare trespin and bomapin expression in this tissue. Similar to Fig. 4B, trespin is expressed in rat lung, whereas the human lung blot was negative for bomapin expression, consistent with previously published results (42). As control for the bomapin probe and hybridization conditions, we performed Northern blot analysis of two leukemic cells lines of rat basophilic and human monocytic origin, RBL-1 and THP-1, respectively (Fig. 4D). THP-1 demonstrated high bomapin expression, as suggested by Riewald et al. (49), and trespin expression was present in RBL-1 (Fig. 4C). Evidence suggests that bomapin is not expressed in nonmonocytic leukocytes (such as basophils), and trespin's expression in RBL-1 provides further support that it is not the rat homologue of bomapin.
Ov-serpins have unique cellular localization, and many are secreted into the extracellular matrix. The nuclear versus cytoplasmic distribution of trespin was determined by fractionation of HEK293 cells stably expressing FLAG-trespin, followed by Western blot analysis using an anti-FLAG M1 antibody. Expression of FLAG-trespin in duplicate cultures was assayed (Fig. 5A, upper panel) in a nuclear fraction compared with whole cell extracts. The same number of nuclei and whole cells were analyzed to allow for direct comparison; the Western blot detection of proliferating cell nuclear antigen is shown as loading control (Fig. 5A, lower panel). These analyses indicated that FLAG-trespin is primarily localized to the cytoplasm, with only a minor species in the nucleus, which greatly contrasts with bomapin, a predominantly nuclear protein.
To measure if trespin was secreted, the total conditioned medium (Fig. 5B) and whole cell lysate from a confluent 150-mm dish of HEK293-FLAG-trespin cells were each purified with immobilized anti-FLAG M2, and an equal volume of each eluted fraction was analyzed by Western blot. No FLAGtrespin was detectable in the conditioned medium from the HEK293-FLAG-trespin clones, suggesting that trespin is not constitutively secreted.
Survey for Target Proteinases-Serpins normally form complexes with target proteinases that are resistant to reducing agents, SDS and heat (50,51). The presence of an arginine at the reactive center ( Fig. 2A) suggests that trespin inhibits trypsin-like proteinases. To discover targets that trespin binds and potentially regulates, we screened a panel of serine proteinases (subtilisin A, trypsin, chymotrypsin, thrombin, papain, elastase, and plasmin) for the ability to form SDS-stable, DTT-stable, and thermostable complexes with trespin. Trespin (45.2 kDa) was in vitro transcribed/translated (trespin IVTT ) and combined with the indicated proteinases in Tris-buffered saline. The samples were heated to 100°C in the presence of SDS and DTT (2% and 100 mM final concentrations, respectively) before SDS-PAGE analysis, transfer to nitrocellulose, and exposure to a phosphorscreen. Elastase (32 kDa) and plasmin (34 kDa) formed SDS-stable, DTT-stable, and thermostable complexes with trespin IVTT with the expected bands at 76 and 77 kDa, respectively (Fig. 6A).
Although the tissue expression patterns and cellular localization of trespin and bomapin are different, we wanted to compare the proteinase specificity of these serpins to further substantiate that these proteins are not homologues. As mentioned above, trespin forms complexes with elastase and plasmin; data from Riewald and Schleef (42) show that bomapin complexes with thrombin and weakly with trypsin. To confirm the reported bomapin results and directly compare the protein- FIG. 1. Cloning of TRG-13, a novel rat gene transcriptionally down-regulated by TGF-␤1 in a manner inhibited by insulin. A, differential display RT-PCR detection of TRG-13. Differential display RT-PCR was performed using total RNA prepared from either untreated NRP-152 cells (control) or cells treated with TGF-␤1 (10 ng/ml) in the absence or presence of insulin (5 g/ml). The 300-bp TRG-13 cDNA fragment shown above was amplified in reactions using the E 2 AP 13 sense and ET 12 MA anchor antisense primers. Each treatment was done in duplicate. B, Northern blot analysis confirms that TGF-␤1 decreases TRG-13 gene expression. TGF-␤1 (10 ng/ml) was added at different time periods to NRP-152 cells grown under low serum conditions. For cultures treated with insulin, insulin (5 g/ml) was added to the cells 24 h before the addition of TGF-␤1. Each blot is representative of three independent experiments. C, nuclear run-on analysis of nuclei isolated from untreated NRP-152 cells (control) or NRP-152 cells treated with 10 ng/ml TGF-␤1 for 24 h. [ 32 P]dUTP-labeled trespin and ␤-actin mRNAs were extracted from nuclei and allowed to hybridize to blots containing their respective cDNA and pcDNA3 DNA control. Message levels were quantified by a phosphorimager and expressed as the ratio of trespin mRNA/␤-actin mRNA with respect to control. The trespin/␤-actin ratio for the control group has been set at 1. ase specificity of trespin with bomapin, we performed a parallel proteinase binding survey. Bomapin formed SDS-stable, DTTstable, and thermostable complexes with thrombin and trypsin, as expected, with no detectable binding to elastase and plasmin, unlike trespin (Fig. 6B). These results support the notion that bomapin and trespin are functionally different proteins.
We next characterized the dose dependence of the trespin interactions. Plasmin demonstrated a classical dose-response profile with complex formation (Fig. 7A) detectable at 1 pmol of plasmin, increasing linearly (data not shown) up to 100 pmol of plasmin. At the highest mass of plasmin (100 pmol), ϳ100% of trespin IVTT was complexed at near 1:1 stoichiometry. Also, it appears that trespin IVTT may serve as a plasmin cleavage substrate in the presence of high plasmin activity, which may occur before or after complex formation. In contrast, elastase did not exhibit a classical dose-response inhibition profile (Fig 7B), with the majority of complex formation detectable at 20 pmol of elastase and little to zero detectable complex formation at Ͻ20 pmol elastase. Even so, the majority of trespin IVTT was cleaved at 100 pmol of elastase (with minimal complex observed), suggesting that it preferentially serves as an elastase substrate.
Trespin Inhibits Plasmin Activity in a Dose-dependent Manner-Due to the classical dose-response profile between plasmin and trespin, we wanted to observe if purified trespin could inhibit plasmin activity in vitro. We developed a stable mammalian expression system (pCEP4-FLAG-trespin) in HEK293 cells and purified FLAG-trespin from these cells by large scale immunoprecipitation with anti-FLAG-agarose and gentle elution with 100 g/ml competitor FLAG peptide. The eluted FLAG-trespin was highly pure (ϳ95%) with minimal nonspecific bands detectable by SDS-PAGE and Coomassie Blue staining (Fig. 8A). Also, no other proteins in the eluted material were recognized by the FLAG antibody (Fig. 8B), as demonstrated by the control immunoprecipitation lacking FLAG immunoreactive bands. To ensure that the purified FLAG-trespin was in its active conformation, we obtained a thermal denaturation profile. Native serpins, due to their metastable conformation, undergo a temperature-induced transition (i.e. precipitate out of solution) with a melting temperature of T m ϭ ϳ60 -70°C and no other detectable transition up to 125°C (1,52). Cleaved or latent forms exhibit a sharp, highly cooperative unfolding transition at a much higher temperature of T m ϭ 124°C. Fig.  8C shows that FLAG-trespin rapidly precipitates at ϳ60 -70°C, suggesting that it is properly folded and in the appropriate conformation for further analyses.
Classical serpins exhibit a stoichiometry of inhibition (SI) of 1:1 with their target proteinase. The SI ((k s ϩ k i )/k i )) determines whether the serpin-proteinase complex partitions down the pathway leading to the formation of a covalent inhibitory complex (k i ) or if parallel substrate pathways (k s ) predominate (53). A SI of 1 suggests that the formation of inhibitory complexes predominates, whereas a SI of Ͼ1 indicates the substrate pathway exceeds the k i . To determine the SI for trespin and plasmin, the indicated concentrations of FLAG-trespin (0 -168.8 nM) were combined with 100 nM plasmin in PBS and incubated for 20 min at 37°C. Following the addition of plasmin substrate, D-Val-Leu-Lys-pNA (final concentration 1 mM), . Linear regression analysis was performed to extrapolate the inhibitor and enzyme ratio resulting in 100% inhibition (x intercept), and we obtained an SI of 0.85 (Fig. 9A), similar to the expected SI of 1:1. The inset (same x and y axes) contains data points from above and includes [I] 0 /[E] 0 ratios of Ͼ1. Data points with [I] 0 /[E] 0 ratios of Ͼ1 are on the x axis (due to 100% inhibition) and bias the linear regression analysis; therefore, they were omitted in determining the stoichiometry of inhibition.
To further substantiate that trespin inhibits plasmin through irreversible inactivation, we performed a progress curve method analysis (54,55). 100 nM plasmin was simultaneously combined with VLK-pNA (1 mM final substrate) and the indicated concentrations of FLAG-trespin (0 -84.4 nM) in PBS. Plasmin activity (A 405 ) was recorded using a microplate reader and plotted against time (s). Fig. 9B demonstrates that FLAG-trespin inhibits plasmin activity in a dose-dependent manner. Furthermore, the initial inactivation (Ͻ300 s) proceeds at a linear rate; however, the inactivation pathway (i.e. plasmin inhibition by trespin) predominates as the rate of product formation is followed. These results demonstrate that trespin inhibits plasmin by directly binding to plasmin's active site, resulting in functional inactivation, and not through an allosteric mechanism. DISCUSSION In this report, we have identified a novel member of the rat ov-serpin family that is down-regulated by TGF-␤1 using differential display RT-PCR. Trespin is expressed in a variety of tissue types and forms SDS-stable, DTT-stable, and thermostable complexes with elastase and plasmin. FLAG-trespin ex-hibits a classical dose-response inhibition profile with plasmin (Fig. 9B), and complex formation between FLAG-trespin and plasmin is detectable at 1 pmol (Fig. 7A).
Three-dimensional structure prediction by a Swiss Model Protein Data Bank homology screen (Fig. 3) and steric considerations demonstrated that trespin has a classical serpin fold with nine ␣-helices, three ␤-sheets, and canonical carboxylterminal RSL. Trespin lacks an amino-terminal signal sequence, suggesting that it is not part of the secretory pathway, and Fig. 5B further suggests that trespin is not normally secreted. However, analysis of the amino acid sequence revealed the presence of potential N-linked glycosylation sites (NX(T/S)) at Asn 177 , Asn 201 , Asn 209 , Asn 321 , Asn 324 , and Asn 383 that may allow trespin, like PAI-2 (29,30), to be secreted under certain conditions. The stimuli that promote potential cellular relocalization and/or secretion of trespin are under investigation.
Protein motif analysis indicated that trespin shares an identical nuclear targeting sequence (KKRK) with bomapin (16) at amino acids 74 -77. However, analysis of the trespin sequence by PSORT (56) predicted trespin to be primarily cytoplasmic with minor percentages targeted to nuclear and mitochondrial compartments; in contrast, PSORT prediction and data suggest that bomapin is localized to the nucleus (16). Cellular fractionation experiments (Fig. 5A) in our laboratory demonstrate that FLAG-trespin is predominantly cytoplasmic in HEK293 cells, consistent with PSORT predictions and preliminary data with endogenous trespin in NRP-152 cells (data not shown).  A and B) or FLAGtagged bomapin (B) was in vitro transcribed and translated from pcDNA3-FLAG-trespin using a coupled TnT kit with T7 RNA polymerase. 20 pmol of indicated proteinase, IVTT product, and Tris-buffered saline (volume to 15 l) were incubated at 37°C for 15 min. In vitro translation from pcDNA3 empty vector was used as a control (control IVTT ). Samples were heated to 100°C for 5 min in the presence of 2% SDS and 100 mM DTT. Proteins were resolved in a 4 -12% NuPAGE gel with 1ϫ MOPS buffer and transferred to nitrocellulose. Nitrocellulose was exposed to a phosphorscreen for 72 h. complex 1 and complex 2 , trespin and bomapin complexes, respectively. Analyses are representative of three independent experiments.
It is important to note that although functional domains between serpins are highly conserved, this similarity does not indicate similar function or homologous genes and is not a useful method to predict homologous serpins. Several human serpins within the ovalbumin and heat shock 47 clades exhibit high degrees of RSL homology, such as the squamous cell carcinoma antigen serpins 1 and 2 (65% homologous RSLs) and SERPINH1 and SERPINH2 (100% homologous RSLs); furthermore, trespin and bomapin have 67% RSL homology, which does not suggest that these proteins are homologues. Accordingly, interspecies homologues of the same serpin, such as PAI-2, usually demonstrate 100% RSL conservation, whereas the structural backbone may vary by 25%. Serpins evolved through gene duplication, which results in highly homologous proteins with specificity not necessarily conferred through sequence but rather by tissue-specific expression and regulation. A recent review by Silverman et al. (1) notes that a linear relationship between human and murine (and assumed other species) is not likely and that once all human serpins have been identified, proper nomenclature of nonhuman serpins (i.e. following the established rules of serpin nomenclature by the Second International Symposium on the Structure and Biology of Serpins and the HUGO Gene Nomenclature Committee), including rat trespin, will ensue.
Each member of the ov-serpin family has a unique and sometimes limited tissue expression. Trespin is likely to function in multiple tissues as indicated by its expression pattern, in contrast to human bomapin, which has been detected only in the bone marrow by multiple tissue Northern blot analyses and RT-PCR screens (49). Results suggest that primary cells and established cell lines from monocytic lineage have significant bomapin expression (49), with elevations in patients with acute monocytic leukemia and no detectable expression in the peripheral blood. On the contrary, we detected trespin expression in RBL-1, a basophilic cell line, which is representative of one cell type in peripheral blood, and in other tissues (prostate, lung, etc.). This result, in conjunction with proteinase specificity (Fig. 6B) and other mentioned differences, strongly suggests that trespin is not the rat homolog of bomapin but instead is a novel rat serpin.
Alignment of the trespin amino acid sequence with bomapin and other serpins (PAI-2, PI-9, and MENT) indicated a unique reactive center of trespin between amino acids 346 (P17) and 367 (P5Ј). Within the reactive site loop, the amino acid sequence of the hinge region (P17-P8) is highly conserved among serpins shown to inhibit serine proteinase activity (57)(58)(59). Trespin is identical to the consensus sequence for the hinge region at every residue except for the P9 position, suggesting that it functions as an inhibitory serpin. Of the amino acids present in the reactive site, the P1 residue is most critical in determining the proteinase specificity of an inhibitory serpin (2,60). Trespin, like bomapin and PAI-2, contains arginine at its P1 position, suggesting that it interacts with trypsin-like serine proteinases. We analyzed a variety of trypsin-like serine proteinases and determined that elastase and plasmin formed stable complexes with trespin (Fig. 6A). Although elastase formed a stable complex with trespin, the profile obtained (Fig.  7B) did not indicate a dose-dependent inhibitory response. This may be the result of trespin binding to a region of elastase other FIG. 7. Dose-dependent effects of plasmin and elastase on trespin complex formation. Full-length FLAG-tagged trespin was in vitro transcribed and translated using a coupled TnT kit with T7 RNA polymerase. The indicated amounts (pmol) of plasmin or elastase, FLAG-trespin IVTT , and Tris-buffered saline (volume to 20 l) were incubated at 37°C for 15 min. Samples were heated to 100°C for 5 min in the presence of 2% SDS and 100 mM DTT. Proteins were resolved in a 4-12% NuPAGE gel with 1ϫ MOPS buffer and transferred to nitrocellulose. Nitrocellulose was exposed to a phosphor screen for 72 h. These analyses are representative of three independent experiments.

FIG. 8. Purity and thermal denaturation analyses of purified FLAG-trespin.
A, immunopurified FLAG-trespin (1 g) was subjected to SDS-PAGE and Coomassie staining. Lanes 1 and 2 are molecular weight standards and purified FLAG-trespin, respectively. B, eluted material (400 ng of total protein) from HEK293 cells expressing pCEP4-FLAG-trespin or pCEP4 control (pCEP4 control immunoprecipitation (IP)) was subjected to Western blot analysis and detected with an anti-FLAG M1 antibody. Total lysate from HEK293-FLAG-trespin cells is shown as a positive control (lane 3). C, thermal denaturation of 0.5 g of FLAG-trespin in PBS was performed at the indicated temperatures. Samples were then centrifuged at 4°C, and supernatants containing soluble FLAG-trespin were detected by Western blot analysis using anti-FLAG M1 antibody. These analyses are representative of two independent experiments. than the active site or proteolytic degradation of trespin by elastase, which generates trespin fragments with altered conformation(s) or affinities for elastase. The dose-dependent degradation of FLAG-trespin by elastase is obvious in Fig. 7B, which suggests that trespin may be a potential elastase substrate.
Most serpins characterized inhibit proteinases through a suicide substrate-like mechanism with the exception of maspin (61). From a screen to find target proteinases that trespin binds, only plasmin exhibited dose-dependent complex formation (Fig. 7A). Serpins usually exhibit a 1:1 stoichiometry with their target proteinase, and deviations from this ratio are often observed by altering the length of the RSL (62). Enzyme catalysis studies demonstrate trespin neutralizes plasmin with a stoichiometry of inhibition of 0.85 (Fig. 9A). The 15% deviation from the ideal 1:1 ratio may be due to the inherent error of the assays used to determine the concentration of immunopurified FLAG-trespin or active plasmin. Further support for the possibility that FLAG-trespin inhibits plasmin was obtained by a progress curve method analysis. This analysis demonstrates that FLAG-trespin inhibits plasmin in a dose-dependent manner (Fig. 9B) and that the FLAG-trespin inactivation pathway predominates as catalysis proceeds.
We have not yet determined whether trespin is expressed in peripheral blood (except for the basophilic cell line, RBL-1; Fig.  4D) or in other tissues where plasmin activity is evident. However, our data indicating that trespin inhibits plasmin provide fresh insight to the active site(s) of proteinases which trespin may regulate. Due to the significant degree of amino acid homology to bomapin, we were interested in determining whether trespin exhibited similar functional characteristics to bomapin, in light of the differences in tissue expression and cellular localization. Bomapin has been shown to form stable complexes with thrombin and trypsin (42); however, we cannot detect the formation of any trespin complex with these proteinases (Fig. 6, A and B). Also, the only physiological role of bomapin described is resistance to tumor necrosis factor ␣-induced apoptosis (22) in HeLa cells, a cell line that does not express endogenous bomapin. Even so, we investigated a similar role for trespin and found no protection from tumor necrosis factor ␣-induced apoptosis in HeLa cells by both transient transfection and stable clone assays (data not shown), further substantiating that bomapin and trespin have distinct physiological roles.
In multiple tumor types through genetic and epigenetic mechanisms, the TGF-␤ pathway is susceptible to loss or mutation of either TGF-␤ receptors or Smad proteins (63). The consequence of these alterations is the escape from the tumorsuppressive effects of TGF-␤, namely cell cycle arrest, apoptosis, and perhaps increased tumor burden induced by local immune suppression, enhanced metastatic potential, or angiogenesis (64,65). In the nontumorigenic basal prostatic epithelial cell line, NRP-152, our laboratory in collaboration with Dr. Lalage Wakefield's group has demonstrated a potential tumor suppressor role for the TGF-␤1 type II receptor (66). NRP-152 cells respond to TGF-␤ with classical effects: extracellular matrix secretion, cell cycle arrest, and apoptosis (32). Other unique and well characterized properties of these cells make them an ideal model for prostatic carcinogenesis and to explore novel pathways regulated by TGF-␤.
The literature describes few cases of serpin regulation by TGF-␤1, such as PAI-1, a direct transcriptional target of TGF-␤1 (44,45,67), and PAI-2 (68,69). PAI-1 deposition into the extracellular matrix is increased by TGF-␤1, along with a down-regulation of the PAI-1 targets: urokinase-type and tissue-type plasminogen activators (45). The protein level of PAI-2 can be regulated by TGF-␤1, but data suggest that this effect may be cell type-dependent and occurs through nontranscriptional mechanisms (68,69). The overall effect of the above TGF-␤1 responses is a dramatic decrease in extracellular proteolytic activity. On the contrary, trespin is transcriptionally down-regulated by TGF-␤1 (Fig. 1C) and is found to be only intracellular (Fig. 4C). Trespin may protect against cellular injury induced by a variety of stimuli such as leakage of potent proapoptotic enzymes like granzyme B or cathepsin G in the cytoplasm or mispackaged lysosomal or secretory granule enzymes. Trespin's ability to complex with elastase and plasmin, along with its expression in RBL-1, suggests a potential role in fibrinolysis or similar cascades. Of interest to our laboratory is the impact trespin will have on the biology of TGF-␤ signal transduction and responses.

FIG. 9. Trespin inhibits plasmin in a dose-dependent manner.
A, the stoichiometry of inhibition for FLAG-trespin with plasmin was determined by preincubating 100 nM plasmin with the indicated concentrations of FLAG-trespin (0 -88.4 nM) for 37 min at 37°C. VLK-pNA (1 mM final concentration) was added, and the reaction was incubated for 20 min at 37°C. Residual plasmin activity was recorded using a microplate reader (A 405 ). The fractional activity was the product formation of the inhibited reaction (V i ) divided by the control reaction (V o ). Linear regression analysis was performed to extrapolate the inhibitor and enzyme ratio, resulting in 100% inhibition. The inset (same x and y axes) contains data points from above and includes [I] 0 /[E] 0 ratios of Ͼ1. Data points with [I] 0 /[E] 0 ratios of Ͼ1 are plotted on the x axis (due to 100% inhibition) and bias the linear regression analysis; therefore, they are omitted in determining the stoichiometry of inhibition. B, a progress curve method analysis was used to demonstrate that FLAG-trespin inhibits plasmin in a dose-dependent manner. 100 nM plasmin was simultaneously combined with VLK-pNA (1 mM final substrate) and