SERPINB11 Is a New Noninhibitory Intracellular Serpin

SERPINB11, the last of 13 human clade B serpins to be described, gave rise to seven different isoforms. One cDNA contained a premature termination codon, two contained splice variants, and four contained full-length open reading frames punctuated by eight single nucleotide polymorphisms (SNPs). The SNPs encoded amino acid variants located within the serpin scaffold but not the reactive site loop (RSL). Although the mouse orthologue, Serpinb11, could inhibit trypsin-like peptidases, SERPINB11 showed no inhibitory activity. To determine whether the human RSL targeted a different class of peptidases or the serpin scaffold was unable to support inhibitory activity, we synthesized chimeric human and mouse proteins, in which the RSLs had been swapped. The human RSL served as a trypsin inhibitor when supported by mouse scaffold sequences. Conversely, the mouse RSL on the human scaffold showed no inhibitory activity. These findings suggested that variant residues in the SERPINB11 scaffold impaired serpin function. SDS-PAGE analysis supported this notion as RSL-cleaved SERPINB11 was unable to undergo the stressed-to-relaxed transition typical of inhibitory type serpins. Mutagenesis studies supported this hypothesis, since the reversion of amino acid sequences in helices D and I to those conserved in other clade B serpins partially restored the ability of SERPINB11 to form covalent complexes with trypsin. Taken together, these findings suggested that SERPINB11 SNPs encoded amino acids in the scaffold that impaired RSL mobility, and HapMap data showed that the majority of genomes in different human populations harbored these noninhibitory SERPINB11 alleles. Like several other serpin superfamily members, SERPINB11 has lost inhibitory activity and may have evolved a noninhibitory function.

SERPINB11, the last of 13 human clade B serpins to be described, gave rise to seven different isoforms. One cDNA contained a premature termination codon, two contained splice variants, and four contained full-length open reading frames punctuated by eight single nucleotide polymorphisms (SNPs). The SNPs encoded amino acid variants located within the serpin scaffold but not the reactive site loop (RSL). Although the mouse orthologue, Serpinb11, could inhibit trypsin-like peptidases, SERPINB11 showed no inhibitory activity. To determine whether the human RSL targeted a different class of peptidases or the serpin scaffold was unable to support inhibitory activity, we synthesized chimeric human and mouse proteins, in which the RSLs had been swapped. The human RSL served as a trypsin inhibitor when supported by mouse scaffold sequences. Conversely, the mouse RSL on the human scaffold showed no inhibitory activity. These findings suggested that variant residues in the SERPINB11 scaffold impaired serpin function. SDS-PAGE analysis supported this notion as RSL-cleaved SERPINB11 was unable to undergo the stressed-to-relaxed transition typical of inhibitory type serpins. Mutagenesis studies supported this hypothesis, since the reversion of amino acid sequences in helices D and I to those conserved in other clade B serpins partially restored the ability of SERPINB11 to form covalent complexes with trypsin. Taken together, these findings suggested that SER-PINB11 SNPs encoded amino acids in the scaffold that impaired RSL mobility, and HapMap data showed that the majority of genomes in different human populations harbored these noninhibitory SERPINB11 alleles. Like several other serpin superfamily members, SERPINB11 has lost inhibitory activity and may have evolved a noninhibitory function.
Serpins are a superfamily of serine and cysteine peptidase inhibitors that contain ϳ1500 family members and are found in all domains of life (Eukarya, Eubacteria, and Archaea) as well as many Poxviridae (reviewed in Refs. [1][2][3][4]. Unlike canonical inhibitors, inhibitory serpins employ a unique suicide substrate-like inhibitory mechanism to neutralize their target peptidases. The surface-exposed reactive site loop (RSL) 4 serves as a pseudosubstrate and binds to the active site of the peptidase. Upon cleavage of the RSL, the metastable serpin molecule undergoes a major conformational rearrangement and traps the covalently attached peptidase in a distorted, inactive form (5). A small proportion of serpins are noninhibitory and aid in diverse functions, such as hormone transport (e.g. SERPINA7/ thyroxine binding globulin) and protein folding (e.g. SER-PINH1/HSP47) (6,7).
Serpins are classified into 17 clades based on phylogenetic relationships (8). So far, 36 human serpins from nine clades (A-I) have been identified (2). The majority of serpins are plasma proteins that serve as critical regulators of important physiological processes, such as blood coagulation, fibrinolysis, and inflammation. In contrast, clade B serpins exist predominantly, but not exclusively, as intracellular proteins with a cytoplasmic or nucleocytoplasmic distribution (9,10). Other distinct features of clade B serpins include the absence of both cleavable N-terminal signal peptides and terminal extensions (N-or C-) relative to the prototypical serpin, ␣1-antitrypsin (SERPINA1) (reviewed in Refs. 4 and 11). Although the functions of many clade B serpins are unknown, inhibitory members appear to protect cells from endogenous (e.g. lysosomal or gran-ule enzymes) or exogenous (e.g. microbial or inflammatory cell) peptidase-mediated injury (4,12). To date, a single intracellular serpin, SERPINB5, does not function as an inhibitor, is unable to undergo the stressed to relaxed (S-R) transition (13,14), and performs a crucial role in preventing the development of invasive breast and prostrate cancers (15,16).
Thirteen clade B serpin genes reside in the human genome (4). Three map to 6p25, and 10 map to 18q21.3 (4). The syntenic regions in the mouse genome are 13A3.2 and 1D1, respectively. The clade B cluster at 13A3.2 is greatly expanded (n ϭ 15) relative to that of the human locus at 6p25 (17). Except for the SCCA locus, which contains SERPINB3 and -B4 in humans and Serpinb3a, -b, -c, and -d in mice, the remaining eight clade B serpins mapping to human 18q21 and mouse 1D1 show 1:1 orthology and are highly conserved in terms of gene structure, gene order, and amino acid sequence (18). These correlations suggest that the functional activity of the clade B serpins has been retained throughout mammalian evolution and that knowledge gained about any one of these genes should help define the overall activity of each orthologous pair. Amino acid analysis of the human and mouse RSLs confirms that these orthologous pairs are likely to demonstrate identical inhibitory profiles (18). Of this set of eight orthologous clade B serpins, only human SERPINB11 and mouse Serpinb11 have yet to be characterized. The goal of this study was to use recombinant human and mouse proteins to define the biochemical activity of this pair of clade B serpins. Although Serpinb11 inhibited trypsin-like peptidases, single nucleotide polymorphisms (SNPs) in the human gene generated several different amino acid variants within the serpin scaffold that precluded proper inhibitory function in vitro. Based on current haplotype data, the majority of SERPINB11 alleles encoded proteins that lack detectable inhibitory activity.

EXPERIMENTAL PROCEDURES
Identification of SERPINB11 and Its Mouse Orthologue-SERPINB11 (per the nomenclature guidelines, human genes (italicized) and gene products (not italicized) are uppercase, and mouse gene and gene products are lowercase after the first letter) was identified by electronic hybridization using the BLAST2 algorithm. Serpinb11 was identified using the SER-PINB11 cDNA sequence and BLAST programs to query the mouse genomic and EST data bases as reported (18).
Site-directed Mutagenesis-SERPINB11 cDNA sequences were mutated using the QuikChange kit per the manufacturer's instructions (Stratagene). All mutations were confirmed by DNA sequencing, as described (21).
RSL Swap Mutations-The RSL of Serpinb11 was flanked by PstI sites located 5Ј in the proximal hinge region and 3Ј in the pGEX-6P-1 backbone. To generate a PstI site in the identical position within the proximal hinge region of SERPINB11d, an A1014T transversion was introduced by sitedirected mutagenesis. To facilitate the RSL swap, a second PstI site located in the proximal portion of SERPINB11d (site not present in the mouse sequence) was eliminated also by a sitedirected transversion (A153T). The PstI RSL fragments of SERPINB11d and Serpinb11 were isolated after restriction endonuclease digestion and agarose gel electrophoresis. The fragments were ligated into PstI-digested Serpinb11-pGEX-6P-1 and SERPNB11d-pGEX-6P-1, respectively. Correct insertions of the human and mouse RSLs into the mouse and human serpin scaffolds, respectively, were confirmed by DNA sequence analysis.
Determination of Protein Concentrations-Trypsin and plasmin were active site-titrated using 4-methylumbelliferyl-para-guanidinobenzoate (Sigma) as described (25). The concentrations of recombinant serpins were determined by using the Bio-Rad protein assay kit II (Bio-Rad).
Screening Assays for Enzyme Inhibition-The inhibitory activities of Serpinb11 and SERPINB11 were determined initially by mixing the inhibitor with various peptidases in appropriate buffer, incubating for 30 min at 25°C, and measuring residual enzyme activity as described (22). Residual enzyme activity was determined by adding the appropriate substrate and measuring hydrolysis over time (velocity) using either a THERMOmax (in the case of UV-visible substrates) or fmax (in the case of fluorescent substrates) microplate reader (Molecular Devices, Sunnyvale, CA). All screening assays were performed at an inhibitor/enzyme ratio of Ͼ5 as listed in Table 1.
Stoichiometry of Inhibition (SI)-Assays were performed in a final volume of 100 l in a 96-well microtiter plate (Costar, Cambridge, MA). Varying amounts of inhibitor were incubated with enzyme for 30 min at 25°C. Substrate, at a final concentration of 1 mM VLK-pNA for plasmin and 0.2 mM H-Glu-Gly-Arg-pNA for trypsin, was added to each well, and the velocity of substrate hydrolysis was determined by measuring the A 405 using the THERMOmax microplate reader.
Enzyme Kinetics-The apparent second order rate constants for the interaction of trypsin or plasmin with serpin were determined by the progress curve method under pseudo-first order conditions (26). A constant amount of enzyme (2.0 nM trypsin or 10 nM plasmin) and substrate (final concentration 2.0 mM AAPR-pNA or 0.5 mM VLK-pNA, respectively) were mixed with different concentrations of inhibitor (0 -1000 nm), and product formation was measured over time. Since the inhibition of trypsin or plasmin is assumed to be irreversible, the course of the reaction can be fit via nonlinear regression, where product formation (P) proceeds at an initial velocity (v z ) and is inhibited over time (t) at a rate (k obs ).
For each combination of enzyme and inhibitor, a k obs was calculated by nonlinear regression analysis using Equation 1. A second order rate constant (kЈ) was determined by plotting a series of k obs versus the respective inhibitor concentration and measuring the slope of the line (kЈ ϭ ⌬k obs /⌬[I]). Since the inhibitor was in competition with the substrate, the second order rate constant (kЈ) was corrected (k a ) for the substrate concentration and the K m of the enzyme for the substrate.
The K m of human trypsin for AAPR-pNA in PBS was 50 M, and that of plasmin for VLK-pNA in PBS was 230 M. All kinetic studies were repeated 2-4 times.
Partial Digestion of Serpin Proteins-Clostripain was diluted to 0.15 units/ml in 3.3 mM dithiothreitol and 0.5ϫ PBS, pH 7.4, and activated by incubation for 2 h at 25°C. For partial digestion, 5 g of recombinant serpin in 51 l of PBS, pH 7.4, was incubated 20 min at 25°C with 51 l of activated clostripain in 10 mM dithiothreitol. The amount of clostripain in each reaction varied with the serpin and was adjusted to cleave ϳ50% of the RSL substrates: 0.00024 units for Serpinb3b, 0.00006 units for Serpinb11 and Serpinb11-hRSL, and 0.00003 units for SERPINB11d and SERPINB11d-(E90L,S91F,P303S). Digestion was terminated by the addition of EDTA pH 8.0 to a final concentration of 25 mM.
Thermostability Assays-Assays were performed as described (21). Briefly, aliquots containing 1 g of undigested or partially digested recombinant serpin were incubated at temperatures ranging from 37 to 95°C, transferred to ice for 10 min, and then centrifuged at 12,000 ϫ g for 10 min at 4°C. Supernatants were analyzed by SDS-PAGE and immunoblotting.
Matrix-assisted Laser Desorption Ionization Mass Spectroscopy (MALDI-MS)-Human trypsin (0.2 g) or plasmin (0.6 g) was mixed with Serpinb11 (4.4 g) in 20 l of PBS. Following either a 3-or 30-min incubation at 25°C, the protein mixtures were frozen and stored at Ϫ80°C. The mixture components were separated by MALDI-MS at the Wistar Protein Microchemistry Facility (Philadelphia, PA).

JOURNAL OF BIOLOGICAL CHEMISTRY 24951
Statistical Analysis-The Student's t test was used to compare the differences between the means of second order rate constants.

Identification of a Novel Human Clade B Serpin and Its
Mouse Orthologue-To identify novel clade B genomic sequences, we used the BLAST2 algorithm and exon 8 of SER-PINB4 to query the human genomic DNA sequence deposited in GenBank TM (25). This exon was chosen, since it contained the highly conserved hinge region of the RSL (28) and the PROSITE serpin signature motif. A significant match was detected in the DNA sequence from the human 18q21 contig clone RP11-69807 (GenBank TM accession number AP001583). Using a combination of gene prediction programs, a novel serpin gene was identified that had the typical features of clade B serpins, including eight exons and the absence of an N-terminal hydrophobic signal sequence (4). The gene was designated SER-PINB11 by the HUGO Gene Nomenclature Committee. The genomic sequence of SERPINB11, however, contained an early termination codon in exon 4. Since this mutation would lead to either a truncated and dysfunctional peptide of 89 residues or an mRNA targeted for decay (29,30), we sought to determine whether the reference SERPINB11 in GenBank TM was a pseudogene or a simple sequence variant with a nonsense mutation.
We designed oligonucleotide primers encompassing the putative start and stop codons of SERPINB11, and used these reagents to amplify DNA fragments (full-length open reading frame ϳ1200 bp) from both lung and prostate first-strand cDNAs. DNA fragments were subcloned into pBluescript or pCR2.1-TOPO and analyzed by DNA sequencing. Seven different transcripts were identified; four contained full-length coding sequences that could be distinguished by different combinations of eight different SNPs (designated SERPINB11a, -b, -c, and -d; Fig. 1, A and B), two were splice variants (designated SERPINB11e and -g; Fig. 1B), and one contained a nonsense mutation at the same position identified in the original genomic DNA sequence (SERPINB11f; Fig. 1B). The SNPs in SERPINB11a to -d were not due to PCR or DNA sequencing artifacts, since they were detected in multiple independent cDNA clones, and most have been verified independently by other sequencing efforts and documented in dbSNP ( Table 1). The putative RSL amino acid sequences of the SERPINB11 gene and the SERPINB11a to -d cDNAs were identical and revealed a typical proximal hinge region (P15-P8) and Lys and Ser residues at the canonical reactive center (P1 and P1Ј; Fig. 1A). Based on the RSL sequence of SERPINB11a to -d, SERPINB11 would be predicted to encode an inhibitory type serpin. However, we reasoned that the amino acid variations, which

SERPINB11
Human genomic The designation of conservative (italic type) and nonconservative (bold face type) amino acid variants in SERPNB11d are relative the consensus motif.
mapped to important structural elements within the serpin scaffold, could interfere with functional activity. Thus, biochemical analysis was needed to determine whether SER-PINB11 was a peptidase inhibitor. Of note, the GenBank TM genomic sequence of SERPINB11 was one of the first to document a nonsense mutation in a human clade B serpin gene. One Serpinb11 cDNA (mouse) was obtained previously, and this transcript encoded Lys-Arg at the reactive center ( Fig. 1A) (18). The amino acid sequences of Serpinb11 and SERPINB11 were 67% identical and had a predicted M r of ϳ43,478 and ϳ44,095 and a pI of 9.16 and 8.63, respectively (18).
Surveying for SERPINB11 and mSerpinb11 Inhibitory Activity-The predicted amino acid sequences of SERPINB11a to -d were highly similar to that of the genomic sequence except for the absence of the stop codon at position 90 and several nonconservative amino acid variants at positions 51, 91, 148, 181, 188, 267, and 303 (Fig. 1, A and B, and Table 1). Of all of these variants, the Trp residue at amino acid position 188 (equivalent to position 194 using SERPINA1 numbering) is conserved absolutely among all of the clade B serpins and is present in nearly all inhibitory type serpins throughout evolution (8,31). In SERPINA1, Trp-194 resides at the top of strand 3A and packs into a hydrophobic pocket in the breach region, where it forms interactions with conserved residues, Phe-198 (strand 4C) and Tyr-244 (strand 2B) (8). The latter residues (Phe-192 and Tyr-239) were conserved also in all of the available deduced amino acid sequences of SERPINB11. Thus, we reasoned that the Arg-188 (SERPINB11 and SERPINB11a, -b, and -c), instead of Trp-188 (SERPINB11d) would lead to catastrophic misfolding of the protein, since the charged nature of the Arg residue would preclude packing into the hydrophobic core. Thus, SERPINB11d was selected for expression in Escherichia coli. However, incubation of purified SERPINB11d with a variety of serine and cysteine peptidases failed to indicate any inhibitory activity against any of the peptidases listed in Table 2. Of note, recombinant SERPINB11a, -b, and -c also failed to show inhibitory activity against a more limited panel of peptidases (not shown).
In contrast to SERPINB11d, when recombinant Serpinb11 was tested against the same panel of cysteine and serine peptidases, inhibitory activity against the model peptidases, trypsin and plasmin, was detected ( Table 2). Serpins interacting with their target peptidases typically show 1) a low stoichiometry of inhibition, 2) a second order rate constant of ϳ10 4 M Ϫ1 s Ϫ1 , 3) a covalent serpin-peptidase complex by SDS-PAGE, and 4) an RSL cleavage site at or near the canonical P1 position. Our data suggested that Serpinb11 was capable of functioning as a bona fide serpin-like inhibitor. The molecule exhibited an SI and k a of 3.5 and 2.0 Ϯ 0.4 ϫ 10 4 M Ϫ1 s Ϫ1 , respectively with trypsin (Fig. 2, A and B) and 3.9 and 1.3 Ϯ 0.4 ϫ 10 4 M Ϫ1 s Ϫ1 , respectively with plasmin (Fig. 2, A and C). Further, Serpinb11 formed SDS-stable, high molecular mass complexes with both both trypsin (Fig. 3A) and plasmin (data not shown). Finally, MALDI-MS analysis revealed that the Lys-Arg at the canonical P1-P1Ј positions within the RSL served as the reactive center for the Serpinb11 interaction with both peptidases (Fig. 3,  B-D). This result was expected, since Lys and Arg residues at the putative reactive center (P1 and P1Ј, respectively) predicted inhibitory activity against trypsin-like peptidases. Serpinb11 also showed some inhibitory activity against the cysteine peptidases, cathepsin L, K, and V, and the serine peptidase, ␥-tryptase. However, subsequent SDS-PAGE analysis showed that these results were due to simple competition reactions, with the serpin serving as a substrate rather than as an inhibitor of the enzymes (not shown).
The RSL of SERPINB11 Exhibits Inhibitory Activity-The discrepancy in inhibitory activity between Serpinb11 and SERPINB11d could be attributed to at least two testable hypotheses; either the human and mouse RSLs had diverged sufficiently such that the human serpin was no longer recognized by any of the peptidases listed in Table 2, or the human SNPs encoded a combination of nonconservative amino acid changes within the SERPINB11d scaffold that impaired proper folding and/or inhibited the conformational rearrangement required for peptidase trapping. To distinguish between these possibilities, we first reexamined the degree of conservation among the human, mouse, chimpanzee, dog, and rat Serpinb11 RSL amino acid sequences (Fig. 3D). Remarkably, these RSLs were highly conserved, with the only appreciable difference that could affect peptidase binding being the P1Ј Arg of mice, rats, and dogs as compared with the Ser of humans and chimpanzees. To determine if the Ser at the P1Ј position could deter the binding of trypsin-like peptidases, we generated an Arg-355(P1Ј) 3 Ser (R355(P1Ј)S) mouse Serpinb11 mutant by sitedirected mutagenesis. This mutant inhibited trypsin in a similar fashion to wild-type protein (SI and k a of 2.3 and 4.8 Ϯ 2.1 ϫ 10 4 M Ϫ1 s Ϫ1 , respectively; Fig. 2, A and D). Thus, it was unlikely that a Ser residue at the P1Ј position accounted for the absence of   AUGUST 24, 2007 • VOLUME 282 • NUMBER 34

JOURNAL OF BIOLOGICAL CHEMISTRY 24953
SERPINB11d inhibitory activity. We also performed an RSL swap between mouse and human cDNAs by substituting terminal PstI fragments. The resulting mouse and human constructs encoded the opposite species amino acid sequence from the P11 residue to the C terminus (see amino acid alignment in Fig.  1A). Again, recombinant Serpinb11-human RSL (hRSL) inhibited trypsin in a fashion comparable with wild-type protein (SI and k a of 7.2 and 2.0 Ϯ 1.8 ϫ 10 4 M Ϫ1 s Ϫ1 , respectively; Fig. 2, A and E). Serpinb11-hRSL also formed a covalent complex with trypsin (Fig. 3A). In contrast, the reverse RSL swap, SER-PINB11-mouse RSL, showed no detectable inhibitory activity (not shown). Analysis of the SERPINB11d Scaffold-The RSL swap mutants showed that the sequence of the human RSL of SERPINB11, like the mouse RSL, was capable of mediating peptidase inhibition and supported the notion that amino acid variants/ mutations within the SERPINB11d scaffold were responsible for the lack of inhibitory function. To test this hypothesis, we first compared the eight amino acid variants of SERPINB11 with those amino acids present at the identical positions in the mouse, chimpanzee, dog and rat orthologues and all of the human clade B paralogues (Table 1). Based on these comparisons, SERPINB11d harbored three amino acids, Glu-90, Ser-91, and Pro-303, that differed significantly from the other serpins (Table 1). In particular, Glu-90 and Ser-91 were located at critical residues in helix D that pack up against the shutter region. The shutter region, which is centered on the top of helix B and includes parts of helix D, contains important determinants for the mobility of ␤-sheet A. This region is highly sensitive to change, since mutations in helix B abolish the inhibitory activity of thermopin without affecting its heat denaturation profile (32), and mutations in helix D cause spontaneous conformational changes in SERPINC1/antithrombin III (33). Mutations in helix I are less common, but an S349P mutation of SERPINC1 leads to increased thrombosis and a type 2 deficiency (normal amounts of circulating protein but decreased activity) (34).
To determine whether these three amino acid variants in SERPINB11d could account for the lack of inhibitory activity, we generated three revertants by site-directed mutagenesis: SERPINB11d-(E90L,S91F), SERPINB11d-(P303S), and SERPINB11d-(E90L,S91F,P303S). Under the assay conditions that we used to measure second order rate constants, none of the revertants showed detectable inhibitory activity against trypsin (not shown). The absence of inhibitory activity suggested that the variant amino acids in the scaffold either caused the protein to misfold (for example, the serpin was in the latent or polymeric state) or interfered with the inhibitory mechanism (e.g. retarded RSL insertion) (35). To help differentiate among these possibilities, we first examined the reaction products of the SERPINB11d-trypsin interaction. SERPINB11d was incubated with trypsin at a 20:1 ratio for 1, 5, and 20 min. Samples were incubated at 95°C for 3 min in the presence of 2% SDS and 1% ␤-mercaptoethanol, separated by SDS-PAGE, and immunoblotted with GST antibody. Unlike with the Serpinb11-trypsin interaction, no covalent SERPINB11d-trypsin complex was detected (Fig. 4A). These findings are consistent with the inability of SERPINB11d to block substrate conversion by trypsin in vitro. However, a typical RSL-cleaved formed of SERPIB11d was detected just below the wild-type band. This finding suggested that the RSL of SERPINB11d was available for target peptidase interaction, but the enzyme escaped before being trapped by the inhibitor. Consistent with the lack of trypsin inhibition, the enzyme degraded the serpin with increasing incubation times. Similar findings were observed with the SERPINB11d-(E90L,S91F) revertant (Fig. 4A). Interestingly, a small amount serpin-trypsin complex was detected using the SERPINB11d-(P303S) revertant (Fig.  4A). This finding suggested that a single amino acid change enhanced RSL insertion such that a fraction of the serpin molecules could trap the enzyme as a covalent intermediate. However, this fraction was insufficient to measure an SI or a k a . Of note, the loss of serpin-trypsin complex after the 20-min incubation is consistent with previous findings showing that these types of complexes are susceptible to degradation in enzyme excess (36). SERPINB11d-(E90L,S91F,P303S) also formed covalent complexes with trypsin (Fig. 4A). The SI for this interaction was ϳ40 (Fig. 4B). Due to the high SI, we were unable to determine the k a for this interaction.
To further assess the conformations of SERPINB11d and the revertants, we next examined their ability to undergo the stressed (S) to relaxed (R) transition typical of serpins with mobile RSLs, a requisite function for inhibitory activity (14,(37)(38)(39). Serpins in the S-conformation (RSL noncleaved and exposed) are metastable and in most cases precipitate upon heating to ϳ55-65°C. In contrast, serpins in the R-conformation (RSL cleaved and inserted into ␤-sheet A) are stable and remain in solution at elevated temperatures. Thus, the S-and R-forms can be differentiated by differences in their thermal stability. To cleave the RSL, we performed limited digestion with clostripain, a cysteine peptidase that cleaves after the P1Ј, P5Ј, and P1 Arg residues of Serpinb11, SERPINB11, and Serpinb3b (control), respectively. As expected, limited digestion of Serpinb3b with clostripain yielded a cleaved serpin that was resistant to heat denaturation up to 95°C (Fig. 4C). Most of the uncleaved Serpinb3b precipitated from solution at this tem-  Inhibition rates and the stoichiometry of inhibition were calculated as described in Fig. 2A. C, thermostability and assessment of the stressed-torelaxed conformational change of Serpinb11, SERPINB11d, and SERPINB11d-(E90L,S91F,P303S). Serpins were subjected to limited digestion with clostripain to yield a mixture of uncleaved and RSL-cleaved serpins. The reaction was terminated by the addition of EDTA to 25 mM. Samples were then incubated at the indicated temperatures for 10 min, chilled on ice, and centrifuged. The supernatant was analyzed by SDS-PAGE and immunoblotting as in A. Full-length and RSL-cleaved serpins are indicated by white and gray arrowheads, respectively. perature, although there appeared to be a small fraction of this preparation that was in the stable, RSL (noncleaved)-inserted latent conformation. Native Serpinb11 and Serpinb11-hRSL, like many serpins, unfolded and precipitated from solution by incubation at 50 -56°C (Fig. 4C). Interestingly, although the RSL-cleaved fractions were more stable, they both precipitated at 75°C (Fig. 4C). This finding suggested that the R-form of Serpinb11 was not as stable as those of Serpinb3b and other serpins, which typically retain their solubility within this temperature range (14,(37)(38)(39). Uncleaved SERPINB11d showed a thermal denaturation profile similar to Serpinb11 (Fig. 4C), suggesting that it adopted a typical metastable heat-labile conformation. Remarkably, however, RSL-cleaved SERPINB11d was more susceptible to heat denaturation than the uncleaved form (Fig. 4C). This phenomenon has only been observed with angiotensiongen, a noninhibitory serpin (39). Although the RSL was available for peptidase cleavage, it appeared that the RSL of SERPINB11d was incapable of assuming the more stable, R-conformation. Uncleaved and cleaved SERPINB11d-(E90L,S91F,P303S) gave a thermal denaturation profile similar to Serpinb11 and suggested that these three amino acid changes helped to restore stability and functionality to the SERPINB11d molecule (Fig. 4C). Taken together, these findings suggested that 1) the cleaved forms of mouse and human serpinb11 were inherently more unstable than those of other serpins and 2) the amino acid variants in SERPINB11d impaired inhibitory activity by either impeding proper loop insertion and conversion to the more stable R-form or by accommodating the loop in a fashion that led to further destabilization of the molecule. Structural studies may be required to resolve this latter issue.
Allelic and Haplotype Analysis of SERPINB11 in Human Populations-Mutational analysis suggested that variants in the SERPINB11 scaffold impaired in vitro peptidase-inhibitory activity. To determine the frequency of these variants within different ethnic populations, we analyzed DNA sequencing data deposited into dbSNP and HapMap (40,41). The current HapMap data are derived from several populations: Yoruba in Ibadan, Nigeria; Japanese in Tokyo; Han Chinese in Beijing; and CEPH (Utah residents with ancestry from northern and western Europe) (40,41).
To simplify the analysis, we focused on SNPs encoding amino acids at positions 90, 188, and 303, since these variants were likely to have the greatest impact on functional activity. Nucleotide variants leading to either a termination instead of a Glu at 90, an Arg instead of a Trp at 188, or a Pro instead of a Ser at 303 resulted in a dysfunctional allele in terms of inhibitory activity (see above). The frequency of each of these deleterious variants (i.e. mutations) per gene ranged from 23 to 68%, 32 to 77%, and 70 to 92%, respectively (Table 3). Depending on the ethnic population, the frequency of individuals homozygous for each of these deleterious variants ranged from 8 to 50%, 8 to 63%, and 48 to 83%, respectively (Table 3). From these data alone, the percentage of the population possessing the best inhibitory allele (Glu-90, Trp-188, and Ser-303; equivalent to SERPINB11d-(P303S)) could be no greater than 17-53%. However, if we examined the genotypic patterns of variants at the three positions in cis (i.e. the haplotype) and trans (i.e. homozygosity versus heterozygosity), we were able to more accurately assess the number of individuals with an inhibitory allele. Twohundred seventy-four individuals (ϳ80%) were homozygous for at least one of the deleterious SNPs (Table 4, partial haplotypes a, b, d-g, i, j, and k). Thus, none of these individuals (ϳ80% of the population) carried a copy of an inhibitory type SERPINB11. Of the remaining three partial haplotypes (Table  4, haplotypes c, h, k, and m; note that one individual from the JPT population could not be haplotyped (m) but was included to give a maximal estimate of individuals with inhibitory type copies of SERPINB11), 67 (20%) individuals could carry an allele with Glu-90, Trp-188, and Ser-303 in cis. However, since the precise phasing (haplotyping) was not determined in these individuals, the number of individuals heterozygous for an inhibitory allele could range from 20 to 0%. It is also noteworthy that no individuals were homozygous for all three favorable amino acids as positions 90, 188, and 303 (Table 4, haplotype n), and of 14 sequences deposited in GenBank TM (two genomic, eight full-length cDNAs, and four partial cDNAs; Table 1), each contained at least one codon (stop90, Arg-188, or Pro-303) that would designate it as a noninhibitory allele. Taken together, there was no evidence to suggest that an inhibitory type SERPINB11 allele resides within the human genome. Tissue Expression Pattern of SERPINB11 and Serpinb11-To determine the tissue expression pattern of SERPINB11 and Serpinb11, first-strand cDNA prepared from a panel of human and mouse tissue samples were screened by PCR, respectively. SERPINB11 transcripts were detected in a restricted number of tissues, including lung, placenta, prostate, and tonsil (Fig. 5A). Serpinb11 showed an expanded expression profile with transcripts present in samples prepared from the eye, lung, lymphocytes, thymus, stomach, uterus, heart, brain, liver, skeletal muscle, and whole embryonic day 7, 15, and 17 embryos (Fig. 5B).

DISCUSSION
In silico analysis of human genomic DNA predicted the existence of 13 clade B serpin genes (4). Experimental evidence had confirmed the existence of 12 family members, but heretofore data on SERPINB11 was lacking. Moreover, the ability to infer biochemical function based solely on the genomic sequence of SERPINB11 was confounded by the presence of a premature termination codon in exon 4, which, based on known structurefunction relationships, would yield a null mutant in terms of inhibitory activity (42,43). Thus, the goal of this study was to determine whether human SERPINB11 was a pseudogene or a bona fide peptidase inhibitor. The absence of a premature stop codon in the highly conserved mouse orthologue Serpinb11 suggested the latter (18). To resolve this issue, we cloned and sequenced seven SERPINB11 cDNAs. One cDNA contained a premature stop codon and corresponded to the original genomic sequence. Two other cDNAs contained spliced variants that would lead to the synthesis of truncated proteins with no discernible serpin function. However, we also identified four different cDNA clones with intact, full-length open reading frames. The full-length SERPINB11 cDNAs were designated SERPINB11a, -b, -c, and -d based on combinations of eight SNPs. These SNPs encoded nonconservative amino substitutions relative to other clade B serpins, and these variant residues mapped to different structural elements located within the main serpin scaffold. Interestingly, none of the amino acid variants mapped to the RSL, which is the motif that is required for inhibitory function and target peptidase specificity.
One SNP that was present in three cDNA clones (SERPINB11a-c) encoded an Arg instead of a Trp residue in strand 3A at position 188 (position 194 by SERPINA1 numbering). A search of dbSNP failed to reveal a variant at this position in any of the other clade B serpins. Also, analysis of 219 serpins distributed among all the clades showed that Trp-194 was conserved in 94% of serpin amino acid sequences and in virtually all inhibitory type molecules (8,31). This residue is important for FIGURE 5. Tissue expression of SERPINB11 and Serpinb11. A, PCR was performed using first-strand cDNA templates from 18 adult human tissues with SERPINB11 (expected product size 490 bp) and ␤-actin (ACTB) primers (expected product size 616 bp). B, PCR was performed using first-strand cDNA samples from 20 adult mouse tissues and mouse embryos with Serpinb11 (expected product size 554-bp) and glyceraldehyde-3-phosphate dehydrogenase (G3PDH) primers (expected product size 983 bp). the stabilization of the covalent enzyme-inhibitor complex, since a Trp to Phe mutation in plasminogen activator inhibitor-1 (SERPINE1) had no effect on peptidase binding but decreased significantly the rate at which t-PA was inhibited irreversibly by SERPINE1 (44). Since one cDNA clone, SERPINB11d, harbored the conserved Trp residue, we chose this clone for further analysis. However, recombinant SERPINB11d, which had a thermal denaturation curve similar to that of other inhibitory type serpins (implying that the protein folded into the loop-exposed, metastable conformation typical of inhibitory serpins), failed to show any inhibitory activity. This finding contrasted with the mouse protein, which showed modest inhibitory activity against trypsin-like peptidases. The apparent discrepancy between the human and mouse proteins could be due to the minor differences between their RSL amino acid sequences. Although both RSLs contained a Lys residue at the P1 position (the most critical determinate of serine peptidase specificity), they differed at the P1Ј position, with the human sequence encoding for a Ser and the mouse encoding for an Arg. Although the S1 subsite of trypsin-like enzymes preferentially accommodates Arg and Lys residues at the P1 position, the S1Ј subsite has broader specificity and is quite tolerant of Ser at the P1Ј position (45,46). Indeed, a Ser-pinb11-R355(P1Ј)S mutant retained inhibitory activity. Based on this knowledge, we were hard pressed to attribute the lack of SERPINB11d antitrypsin activity to a Ser instead of an Arg at the P1Ј position. In addition, we generated expression constructs in which RSLs of SERPINB11d and Serpinb11 were swapped. Recombinant Serpinb11-human RSL demonstrated antitrypsin activity comparable with that of Serpinb11, whereas SERPINB11-mouse RSL was not inhibitory. Together, these data suggested that it was amino acid sequences in the SERPINB11d scaffold rather than those of the RSL that hindered inhibitory function.
We hypothesized that the inability of SERPINB11d to support inhibitory activity was due to amino acid variants in the serpin scaffold that either impaired proper folding or restricted the ability of the protein to undergo the S-R transition conformational rearrangement characteristic of the peptidase trapping mechanism (5). When we compared the SERPINB11 SNPencoded amino acid variants with those amino acids located at identical positions in four orthologous mammalian genes (mouse, rat, dog, and chimpanzee) and 12 paralogous human clade B genes, conflicts at key positions within hD and hI became apparent. A portion of hD comprises the shutter region, a functional domain that regulates ␤-sheet A movement and permits RSL loop insertion upon peptidase cleavage (42,43,47). The presence of a charged and polar residue at positions 90 and 91, respectively, instead of two hydrophobic residues suggested that these amino acid variants could be contributing to serpin scaffold dysfunction. However, when we converted the variant amino acids at positions 90 and 91 to the conserved hydrophobic residues, inhibitory activity of SERPINB11d was not detected, and this revertant was unable to form a covalent complex with trypsin. Reversion of the hI breaking Pro-303 to the well conserved Ser residue in the presence or absence of Leu-90 and Phe-91 partially restored the SERPINB11d-inhibitory mechanism, as shown by the presence of a covalent com-plex on SDS-PAGE. Relative to other inhibitory serpins, however, this revertant was still a poor inhibitor, with an SI of ϳ35. This high SI suggested that either a small fraction of the inhibitor was in the RSL-exposed conformation and available for peptidase interaction or that most of the inhibitor was available for peptidase interaction but an impaired loop insertion mechanism precluded enzyme trapping. Studies examining the thermal stability of RSL uncleaved and cleaved SERPINB11d and its revertants suggested that the overall thermal stability of SER-PINB11 was less than that of other clade B serpins. In addition, it appeared that the majority of the proteins folded into a metastable conformation with the RSL exposed and available for peptidase cleavage and not the noninhibitory, RSL loop-inserted latent or polymerogenic forms (35,48). Moreover, the inability of RSL-cleaved SERPINB11D to undergo the S-R transition and the partial restoration of this activity in a SERPINB11d revertant, strongly suggested that lack of inhibitory activity was secondary to structural alterations in the serpin scaffold that impaired RSL insertion into ␤-sheet A. Impaired RSL mobility is typical in other noninhibitory serpins, such as ovalbumin, but in these cases, amino acid alterations in the RSL proximal hinge region, rather than the serpin scaffold, account for the loss of movement and lack of inhibitory function (49).
To further examine the functional effects of the hD and hI variants, we projected the Glu-90 and Pro-303 residues onto a model of the serpin scaffold. PSI-BLAST (50) searches of the protein data bank with SERPINB11 (51) showed that chicken ovalbumin was the closest putative native serpin homologue with a known structure (39% identity over 390 residues; Protein Data Bank identifier 1OVA (52)). The sequence alignment between SERPINB11 and ovalbumin revealed that the region around Glu-90 and Pro-303 was well conserved, with no predicted insertions or deletions. Thus, we were able to identify with confidence the residues equivalent to Glu-90 and Pro-303 within the ovalbumin template (Ile-99 and Ala-305, respectively (Fig. 6A)). By superimposing the Glu-90 of SERPINB11 onto the ovalbumin template, this residue was positioned halfway up hD with its side chain pointed into a solvent-exposed cleft bounded by hA, hB, hD, and the loop between s4B and s5B (Fig. 6, A and B). In other serpins, this position corresponds to a highly conserved Leu residue (Ͼ70%) (8). Exceptions include neuroserpin, which also contains a Glu at the position equivalent to Glu-90 in SERPINB11. Although the structure of cleaved neuroserpin has been determined (53), this region is not visible in electron density. Given the prediction that the side chain of Glu-90 was solvent-exposed, we did not anticipate that this substitution would alter the overall structure of the native serpin fold. However, Glu-90 was located close to the shutter region, and as demonstrated with the Thermobifida fusca serpin, thermopin, even minor substitutions in the shutter can lead to a profound loss of inhibitory activity (32). Residue Pro-303 was predicted to be located at the C-terminal end of hI (Fig.  6, A and C). The presence of a Pro in this position would distort the C-terminal portion of this short helix. Analysis of conformational mobility in serpins during the S-R transition reveals significant movement in hI and the C-terminal loop during RSL insertion (42,43). Indeed, this region abuts other mobile regions of the serpin scaffold, including hE. Thus, in addition to distortion of hI, the model supported the notion that the Pro-303 substitution affected inhibitory function by interfering with the S-R transition. These observations were consistent with the SDS-PAGE data showing that substituting a Ser residue partially restored inhibitory activity. Although CD spectroscopy and measurements of internal tryptophan fluorescence may help confirm that SERPINB11d failed to undergo the S-R transition, a comparison between RSL uncleaved and cleaved crystal structures may be required to definitely determine how these variants affect serpin function.
Analysis of SERPINB11 showed that either a stop codon instead of a Glu residue at position 90 (helix D), an Arg instead of the well conserved Trp residue at position 188 (strand 3, ␤-sheet A), or a helix-breaking Pro instead of a Ser residue at position 303 (helix I) led to a loss of inhibitory activity. Based on the codon variations at positions 90, 188, and 303, data from dbSNP and the HapMap project (40,41) showed that at least 80% of the population was homozygous or compound heterozygous for noninhibitory copies of SERPINB11. Of the remaining 20%, none were homozygous for an inhibitory allele. Since all of these individuals were heterozygous for all three codons at positions 90, 188, and 303, it was conceivable that each carried one inhibitory allele. However, since the phasing of these residues in cis was unknown, and if we consider that none of the 14 genomic or cDNAs nucleotide sequences deposited in GenBank TM carry the inhibitory allele, the fraction must be substantially less than 20%. Moreover, if we acknowledge that the "inhibitory" allele (i.e. SERPINB11d-(P303S) encoded a protein lacking measurable kinetic activity (evidence for inhibition was only detected by complex formation on SDS-PAGE), there was little evidence to suggest that the human genome harbors a physiologically relevant, inhibitory type SER-PINB11 gene. Thus, we concluded that SERPINB11 has lost its inhibitory activity. To date, only one noninhibitory intracellular human serpin, SERPINB5/maspin, has been identified. This molecule is required for early embryonic development and serves as a tumor suppressor gene by facilitating cell adhesion and preventing cancer metastasis (16,54). Here, we also suggest that SERPINB11 performs a role outside of peptidase inhibition. Although this function has yet to be characterized, clues may come from the identification of binding partners in tissues where SERPINB11 was expressed, such as lung, placenta, or prostate.
If we consider that Serpinb11, but not SERPINB11, was inhibitory, yet both human and murine proteins possessed inhibitory type RSLs, SERPINB11 may have begun its transition to a noninhibitor relatively recently. Moreover, the amino acid variations in SER-PINB11 may provide a model for understanding how noninhibitory serpins evolve. Of particular interest was the finding that only a few naturally occurring mutations within the scaffold (rather than the RSL) were sufficient to interfere with the S-R transition and abolish inhibitory activity. Given that the newly acquired noninhibitory function of SER-PINB11 must be independent of the conformational change associated with loop insertion, the scaffold becomes a malleable structure, accommodating a larger number of SNPs, until its noninhibitory function has been optimized. Considering that mouse Serpinb11 was only a modest inhibitor of trypsin-like peptidases with a relatively high SI, it is conceivable that the mouse protein may be evolving toward this noninhibitory function as well.
Since most of the human clade B serpins harbor SNPs encoding for nonconservative amino acid changes (supplemental Table 1), it is conceivable that the association between different combinations of SNPs and alterations in biochemical activity is a more common occurrence than originally suspected. Although we are unaware of any natural mutation of a clade B serpin that leads to a disease phenotype in humans at this time, these SNPs are likely to contribute to quantitative traits, such as chronic obstructive pulmonary disease and asthma (55,56). With the advent of SNP genotyping, this hypothesis is readily testable and might identify a clade B serpin haplotype associated with disorders manifested by significant alterations in peptidase-inhibitor imbalance.