Pepsin-inhibitory activity of the uterine serpins

Abstract

Among the major products secreted by the uteri of cattle, sheep, and pigs during pregnancy are glycoproteins with amino acid sequences that place them in the serpin (serine proteinase inhibitor) superfamily of proteins. The inferred amino acid sequences for bovine uterine serpin (boUS-1) and ovine uterine serpin (ovUS-1) exhibit about 72% sequence identity to each other but only about 50% and 56% identity, respectively, to two distinct porcine uterine serpins (poUS-1 and poUS-2). Despite these differences in primary structure, the uterine serpins possess well-conserved reactive center loop regions that contain several motifs present in the propeptide regions of pepsinogens. One such motif, VVVK, aligns with the first 4 amino acids of the aspartic proteinase inhibitor pepstatin. Although no inhibitory activity toward any serine proteinase has been found, at least one of the uterine serpins, ovUS-1, can bind specifically to immobilized pepsin A and can weakly inhibit the proteolytic activities of pepsin A and C (but not cathepsins D and E). OvUS-1 is the first specific inhibitor of aspartic proteinases to be identified in vertebrates and provides another example of a serpin with “crossover” activity. The pregnancy-associated glycoproteins (PAGs), which are secreted by the trophoblast layer of the placentas of ungulate species and are inactive members of the aspartic proteinase family, can also bind ovUS-1 and may be the natural target partners for the uterine serpins.

The porcine uterus produces large quantities of several proteins in response to progesterone, the hormone of pregnancy (1, 2). These proteins are secreted into the uterine lumen and, during pregnancy, contribute to the so-called histotrophe or uterine milk that bathes the conceptuses. Among them are uteroferrin (3) and a retinol-binding protein (4, 5) (both of which probably have nutritional roles), growth factors, growth factor binding proteins (6), a group of low molecular weight proteinase inhibitors belonging to the Kunitz family (7), and three related basic glycoproteins known collectively as the uteroferrin-associated basic proteins (8, 9). The latter (M_r = 50,000, 48,000, and 42,000, respectively) are in the serpin superfamily and arise by proteolytic processing and differential glycosylation of a larger precursor molecule (9).

The uterus of the ewe also synthesizes abundant amounts of progesterone-induced secretory protein composed largely of two basic glycoproteins, the so-called uterine milk proteins, which have a M_r of 57,000 and 55,000, respectively (10, 11, 12, 13), and which are also in the serpin superfamily (14). These ovine uterine serpins (now called ovUS) have been reported to be immunosuppressive (15, 16) and may prolong the ability of skin grafts to survive within the uterus (17). They have not been shown to possess any antiproteinase activity (14).

Although pigs and sheep are both ungulate species, their ancestors diverged at least 55 million years ago (18, 19). Moreover, the types of placentation they exhibit are quite distinct (20). Therefore, it was of considerable interest that both species should produce large quantities of structurally similar progesterone-inducible products during pregnancy. Hence the studies on uterine serpins have been extended. Herein we demonstrate that these uterine serpins interact with members of the aspartic proteinase family rather than with serine proteinases. They provide another example of serpins with crossover function.

Because various acronyms were used for these uterine serpins before their general relatedness was revealed by molecular cloning studies, it is proposed that the previous designations [e.g., uteroferrin-associated basic protein (UABP) and uterine milk protein (UTMP)] be abandoned and instead they should be named uterine serpins (or US) preceded by the species name, e.g., bovine (bo), ovine (ov), and porcine (po).

MATERIALS AND METHODS

Materials.

Porcine gastric pepsin A, bovine spleen cathepsin B, N^α-CBZ-l-lysine p-nitrophenyl ester hydrochloride (CBZ is carbobenzoxy), bovine hemoglobin, and CNBr-activated Sepharose-4B were purchased from Sigma. Recombinant human and porcine cathepsin D were donated by G. Conner, University of Miami. Recombinant cathepsin E was a gift from B. Dunn, University of Florida, Gainesville, FL. Pepsin C was provided by J. Tang, University of Oklahoma, Oklahoma City, OK. [¹⁴C]Formaldehyde was obtained from American Radiolabeled Chemicals, St. Louis. [¹⁴C]Hemoglobin was prepared by the procedure described by Means and Feeney (21).

Screening of Porcine Endometrial cDNA Library.

About 40,000 recombinant phages from a porcine endometrial cDNA library were screened with a random-primed poUS-1 cDNA probe (9, 22).

Phage DNA was isolated from 10 positive plaques. The sizes of cDNA inserts ranged from 400 bp to 1250 bp. Clone 12.1 cDNA (1250 bp) was subcloned and sequenced (23). To obtain a full-length cDNA (1400 bases), the sequence of clone 12.1 cDNA was merged with the 5′ sequence of a previously reported poUS clone (2.1) (9). This sequence was confirmed from the genomic sequence of poUS-2 (data not shown).

Screening of Bovine Endometrial cDNA Library.

The bovine library was constructed from day 17 pregnant cow endometrial poly(A)⁺ RNA in λZAP vector (Stratagene). The library was amplified in XL1-blue cells, and about 10⁶ plaques were screened with a ³²P-labeled full-length ovUS-1 cDNA. About 20 positives were plaque-purified, and ones with the largest inserts were identified by PCR with M13r and M13f primers. Clone 1.38 (1.4 kb) was selected for further characterization. The plasmid was excised from the phage by in vivo excision and sequenced in both directions.

Purification of ovUS.

Sheep uterine milk was collected from unilaterally pregnant ewes (12). The basic protein fraction of the uterine secretions, which is predominantly ovUS-1, was obtained by chromatography on CM-cellulose at pH 8.2 (12). Protein was eluted with 0.5 M NaCl in 10 mM Tris·HCl (pH 8.2) and dialyzed for 6 h against three changes of 0.9% NaCl at room temperature. About 1.0 mg of the eluted protein from CM-cellulose was further chromatographed on a Superose-12 column (1 × 30 cm, Pharmacia), equilibrated with 0.9% NaCl and eluted at a flow rate of 0.5 ml/min. Samples (50 μl) from each fraction were assayed for pepsin A inhibitory activity. Samples (10 μl) from the peak protein fractions were analyzed by gel electrophoresis in 12.5% polyacrylamide gels in presence of SDS (12).

Enzyme Inhibitory Activity Measurements.

Inhibitory activity of ovUS-1 toward pepsin A and pepsin C (gastriscin) was determined by using [¹⁴C]methyl-hemoglobin as substrate (24). Increasing protein concentrations of purified ovUS-1 (1–250 μg), bovine serum albumin, or ovalbumin were preincubated with 0.5 μg of pepsin A or 5.0 μg of pepsin C in 0.9% NaCl in water in a total volume of 50 μl at 37°C for 15 min. After the preincubation, 0.1 ml of hemoglobin [0.25% labeled hemoglobin in 0.2 M sodium citrate (pH 2.0) or in 0.2 M sodium acetate (pH 4.5)] was added, and the incubation continued for 30 min. At 30 min, the reactions were terminated by addition of 20 μl of 1% bovine serum albumin and 0.23 ml of 10% trichloroacetic acid. After centrifugation (10,000 × g; 10 min), 0.2 ml of the supernatant solutions were removed and their content of ¹⁴C measured. Controls included reactions containing no enzyme (to provide background cpm) and ones without substrate. Inhibition assays on recombinant cathepsin D and recombinant cathepsin E were carried out by the same procedure in 0.2 M sodium acetate (pH 4.5) (24, 25).

Inhibition of Proteolysis of Bovine Serum Albumin.

Pepsin A (0.5 μg) in 50 μl of saline was incubated for 15 min in the presence or absence of purified ovUS (50 μg). Bovine serum albumin (0.45 ml of a 1% solution in 0.1 M HCl) was then added and the incubation continued at 37°C (26). Samples (75 μl) were removed at 0, 10, 20, and 30 min, proteolysis was stopped by addition of 10 μl of 6 M NaOH, and the digestion products were analyzed by electrophoresis in 15% polyacrylamide gels (27).

Affinity Chromatography of Sheep Uterine Secretions.

Affinity matrices (bovine serum albumin, ovalbumin, and porcine pepsin) were prepared by coupling 35 mg of each protein to 1.5 g of CNBr-activated Sepharose 4B (Pharmacia). Coupling efficiency was between 95 and 97%. The immobilized pepsin retained about 10% of its original activity. Glycine (0.2 M) was used for preparing the glycine-Sepharose column. The columns were equilibrated with phosphate-buffered saline (PBS; 10 mM, pH 7.4, and 0.15 M NaCl). Ovine uterine secretions that had been dialyzed against PBS at room temperature (≈1 mg of protein) were applied to the column, which was then washed with PBS until absorption at 280 nm was zero. Bound protein was eluted with sodium phosphate (50 mM, pH 7.4) containing 1 M NaCl. Eluted protein fractions (1 ml) were pooled, desalted, freeze-dried, and analyzed by electrophoresis.

Western Blot Analysis.

The proteins from SDS/PAGE gels were transferred to nitrocellulose filters and incubated with 5% nonfat dry milk in Tris-buffered saline (TBS, 10 mM, pH 8.0) containing 0.15 M NaCl and 0.05% Tween 20 for 2 h at 25°C. The blot was washed three times with TBS and incubated with rabbit antiserum to ovUS-1 (13, 14) at a dilution of 1:5000 in TBS for 4 h at 25°C. The blot was washed again with TBS and incubated with anti-rabbit immunoglobulin G conjugated to alkaline phosphatase (Promega) in TBS for 2 h at 25°C. After incubation with the second antibody, the blot was briefly washed with TBS and placed in a buffer suitable for display of alkaline phosphatase activity (100 mM Tris·HCl, pH 9.5/100 mM NaCl/5 mM MgCl₂) (22).

Microtiter Plate Binding Assay.

Proteins (1 μg in 0.2 ml of PBS) were allowed to bind to microtiter well surfaces for 12 h at 25°C. Remaining sites were blocked with 2% nonfat dry milk for 1 h. OvUS-1 (1 μg in 0.2 ml of PBS) was then allowed to bind to the adsorbed proteins in presence or absence of control proteins (bovine serum albumin or uteroferrin; 5 μg per well for 1 h). Bound ovUS-1 was detected by using the anti-ovUS-1 antiserum described above (1:10,000 dilution) followed by a second antibody (goat anti-rabbit immunoglobulin G conjugated to alkaline phosphatase). Bound enzyme was assayed by using p-nitrophenyl phosphate as substrate (Sigma).

OvUS Affinity Chromatography of Ovine Placental Secretions.

Secretory proteins were collected by in vitro incubation of explants prepared from day 100 sheep placenta (28). Purified ovUS-1 (25 mg) was coupled to CNBr-activated Sepharose 4B. About 1 mg of placental secretory proteins were passed over the ovUS-1-Sepharose column. Equilibration, washing, and elution of the affinity column were carried out as described earlier for the pepsin affinity column. Bovine serum albumin- and ovalbumin-Sepharose matrices were used as controls. Polypeptides in the flow-through and eluted fractions were analyzed by SDS/PAGE and Western blot analysis with a rabbit antiserum (diluted 1:1000) raised against recombinant bovine pregnancy-associated protein (PAG) 2 (29). The bands were visualized as described for Western blot analysis.

Determination of Amino-Terminal Sequences.

The protein samples were applied to Prospin cartridges (Applied Biosystems) and sequenced on an Applied Biosystems model 470 protein sequencer with on-line analysis for phenylthiohydantoin derivatives.

RESULTS

Identification and Cloning of a Second cDNA for Porcine Uterine Serpin (poUS-2).

When the porcine cDNA library was screened with a poUS-1 cDNA probe (9), several clones containing a related but clearly distinct cDNA (poUS-2) were identified (sequence not shown, GenBank accession no. X62845X62845). The poUS-2 cDNA was 1400 bp in length and longer than poUS-1 by 9 bases due to the presence of three additional codons that resulted in the insertion of a Leu, a Phe, and a Lys in the mature protein at positions, 111, 118, and 124, respectively. The 5′ and 3′ untranslated regions of poUS-1 and poUS-2 cDNA were identical in length (63 and 77 bases, respectively) and in sequence. Similarly, the primary structures of the two signal sequences and the predicted sites (30) for signal sequence cleavage (Cys⁻²⁶–Glu⁺¹) were the same. The predicted molecular weights of poUS-1 and -2 were 45,123 and 46,009, respectively, values that were close to that (M_r 45,000) of the major in vitro translation product of the poUS mRNA (8). Overall the two cDNA exhibited 97% identity in nucleotide sequence and 89% identity in inferred amino acid sequence. While poUS-1 possessed four sites for potential N-glycosylation (positions 107, 197, 243, and 315), poUS-2 lacked the one at Asn¹⁰⁷.

Identification and Cloning of a cDNA for Bovine Uterine Serpin.

The longest cDNA isolated was 1463 bases in length [excluding the poly(A) tail] and had short 5′ and 3′ untranslated regions (18 and 68 bases, respectively) flanking an open reading frame of 1377 bases (data not shown; GenBank accession no. L11627L11627). As has been observed with all the other uterine serpins (9, 14), the cDNA had two in-frame initiation codons (one beginning at base 19 and the second at base 34).

The boUS cDNA encoded a polypeptide of 459 amino acids and would include a 25-residue signal sequence. If the first in-frame methionine were used for initiation, the molecular weight of the mature protein would be 52,371. Unlike the other uterine serpins, there was only a single site for potential N-glycosylation (Asn²⁴³). This site was conserved in poUS-1, poUS-2, and ovUS-1.

Comparison of the Inferred Amino Acid Sequences of the Uterine Serpins.

As expected, bo- and ovUS-1 exhibit considerable amino acid sequence similarity (72%), presumably reflecting the relatively recent evolutionary divergence of cattle and sheep (18, 19). The differences between them and poUS-1 and -2 are more considerable. All the uterine serpins have closely similar signal sequences and predicted signal sequence cleavage sites. Several amino acid residues that are critical for maintaining serpin backbone structure of α₁-antitrypsin (31) are also present in uterine serpins in identical locations (data not shown). The basic architecture of serpins is highly conserved across evolutionarily diverse organisms (31). Therefore, it is possible to recognize the reactive center loop, the region that potentially interacts with proteinase, even in serpins that appear to lack inhibitory activities. In Fig. 1A, the putative reactive center loops of the uterine serpins have been aligned with those of six other serpins, three of which are known to be inhibitory and whose P₁–P′₁ bond has been defined. The uterine serpins possess the small conserved Thr and Ala residues found at positions P₁₄ and P₁₀ (relative to the P₁–P′₁ bond) found in the majority of inhibitory serpins. However, the loop regions of the uterine serpins are clearly unusual. All have an invariant KVP sequence, which is found in the propeptide of several aspartic proteinases, including pepsinogens A and C (33) (Fig. 1B). Similarly there is a conserved KEVPVVVK sequence downstream of the KVP motif, found in poUS-1 and -2 and in ovUS-1 (Fig. 1A). The VVVK portion of this motif aligns almost perfectly with the first 4 amino acids of pepstatin (isovaleryl-Val-Val-statine-Ala-statine; ref. 34). The same valine-rich motif is present in boUS-1 but is preceded by a 39-residue insertion, which incorporates three imperfect KEVPVVVK repeats. In addition, the VVKVP sequence repeated three times in the insert on boUS-1 is present in the propeptide sequences of several pepsinogens (Fig. 1B). The distal end of the reactive center loop is represented by the sequence RPF, which is part of a conserved motif, RPF(F/L)LFV that has been implicated in binding of serpin–proteinase complexes to clearance receptors (35).

Figure 1.

(A) Reactive center loop regions of the uterine serpins and of various inhibitory and noninhibitory serpins. The 39-residue insert sequence present in the reactive center loop of boUS-1 is shown separately to compare all serpin sequences. Conserved residues of uterine serpins are indicated by boldface type. The following three (human plasminogen activator inhibitor-2, PAI-2; human α₁-antitrypsin, huα₁-AT; and human antithrombin, AT III) are sequences of inhibitory serpins. Ovalbumin, human angiotensinogen (huAng); and corticosteroid binding globulin (rabbit CBG) are noninhibitory serpins. The arrowhead indicates the P₁-P′₁ peptide bond of the inhibitory serpins. (B) Alignment of propeptide amino acid sequences of aspartic proteinases from various species (32). The conserved KVP residues and VVKVP motifs are indicated by boldface type. These motifs (underlined sequences in A) are present in the reactive center loop region of uterine serpins. The sequences were obtained from Swiss-Prot data bank.

Inhibitory Activities of Uterine Secretions.

Uterine secretory proteins flushed from the uteri of cattle, sheep, and pigs at times when the uterus was under long-term progesterone maintenance were tested for their ability to inhibit porcine pepsin. Crude uterine flushings of cattle and sheep were able to inhibit pepsin activity in a dose-dependent manner (data not shown). Inhibitory activity was only noted in occasional porcine samples, was unstable, and was lost upon standing at 4°C for more than a day.

Freshly purified ovUS-1 was able to inhibit pepsin A and chymosin (pepsin C) activity at both pH 2.0 and 4.5 in the standard assay employing ¹⁴C-labeled hemoglobin as substrate (Fig. 2). Neither bovine serum albumin nor ovalbumin had any inhibitory activity. A 50% inhibition of 0.5 μg of pepsin A activity at pH 2.0 required about 20 μg, an approximately 35-fold molar excess of ovUS-1, whereas a 50% inhibition of pepsin C was achieved with about an 8-fold molar excess of ovUS-1. The ability of ovUS-1 to inhibit pepsin is also illustrated in Fig. 2C, where preincubation of pepsin (0.5 μg) with 50 μg of ovUS-1 effectively prevented proteolysis of albumin.

Figure 2.

Concentration-dependent inhibition of pepsin A and pepsin C activities by purified ovUS-1. (A) pepsin (0.5 μg) activity in presence of increasing concentrations of ovUS-1 at pH 2.0 (○) or pH 4.5 (•). Control protein (ovalbumin, □) gave no inhibition. (B) Same as A, except pepsin C (5 μg) was used as proteinase. (C) Electrophoretic analysis of the silver-stained products formed during proteolysis of bovine serum albumin by pepsin A in presence (Right) or absence (Left) of purified ovUS. Bovine serum albumin (4.5 mg) was incubated with pepsin A (0.5 μg) that had been preincubated for 15 min at 37°C in presence or absence of 50 μg of ovUS. Samples (20 μg of protein) were removed at specified time points and analyzed by SDS/PAGE.

The activity of purified ovUS-1 was unstable. Either dialysis or storage of the samples in 0.9% NaCl or PBS at 4°C for 48 h led to complete loss of antipepsin activity. Freeze–thawing was also deleterious. A visible precipitate of protein became evident in each stored sample. Therefore, dialysis and purification were carried out at room temperature, and freshly purified protein was used in all experiments. No inhibitory activity of uterine secretions was observed toward recombinant human cathepsin D, porcine cathepsin D, or recombinant cathepsin E. Activity of the cysteine proteinase cathepsin B was also unaffected (data not shown).

Association of Antipepsin Activity with Purified ovUS.

Sheep uterine flushings were passed through CM-cellulose at pH 8.2 to bind ovUS-1 selectively (11, 12, 13). The ovUS-1 was eluted with 0.5 M NaCl (12) and subjected to gel filtration on a Superose-12 column. Individual fractions were tested for their ability to inhibit pepsin (Fig. 3A). Antipepsin activity was associated with the small excluded volume peak, which was probably aggregated ovUS-1 (Fig. 3B, lane 2), and with the leading shoulder of the main peak (Fig. 3B, lane 3). The latter, containing the majority of inhibitory activity, provided a single band of protein of M_r 55,000 upon electrophoresis, while fractions within the center and trailing edge of the peak contained polypeptides of smaller size (Fig. 3B, lane 4). Amino acid sequencing of the total protein in each of the three peaks gave the identical sequence EKQQHS, which corresponded to the amino terminus of ovUS-1. No significant secondary signals were observed. Moreover, all protein bands within peak 3 that were detectable by silver staining also reacted with ovUS-1 antiserum on Western blots (data not shown). Thus, the pepsin inhibitory activity of ovine uterine secretion is associated with a specific subfraction of ovUS-1. The material in the trailing edge of the peak was presumed to be ovUS-1 that was partially denatured (11, 12), already cleaved (36, 37, 38, 39), or present in a stable but latent form, as has been observed for plasminogen activator inhibitor 1 (40).

Figure 3.

(A) Superose-12 gel filtration chromatography of ovUS-1 purified from ovine uterine secretions and the distribution of pepsin A inhibitory activity. The column was equilibrated and eluted with 0.15 M NaCl. Approximately 1.0 mg of protein was analyzed, and absorption at 280 nm in the eluant was followed. Arrows (I–IV) represent the elution positions of blue dextran, bovine serum albumin, ovalbumin, and cytochrome c. (B) Silver-stained SDS/PAGE analysis of peak protein fractions of Superose-12 chromatography. Lanes: 1, molecular weight markers; 2–4, 50-μl aliquots of peak protein fractions 1–3; 5, 10 μg of ovUS-1 before Superose-12 chromatography. (C) Silver-stained SDS/PAGE analysis of proteins purified from ovine uterine secretions by affinity chromatography on a pepsin A-Sepharose affinity column. Lanes: 1, molecular weight markers; 2, 2 μg of unfractionated uterine secretions; 3–6, protein fractions eluted from bovine serum albumin, ovalbumin, glycine, and pepsin affinity columns, respectively. Western blot analysis of protein eluted from pepsin affinity column developed with ovUS-1 antiserum is shown in lane 7.

Inhibitory serpins form 1:1 complexes with their partner proteinases, which are often so stable that they can be resolved electrophoretically in presence of SDS (41). OvUS-1 complexed with pepsin could not be detected in this manner (data not shown).

Pepsin Affinity Column Chromatography.

Sheep uterine flushings were passed over a pepsin affinity column. Electrophoretic analysis of the protein that bound revealed a single band (M_r ≈ 55,000) that was recognized by ovUS-1 antiserum (Fig. 3C, lane 7). The control affinity matrices did not retain any proteins under these conditions. The amino-terminal sequence of the eluted protein was that of ovUS-1 (EKQQHS). A second signal (TDNLLKV) of about two-thirds the intensity of the first was also detected and corresponded to a region within the reactive center loop (Fig. 1A). The cleaved C-terminal fragment appears to be stably associated with ovUS since there was no shift in the electrophoretic mobility of the eluted ovUS-1. However, it remains unclear whether the second peptide resulted from specific cleavage of the P₁–P′₁ bond (in which case P₁ would be a histidine residue) or from less specific proteolysis occurring after partial dissociation of the inhibitory complex. The reactive center loop of serpins is notoriously susceptible to nonspecific proteolytic cleavage (41).

Binding of ovUS-1 to PAG.

PAGs are among the major secretory products of the placenta of ungulate species (28, 29, 42, 43, 44) and, despite lacking evident proteolytic activity, belong to the aspartic proteinase gene family. To determine whether PAGs were possible partners for ovUS-1, their binding to the serpin was examined. Fig. 4A illustrates that both pepsin and proteins secreted by ovine placental explants were able to bind ovUS-1, whereas ovalbumin and bovine serum albumin were not. Electrophoretic analysis of the products from explant cultures that were selectively adsorbed by an ovUS-1 affinity column revealed several protein bands in the range of 70–50 kDa, typical of PAGs (28, 29, 43, 44), two of which cross-reacted with the anti-bovine PAG-2 antiserum (Fig. 4B).

Figure 4.

(A) Binding of ovUS-1 to proteins immobilized on surface of microtiter wells. Proteins used for coating on the surface of microtiter plate are printed vertically. Proteins used for binding are shown on the horizontal axis. (B) OvUS-1 specifically binds polypeptides of the aspartic proteinase family released by ovine placental explant cultures. Silver-stained SDS/PAGE analysis of proteins purified from ovine placental secretions by ovUS-1 affinity column. Lanes: 1, molecular weight markers; 2, 20 μg of placental secretory proteins; 3–5, eluates from bovine serum albumin, ovalbumin, and ovUS-1 affinity columns, respectively. Lanes 6 and 7 are Western blots of protein samples in lanes 2 and 5 developed with antiserum to recombinant bovine PAG-2 (44). Arrowheads indicate the protein bands that are cross-reacting with PAG antiserum.

DISCUSSION

The serpin family is so named because, among its members, it contains a range of serine proteinase inhibitors with a similar structural organization. However, earlier failures by this laboratory to demonstrate inhibitory activity of uterine serpins toward the commoner serine proteinases such as trypsin, chymotrypsin, and elastase were not unexpected because the reactive center loop region sequences of uterine serpins were atypical. The decision to examine whether uterine serpins could inhibit aspartic proteinases was prompted by three related observations. The first was the KVP motif, present in the inhibitory propeptide region (residues 4, 5, and 6) of pepsinogens (ref. 32; Fig. 1B). The second was a sequence (VVVK) just distal to KVP motif that can probably assume a configuration similar to that of pepstatin (45). Third, the VVKVP sequence, which is repeated three times in the insert of boUS-1 reactive loop region, is also found, with only minor variations, in most mammalian pepsinogens (44). Exactly how these amino acids contribute to pepsin inactivation, if indeed they do, is not clear, but the uterine serpins now provide, to our knowledge, the first known examples of inhibitors of aspartic proteinases in vertebrates, and it seems likely that the peptide regions discussed above occupy the substrate-binding cleft and prevent access of substrate. The inhibitor from the intestinal nematode Ascaris, which inhibits cathepsins E and D as well as pepsin, is not a serpin and the presumed inhibitory region possesses none of the amino acid motifs discussed above (46).

Serpin inhibitory activity is not confined to serine proteinases. A serpin coded by virulent pox viruses has been demonstrated to be an inhibitor of the cysteine proteinase that processes the precursor of interleukin β1 (47). Thus, the reactive center loop region, which is quite variable in sequence, may have become adapted for a range of very different specificities.

The structures of several serpins are now known through x-ray diffraction analysis (36, 37, 38, 39, 40, 48, 49), and there are now sufficient data to allow some serpin structures to be modeled (50) and the reactive center loops to be identified (see Fig. 1). The P₁₄ threonine, which is conserved in the uterine serpins, is the critical hinge residue of the reactive center loop of inhibitor serpins (40). When poised in the inhibitory conformation, the region containing this small amino acid is partially inserted into the β-sheet superstructure that makes up the bulk of the molecule (51). The mechanism of serine proteinase inhibition involves the formation of a covalent bond between inhibitor and proteinase (52), with hydrolysis of the peptide bond between the P₁ and P′₁ residues occurring only slowly (38). As cleavage occurs, the loop amino acids on the amino-terminal side of the cleavage site undergo a major conformational change as the segment pivots about the hinge and swings downward to become inserted in the body of the molecule (39, 40, 53). Between binding and subsequent release from the pepsin affinity column, the loop region of ovUS-1 had clearly become cleaved between His and Thr (Fig. 1), but it is unclear whether this bond corresponded to the P₁–P′₁ warhead site and whether strand insertion had occurred.

The amount of ovUS-1 needed to inhibit pepsin clearly exceeds that required to provide a 1:1 stoichiometry. This may be due to inactivation of the ovUS-1 by pepsin. Fig. 3A suggests that only a fraction of the ovUS-1 isolated from uterine secretions retains any inhibitory capability. A high proportion may already be cleaved (Fig. 3B). Serpins, in general, seem to be excellent substrates for their companion proteinase, most likely because the correct conformation leading to a stable 1:1 complex is not continually maintained (31, 41). In addition, conducting the proteinase assays at low pH probably shortened the half-life of the inhibitory complex. In any case, pepsin is unlikely to be a proteinase encountered by the uterine serpins.

Based on their apparent preference for aspartic proteinases and their site of synthesis, the most likely natural partners for the uterine serpins are the PAGs, a heterogeneous group of macromolecules secreted in abundance by the placentas of sheep, pig, and probably other ungulate species (28, 29, 42, 43, 44). Although related structurally to pepsin and possessing a functional peptide binding cleft, the majority of PAGs appear to be catalytically inactive (28, 29, 43, 44, 54). Why the mother would find it necessary to complex PAG with a serpin is unclear, but it may reflect her need to limit the activities of the trophoblast during pregnancy.