| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Regular Article |
a Departments of Animal Sciences
b Biochemistry, University of Missouri-Columbia, Columbia, Missouri 65211
ABSTRACT
A murine aspartic proteinase, described herein, is intermediate in amino acid sequence identity between the placentally produced pregnancy-associated glycoproteins (PAGs) and gastric pepsins. While PAGs are secreted products of placental trophoblast tissue of ungulates and most are not believed to function proteolytically, pepsins are digestive enzymes. The cDNA for this aspartic proteinase was amplified by reverse transcription-polymerase chain reaction from RNA extracted from murine placentas and neonatal stomachs. The open reading frame encoded a 387-amino acid polypeptide with a 15-residue signal sequence. The enzyme most resembled pepsinogen F (a protein identified in the stomachs of neonatal rabbits and rats) and PAG-like proteins cloned from equine and feline placentae. In the stomach, both its mRNA and protein were expressed in gastric chief cells of preweaned neonates. Within the placenta, its mRNA was present in both the parietal and visceral yolk sacs. However, the protein was most prevalent in the visceral yolk sac, with little detectable in the parietal yolk sac. The recombinant protein was expressed in Escherichia coli. This protein was capable of self-activation and exhibited proteolytic activity toward casein. The presence of this enzyme in two organs involved in the selective transcellular transport of proteins suggests that it has specialized digestive functions.
placenta, pregnancy
INTRODUCTION
Aspartic proteinases represent one of the four major proteinase families. They have been found in a broad range of organisms including retroviruses [1], fungi, plants [2], and vertebrates [3]. Eukaryotic aspartic proteinases share a common structure comprised of two symmetrical lobes [4]. Each lobe contains a conserved aspartic acid residue that mediates catalysis [5]. Most aspartic proteinases are maximally active at an acidic pH and are inhibited by pepstatin A [6, 7]. These enzymes are synthesized as an inactive precursor with an N-terminal propeptide that occupies the substrate binding cleft and is important for proper protein folding and inhibition of the zymogen [8–10].
Proteins related to the aspartic proteinase family have been detected in the plasma of pregnant cows and other ruminant species [11, 12]. These proteins are known collectively as pregnancy-associated glycoproteins (PAGs). The PAGs comprise a large group of proteins expressed exclusively in the chorionic epithelium of the placenta [13]. To date, 21 bovine [14, 15], nine ovine [14], 11 caprine [16, 17], and two porcine cDNA clones [18] have been identified in even-toed ungulate species. Some PAGs are enzymatically inactive due to mutations in conserved amino acid residues that comprise the catalytic center [14, 18]. However, computer modeling indicated that their structures are similar to those of active aspartic proteinases, such as pepsin [19]. These findings suggested that PAGs have retained the ability to bind ligands but not the ability to cleave them. Indeed, affinity chromatography with pepstatin A has been used to purify different classes of PAGs [20]. The function of these proteins is unclear.
Recently, PAG-like molecules have been characterized outside the Artiodactyla order. A protein, known as equine PAG (ePAG) was cloned from a Day 25 placental cDNA library [21]. The ePAG cDNA shared 65–70% identity with ruminant PAG and 76% with a related protein, feline PAG (fPAG), cloned by reverse transcription-polymerase chain reaction (RT-PCR) amplification from a Day 30 feline placenta (GenBank accession number AF036953). Unlike the diverse PAG family of the ruminant ungulates, ePAG is probably represented by only a single gene [21], and, in contrast to many of the bovine and ovine PAGs, it has conserved amino acid residues around the catalytic site. Equine PAG was also able to degrade denatured hemoglobin [21], suggesting that ePAG functions as a proteinase. Equine and fPAG are most similar in sequence to gastric aspartic proteinases expressed in the neonatal stomachs of the rabbit and rat, known as pepsinogen F or pepF (F representing fetal) [22, 23]. As a group, the pepF proteins more closely resemble PAG family members than other aspartic proteinases, including pepsin A and cathepsin D [21, 23]. Similar to the ruminant PAGs, the function of pepF is not known. The proteolytic activity of the pepF group suggests that they are functionally distinct from the ruminant PAGs.
The presence of PAG-like proteins in orders as diverse as Perissodactyla, Carnivora, Lagomorpha, and Rodentia suggests that members of this group are likely to be represented in all placental animals, including primates. We sought to determine whether there are PAG-like genes expressed in the placenta of the mouse, a species with a hemochorial placenta that is quite distinct from the placentae of the other two species (equids and felines) known to express PAG-like proteins. Furthermore, the mouse is an animal model in which the function of a gene product can be examined by genetic manipulation.
MATERIALS AND METHODS
Materials
Avian myeloblastosis virus (AMV) reverse transcriptase and restriction enzymes were purchased from Promega (Madison, WI). [
-32P]Deoxy-ATP, [-35S]deoxy-ATP, and UTP (1000–1500 Ci/mmol) were purchased from NEN Research Products (Boston, MA). Taq polymerase was obtained from Gibco BRL (Grand Island, NY). Sequencing kits were purchased from Epicentre (Madison, WI). Qiax II gel-purification kits were purchased from Qiagen (Valencia, CA). The pGEM-T easy vector and restriction enzymes were purchased from Promega. Magna Graph nylon membranes were purchased from Micron Separations Inc. (Westborough, MA). Paraformaldehyde was purchased from Fisher Scientific (Pittsburgh, PA). All other chemicals were purchased from Sigma Chemical Co. (St. Louis, MO) unless specified otherwise.
Animals
All mice used in these experiments were maintained and handled according to protocols approved by the animal care and use committee at the University of Missouri-Columbia, which follow the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Sexually mature cross-bred mice (129/SvEv; Jackson Labs, Bar Harbor, ME) were fed mouse chow formulation 5008 (Purina, St. Louis, MO) and water ad libitum. A 12L:12D cycle was maintained with lights on at 0700 h. The health of the mice was monitored and recorded daily. Males and females were paired together in the evenings, and the next day females were examined for the presence of a vaginal plug. The day in which a plug was observed was considered Day 0.5 of pregnancy. Some of the pregnant mice were allowed to deliver pups. These pups were weaned 21 days after birth.
Tissue Collection
Stomachs from different ages of mice, 18 days postcoitus (dpc), Days 1–30 after birth and adults (3–4 mo), were collected. For RNA isolation, stomachs were isolated (n = 10–15 for those from Postnatal Days 1–5; n = 5–10 for those from Postnatal Days 6–17; n = 2–5 for those after Postnatal Day 17), frozen in liquid nitrogen, and stored at -80°C. For histology, the stomach contents were removed, and the tissues fixed in 4% paraformaldehyde solution, dehydrated, and embedded in paraffin wax. For placental RNA isolation, extraembryonic membranes, including the yolk sac, were isolated from pregnant mice at different stages (11.5 dpc until term; 1-day intervals). Four to eight placentae from the same dam were isolated. For histological purposes, pregnant mice were infused via cardiac and uterine perfusion with 4% paraformaldehyde or Bouin fixative (Sigma). Whole pregnant uteri were isolated and separated into several segments, each of which contained an entire fetus and its associated membranes. These tissues were held in fixative for 12 h, dehydrated, embedded in paraffin, and sections (3–4 µm) were cut on a Microm microtome (Carl Zeiss Inc., Thornwood, NY).
RNA Isolation and RT-PCR
Total RNA samples were isolated from frozen tissues with STAT-60 (Tel-Test Inc., Friendswood, TX). The mRNA in 1 µg total RNA was reverse-transcribed to cDNA by AMV reverse transcriptase with a poly(dT) oligonucleotide as primer. Two sequences were selected from mouse expression sequence tags (ESTs) (accession numbers w30491 and w16236) because of their high identity with ePAG and rabbit pepF. The N-terminal primer, mpepFe1 (5'-GAAGGGATCATGAAGTGGCTC-3') and the C-terminal primer, mpepFe9r (5'-GGGCATGCCCCGCCATTC-3'), used for PCR amplification, represented sequences within the w30491 and w16236 EST, respectively. PCR amplification was performed (98°C, 10 sec; 48°C, 20 sec; 72°C, 30 sec; 35 cycles), and amplified products were separated by electophoresis on 1% agarose gels.
Gene Cloning, Sequencing, and Analysis
PCR products of the expected size (
1.2 kilobase pairs [kbp]) were purified by using a Qiax II gel-purification kit (Qiagen). Purified DNA was ligated into the pGEM-T easy vector (Promega) and transformed into competent Escherichia coli JM109 cells (Promega). Subclones were analyzed by EcoRI restriction digestion. Plasmids containing inserts of the expected size were subjected to complete bidirectional dideoxynucleotide sequencing with both vector and internal primers on an ABI 377 sequencer (DNA Core Facility, University of Missouri, Columbia).
The amino acid sequence, inferred from the open reading frame of the cDNA, was compared with those of other mammalian aspartic proteinases by using the Pileup Program (Genetics Computer Group [GCG], Madison, WI). All sequences were then assembled into a phylogenetic tree (GROWTREE program; GCG) based on amino acid sequence differences.
Northern Analysis
Total RNA (20 µg for each sample) was separated by electrophoresis on 1% formaldehyde-agarose gels. The RNA was transferred onto nylon membranes and immobilized by UV crosslinking. The full-length murine cDNA was labeled with [-32P]dATP by PCR. The nylon membranes containing total RNA were incubated with the cDNA probe in hybridization buffer (50% formamide, 5x saline-sodium citrate [1x SSC is 0.15 M NaCl and 0.015 M sodium citrate], 0.5% SDS, 5x Denhardt solution, 0.1 mg/ml herring sperm DNA) at 42°C [24]. The filters were washed three times for 20 min each with 0.1x SSC and 0.5% SDS at 45°C to remove nonspecific binding and exposed to film for 3–36 h, depending on the amount of radioactivity that remained bound to the blots. Each membrane was subsequently washed at 90°C and rehybridized to a murine ribosomal protein large fragment (RPL7) probe (250 bp) to account for signal differences due to RNA loading.
In Situ Hybridization
The procedures for in situ hybridization were performed as described elsewhere [25] with some modification. The pGEM-T easy vector (Promega) containing the full-length murine cDNA was linearized by either SacI for sense probes or by SacII for antisense probes. The antisense and sense 35S-labeled UTP cRNA probes were transcribed from linearized templates by using SP6 transcriptase for antisense and T7 transcriptase for sense probes. Paraffin wax was removed with xylene (Fisher Scientific) or Clearite (Richard Allen Scientific, Kalamazoo, MI), and the sections were rehydrated through decreasing concentrations of ethanol solutions. Prehybridization, hybridization, and washing conditions were standard [25]. After treating the slides with Kodak NTB-2 emulsion, the slides were exposed for either 6 days (for stomach sections) or 15 days (for placenta and yolk sac sections) at 4°C. The slides were developed and counterstained with Harris hematoxylin (Fisher Scientific) and eosin (Fisher Scientific). At least four different tissue samples for stomachs and placentae at each stage were examined.
The murine aspartic proteinase mRNA was also localized by using a nonradioactive in situ hybridization technique. The pGEM-T easy vector (Promega) containing the aspartic proteinase cDNA was used to generate both antisense and sense probes, as described previously. However, for this particular method, the probes were labeled with biotin-16-UTP (Roche Molecular Biochemicals, Indianapolis, IN) instead of [35S]UTP. Pre- and posthybridizations were performed as described previously [25]. Regions of the stomach and yolk sac that hybridized to the biotinylated probes were detected by strepavidin alkaline phosphatase followed by the chromogens, 5-bromo-4-chloro-3-inoyl phosphate (5-BCIP) and nitroblue tetrazolium (NBT) (Dako Corp., Carpinteria, CA). The slides were counterstained with nuclear fast red (Fisher Scientific).
Immunohistochemistry
Immunohistochemistry was performed as described previously [26]. The sections were deparaffinized, rehydrated, and treated with 3% hydrogen peroxide:methanol to quench endogenous peroxidase activity. Normal goat serum was used to reduce nonspecific binding. Western blotting of stomach tissue proteins demonstrated that the ePAG antiserum [21] recognized this murine PAG-like aspartic proteinase. Serial dilutions of primary antiserum were used to obtain the optimal dilution for antigen recognition in the sections. Sections were incubated overnight at 4°C with either anti-ePAG serum, anti-ePAG serum that had been preadsorbed with recombinant ePAG, or with nonimmunized rabbit antiserum. The following day, slides were washed three times (5 min each) with Tris-buffered saline. The secondary anti-rabbit IgG antibody, followed by avidin and biotin (Dako Corp.) were incubated on the tissues for 30 min each. The chromagen 3,3'-diaminobenzidine (Dako Corp.) was used to detect binding. The slides were counterstained with Harris hematoxylin for 30 sec. At least three slides for each specimen were used. The procedure was repeated at least once for each stage of development.
SDS-PAGE and Western Blotting of Gastric Extracts
Neonatal (2 days old, n = 12) and adult (>3 mo, n = 2) mouse stomachs were homogenized in 20 mM Tris, pH 8.0, 2 mM EDTA, 40 mM NaCl, 0.02 mM PMSF, 0.02% NaN3. The extracts were centrifuged at 5000 x g to remove debris and passed through a 0.2-µm filter. Foal glandular stomach (obtained from a necropsy performed at the University of Missouri Veterinary Diagnostic Laboratory) was extracted and processed in the same manner. The soluble proteins (5 µg) in the extracts were separated by electrophoresis through 12.5% SDS-polyacrylamide gels and either stained by Gel-Code Blue (Pierce, Rockford, IL) or electrophoretically transferred to nitrocellulose membranes. For the Western blots, the membranes were blocked in 2% BSA, 1% nonfat dry milk in 0.1 M NaHCO3, pH 9.0, reacted with ePAG antiserum (1:1000) or preimmune rabbit serum (1:1000), washed with TBS-Tween 20 (0.05%) and reacted with alkaline-phosphatase-conjugated anti-rabbit IgG (1:2000) (Promega). The blots were washed and stained in a mixture of NBT and 5-BCIP (Promega).
Expression of the Recombinant Protein
The primers, mpepFpET11a-f (5'-GCCCATATGTTGGTCAAAATCCCTCTG-3') and mpepFpET11a-r (5'-GGCCATATGCTATCATGCAGCAGGAGCCAG-3'), were designed for cloning the cDNA into the pET11a expression vector (Novagen, Madison, WI). The PCR product, minus the signal peptide sequence, was digested with NdeI, subcloned into the corresponding cloning site in the pET11a vector, and the entire cDNA insert was sequenced. This construct was transformed into BL21(DE3)pLysS cells. Expression was induced at midlog phase by adding isopropylthiogalactoside (1 mM). The cells were harvested by centrifugation 6 h after induction and resuspended in 1x PBS. The cells were lysed by passage through a French pressure cell, and inclusion bodies were separated from soluble materials by centrifugation. The inclusion bodies were solubilized with 5 M guanidine, 1 mM glycine, 20 mM Tris, and 1 mM EDTA, pH 7.5, and the presence of recombinant protein confirmed by SDS-PAGE.
Solubilized, recombinant protein was refolded as described previously [27] with some minor modifications. The solution was first diluted to 1 mg/ml in 6 M urea, 1 mM glycine, 20 mM Tris, and 1 mM EDTA, pH 7.5. This solution was then added dropwise to 100 volumes of 20 mM Tris, 1 mM EDTA, 1 mM glycine, 2 mM PMSF, and 5% glycerol, pH 9.5, with rapid stirring. After 10 min at room temperature, the pH was adjusted to 7.5 with 1 M HCl and held overnight at 4°C. This solution was concentrated in an ultrafiltration cell to 1/20 the original volume with a YM30 membrane (Amicon, Beverly, MA). Precipitated proteins were removed by centrifugation. The presence of refolded protein was confirmed by SDS-PAGE.
Proteolytic Activity Against Caseins
A turbidimetric milk-clotting assay was performed to determine if this murine aspartic proteinase was proteolytically active. The procedure was performed as described previously [28] with some modifications. Fresh skim milk was purchased from a local grocery store. A stock solution (0.2 M sodium acetate, 0.01 M calcium chloride, 2% skim milk, pH 5.3) was prepared for each assay. Spectrophotometric measurements were performed in 1 ml of reaction mixture at a wavelength of 480 nm in a plastic cuvette. The milk solution without enzyme was used as the blank. Samples tested included pepsin, pepsin plus pepstatin A, the recombinant murine aspartic proteinase described herein, recombinant enzyme plus pepstatin A, recombinant enzyme after low pH treatment, and recombinant enzyme (low pH treated) plus pepstatin A. Buffer alone and BL21 bacterial lysate were used as additional negative controls. The reaction was performed at 28.5°C and the absorbance was measured every 15 min. Each assay was performed three times in duplicate.
RESULTS
Cloning of the Murine Aspartic Proteinase cDNA
In order to identify PAG-like proteins in the mouse, the mouse EST database (National Center for Biotechnology, Bethesda, MD) was searched for close identity with ePAG [21] and rabbit pepF [22]. Two EST sequences (w30491 and w16236) were found to have high identity with the 5' and the 3' ends of both ePAG and rabbit pepF cDNA. Primers based on these sequences were used in RT-PCR with RNA extracted from placentas (16 dpc) and neonatal stomachs (1.5 days after birth). An 1.2-kbp product was purified and subcloned from each tissue. Partial sequencing of nine subclones indicated that they were all identical where the sequences overlapped. Two of the clones were sequenced fully in both directions. The putative start codon ATG was determined by sequence comparisons to known aspartic proteinases [21]. The putative start codon was not flanked by the usual Kozak consensus sequences [29], but the anticipated adenine located at the -3 position from the 5' end of the proposed start site was present. On this basis, the open reading frame of the cloned cDNA consisted of 1164 bp and encoded a polypeptide of 387 amino acids with a theoretical molecular weight of 42 787 (Fig. 1). The putative position for signal peptide cleavage was inferred by comparison with PAG [13], pepsinogen A [30], and from the criteria described by von Heijne [31]. Cleavage was predicted to be 15 amino acids from the start site (Fig. 1). No sites for potential N-glycosylation were present within the entire sequence.
|
The deduced protein possessed the hallmarks of enzymatically active eukaryotic aspartic proteinases, including the presence of a putative propeptide and the hydrophobic-hydrophobic-D-T-G-T/S-T/S consensus sequences around the two catalytic aspartic acid residues located at positions 32 and 215 (porcine pepsin numbering). The hydrophobic-hydrophobic-G residues that comprise the psi loop [5] in each lobe were conserved, as were the tyrosine-75 and glycine-76 residues (porcine pepsin numbering) within the flap region [5] (Fig. 1).
Sequence Identity to Related Proteins
The cloned murine aspartic proteinase shared 94% and 77% nucleotide sequence identity, and 93% and 75% amino acid sequence identity, with rat and rabbit pepF, respectively [22, 23]. There was also high sequence conservation with ePAG and fPAG (79% nucleotide; 71% amino acid), two proteins that are expressed in the placenta [21]. Sequence identities with PAG from sheep and cattle were more distant and about the same as that for pepsinogens. An unrooted neighbor-joining tree (GCG), based on pairwise amino acid sequence comparisons representing a range of mammalian aspartic proteinases, confirmed the close relationship of ePAG, fPAG, rabbit pepF, rat pepF, and the mouse clone described herein. For simplicity, the aspartic proteinase from the mouse will be referred to as murine pepF (mpepF) throughout the remainder of the text. Together, these five proteins formed a separate cluster intermediate between the PAG and other aspartic proteinases (Fig. 2).
|
Northern Analysis of mpepF mRNA Expression in the Mouse
Kageyama et al. [22, 23] determined that the expression of rabbit and rat pepF in the neonatal stomach was elevated immediately after birth but was later replaced by mature pepsinogens. To determine if mpepF had a similar pattern of expression, Northern blotting was performed upon RNA extracted from murine stomachs at different stages of development. Expression was maximal during the first 2 wk after birth (Fig. 3A). A sharp decrease in expression was observed between Days 17.5 and 20.5, a period that corresponded to when the pups were weaned. Expression was consistently low in the adult stomach (Fig. 3A).
|
Apparent orthologs of mpepF are expressed in placentae of the horse [21] and cat [32]. Therefore, the expression of mpepF in the placenta was characterized by Northern blotting with RNA samples extracted from whole placentae, including yolk sac, from Day 11.5 of gestation until term. Expression was low at Days 11.5 and 12.5 of gestation, but increased at Day 13.5 and remained constant from Day 14.5 to term (Fig. 3B).
The presence of mpepF in other tissues was assessed by Northern blotting of RNA from a wide sampling of adult (3 mo old) and fetal tissues that included brain, heart, intestine, kidney, liver, lung, muscle, ovary, prostate, skin, spleen, testis, and uterus (data not shown). Murine pepF mRNA was not detectable in any of the aforementioned tissues.
Localization of mpepF mRNA Expression by In Situ Hybridization
Because mpepF was restricted to neonatal stomach and placenta, the spatial localization of mpepF mRNA in these two organs was assessed. Two methods were used for in situ hybridization, a radioactive procedure carried out on frozen sections of tissue that allowed low magnification scans to be made under darkfield illumination and a nonisotopic procedure on tissues fixed either with Bouin or paraformaldehyde fixative before sectioning. The latter method provided improved histological preservation of tissues, particularly of the fragile yolk sac tissue.
Both procedures gave comparable results in sections of stomach. At Day 4.5 after birth, a strong signal was present in the chief cells of the stomach, the same cells known to produce adult pepsinogens (Fig. 4, A–D) [33]. Other cell types of the stomach, including the stratified squamous epithelium lining the nonglandular regions (Fig. 4A) and the smooth muscle of the muscularis mucosa and muscular externa (Fig. 4C) contained no detectable mRNA for mpepF. Only background signal was present in sections hybridized to the sense probe (data not shown).
|
In situ hybridization was also used to examine the spatial expression pattern of mpepF mRNA in the placenta of the mouse (Fig. 4, E–H). In the first series of experiments in which a radioactive probe was used on frozen sections, it was apparent that at Day 16.5 of gestation, transcripts for mpepF were absent from the majority of placental tissues, including the trophoblast giant cells and other trophoblast derivatives. Silver grains were instead localized to the intraplacental yolk sac (sinus of Duval) and the parietal yolk sac adjacent to Reichert's membrane. Figure 4, E and F, represents brightfield and darkfield images, respectively, of a frozen section from a Day 16.5 placenta showing the sinus of Duval, which is a cavity formed by the confluence of the visceral and parietal yolk sac membranes (see Fig. 5). In Figure 4, E and F, the sinus of Duval (within the chorioallantoic placenta), parietal and visceral endoderm layers can be observed in the same panel (see Fig. 5 insert for orientation). Minimal to no label was associated with the visceral yolk sac on the opposing side of the yolk sac cavity. Transcripts for mpepF could be detected in parietal yolk sac endoderm as early as Day 9.5, with signal intensity increasing as pregnancy proceeded (data not shown).
|
Nonradioactive in situ hybridization experiments conducted on fixed specimens showed that although the hybridization signal for the presence of mpepF mRNA was present in the parietal cells, there was also signal in the visceral endoderm of the yolk sac cavity in the samples from late pregnant mice (Fig. 4G). The parietal and visceral yolk sac endodermal cells lining the yolk sac cavity are demarcated with arrows. We have no good explanation for the discrepancy between the two procedures. Possibly the nonradioactive method was more sensitive or chemical fixation provided better preservation of the mRNA in the visceral endoderm. To help orient the reader, a drawing is provided showing the relative positions of the extraembryonic membranes of the rodent yolk sac placenta (Fig. 5).
Localization of mpepF by Immunolocalization
Immunohistochemistry was performed with a polyclonal antiserum raised in rabbits against recombinant ePAG (the equine ortholog of mpepF) in order to determine whether the distribution of the mpepF protein mirrored that of the mRNA (Fig. 6, A–F).
|
Immunohistochemical staining in the stomach was restricted to the glandular chief cells, consistent with the results from in situ hybridization (Fig. 6A and insert). Preimmune serum or serum that had been adsorbed with recombinant ePAG protein did not generate any specific staining (Fig. 6, B and C).
When immunostaining was used to localize mpepF in extraembryonic membranes, the strongest signal was not, as expected, in the parietal yolk sac or the sinus of Duval. Rather, mpepF protein was localized predominantly to the visceral yolk sac (Fig. 6D). The region shown in these sections corresponds to the upper region of the yolk sac nearest the allantochorionic placental membranes (Fig. 5) where the villous-like folds of the absorptive visceral endoderm are most evident. The controls, which employed both a preimmune serum (Fig. 6, B and E) and anti-ePAG that had been preabsorbed with ePAG recombinant protein (Fig. 6, C and F), exhibited no specific staining. The immunostaining results contrasted with those observed by in situ hybridization where expression was predominantly in the parietal endoderm.
Western blotting revealed that this antiserum was able to recognize both recombinant mpepF protein and two protein bands (a faint band of 40 000 molecular weight and a predominant band of 35 000 molecular weight) in extracts of neonatal mouse stomach (Fig. 6H, lane 1). Presumably, these immunoreactive bands represented the zymogen and mature form (lacking the propeptide) of native mpepF [10]. A similar pattern of staining is observed in proteins extracted from foal stomach (Fig. 6H, lane 4). The antiserum did not recognize any proteins extracted from adult mouse stomach (Fig. 6H, lane 2).
Proteolytic Activity of the mpepF
Some PAG molecules are proteolytically inactive because of mutations in critical residues related to the catalytic mechanism [13–15, 19]. However, the equine ortholog of mpepF, ePAG, is an active proteinase [21]. No enzymatic activity has as of yet been attributed to rabbit or rat pepF [22, 23]. To determine if mpepF is enzymatically active, sensitive milk-clotting assays were performed with recombinant mpepF produced in E. coli. The expressed product was deposited as inclusion bodies. After solubilization, the protein was refolded by rapid dilution and concentrated. Refolded mpepF showed proteolytic activity against casein. A representative assay is shown in Figure 7. This activity could be inhibited by pepstatin A, a specific inhibitor of aspartic proteinases. Preincubation of mpepF at an acidic pH (4.5) resulted in higher activity in the assay than mpepF that had not been preincubated (Fig. 7). Presumably, this increase in activity was due to cleavage of the propeptide in a manner similar to that for other aspartic proteinases [10, 21, 34].
|
DISCUSSION
This paper describes an aspartic proteinase (mpepF) in the mouse that is expressed exclusively in the placental yolk sac and the neonatal stomach. Murine pepF expression in the stomach is limited to a few days before birth (data not shown) and to preweaned pups. A structurally and antigenically related protein has been described in the horse (ePAG) [21], where it was found associated with the outer layer (chorion) of the placenta. Equine PAG has also been detected in the neonatal stomach of foals (Fig. 6H, lane 4). A somewhat similar aspartic proteinase has been cloned from the placenta of the cat (fPAG), but no systematic study has been made of its expression [21, 30].
Equine PAG and fPAG were initially described as belonging to the PAG grouping because of their placental expression and slightly closer sequence identity to the placental PAGs than to pepsinogen A [21]. The clustering within the phylogenetic tree of the equine and feline proteins, the murine aspartic proteinase described here, and pepF from the rabbit and rat (Fig. 2) suggest that the five are orthologs and are distinct from the PAGs. Although the designation pepF is not satisfactory, as it implies that the protein is exclusively expressed during the fetal stage and restricted to the gastric mucosa, it would seem to be preferable to PAG as the protein is neither restricted to pregnancy nor apparently glycosylated (data not shown). For now, we will refer to the murine enzyme as pepF to indicate its relationship to the rabbit and rat enzymes. Presumably, these enzymes will be given a more appropriate name once a function has been established.
The presence of the pepF aspartic proteinase in four distinct orders (Perrisodactyla, Carnivora, Lagomorpha, and Rodentia) that diverged at least 90 million yr ago [35] is suggestive that the pepF gene is present in all eutherians, including primates. Recent calculations showed that the PAG had their origins approximately 85 million yr ago [36], around the time the lineage leading to the modern-day artiodactyl species diverged from other hoofed mammals [35]. It seems likely that the PAG arose from a pepF gene by a duplication event around that time [36]. Whereas, the pepF group has retained considerable structural conservation and appears not to have undergone further rounds of duplication, the PAG duplicated extensively and seem to be undergoing rapid functional diversification, including loss of proteolytic activity. It is curious that ePAG and fPAG, along with all the ruminant PAG are expressed in trophectoderm, while in the mouse, placental expression is confined to yolk sac, a tissue of endodermal origin. In this regard, it was recently demonstrated that transgenic mice carrying the hCG-ß locus express the chorionic gonadotropin-ß subunit in parietal yolk sac, not in trophoblast derivatives [37].
The data presented here show that murine pepF is an active proteinase that is expressed transiently in the stomach and yolk sac. Notably, after rodent and rabbit pups are weaned, pepF mRNA expression is downregulated, and other adult-type pepsinogens predominate [22, 23]. The switching of expression of pepsinogens during development is not unusual. In ungulates [38–41] and carnivores [42], chymosin is a major neonatal gastric enzyme that is expressed for a short time after birth. Recently, chymosin has been cloned from neonatal rat stomach [23] and chymosin-like ESTs from the mouse have been deposited into Genbank (AI326975, AI322423, AI324867), suggesting that, like the ungulates, rodents also require this enzyme. Possibly in rodents, pepF and chymosin are both involved in fetal and neonatal nutrient acquisition.
The presence of pepF in the neonatal stomach and the placental yolk sac of the mouse suggests a common requirement of both tissues for this enzyme. Indeed, there are functional similarities between the gastrointestinal tract (particularly that of the newborn) and the rodent placental yolk sac. Both tissues are involved in protein digestion and both appear to spare immunoglobulins from destruction [41, 43]. One testable hypothesis is that pepF functions in both tissues to selectively degrade proteins to produce small, digestable peptides, while having little activity toward maternal immunoglobulins and certain other resistant proteins [40, 44]. The young of many mammalian species, including cattle, horses, and rodents, are passively immunized by maternal immunoglobulins present in colostrum [41, 45]. These immunoglobulins are able to escape destruction in the stomach selectively and are absorbed through the wall of the small intestine [41, 45–47]. Progeny that do not receive these immunoglobulins are more likely to succumb to deadly pathogens. During the second half of pregnancy in rodents (but not ungulates), maternal immunoglobulins are also transferred across the visceral yolk sac to the fetus, while accompanying proteins are degraded [48]. The production of pepF by the parietal yolk sac places it in a position to process proteins proteolytically in the yolk sac cavity. Presumably, the destruction of immunoglobulins is avoided because of the specificity of this enzyme. The results in Figure 6D suggest that pepF accompanies endocytosed materials into the visceral endoderm [49]. Indeed, the environment (lower pH) of the endocytotic vesicles and secondary lysosomes is probably more conducive to the enzymatic activity of pepF as aspartic proteinases tend to function optimally in acidic conditions. Currently, experiments are ongoing to determine if mpepF exhibits selective proteolysis of protein substrates in order to test the hypothesis that it minimally degrades immunoglobulins, so as not to interfere with the passive immunization of the fetus or neonate.
In conclusion, mpepF is a distinct member of the aspartic proteinase family and is expressed solely in the placental yolk sac and neonatal stomach. Unlike many of the ruminant placentally expressed PAGs, pepF is proteolytically active. Consequently, this placental and gastric protein probably has a function distinct from the ruminant PAG. In both organs it may function as a specialized digestive enzyme, sparing some proteins, while degrading others.
ACKNOWLEDGMENTS
The authors thank Jim Bixby for his help in comparing the aspartic proteinase sequences and generating the neighbor-joining tree, Dr. Jiazhong Liu (ABC Laboratory, Columbia, MO) for his help on in situ hybridization, Melissa Larson and Dr. Xin-an Pu for breeding and maintaining mice, personnel in the Histology Laboratory in the Department of Veterinary Pathobiology for their help in sectioning the tissue, Jessica Wagner and Shawn Bailes at the University of Missouri Cytology Core for their help in compiling the images and drawing the diagram of the murine placental yolk sac, and Carrie Neville for her help in the preparation of the manuscript.
FOOTNOTES
First decision: 12 February 2001.
1 This work was supported by the U.S. Department of Agriculture National Research Initiative grant 96-35203-3257. 
2 Correspondence: Jonathan A. Green, University of Missouri, 158 ASRC, 920 East Campus Drive, Columbia, MO 65211. FAX: 573 882 6827; greenjo{at}missouri.edu
Accepted: May 22, 2001.
Received: January 17, 2001.
REFERENCES
-hydroxylase, and cytochrome P450 aromatase in bovine follicle during the first follicular wave. Endocrinology 1995; 136:981-989[Abstract]This article has been cited by other articles:
|
|
 |
|
  A. L. Hughes, J. A. Green, H. Piontkivska, and R. M. Roberts Aspartic Proteinase Phylogeny and the Origin of Pregnancy-Associated Glycoproteins Mol. Biol. Evol., November 1, 2003; 20(11): 1940 - 1945. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. S. Rosenfeld, C.-S. Han, A. P. Alexenko, T. E. Spencer, and R. M. Roberts Expression of Interferon Receptor Subunits, IFNAR1 and IFNAR2, in the Ovine Uterus Biol Reprod, September 1, 2002; 67(3): 847 - 853. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
