| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Regular Article |
a Department of Animal Science, University of Wyoming, Laramie, Wyoming 82071
ABSTRACT
An interferon (IFN)-stimulated gene (ISG) encodes a bovine 17-kDa protein (bISG17) that is released from endometrial cells but also conjugates to intracellular proteins through a ubiquitinlike mechanism. During early pregnancy in ruminants, conceptus-derived IFN-
induces endometrial ISG17. The present experiments were designed to generate bioactive recombinant (r) bISG17. The Pichia pastoris yeast expression system was used because previous experiments expressing the human ISG15 ortholog in bacteria were confounded by inherent carboxypeptidase activity that cleaved C-terminal residues resulting in an inactive protein. In a series of extensive yeast culture experiments using shaker-bath and fermentation approaches, optimal conditions were determined for a transformant containing a multi-ISG17 gene insertion. Recombinant bISG17 was purified. Carboxy-terminal sequencing revealed that rbISG17 retained the C-terminal Gly that is potentially critical for the first step in covalent attachment to targeted intracellular proteins. The rISG17 induced (P < 0.0001) IFN-
mRNA (reverse transcription–polymerase chain reaction) and release of IFN- protein (ELISA) by bovine peripheral blood mononuclear cells. The IFN- mRNA also was upregulated (P < 0.0001) in endometrium from pregnant (Day 18) when compared with nonpregnant (Days 14 and 18) cows. It is concluded that rbISG17 generated in a yeast expression system retains cytokine/hormonal activity. This is the first description coupling the biology of two distinct IFNs ( and ) through the intermediary ubiquitin homolog ISG17.
pregnancy, uterus
INTRODUCTION
Bovine (b) ISG17, a 17-kDa protein encoded by an interferon (IFN)-stimulated gene is produced by the endometrium in response to conceptus-derived IFN- between Days 15 and 26 of the peri-implantation period [1]. Bovine ISG17 is found as an intracellular protein covalently linked to a subset of targeted proteins [2] and as a released product in the histotroph [1]. The bISG17 cDNA [3] and gene [4] have been isolated and sequenced. Bovine ISG17 shares 31% amino acid sequence identity with a tandem ubiquitin repeat and 68% amino acid sequence identity with the human (h) ortholog ISG15. The bISG17 gene promoter contains five IFN-stimulated response elements (ISRE) that are common to other type 1 (
, ß, 
, ,) IFN-inducible genes [4]. Western blot analysis and electrophoretic mobility shift assay were used to study bISG17 gene expression. Interferon- induces phosphorylation of signal transducers and activators of transcription, which in turn induce transcription of IFN regulatory factor genes [4]. Interferon regulatory factor-1 directly interacts with an ISRE within the bISG17 gene promoter [4].
The most functionally significant amino acid identity between bISG17, ubiquitin, and other ubiquitin-like proteins is the carboxyl terminal Leu-Arg-Gly-Gly motif [5]. This motif is required for covalent conjugation of ubiquitin and related proteins to the 
amino group of lysine residues on targeted intracellular proteins [6, 7]. Conjugation of bISG17 to intracellular proteins is distinct from ubiquitylation during bovine pregnancy and in response to IFN- [2].
Bovine ISG17 mRNA and protein are first detected on Day 15 of pregnancy, increase to highest levels on Day 18 of pregnancy, then decline to low levels by Day 26 of pregnancy [1, 8]. Low levels of bISG17 mRNA and negligible levels of protein are found in endometrium from nonpregnant cows. The bISG17 mRNA [9] and protein [1, 2, 10] are induced by recombinant (r) bIFN- in cultured endometrial explants from nonpregnant cows and primary bovine endometrial (BEND) cells. Bovine ISG17 mRNA [9] and protein are induced 3 h after treatment with rbIFN-. Conjugation of bISG17 to intracellular proteins also occurs within 3 h following induction by rbIFN- in BEND cells (unpublished result). Bovine ISG17 mRNA is localized to glandular epithelium and stromal cells on Day 18 of pregnancy but is absent or in very low amounts in endometrial sections from nonpregnant cows [9].
Because bISG17 was found in uterine flushings from Day 18 pregnant cows, it is likely that an extracellular role exists for this protein. Human ISG15 is released in response to type 1 IFNs [11] and induces the secretion of IFN- by T lymphocytes [12]. Human ISG15 is a unique immunomodulatory molecule of the cytokine cascade in that it is the first protein shown to be secreted in response to IFN- or IFN-ß in peripheral blood mononuclear cells (PBMCs) [13], induces the production of IFN- from T cells, augments natural killer (NK) cell proliferation, activates nonmajor histocompatibility complex-restricted cytolytic lymphocytes, and activates monocytes and macrophages via the induction of IFN- [12].
The methylotrophic yeast Pichia pastoris is extensively used for heterologous protein expression as an alternative to Saccharomyces cerevisiae protein expression systems. [14]. Complementary expression vectors are designed to be inserted within or to replace the alcohol oxidase I (AOXI) gene. Protein expression is driven by the inducer methanol, which is normally utilized as a carbon source. This system has the advantages of growth to high cell density, proper protein folding, and secretion of heterologous proteins over bacterial expression systems. It also minimizes hyperglycosylation, an expression artifact of S. cerevisiae [15]. Additionally, P. pastoris can be grown on an industrial scale in a very simple, defined medium, presumably making purification of secreted recombinant proteins easy and efficient. Here we report for the first time the expression, secretion, purification, and biological activity of rbISG17, formerly known as ubiquitin cross-reactive protein [1]. This protein will provide an invaluable tool to investigate the physiological and pharmacological functions of bISG17 during pregnancy. It was hypothesized that rbISG17 would induce IFN- mRNA and protein expression by PBMCs and that the induction of IFN- could be used as a functional bioassay. Because bISG17 is identified in uterine endometrium and flushings from pregnant cows and other ubiquitin-like proteins are components of the cytokine cascade, we hypothesized that IFN- would be upregulated or induced in the endometrium of pregnant when compared to nonpregnant cows.
MATERIALS AND METHODS
Materials
All enzymes and kits were used according to manufacturer's specifications. Restriction endonucleases, T4 DNA ligase, and Vent DNA polymerase (New England Biolabs, Beverly, MA), Random Primers DNA Labeling System, DH5 Escherichia coli, and streptavidin–alkaline phosphatase (GIBCO-BRL, Rockville, MD), Sequenase v 2.0 sequencing kit, -[32P]dCTP, reagents for purification of protein (Amersham Pharmacia Biotech, Piscataway, NJ), Zero Blunt PCR Cloning kit, P. pastoris GS115 yeast strain, zeocin, yeast nitrogen base, biotin, and the pPICZA expression vector (Invitrogen, Carlsbad, CA) were purchased from commercial sources. All other reagents and chemicals were obtained from Sigma Chemical Co. (St. Louis, MO) or Fisher Scientific (Pittsburgh, PA).
Oligonucleotides were synthesized at the DNA core facility at the University of Missouri, Columbia. Carboxyl terminal sequencing [16] of rbISG17 was completed at the Macromolecular Structure Facility in the Department of Biochemistry at Michigan State University (Dr. Joseph Leykam, East Lansing, MI). Fermentation of yeast was completed at the University of Nebraska Fermentation Facility (Mr. Brad Plantz, Lincoln, NE).
Construction of the pPRU14 Expression Vector
The bISG17 coding sequence was amplified from the KA16 plasmid [3] using polymerase chain reaction (PCR) and the following oligonucleotide primers: 5'-GAATTCATGGGCGGGGACCTGACG and 5'-GGTACCCTACCCACCCCGAAGACGT. Denoted in bold lettering are EcoRI (5') and KpnI (3') restriction sites engineered in the primers to allow directional subcloning of the bISG17 coding sequence into the expression vector. These restriction sites were necessary for in-frame integration into the pPICZA expression vector. The PCR used 20 pmoles of each primer and 1 ng of KA16 in reaction buffer containing 200 µM dNTPs and 0.8 units of Vent DNA polymerase (30 cycles of 90°C for 1 min, 72°C for 2 min). Amplified cDNA was purified by 1% agarose gel electrophoresis and electroelution, and blunt-end ligated into the pCR blunt cloning vector. Cloning steps were confirmed by the Sanger dideoxy sequencing method. This cloning strategy incorporated two codons encoding the amino acids glutamic acid and phenylalanine at the amino-terminus that were lacking in native bISG17. The bISG17 coding sequence was released from the pCR Blunt cloning vector by EcoRI/KpnI digestion and subcloned into the pPICZA expression vector. Engineered junctions were verified to be in frame by DNA sequencing. The pCR Blunt cloning vector was amplified in One Shot Top10 E. coli cells and the pPICZA and pPRU14 (pPICZA containing the bISG17 coding sequence) expression vectors were amplified in DH5 E. coli cells.
Transformation of P. pastoris with pPICZA and pPRU14
The GS115 strain of P. pastoris was transformed with pPICZA or pPRU14 according to manufacturer's specifications. Positive transformants were selected on YPD (1% yeast extract, 2% peptone, and 1% dextrose) plates containing 100 µg/ml zeocin after 2 days at 30°C and confirmed to contain the bISG17 cDNA sequence via PCR analysis. Primers used in PCR were complementary to regions of the 5' AOXI (5'-GACTGGTTCCAATTGACAAGC) and 3' AOXI (5'-CCTACAGTCTTACGGTAAACG) DNA sequences. Linearization with PmeI yielded fragments that were integrated by single or double crossover events into the AOXI locus, thus generating transformants in which the AOXI gene remained active (Mut+) or was disrupted (Muts), respectively. The Mut phenotype of recombinant clones was determined by growth on YNB minimal medium plates (1.34% [w/v] yeast nitrogen base, 4 x 10-5% biotin, 2% histidine, and 0.5% [v/v] methanol). The Muts transformants had a slow or absent growth rate. The GS115 cells transformed with pPICZA were used as a control during selection, protein expression, and assessment of biological activity.
Identification of pPRU14 Multicopy Transformants
Southern blot analysis was used to identify transformants containing tandem insertions of the pPRU14 expression vector at the AOXI locus. Genomic DNA was isolated from 30 pPRU14 transformants, 10 pPICZA transformants, and from untransformed GS115 using a modification of the method of Cryer et al. [17]. Briefly, cells were isolated from single colonies on YPD-zeocin plates and were grown at 30°C to an OD600 of 5–10 in 10 ml nonbuffered glycerol-complex medium (MGY) (1% yeast extract, 2% peptone, 1.34% YNB, 1% glycerol, 4 x 10-5% biotin). Cells were collected by centrifugation (1500 x g, 10 min), washed with 10 ml sterile H2O, and collected a second time by centrifugation. Cells were resuspended in 2 ml of SCED buffer, pH 7.5 (1 M sorbitol, 10 mM sodium citrate, 10 mM EDTA) and incubated at 37°C for 50 min in the presence of 0.2 mg zymolyase to achieve <80% spheroplasting. Cell membranes were disrupted by incubation in 1% SDS on ice for 5 min. Potassium acetate (5 M, pH 8.9) was added and mixed gently followed by centrifugation at 10 000 x g for 10 min at 4°C. Genomic DNA was precipitated from supernatant and digested with PvuI or ClaI using standard procedures. DNA was separated electrophoretically, denatured, transferred, and cross-linked to nylon membranes (0.2 µM) and hybridized with bISG17 cDNA randomly primed with 50 µCi -[32P]dCTP. Southern blots were exposed to x-ray film for 42 h. Transformants containing multiple insertions of the pPRU14 expression vector were identified and used for all subsequent experiments.
Expression of rbISG17
Transformants harboring multiple copies of the pPRU14 expression vector were grown to midlogarithmic phase at 30°C in MGY medium to generate cell numbers sufficient for induction by methanol. Cells were collected by centrifugation, resuspended in 10 ml fresh MGY medium, and used to inoculate 1 L of methanol-complete medium (MMY) (1% yeast extract, 2% peptone, 1.34% YNB, 4 x 10-5% biotin, 0.5% methanol) to a final OD600 of 0.4. A 1-ml aliquot of culture was removed every 12 h for a period of 96 h to monitor cell viability, protein expression, and integrity of rbISG17. All subsequent cultures were incubated for 48 h after induction by methanol as it was determined that this time yielded the highest expression of rbISG17 without substantial proteolysis or accumulation of other contaminating yeast proteins. The JPY13 clone was sent to the University of Nebraska Fermentation Facility (Mr. Brad Plantz) for expression under fed-batch fermentation conditions. Five liters of basal salts medium (proprietary) or MMY were inoculated with JPY13 and allowed to grow (~36 h) to an OD600 = 3.5–6.1 prior to induction with methanol. Oxygen saturation was held at 40%. Glycerol was continuously fed to the culture for ~36 h until induction by methanol. The glycerol fed-batch phase was gradually eliminated by decreasing glycerol after the addition of methanol. Cultures were terminated at 9, 24, 40, or 52 h following induction by methanol. Wet cell weight (mg/ml) was determined for growth analysis, and rbISG17 expression was determined at each time point by Western blotting using anti-ISG17 peptide antibody and Coomassie staining [2]. Expression of rbISG17 was optimized through altering pH, oxygen saturation, time of induction by methanol, OD at time of induction, type of medium, and presence or absence of 1% casamino acids.
Purification of rbISG17
Medium was centrifuged at 3000 x g for 8 min to remove cells, and supernatant was filter sterilized through a 0.2-µm filter fitted to a sterile vacuum flask to remove residual cells and debris. Then PMSF was added (1 mM) to the filtered media to inactivate carboxypeptidase Y, a major secreted yeast peptidase. Medium was dialyzed versus 20 mM Tris, pH 8.0 and loaded onto a DEAE sephacel column (40 ml bed volume). Proteins in the flow through were precipitated from solution using ammonium sulfate at 70% saturation followed by resuspension in 10 ml buffer (20 mM Tris, pH 8.0). Proteins were immediately loaded onto a G-75 gel-filtration column (1.5 cm x 80 cm). Fractions containing rbISG17 following G75 gel filtration were loaded onto a high subbing (ligand density = 40 µmole/ml) phenylsepharose column (7 ml bed volume) in 1.6 M ammonium sulfate and eluted with a reverse gradient. Fractions containing rbISG17 were pooled and concentrated using Centricon centrifugation capsules (Amicon, 3000 Mr cutoff). Two nanomoles of rbISG17 (>90% pure based on silver stain, not shown) were sent to the Macromolecular Structure Facility at Michigan State University for carboxyl terminal sequencing [16] using thiohydantoin derivitization to confirm retention of the Leu-Arg-Gly-Gly motif.
Bioassay for rbISG17
All experiments involving animals were approved by the University of Wyoming Animal Care and Use Committee (A3216-01). Bovine PBMCs were enriched from heparinized blood by density gradient centrifugation using Histopaque according to manufacturer's specifications. The PBMCs (25 x 106) were cultured in RPMI-1640 medium supplemented with 10% fetal calf serum and 100 IU penicillin–streptomycin. The PBMCs were untreated (control) or treated in triplicate for 24 or 48 h with concanavalin A (10 µg/ml), rbISG17 (50 ng/ml or 500 ng/ml), or medium (equal volume with 500 ng/ml rbISG17) generated from a P. pastoris clone that expressed serum albumin instead of rbISG17 (negative control). These concentrations of rbISG17 were selected to be in the low nanomolar range, which is within the range of dissociation constants for most receptors. Also, these concentrations of rbISG17 were below levels of ISG17 (i.e., 100 nM; unpublished results) found in uterine luminal flushings from pregnant cows. Total cellular RNA was isolated and served as template for reverse transcription (RT)-PCR (1 µg RNA template; 35 cycles of 95° for 1 min and 60°C for 1 min; 72°C for 7 min) using the bIFN- primers 5'-TATGGCCAGGGCCAATTTTTTAGAGAAATAG and 5'-TACGTTGATGCTCTCCGGCCTCGAAAGAG (GenBank accession number M29867). A plasmid containing a bIFN- cDNA [18] was generously provided by Dr. Joy L. Pate (Ohio State University, Wooster, OH).
An IFN- double-sandwich ELISA [19] (Veterinary Infectious Disease Organization, VIDO, Saskatoon, Saskatchewan, Canada) was standardized to detect the presence of IFN- after treatment of PBMCs. The ELISA plates (96-well Immulon-2) were coated with monoclonal antibody (1:4000, Clone #2-2-1) in carbonate coating buffer (68 mM NaHCO3, 32 mM Na2CO3, pH 9.6) and incubated overnight at 4°C. After washing (6x in PBST), 100 µl of RPMI-1640 medium from the various treatments was added in duplicate to the wells and incubated 2 h at room temperature. Wells were washed as before and 100 µl rabbit polyclonal anti-IFN- (1:5000 in PBST) was added to each well and incubated 1 h at room temperature. After washing, biotinylated goat antirabbit IgG (Zymed, San Francisco, CA; 1:10,000 in PBST) was added and incubated for 1 h at room temperature. Wells were washed again and streptavidin–alkaline phosphatase (1:2000 in PBST) was added and incubated 1 h at room temperature. After washing, p-nitrophenylphosphate substrate was added and incubated at room temperature for 30 min. Absorbance (405/490 nm) was determined using a BioRad model 550 Microplate Reader (BioRad, Hercules, CA). Medium also was tested for reaction with the antibodies. A standard curve was established by serial dilutions of rbIFN- from 1 ng/ml to 2 pg/ml in duplicate.
Detection of IFN- in Endometrium from Pregnant Cows
This experiment tested the hypothesis that pregnancy (i.e., IFN- and/or ISG17) upregulated endometrial IFN- mRNA or protein. Estrous cycles in cross-bred Angus cows were synchronized using prostaglandin F2 (Lutalyse, Upjohn Company, Kalamazoo, MI). Nonpregnant cows were not exposed to semen. Pregnant cows were artificially inseminated 12 h following detection of sexual behavior. Pregnancy was confirmed at the time of collection of the uterus by the presence of a conceptus. Uteri were collected via hysterectomy surgery on Day 18 of pregnancy or on Days 14 and 18 of the estrous cycle. Endometrium was collected on Day 18 of pregnancy because this is a time of maximal uterine exposure to conceptus-derived IFN-, and levels of bISG17 are elevated in uterine flushings [1]. Serum progesterone is elevated in Day 14 nonpregnant and Day 18 pregnant cows. The primary difference between these two groups of cows is presence or absence of an embryo and time of exposure to progesterone. This can be contrasted with the Day 18 nonpregnant cow that also lacks a conceptus but also is undergoing corpus luteum regression and a switch in steroid dominance from progesterone to estrogen (i.e., preovulatory). Endometrial explants were cultured for 24 h as previously described [1, 2]. Release of IFN- into culture medium was determined using 1D-PAGE and Western blot (rabbit anti-IFN antibody; 1:10 000; VIDO). Total cellular RNA was isolated immediately following hysterectomy, stored at -80°C, and subsequently analyzed using the RT-PCR approach for detecting IFN- mRNA described under Bioassay for rbISG17.
Statistics
Interferon- RT-PCR, Western blot, and ELISA results were analyzed using factorial analysis of variance. Main effects and interactions shown to be significant (P < 0.05) were further examined using a protected least-square difference test.
RESULTS
Construction of the pPRU14 Expression Cassette and Transformation
A single plasmid, referred to as pPRU14, was constructed for the heterologous expression of rbISG17 (Fig. 1). Subcloning of the bISG17 coding sequence required addition of cloning sites at either end of the cDNA. The 5' site resulted in the addition of two amino acids at the amino-terminus. Because the 3' site was downstream of the stop codon it had no effect on the mature protein. The cloning sites were necessary to place the bISG17 cDNA in the correct reading frame with the -factor mating signal sequence. Sequencing of DNA confirmed in-frame ligation and Western blotting confirmed secretion and proper processing of rbISG17.
|
The transformation strategy yielded an approximate 20% MutS phenotype as determined by slow growth on YNB minimal medium plates. The transformation procedure was a modified version of that described for S. cerevisiae and resulted in a transformation efficiency within the expected range of 102 to 103 CFU/µg linearized DNA. The transformation efficiency did not differ between GS115 cells transformed with the vector pPRU14 or pPICZA. For all subsequent experiments, Mut+ clones were selected due to comparatively rapid growth. The construct was engineered to utilize the TAG stop codon from the bISG17 coding sequence. It was important to terminate transcription at this junction because modification to the C-terminus by the addition of the myc epitope or polyhistidine tag amino acids (see Fig. 1) might render the recombinant protein biologically inactive.
Selection of Stable Multicopy Clones
Positive transformants were initially selected via zeocin resistance. To ensure stable integration of the expression cassette, PCR was performed on genomic DNA isolated from Mut+ clones using oligonucleotides complementary to the pPICZA 5' and 3' AOXI priming sites (positions 855–875 and 1423–1443, respectively). Lanes 2–5 of Figure 2A show PCR-amplified products from four different positive transformants. The large arrow at the right denotes the 1062-bp fragment found in all positive transformants and corresponded to 588 bp of the pPICZA expression vector between nucleotide positions 855 and 1443, and the 474-bp bISG17 coding sequence with EcoRI and KpnI restriction sites. The PCR amplification of genomic DNA obtained from yeast transformed with the pPICZA vector alone (lane 6) or untransformed yeast (lane 7) did not produce the 1062-bp fragment. The PCR fragments amplified from the vectors pPRU14 (lane 8) and pPICZA (lane 9) were used as positive and negative controls, respectively.
|
Southern blot analysis was used to identify transformants possessing multiple insertions of the pPRU14 expression cassette. Genomic DNA was isolated from 30 transformants previously shown by PCR analysis to contain the pPRU14 expression cassette. Five clones containing multiple, tandem pPRU14 expression cassettes (Fig. 2B) were identified by Southern analysis using 32P-labeled bISG17 cDNA. Lane 2 of Figure 2B contains a ClaI-digested DNA fragment harboring seven tandem pPRU14 expression cassettes based on size comparison with single copy transformants (large arrow). This clone, referred to hereafter as JPY13, was used for subsequent large scale production of rbISG17 (also see Fig. 2C). As expected, ClaI DNA fragments isolated from GS115 transformed with pPICZA (lane 6) or untransformed GS115 (lane 7) did not hybridize with the radiolabeled probe.
To test the hypothesis that heterologous protein expression is proportional to gene dosage, four independent cultures were completed under identical conditions as described in Materials and Methods. Figure 2C is a representative Western blot detecting rbISG17 in crude medium inoculated at OD600 = 0.4 with JPY clones 3, 7, 13, and 24. Western blotting and Coomassie staining revealed the following levels of expression of rbISG17 48 h after induction with methanol: JPY13 clone (6 mg/L), JP3 and JP24 clones (2.5 mg/L), and the single copy JPY7 clone (0.5 mg/L).
Selection of Growth Medium, Induction by Methanol, and Purification of rbISG17
Several attempts were made initially to use the manufacturer's suggested media recipes for heterologous protein expression. Modifications in pH, growth temperature, time of induction by methanol, inoculation density, and media type were varied to optimize protein expression. All attempts at growth in minimal YNB medium resulted in proteolysis of the apparently hypersensitive rbISG17 before intact protein could be observed at 24 h. Growth at low pH and/or in the presence of 1% casamino acids also proved unfruitful. Optimal growth and protein expression with minimal proteolysis occurred in nonbuffered MMY medium at 28°C starting with an inoculation OD600 = 0.4. The authors are unable to explain why expression of rbISG17 was successful under these conditions, especially because phosphate was only 50% of the recommended level required for proper protein expression (B. Plantz, personal communication).
Plotted in Figure 3A is secretion of rbISG17 as determined by Western analysis and wet cell weights taken at 12 h intervals for 96 h. Western blots revealed the presence of an immunoreactive band beginning at 24 h after induction by methanol (Fig. 3B). Logarithmic expression of rbISG17 was obtained through 72 h, at which time expression declined throughout the remainder of the time course. Wet cell weight was used as a proliferation index, where cells from 1 ml of MMY were pelleted via centrifugation and weighed after removal of medium. Cell density increased linearly through 48 h and plateaued thereafter. GS115 transformed with pPICZA failed to secrete rbISG17 as did the JPY13 clone when grown in medium lacking methanol (data not shown).
|
Protein degradation, monitored via Coomassie stain and Western blot (not shown), was consistently observed starting at 60 h. The contribution of rbISG17 to total secreted proteins dramatically declined between 60 and 72 h. In Figure 3A rbISG17 expression peaked at 72 h and then declined over the next 24 h. Taken together with the observed plateau in wet cell weight after 48 h, it is likely that cellular viability was compromised by 60–72 h using the aforementioned conditions. It was also noted that prior to 60 h, budding proficiency was consistently greater than 90%. The percentage of budding yeast declined steadily from 60 to 96 h to less than 55%. For these reasons, cultures were terminated 48 h following induction by methanol and crude medium was harvested for purification of rbISG17.
Attempts to express rbISG17 using a fed-batch fermentation approach either failed completely or did not differ from the shake-flask approach. Expression in the basal salts medium consistently ended in proteolysis of rbISG17 regardless of pH (3.5–8.0), presence of casamino acids, OD (3.5–6.1) at time of induction by methanol, oxygen saturation (20–40%), or length of culture time (6–52 h). Protein expression with minimal proteolysis occurred in nonbuffered MMY medium with an oxygen saturation of 40%, an OD of 6.1 at time of induction with methanol, and an incubation time of 6 h. Under these conditions, protein expression did not differ between the shake-flask and fed-batch fermentation conditions.
Crude medium was dialyzed and filtered for DEAE anion-exchange chromatography. Proteins with no affinity to DEAE contained 21 mg of protein compared to 267 mg of protein in crude medium (Table 1). More importantly, rbISG17 was enriched from 2% in crude medium to 23% after DEAE chromatography. Ammonium sulfate precipitation at 70% saturation was an easy approach for concentrating proteins from 1 L to <5 ml while allowing retention of biological activity.
|
Concentrated rbISG17 was loaded onto a G-75 gel-filtration column and eluted in 7-ml fractions 32–58 (Fig. 4A, Western blot inset), while the majority of contaminating proteins were eluted after fraction 54. Recombinant bISG17 was enriched from 23% to 50% by G-75 gel filtration (Table 1). The inset shown in Figure 4A is a representative Western blot used to identify fractions containing enriched rbISG17.
|
Pooled fractions containing rbISG17 from G-75 chromatography were loaded onto phenylsepharose (Fig. 4B) and eluted (fractions 31–39, Western blot inset) with a reverse ammonium sulfate gradient starting at 1.6 M. Recombinant bISG17 is very hydrophobic because it was eluted from the column at the end of the gradient. This was expected based on amino acid composition and hydrophobicity estimates.
Bioassay
The biological activity of rbISG17 was tested by its ability to upregulate IFN- mRNA and to induce protein expression in PBMCs. Peripheral BMCs were used because of ease of collection and the human ortholog ISG15 has previously been shown to upregulate IFN- by PBMCs [11]. Bovine (Angus cross-bred steer) PBMCs were untreated (control) or treated with 50 or 500 ng/ml rbISG17 or concanavalin A (10 µg/ml) for 24 or 48 h. An additional control for this experiment included treatment with an equivalent amount of medium purified in parallel with rbISG17 from the P. pastoris strain GS115 transformed with a serum albumin cDNA (medium control). Figure 5A shows a representative ethidium bromide stain of the 437-bp IFN- RT-PCR products generated from RNA obtained from PBMCs after 48-h treatments. While IFN- RT-PCR products were generated by all treatments, signals were clearly elevated in PBMCs treated with concanavalin A and 500 ng/ml rbISG17. Quantitation of signals (Fig. 5B) revealed that concanavalin A and rbISG17 (500 ng/ml) upregulated (P < 0.0001) IFN- mRNA when compared with controls. Recombinant bISG17 upregulated IFN- mRNA expression to a level similar to the positive control, concanavalin A. Interferon- mRNA expression did not differ between untreated PBMCs and those treated with 50 ng/ml rbISG17 or medium possessing synthetic albumin.
|
Nearly identical results were obtained when examining IFN- protein expression. Using a double-antibody capture ELISA, IFN- was released (P < 0.0001) into culture supernatants from concanavalin A- and rbISG17 (500 ng/ml)-treated PBMCs when compared with negative controls and PBMCs treated with 50 ng/ml rbISG17 (Fig. 5C). Interferon- concentrations were determined by extrapolation from a standard curve in which the linear range of detection (Y = 0.001X + 0.08; R2 = 0.93) was from 2 pg/ml to 1 ng/ml. Antibodies were negative for cross reactivity with medium.
The presence of IFN- has not been investigated in bovine endometrium until the present experiments. It was hypothesized that conceptus-derived IFN- induced release of endometrial ISG17, which then induced transcription of the IFN- gene and release of protein. Based on Western blots, release of IFN- was not detected in endometrial culture medium from Day 18 pregnant or from Day 14 or 18 nonpregnant cows (not shown). However, IFN- mRNA was present in Day 18 pregnant endometrium based on RT-PCR, a time during which bISG17 expression is maximal (Fig. 6A). The IFN- mRNA was not detected (Day 18 of the estrous cycle) or detected in very low amounts (Day 14 of the estrous cycle) in endometrium from nonpregnant cows (Fig. 6B).
|
DISCUSSION
Several proteins have recently been described that belong to the growing ubiquitin family. Most retain the C-terminal residues that are involved with conjugation to intracellular proteins. Some also are released by cells and have an extracellular role consistent with the definition of a cytokine or hormone. Bovine ISG17 retains both intracellular and extracellular function. One related protein is the small ubiquitin-like modifier (SUMO-1) that shares only limited sequence similarity with ubiquitin but in some cases becomes covalently ligated to identical targets of ubiquitin. To date, four proteins have been identified that become conjugated to SUMO-1. These include the RAN GTPase-activating protein RanGAP1 [20], promyelocytic leukemia protein [21], Sp100 [22], and I
B [23]. Proteins that become conjugated to ISG15/17 have not been identified. A number of other ubiquitin-like proteins that possess the C-terminal Leu-Arg-Gly-Gly motif and become covalently conjugated to intracellular proteins have been identified [24–29]. An extracellular role for these proteins has not been proposed. In addition to bISG17 and hISG15, monoclonal nonspecific factor is an ubiquitin-like protein that conjugates to intracellular proteins and serves a cytokine role. While bISG17 and hISG15 are induced by type 1 IFNs, expression, cellular release, and conjugation of the protein monoclonal nonspecific factor are coordinated by IFN- [29]. These cumulative findings suggest that ubiquitin, a protein that exists in both unicellular and multicellular eukaryotes, evolved from a protein that targets proteolysis, to divergent proteins with cytokine function.
Induction of ISG17 by IFN- in the endometrium during early pregnancy was first described in the cow [1]. Subsequently, Bebington and coworkers also described the induction of ISG15 in human decidualized stromal cells in response to pregnancy [30, 31]. These members of the ubiquitin family probably orchestrate endometrial responses to the conceptus during early pregnancy. Bovine ISG17 conjugates to uterine proteins in response to IFN- and pregnancy [2]. It also can be found in uterine flushings and is released by cultured endometrial explants [1, 2, 10]. To address adequately the function of bISG17, it became critical to generate bioactive recombinant protein.
The methylotrophic yeast, P. pastoris, was used to generate biologically active rbISG17. Because the bISG17 coding unit was inserted in-frame behind the -factor mating signal sequence, secretion of rbISG17 was promoted after processing. The 89 amino acids spanning the -factor mating signal sequence from S. cerevisiae are not processed in the endoplasmic reticulum like other signal peptides. Rather they are cleaved within the Golgi by the dibasic KEX2 protease [32]. Alpha-factor signal sequence proteolysis occurs efficiently in the host P. pastoris [33]. In the present experiments, the -factor signal peptide was cleaved efficiently as documented by a final product of the correct estimated molecular weight.
The pPRU14 expression cassette was linearized with PmeI to facilitate transformation. The restriction enzyme PmeI was selected because it linearized the pPRU14 cassette at position 414 within the 5' AOXI promoter and did not cut the bISG17 cDNA, the -factor mating signal sequence or the gene conferring zeocin resistance. As described by Cregg and coworkers [15], insertion of a linearized gene expression cassette occurs as a single crossover event (Mut+) resulting in gene insertion, or a double crossover event (Muts) resulting in gene replacement. Insertion of the bISG17 gene at the AOXI locus allowed wild-type expression of AOXI. Because of comparatively rapid growth, Mut+ clones were selected on YNB minimal medium plates for further analysis. Homologous recombination in yeast results in extreme stability even when present in multiple tandem copies. Intracellular and secreted heterologous protein expression in P. pastoris is dependent on gene dosage and expression is substantially enhanced in clones possessing multiple gene insertions [34]. Clare et al. showed that during transformation multiple tandem insertions occur at the AOXI locus in 16% of the transformants [34]. Transformants with increased gene dosage expressed and secreted up to 30-fold higher levels of protein when compared to single copy transformants. Several multicopy bISG17 clones were identified via Southern blot analysis to contain multiple insertions of the pPRU14 expression vector. These multicopy insertions resulted in enhanced production of rbISG17. These results were expected, at least in part, because each bISG17 gene insert is regulated by an adjacent AOXI promoter. Yet it is recognized that processing of the signal sequence, packaging, or sequestration of rbISG17 in the periplasmic space could affect the overall secretion rate and might differ from one clone to the next. The JPY13 clone, which contained seven tandem insertions, was used for largescale preparation of rbISG17.
One advantage of the P. pastoris expression system is that protein expression can be done in a simple defined medium making purification easy and efficient. The rbISG17 protein was apparently hypersensitive to proteolysis when using simple medium. Intact rbISG17 was never generated using this approach. Numerous pilot studies were conducted to optimize protein expression and limit proteolysis. Induction of the JPY13 clone by methanol in nonbuffered MMY medium inoculated to an OD600 of 0.4 provided the best expression level (6 mg/L). While protein expression continued to increase through 72 h in the MMY medium, proteolysis was first observed at 60 h. For this reason, cultures were terminated 48 h after induction by methanol. Fed-batch fermentation confirmed the extreme sensitivity to yeast protease activity. As with the shaker-flask approach, multiple fermentation conditions were considered in an attempt to increase rbISG17 expression. While intact protein was generated in nonbuffered MMY medium 6 h after induction with methanol, protein levels did not differ from those obtained via the shake-flask method. As shown in Figure 3A, proliferation ceased by 48 h postinduction even when methanol was provided at 12-h intervals. This result, taken together with a decrease in budding proficiency, would suggest that growth conditions were limiting after 48 h. It is likely that under these conditions yeast cells were dying and possibly releasing proteases and other cellular proteins into the medium. Another explanation for the loss of log phase proliferation after 48 h is that, in hindsight, rbISG17 may be toxic to P. pastoris. Human ISG15 expressed in a yeast two-hybrid system is toxic to S. cerevisiae (A.L. Haas, personal communication).
Purification of secreted rbISG17 was hindered by the inability to utilize the polyhistidine tag provided in the pPICZA expression vector (see Fig. 1). To retain biological activity it was important that translation of the rbISG17 cassette be terminated immediately following the Leu-Arg-Gly-Gly motif. As with ubiquitin [35] and other ubiquitinlike proteins [36], it is likely that the Leu-Arg-Gly-Gly motif of bISG17 is critical for covalent linkage to targeted proteins. Several steps were taken to retain the Leu-Arg-Gly-Gly motif. First, the native bISG17 stop codon was used to terminate translation of rbISG17 rather than one of several stop codons provided in the pPICZA expression vector. Second, PMSF was added to crude, filtered medium to minimize carboxypeptidase Y activity [37]. Third, protein purification proceeded rapidly through G-75 gel-filtration chromatography. This purification step was critical in that most proteases secreted by P. pastoris have a molecular weight greater than 30 kDa [38]. Finally, retention of the Leu-Arg-Gly-Gly motif was confirmed by carboxyl-terminal sequencing. Using this system, rbISG17 has been generated at 0.5 mg/L after purification to greater than 90% purity. The average rate of expression and absence of significant amounts of rbISG17 until 48 h can most easily be explained by poor codon usage. The bISG17 coding sequence possesses numerous codons that are complementary to poorly represented anticodons in the P. pastoris tRNA pool. In the future it may become necessary to modify codon usage to improve the rate of expression in P. pastoris.
The authors avoided use of an E. coli expression system initially because of reports of high carboxypeptidase activity and loss of the C-terminal Gly residues of hISG15 [39]. Conjugation of ubiquitin and its homologs is accomplished through an isopeptide bond between the carboxyl glycine and primary amines on targeted proteins. Proteins conjugated to ubiquitin may be regulated or directed to the 26 S proteasome where they are degraded [6, 7]. The fate of proteins conjugated to ISG17/ISG15 is not known. Purification and C-terminal sequencing of rbISG17 generated in the yeast expression system revealed retention of the C-terminal Gly that is thought to be critical for intracellular conjugation to targeted proteins.
Bovine ISG17 contains three cysteine residues that might be involved in intramolecular disulfide bridges as well as forming a dimer through the free cysteine residue in native form. The yeast expression system also was selected in lieu of bacterial expression systems because yeast are more efficient in producing correctly folded and processed recombinant proteins. Yet, because bacterial expression systems produce a moderately high level of recombinant protein, it is planned to compare rate of expression and biological activity of rISG17 generated in both systems in the future. It is possible to block the C-terminal residue of rbISG17 during bacterial expression and then to cleave this residue following purification from bacterial carboxypeptidases (i.e., using a glutathione S-transferase N-terminal and an Arg C-terminal blocking residue).
Human ISG15 induces IFN- release by cultured T cells [12, 13]. The bISG17 only shares 68% amino acid sequence identity with hISG15. Also, bISG17 has three cysteine residues, whereas, hISG15 has only one. Human ISG15 has an insertion of three amino acids in a region that could be considered a hinge between the two domains with identity (~30%) to ubiquitin. However both ISGs retain C-terminal Leu-Arg-Gly-Gly residues. Thus, it was hypothesized that bISG17 would induce release of IFN- from PBMCs in a manner similar to that described for hISG15. Indeed, bISG17 upregulated IFN- mRNA and induced IFN- protein release in cultured bovine PBMCs.
The present experiments are the first to detect IFN- mRNA in bovine endometrium. Interferon- mRNA was detected in higher amounts in endometrium from pregnant when compared with nonpregnant cows. This increase in IFN- mRNA in the endometrium during pregnancy could be mediated directly by IFN- or by bISG17 that is induced by IFN-. Neither IFN- nor bISG17 induced IFN- in cultured BEND cells (results not shown). The BEND cells are a highly enriched epithelial cell line derived primarily from uterine luminal epithelium that retain receptors for IFN- and respond to treatment by IFN- through cytosolic production of bISG17. The IFN- mRNA found in uterine endometrium from pregnant cows may have originated in cells other than the luminal epithelium. For example, cells of the stroma, deep glandular epithelium, or leukocytes may have been the source for IFN- mRNA in endometrium derived from pregnant cows. Future experiments to clarify the source of the mRNA and to confirm the presence of protein will employ in situ hybridization and epifluorescent microscopy approaches.
Interferon- is a potent proinflammatory cytokine that regulates hematopoiesis, differentiation, cytotoxicity, and apoptosis. Yet, the role of uterine-derived IFN- during pregnancy is vaguely understood. In pregnant uteri of mice, IFN- is prominent at early and late stages of gestation and is synthesized by uterine epithelial cells, NK cells, macrophages, and placental trophoblast cells [40]. This would suggest that IFN- is locally programming immune and other cells via autocrine and paracrine pathways. Furthermore, murine endometrial-derived IFN- contributes to the normal health of the midgestational decidua [41]. Within the pig, IFN- and its receptor have been localized to the endometrium and trophoblast, which would imply cytokine cross-talk between embryonic and maternal components of the placenta [42]. In humans, expression of IFN- by endometrial stromal cells would appear to serve as a major switch that controls the local cytokine profile within the endometrium [43]. In long-term immune cell culture, Tuo et al. showed that IFN- caused an upregulation of IFN- in CD4+ T cells [44]. It is possible that within the endometrium bISG17 plays an intermediary role and acts as a switch between the two distinct classes of interferons. Beyond regulating the cytokine profile within the endometrium, IFN- may serve an antiproliferative role in the development and invasiveness of the blastocyst. In an in vitro mouse development model, IFN- inhibited blastocyst spreading without impairing early embryo development [45].
In summary, the P. pastoris yeast expression system generated rbISG17 in amounts consistent with those reported for other recombinant proteins. This method of generating and purifying rbISG17 is labor and cost intensive. However, it does provide biologically active protein that retains the critical C-terminal Leu-Arg-Gly-Gly motif. Future experiments will compare bioactivity of rbISG17 produced in yeast and bacterial expression systems. The rbISG17 upregulated transcription of IFN- mRNA and induced release of IFN- in cultured bovine PBMCs. This is in agreement with previous reports using hISG15 and human leukocytes. The upregulation of IFN- mRNA in response to pregnancy was expected. The bovine conceptus releases IFN-, which then binds a uterine endometrial receptor to induce transcription and release of bISG17. The hypothesis that bISG17 is an intermediary step in the induction of IFN- will be examined in future experiments. The induction of IFN- by rbISG17 in PBMCs is a good correlative indicator that this occurs in the endometrium during pregnancy. One of the most significant outcomes of the present work is the availability of rbISG17 for future experiments. It is now possible to generate more useful anti-bISG17 antibodies, examine function using an in vivo intrauterine infusion model, identify targeted proteins through in vitro conjugation assays, and to clarify both intracellular and hormonal/cytokine function during pregnancy.
ACKNOWLEDGMENTS
The authors thank Dr. A.L. Haas (Medical College of Wisconsin, Milwaukee, WI) and Brad Plantz (University of Nebraska, Lincoln Fermentation Facility) for helpful discussions.
FOOTNOTES
First decision: 1 February 2000.
1 Supported in part by NIH grant R01-32475-6 and USDA 97-02406 awarded to T.R.H. 
2 Correspondence. FAX: 307 766 2355; thansen{at}uwyo.edu
Accepted: March 4, 2000.
Received: December 28, 1999.
REFERENCES
induce conjugation of bovine ubiquitin cross-reactive protein to cytosolic uterine proteins. Biol Reprod 1998; 58:898–904.. J Biol Chem 1991; 147:2617–2623.B inhibits NF-B activation. Mol Cell 1998; 2:233–239.[CrossRef][Medline]This article has been cited by other articles:
|
|
 |
|
  H. Han, K. J Austin, L. A Rempel, and T. R Hansen Low blood ISG15 mRNA and progesterone levels are predictive of non-pregnant dairy cows. J. Endocrinol., November 1, 2006; 191(2): 505 - 512. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. R. Liptak, B. T. Sullivan, L. E. Henkes, M. P.B. Wijayagunawardane, A. Miyamoto, J. S. Davis, B. R. Rueda, and D. H. Townson Cooperative Expression of Monocyte Chemoattractant Protein 1 Within the Bovine Corpus Luteum: Evidence of Immune Cell-Endothelial Cell Interactions in a Coculture System Biol Reprod, May 1, 2005; 72(5): 1169 - 1176. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. M. Joyce, F. J. White, R. C. Burghardt, J. J. Muniz, T. E. Spencer, F. W. Bazer, and G. A. Johnson Interferon Stimulated Gene 15 Conjugates to Endometrial Cytosolic Proteins and Is Expressed at the Uterine-Placental Interface throughout Pregnancy in Sheep Endocrinology, February 1, 2005; 146(2): 675 - 684. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. A. Rempel, B. R. Francis, K. J. Austin, and T. R. Hansen Isolation and Sequence of an Interferon-{tau}-Inducible, Pregnancy- and Bovine Interferon-Stimulated Gene Product 15 (ISG15)-Specific, Bovine Ubiquitin-Activating E1-Like (UBE1L) Enzyme Biol Reprod, February 1, 2005; 72(2): 365 - 372. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. J. Austin, A. L. Carr, J. K. Pru, C. E. Hearne, E. L. George, E. L. Belden, and T. R. Hansen Localization of ISG15 and Conjugated Proteins in Bovine Endometrium Using Immunohistochemistry and Electron Microscopy Endocrinology, February 1, 2004; 145(2): 967 - 975. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. K. Pru, K. J. Austin, A. L. Haas, and T. R. Hansen Pregnancy and Interferon-{tau} Upregulate Gene Expression of Members of the 1-8 Family in the Bovine Uterus Biol Reprod, November 1, 2001; 65(5): 1471 - 1480. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
