Abstract
P- and E-selectin are surface glycoproteins that mediate leukocyte rolling on the surface of endothelium in inflammation. We have cloned porcine P-selectin cDNA and generated a mAb, 12C5, with which to examine P-selectin expression by porcine aortic endothelial cells (PAEC) in comparison with that of E-selectin. Basal expression by PAEC of P-selectin was greater than that of E-selectin, whereas E-selectin expression was more prominently enhanced than that of P-selectin by stimulation with TNF-α or IL-1α. Both human or porcine IL-4 led to an increase in P-selectin expression, with kinetics that were delayed compared with those seen following stimulation with TNF-α or IL-1α, but IL-4 did not stimulate expression of E-selectin. When cells were stimulated with TNF-α in the presence of IL-4, we observed enhanced P-selectin expression with a parallel reduction in E-selectin expression. Finally, the increase in P-selectin expression due to human IL-4 was reduced in the presence of porcine but not human IFN-γ. These observations show that E-selectin and P-selectin expression are differentially regulated in PAEC, and that IL-4 leads to a shift in the relative surface density of the two molecules toward P-selectin. The ability of porcine IFN-γ to inhibit IL-4-induced P-selectin expression suggests that the balance between Th1 and Th2 cytokine production may determine the relative densities of the two selectins in chronic immune-mediated inflammation. Because the increased expression of P-selectin induced by human IL-4 was not inhibited by human IFN-γ, this balance may be shifted toward P-selectin expression in porcine xenografts infiltrated by human lymphocytes.
The selectins are a conserved family of three cell surface lectins involved in early adhesive interactions between leukocytes, platelets, and endothelial cells (EC)3 (1, 2). Each is a single chain type 1 membrane glycoprotein, with the extracellular portion consisting of an N-terminal lectin domain, an epidermal growth factor-like (EGF) domain, and a variable number of complement control protein repeat (CCP) modules similar to those found in complement regulatory proteins. Although L-selectin expression is confined to leukocytes, E-selectin is expressed exclusively by EC, and P-selectin is found on both platelets and EC. The three selectins are thought to be derived from duplication of a single gene (3), and the exact diversification of their functions is a subject of intense ongoing research.
Up-regulated EC expression of adhesion molecules in response to cytokines is critical to the capacity of tissues to recruit leukocytes during subacute and chronic inflammatory responses (4). E-selectin was originally considered to be the only EC selectin to be regulated transcriptionally by cytokines, with P-selectin expression being thought to be primarily controlled by translocation to the plasma membrane of molecules that are constitutively synthesized and stored within intracellular Weibel-Palade bodies (5, 6, 7). More recently, it has become clear that P-selectin expression can also be regulated by cytokines through transcriptionally dependent mechanisms, although there appear to be differences between species. Thus, P-selectin is readily expressed by mouse EC in response to TNF-α (8, 9), whereas there is debate as to whether human EC show any up-regulation of P-selectin expression in response to this cytokine (10, 11). In contrast, both mouse and human EC show increased P-selectin expression in response to IL-4, with kinetics that are delayed compared with the action of TNF-α on mouse EC (12). Recently, IL-4-mediated up-regulation of surface P-selectin has been found to be inhibited by IFN-γ, suggesting that the balance between Th1 and Th2 cell-derived cytokines may control the expression of this adhesion molecule in chronic immune-mediated inflammation (13).
The pig provides a number of inflammatory models that act as a bridge between rodents and humans, and is the preferred species for the development of clinical xenotransplantation. For these reasons, we have investigated mechanisms of inflammation in this species, both by generating cDNA and mAb to adhesion molecules (14, 15) and by developing the use of differentially radiolabeled leukocytes and mAb for the in vivo analysis of EC adhesion molecule expression (16, 17, 18, 19). In view of the possible species variability of its expression and function, we set out to identify porcine P-selectin, and to perform an initial characterization of the control of its expression by porcine aortic EC (PAEC) in response to cytokine stimulation.
Materials and Methods
Complementary DNA
A cDNA library, from PAEC activated for 6 h with LPS, IL-1, and TNF, contained in pKS1 was a kind gift from Dr. Martyn Robinson (Celltech, Berkshire, U.K.) (14, 15), as was the full-length cDNA of porcine E-selectin (3.3 kb). The full-length human P-selectin cDNA (3.1 kb) was a kind gift from Dr. Rod McEver (University of Oklahoma Health Sciences Center, Oklahoma City, OK). The rat GAPDH cDNA (1.6 kb), which cross-reacts with the porcine sequence, was a kind gift from Dr. R. de Martin (Vienna International Research Cooporation Centre, Vienna, Austria).
Abs, cytokines, and other reagents
mAb 1.2B6 (IgG1), which reacts with human E-selectin (20), porcine E-selectin (15), and human P-selectin (21), was generated in our laboratory. Negative control Abs MOPC 21 (mouse IgG1 myeloma protein) and mAb 198 (IgG1, anti-rabbit Mac-1) were gifts from Dr. Martyn Robinson (Celltech, Slough, U.K.). mAb PL-1 (IgG1, anti-human PSGL-1) was a gift from Dr. Kevin Moore (University of Oklahoma Health Sciences Center). mAb TH16 (IgG2a, anti-porcine lymphocyte Ag DQ) was from Veterinary Research and Development (Pullman, CA). Recombinant human (rh)TNF-α was a kind gift from Dr. Gary Jesmok (Bayer, West Haven, CT). rhILα was a gift from Dr. Jean-Jacques Mermot (Glaxo Institute of Molecular Biology, Geneva, Switzerland). rhIL-4 was purchased from Genzyme (Boston, MA). rhIFN-γ was from Biogen (Cambridge, MA). Recombinant porcine (rp) IFN-γ was purchased from Innogenetics (Ghent, Belgium). rpIL4 was from Bioscource (Appligene-Oncor Lifescreen, Watford, U.K.). All cytokines were titrated in preliminary experiments to obtain the optimal concentration. O-Sialoglycoprotease (OSGE) derived from Pasteurella haemolytica A1 was a kind gift from Dr. Allan Mellors (Ontario, Canada).
PCR and Northern analysis
PCRs were performed in 50–100 μl using 10 pg to 100 ng of plasmid or 100 ng to 1 μg of cDNA library DNA as template. Reactions were performed with 10 pmol of each oligonucleotide, 200 μM dNTPs, and 2.5–5 U Amplitaq Gold DNA polymerase (Perkin-Elmer, Cheshire, U.K.) in the manufacturer’s reaction buffer. Thirty to thirty-five cycles were performed at 94°C for 1 min, 50–62°C for 30–60 s, and extension at 72°C for 1–2 min. Samples were analyzed by agarose gel electrophoresis. Northern analysis was performed as described (22). Probes were labeled with [32P]dCTP using the Quickprime kit (Nycomed Pharmacia, Amersham, U.K.), according to the manufacturer’s instructions. Hybridization with radiolabeled probe was conducted at 42°C overnight, after which blots were washed and autoradiographed using Hyperfilm (Nycomed Pharmacia).
Transfection of COS-7 cells
COS-7 cells were cultured in RPMI 1640 with 10% FCS, 100 IU/ml penicillin, 100 mg/ml streptomycin, and 2 mM l-glutamine, until ∼60% confluent, after which they were harvested and electroporated (290 V, 1500 μF) with 30–50 μg cDNA. Cells were then plated onto 9-cm tissue culture dishes and cultured for 24 h, whereupon the medium was replaced and the electroporated cells cultured for an additional 24 h. For some experiments, cells were harvested 24 h after transfection, plated onto 35-mm tissue culture dishes precoated with 1% gelatin, and used in adhesion assays 48 h later.
HL-60-rosetting assay
HL-60 cells were cultured in RPMI 1640 with 10% FCS and antibiotics. Before adhesion assays, HL-60 cells were analyzed by flow cytometry to confirm PSGL-1 expression, and resuspended at 2 × 106 cells/ml in RPMI. HL-60 cells were added in 1 ml to culture dishes containing adherent- transfected COS-7 cells that had been washed three times with cold RPMI. The dishes were rocked on a horizontal platform for 30 min at 4°C, gently washed five times, and then fixed with RPMI containing 1.5% (v/v) formalin. The number of HL-60 cells bound to COS-7 cells was visualized in multiple sections of the dish and counted by phase-contrast microscopy.
Platelets
Fresh pig blood was collected into acid citrate dextrose in polypropylene tubes. Samples were centrifuged at 1100 rpm for 10 min at 22°C, after which the upper layer of platelet-rich plasma was removed and diluted (1/1) with Tyrode’s buffer without calcium, pH 6.5, and containing 1 mM PGE1 (Sigma, Poole, U.K.). Platelet numbers and purity were assessed using a Coulter Counter (Coulter, Miami, FL). To generate ELISA plates for screening hybridoma supernatants, washed platelets were resuspended at 5 × 106/ml, in Tyrode’s buffer, pH 7, stimulated with 10 U/ml thrombin for 5 min at 37°C, and then centrifuged onto 96-well flat-bottom plates (100 μl/well) at 1250 rpm for 5 min. Platelets were then fixed with 2% paraformaldehyde for 45 min at 4°C, after which the fixative was replaced with 100 μl/well 0.05 M Tris/0.1 M glycine for 15 min at 22°C before screening hybridoma supernatants.
Generation of anti-P-selectin mAb
mAbs to porcine P-selectin were generated by immunizing a homozygous P-selectin-deficient mouse (C57BL/6,Selptm1Bay, The Jackson Laboratory, Bar Harbor, ME) (23) with thrombin-activated porcine platelets. Splenocytes were fused with the NSO nonsecreting myeloma cell line, as previously described (20). Hybridoma supernatants were screened initially by ELISA for reactivity with thrombin-activated platelets, using a modification of an assay previously described (20).
Culture, plating, and stimulation of EC for surface Ag expression
PAEC were isolated, as previously described (14). EC were detached from tissue culture flasks with 0.125% trypsin-EDTA, resuspended in growth medium, and cultured for at least 24 h in 24-well plates (2–3 × 105 cells/well) precoated with 1% gelatin, to achieve confluent monolayers. Before stimulation, the medium was replaced with culture medium containing no growth supplements. Stimulation of EC was performed by adding the appropriate volume of the stimulant at 10 times the desired final concentration to minimize disturbance of the cells.
Western blotting
Platelets or TNF-α-stimulated PAEC (10 ng/ml, 4 h) were centrifuged at 1400 rpm at 4°C and resuspended in cell lysis buffer (50 mM Tris-HCl, pH 7.5, 1% Nonidet P-40, 0.25% sodium deoxycholate, 5 mM EDTA, 10 mM sodium fluoride, 1 mM sodium vanadate, 5 μM leupeptin, 100 U/ml aprotinin, and 1 mM PMSF) and rotated at 4°C for 15–30 min. The cell lysate was then centrifuged at 2000 rpm for 5 min, after which the supernatant was carefully removed from the pellet of cell debris. Proteins were resolved by 7.5% SDS-PAGE, immunoblotted onto Millipore membrane, and probed with anti-porcine P-selectin (mAb 12C5, 20 μg/ml) in 0.05% Marvel. The bound mAb was detected by enhanced chemiluminescence for 30 s. Equal loading and transfer were assessed by Coomassie blue staining.
Flow cytometry
PAEC were incubated for 30 min with 100 μl of saturating amounts of the appropriate primary mAb, either hybridoma culture supernatant or 20 μg/ml purified Ab, at 4°C in HBSS, 2.5% FCS, 0.1% azide. Cells were then washed, centrifuged for 10 min at 1400 rpm at 4°C, and resuspended in a fluorescein-conjugated (FITC) rabbit anti-mouse IgG (Dako, Buckinghamshire, U.K.) 1:50 in HBSS/2.5% FCS/0.1% azide at 4°C for 30 min. After washing twice, cells were fixed with 1% paraformaldehyde v/v in PBS. They were then analyzed using a FACScan Epics XL flow cytometer (Coulter). Specific mAb staining was expressed as relative fluorescence intensity (RFI), which was obtained by dividing the mean fluorescence intensity obtained with test Ab by that obtained with an isotype-matched control Ab. Because of variability between PAEC cultures in absolute levels of basal and cytokine-inducible staining, representative experiments are shown in figures⇓⇓⇓⇓⇓⇓⇓ and the overall data are discussed in the text.
Comparison between species of the amino acid sequences of the lectin domain of P-selectin. A multiple sequence alignment of P-selectin from human, pig, mouse, rat, cow, and sheep was created using the human sequence as a reference. Upper case letters signify amino acids in the human sequence, with identity across the multiple alignment shown as dashes in the other species. Lower case letters signify individual differences between the species. There were no gaps in the alignment across the lectin domain. Boxed residues (23–30, 54–63, and 70–79) correspond to peptides that have been shown to mediate Ca2+-dependent ligand binding (25 ).
Adhesion of HL-60 cells to porcine P-selectin. HL-60 cells added at 4°C to COS-7 cells transfected with human P-selectin cDNA (□) or porcine P-selectin cDNA (▪). The plates were then rocked on a horizontal shaker. Binding of HL-60 cells was assessed in the presence or absence of anti-PSGL-1 mAb PL-1 (5 μg/ml), 2.5 mM EGTA, or following incubation of HL-60 cells with OSGE (50 μg/ml in RPMI 1640 for 1 h at 37°C). Values represent the mean ± SD of the number of HL-60 cells bound per COS-7 cell. HL-60-binding/untransfected COS-7 cell and HL-60-binding/empty pKS1-transfected COS-7 cell was 1 ± 0.8 and 1.2 ± 1, respectively. No inhibition of adhesion was seen with a control IgG1 Ab (MOPC 21) (not shown). Results are representative of those obtained in two independent experiments.
Binding of mAb 12C5 to P-selectin-, but not E-selectin-transfected COS-7 cells. COS-7 cells were transfected with cDNA encoding porcine P-selectin, human P-selectin, or porcine E-selectin, and examined by flow cytometry for reactivity with mAb 12C5, or with mAb 1.2B6, which reacts with porcine E-selectin, human E-selectin, and human P-selectin. MOPC 21 was used as an IgG1 control Ab.
Expression of P-selectin and E-selectin on PAEC in response to stimulation with rhTNF-α or rhIL-1α. Primary cultures of PAEC were stimulated with rhTNF-α (10 ng/ml) or rhIL-1α (10 ng/ml); A, cells were analyzed by flow cytometry using mAb 12C5 (anti-P-selectin) and mAb 1.2B6 (anti-E-selectin) at varying times after stimulation. Specific analyses are, respectively, representative of five and two experiments using different PAEC cultures. mAb staining was expressed as RFI, which was obtained by dividing the staining obtained with test Ab by that obtained with an isotype-matched control Ab. B, histograms of Ag expression on unstimulated PAEC (filled histograms) and on PAEC in response to rhTNF-α (10 ng/ml) (open histograms), at the time points of maximal Ag expression (6 h for P-selectin and 4 h for E-selectin). The grey histogram in the upper panel shows the binding of MOPC 21 control IgG1. In this experiment, the histogram for the E-selectin staining of unstimulated cells superimposed on the histogram for the isotype control. C, Western analysis of cell lysates generated from pig platelets or from PAEC stimulated for 6 h with rhTNF-α (10 ng/ml) and probed with mAb 12C5; no other specific bands were recognized in the platelet or PAEC lysates; D, expression of steady state mRNA for P-selectin and E-selectin following incubation with 10 ng/ml rhTNF-α, as determined by Northern analysis with full-length P-selectin and E-selectin cDNAs as probes. Equal RNA loading was assessed by ethidium bromide staining of the gel and by hybridization with rat GAPDH cDNA. The flow cytometry and Northern analyses are, respectively, representative of five and two experiments using different PAEC cultures.
Effect of rhIL-4 on P- and E-selectin expression by PAEC. Primary PAEC were stimulated for the durations indicated with 20 ng/ml rhIL-4. A, Flow cytometry for expression of P- and E-selectins with anti-P-selectin mAb 12C5 and anti-E-selectin mAb 1.2B6. Ag expression is shown at each time point as the RFI compared with the isotype-matched control Ab MOPC 21. This kinetic study is representative of three independent experiments; B, comparison of the effects on P-selectin expression by incubating PAEC with rhIL-4 or prIL-4 for 48 h, as detected by mAb 12C5; C, Northern analysis using full-length porcine P-selectin cDNA as a probe. Equal loading between lanes was assessed by ethidium bromide staining of the gel and by hybridization with rat GAPDH cDNA. This result is representative of two similar experiments.
Surface expression of P-selectin and E-selectin on PAEC in response to human IL-4 and rhTNF stimulation. Primary cultures of PAEC were stimulated for the durations indicated with 20 ng/ml human IL-4 and 10 ng/ml human TNF and then analyzed by flow cytometry for expression of P-selectin (□) and E-selectin (▪) with mAb 12C5 (anti-P-selectin) and 1.2B6 (anti-E-selectin). Ag expression is shown at each time point as the RFI compared with the isotype-matched control Ab MOPC 21. This kinetic is representative of three independent experiments.
Effect of porcine rIFN-γ (prIFN-γ) on P-selectin expression in response to rhIL-4. Primary cultures of PAEC were stimulated for 48 h with rhIL-4 (20 ng/ml), rhIFN-γ (250 U/ml), or prIFN-γ (125 ng/ml), either alone or in combination, and then analyzed by flow cytometry for expression of P- and E-selectins with anti-P-selectin mAb 12C5 and anti-E-selectin mAb 1.2B6. Ag expression is shown at each time point as the RFI compared with the isotype-matched control Ab MOPC 21. This experiment is representative of three independent experiments.
Results
Cloning of porcine P-selectin cDNA
Initially, we generated a 342-bp porcine P-selectin cDNA fragment by PCR, utilizing a cytokine-activated PAEC cDNA library as template and primers designed from the human P-selectin cDNA sequence (forward primer, 5′-2223ATC TGC TCT TTC CAT TGT CTA-3′; reverse primer, 5′-2527GAC TCG GGT GAA ATG CAG CGT TTG-3′) (24), encompassing the transmembrane domain and part of the cytoplasmic domain. The PCR product was sequenced and shown to be 75% identical to the human sequence over this region.
Using the 342-bp porcine P-selectin cDNA fragment as a probe, we isolated from a cytokine-activated PAEC library a 2647-bp cDNA clone encoding an open reading frame of 1938 bp, corresponding to a protein of 646 aa (GenBank accession no. AF163766). The sequence of the porcine P-selectin clone was identical to a GenBank/EMBL sequence submitted during this study (GenBank accession no. L39075). Alignment of the pig sequence with its human counterpart demonstrated strong conservation in the lectin (75.9%) and EGF (85.7%) domains involved in ligand binding. Alignment of the lectin domains of pig, human, mouse, rat, cow, and sheep revealed a high degree of conservation in three regions of the lectin domain corresponding to peptides 23–30, 54–63, and 70–79, which in human P-selectin have been shown to support ligand binding (25) (Fig. 1⇑). Porcine P-selectin cDNA encodes for six CCP modules, which is two less than mouse, rat, and sheep, and the same as cow. Human P-selectin cDNA may encode for eight or nine CCPs, dependent upon alternative splicing (26).
Porcine P-selectin can support the PSGL-1-dependent adhesion of human leukocytes
To establish the functionality of our porcine P-selectin cDNA clone, COS-7 cells were transfected with human or porcine P-selectin cDNA and tested for their ability to bind the human promyelocytic cell line HL-60. As shown in Fig. 2⇑, HL-60 cells bound COS-7 cells transfected with either human or porcine P-selectin cDNA, but did not bind cells transfected with empty vector (not shown). PSGL-1 is a homodimeric leukocyte cell surface glycoprotein that acts as the predominant leukocyte ligand for P-selectin (27). PSGL-1 binding to P-selectin is dependent upon extracellular calcium and is sensitive to digestion of the leukocyte surface with OSGE (28). We found that binding of HL-60 cells to porcine and human P-selectin was inhibited by preincubation with anti-PSGL-1 (mAb PL-1), by the presence of 2.5 mM EGTA, or by predigestion of surface O-linked glycoproteins with OSGE. These data are therefore consistent with human PSGL-1 binding porcine as well as human P-selectin.
Generation of mAbs to porcine P-selectin
To generate mAb to porcine P-selectin, we immunized a homozygous P-selectin-deficient mouse with porcine platelets. Hybridoma supernatants were screened using a cell-based ELISA for reactivity with thrombin-activated porcine platelets. Positive clones were then rescreened by flow cytometry for binding COS-7 cells that had been transfected with porcine P-selectin cDNA. One of nine Abs selected by this approach, mAb 12C5 (IgG1), was chosen for further study. As shown in Fig. 3⇑, mAb 12C5 showed clear reactivity with COS-7 cells transfected with porcine P-selectin, but did not react with COS-7 cells transfected with porcine E-selectin or human P-selectin. In contrast, mAb 1.2B6, which is known to bind human P-selectin in addition to porcine and human E-selectin, showed no reaction with porcine P-selectin. In parallel, we found using flow cytometry that mAb 12C5 does not bind porcine or human PBL, suggesting that this mAb does not cross-react with L-selectin (data not shown).
Comparison of P- and E-selectin expression by porcine EC, constitutively and in response to TNF-α or IL-1α
Assessment of PAEC by flow cytometry with mAb 12C5 showed a low level of basal P-selectin expression, although there was some variation between cultures in the degree of staining (mean ± SD RFI 4.4 ± 2.5, range 1.8–12.5, 25 experiments). In contrast, basal expression of E-selectin was always apparently less than that of P-selectin (mean ± SD RFI 1.9 + 1.1, range 0.5–5.4, 25 experiments). Incubation of PAEC with rhTNF-α or rhIL-1α led to up-regulation of both P-selectin and E-selectin, with maximal P-selectin expression occurring about 2 h after that of E-selectin (6 h compared with 4 h) (Fig. 4⇑A and B). However, in view of the greater basal expression of P-selectin and the relatively modest increase in expression of P-selectin in response to TNF-α, the maximal increase in P-selectin expression was considerably less that of E-selectin (in five experiments, mean ± SD maximal percent increase 142 ± 48.9% for P-selectin compared with 432.6 ± 263.5% for E-selectin. Western blotting of platelets and TNF-α-activated PAEC confirmed that anti-P-selectin mAb 12C5 reacted with a single band, which was ∼130 kDa under reducing conditions and consistent with the molecular mass of porcine P-selectin predicted by the cDNA (Fig. 4⇑C). Northern analysis showed that P-selectin but not E-selectin mRNA was detectable in unstimulated cells, and that the increased surface expression of both selectins in response to rhTNF-α was associated with a transient increase in steady state mRNA encoding each molecule (Fig. 4⇑D). These data therefore indicate that P-selectin, like E-selectin, is inducible on porcine EC in vitro in response to TNF and IL-1, although E-selectin appears to be up-regulated to a relatively greater extent.
IL-4 induces a delayed expression of P-selectin and down-regulates TNF-α-induced E-selectin expression
We next investigated the effects of IL-4 on selectin expression by PAEC. Following stimulation of PAEC with rhIL-4 (20 ng/ml), surface expression of P-selectin was seen to increase with kinetics that were delayed relative to those observed in response to TNF-α or IL-1α, with maximal expression after 48–72 h (in 11 experiments, mean ± SD % increase in RFI after 48 h was 375.1 ± 130.6) (Fig. 5⇑A). In contrast, rhIL-4 reduced the low basal level of E-selectin (in 11 experiments, mean ± SD % decrease in RFI after 48 h was 66.7 ± 27.6). As shown in Fig. 5⇑B, pig rIL-4 induced P-selectin over a similar dose range to that seen with rhIL-4. The increase in P-selectin protein expression in response to rhIL-4 was accompanied by increased steady state P-selectin mRNA (Fig. 5⇑C).
In view of the different kinetics of P-selectin up-regulation in response to IL-4 and TNF-α, we performed an experiment to investigate the effects of the two cytokines in combination. As shown in Fig. 6⇑, IL-4 and TNF-α stimulated P-selectin expression at 8 h additively (RFI 4.1 on unstimulated cells, 7.9 on TNF-α-activated cells, 14.9 on IL-4-activated cells, and 23 on TNF-α- and IL-4-stimulated cells), whereas at 24 and 48 h, the effect of the combination was no greater than that of IL-4 alone. In contrast, IL-4 led to a marked reduction in E-selectin expression at 8 h compared with that seen with TNF-α alone (RFI 2.1 on unstimulated cells, 28 on TNF-α-activated cells, and 16.2 on TNF-α-stimulated cells in the presence of IL-4). Thus, the overall effect of IL-4 was to enhance TNF-α-stimulated P-selectin expression, but to reduce TNF-α-stimulated expression of E-selectin.
Porcine but not human IFN-γ inhibits IL-4-induced P-selectin expression
Evidence has recently been reported suggesting that IFN-γ is able to suppress the induction of P-selectin by IL-4 (13). In a preliminary experiment, porcine but not human IFN-γ was found to increase expression of the porcine DQ Ag, consistent with the species restriction of the actions of this cytokine. When PAEC were cultured for 48 h with the combination of IFN-γ and IL-4, the increase in expression of P-selectin was reduced compared with cultures stimulated with IL-4 alone (in the experiment shown in Fig. 7⇑, RFI 5 on unstimulated cells, 15 on cells stimulated with IL-4, and 9.7 on cells stimulated with IL-4 in the presence of IFN-γ). Thus, in three experiments, the increase in P-selectin expression observed after 48 h in the presence of both 250 U/ml porcine rIFN-γ and 20 ng/ml rhIL-4 was 57.2 ± 9.3% (mean ± SD) that seen in the presence of 20 ng/ml IL-4 alone. As expected from previous reports showing that human IFN-γ is not active on porcine cells (29, 30), human IFN-γ had no effect on the expression of P-selectin in response to IL-4.
Discussion
We have generated cDNA and mAb to porcine P-selectin and used these reagents to characterize the expression of this molecule on PAEC following cytokine stimulation. Examination of the primary structure of P-selectin in the pig with that in other species shows a high degree of conservation of the lectin and EGF domains, particularly in regions involved in ligand binding. It is therefore not surprising that human leukocytes were found to bind porcine P-selectin in a PSGL-1-dependent manner. Our data indicate that TNF-α only weakly stimulates P-selectin expression by PAEC, but that IL-4 leads to a more pronounced and delayed up-regulation. In view of the differences that exist between species with respect to the control of P-selectin expression, these observations are important for understanding the molecular mechanisms of leukocyte-EC interactions in porcine models of inflammation, and also for predicting the nature of the rejection response to porcine xenografts.
We found that freshly isolated PAEC had readily detectable basal P-selectin RNA and surface expression, in contrast to E-selectin, which was minimally expressed at both the message and protein level in the absence of stimulation. We believe that this basal P-selectin expression is likely to be a true reflection of the in vivo situation, since in preliminary work we have found that anti-mouse Ig-coated magnetic beads bind to the endothelial surface of porcine aortae incubated with anti-P-selectin mAb 12C5, but not to the endothelial surface of aortae incubated with anti-E-selectin mAb 1.2B6. P-selectin is known to be constitutively synthesized by human EC, but in HUVEC is programmed to traffic to Weibel-Palade bodies for storage (5, 31, 32). Because the endothelium of the pig aorta is known to have very few Weibel-Palade bodies (33), the basal expression of P-selectin may be due to the default trafficking of constitutively synthesized P-selectin molecules to the cell surface.
The paucity of Weibel-Palade in PAEC can be viewed as an advantage for studying the expression of P-selectin in response to cytokines, because surface expression may be a direct reflection of transcriptional activity. Using PAEC, we found that surface P-selectin was up-regulated both by TNF-α and by IL-1α, although the increase was very modest in comparison with the effects of these cytokines on E-selectin expression. In this respect, pig EC appear to be more similar to human than mouse EC. Thus, on the one hand, TNF and IL-1 stimulate expression of P-selectin in mouse EC that in vivo occurs simultaneously with, and is quantitatively similar to, that of E-selectin expression (34, 35). On the other hand, the human P-selectin promoter lacks an NF-κB-binding element present in the mouse P-selectin promoter, and P-selectin up-regulation by human EC in response to these cytokines is minimal (11). In PAEC, the modest increase both in steady state mRNA and in surface P-selectin protein in response to TNF-α lagged behind those of E-selectin by about 2 h, suggesting differences in the transcriptional mechanisms involved in stimulating the two molecules. It will therefore be interesting to determine whether or not the porcine P-selectin promoter contains the NF-κB-binding element described in the mouse. It is possible that the pig may provide a model for investigating P-selectin expression in vivo that is more representative than the mouse for the human situation.
We found that IL-4 led to a dose-dependent increase in P-selectin expression that was greater and relatively delayed compared with that seen following stimulation with either TNF-α or IL-1α, and comparable with that described in human and mouse EC (12). The first evidence of increased P-selectin expression following stimulation of PAEC with IL-4 occurred at about 8 h, at which point the effect was additive with that of TNF-α. At later time points, the maintained exposure of the cells to TNF-α made no difference to the expression of P-selectin in response to IL-4. In striking contrast, IL-4 markedly reduced expression of E-selectin in response to TNF-α, as reported previously with HUVEC (36, 37, 38). The net outcome, therefore, of stimulating PAEC with the combination of both IL-4 and TNF-α was to change the relative proportions of P-selectin and E-selectin on the cell surface, with P-selectin becoming apparently the more highly expressed selectin.
Our experiments have therefore clearly shown that expression of the two selectins in the pig is differentially regulated, both in terms of basal expression and in response to the cytokines TNF-α, IL-1, and IL-4. These observations raise the possibility that in this species the two selectins may have separate ligands and function to mediate the recruitment of significantly different populations of leukocytes into the tissues. In pig skin, a lack of redundancy between the two selectins is supported by the ability of anti-E-selectin mAb to inhibit the uptake of neutrophils and a population of lymphocytes in the absence of P-selectin blockade (17, 18). Conversely, there is increasing evidence that in some situations, P-selectin may be the predominant endothelial selectin involved in recruiting leukocytes into chronic inflammatory lesions (39, 40, 41). The cytokines responsible for the maintenance of P-selectin expression in chronic inflammation in vivo have not yet been clearly defined, but could include IL-4 and/or oncostatin M (12). In this context, the finding that IFN-γ can inhibit human E- and P-selectin expression suggests an important mechanism for the down-regulation of the quantity of leukocytes entering the tissues during chronic immune-mediated inflammatory responses (13). Moreover, the finding that IFN-γ inhibits the delayed IL-4-mediated expression of human P-selectin raises interesting questions about the role of the Th1-Th2 balance in regulating selectin expression and leukocyte trafficking (42).
In the event that it becomes possible to prevent the acute rejection of porcine organs transplanted into humans, the finding that IL-4 and IFN-γ have opposing actions on P-selectin expression may have a special significance in the context of pig to human xenotransplantation. Since during chronic rejection, porcine endothelium would be exposed to cytokines derived from infiltrating human leukocytes, the responsiveness of porcine EC to human IL-4, but not to human IFN-γ, may lead to a shift in the activation balance toward IL-4-mediated effects. It is predicted, therefore, that P-selectin would become prominently expressed during chronic xenograft rejection, thereby facilitating further leukocyte recruitment. Exactly what types of leukocyte would be recruited in this situation remains to be determined.
Acknowledgments
We thank Dr. Kristian Riesbeck for supplying the porcine P-selectin cDNA probe used to isolate porcine P-selectin cDNA, and Dr. Tony Dorling and Dr. Ravi Rao for helpful discussions.
Footnotes
-
↵1 This study was supported by a grant from the British Heart Foundation.
-
↵2 Address correspondence and reprint requests to Dr. Dorian O. Haskard, British Heart Foundation Cardiovascular Medicine Unit, NHLI, Imperial College School of Medicine, Hammersmith Hospital, London W12 ONN, U.K. E-mail address: d.haskard{at}ic.ac.uk
-
↵3 Abbreviations used in this paper: EC, endothelial cell; CCP, complement control protein repeat; EGF, epidermal growth factor-like; OSGE, O-sialoglycoprotease; PAEC, porcine aortic EC; pr, porcine recombinant; PSGL-1, P-selectin glycoprotein-1; RFI, relative fluorescence intensity; rh, recombinant human.
- Received August 6, 1999.
- Accepted January 3, 2000.
- Copyright © 2000 by The American Association of Immunologists