Parallel induction of epithelial surface-associated ... - CiteSeerX

Parallel induction of epithelial surface-associated chemokine and proteoglycan by cellular hypoxia: implications for neutrophil activation Glenn T. Furuta,*† Andrea L. Dzus,* Cormac T. Taylor,* and Sean P. Colgan* *Center for Experimental Therapeutics and Reperfusion Injury, Brigham and Women’s Hospital; and †Combined Program in Pediatric Gastroenterology and Nutrition, Children’s Hospital, and Harvard Medical School, Boston, Massachusetts

Abstract: Neutrophil-induced damage to the protective epithelium has been implicated in mucosal disorders associated with hypoxia, and such damage may be initiated by epithelial-derived chemokines. Because chemokines can bind to membrane proteoglycans, we hypothesized that chemokines may associate with epithelial surfaces and activate polymorphonuclear neutrophils (PMN). Epithelial hypoxia (pO2 20 torr) resulted in a time-dependent induction of interleukin-8 (IL-8) mRNA, soluble protein, as well as surface protein. Such surface IL-8 expression was demonstrated to be dependent on heparinase III expression, and extensions of these experiments indicated that hypoxia induces epithelial perlecan expression in parallel with IL-8. Finally, co-incubation of post-hypoxic epithelia with human PMN induced IL-8-dependent expression of the PMN ␤2-integrin CD11b/18. These data indicate that chemokines liberated from epithelia may exist in a surface-bound, bioactive form and that hypoxia may regulate proteoglycan expression. J. Leukoc. Biol. 68: 251–259; 2000. Key Words: interleukin-8 䡠 mucosal disorders 䡠 hypoxia

INTRODUCTION The pathological hallmark of many acute and chronic intestinal diseases is the accumulation of polymorphonuclear leukocytes (PMN) adjacent to crypt epithelial cells of the intestine, termed crypt abscesses [1]. Significant tissue damage brought about by PMN migration across intestinal epithelium has been demonstrated in a variety of diseases including Crohn’s disease, ulcerative colitis, hemorrhagic shock, and reperfusion injury [2, 3]. Studies of human mucosae in such diseases suggest that PMN transepithelial migration predates focal breakdown of the epithelial surface [2], and that defective epithelial barrier function also predates structural discontinuities in the mucosa [4]. At present, the pathophysiological factors that lead to the formation of such inflammatory responses are poorly understood. A common feature of many disease processes is tissue hypoxia. Recent studies have shown that hypoxia induces a

number of genes through diverse mechanisms and likely contributes significantly to inflammatory processes [5]. Tissue hypoxia may orchestrate leukocyte recruitment through induction of chemotactic cytokines (chemokines), polypeptides that function as leukocyte activators and chemoattractants in vitro and in vivo [6]. Chemokines can function as soluble mediators or can specifically bind to polysaccharide components of cellsurface and extracellular-matrix proteoglycans [7]. Proteoglycan binding sites on chemokines are generally distinct from the receptor binding site, utilize basic amino acid residues (lysine, arginine, or histidine) for binding, and proteoglycans can bind multiple chemokine molecules [8, 9]. Some studies have suggested that chemokine activity is enhanced when bound to matrix components, particularly the proteoglycan heparan sulfate [7, 10], possibly through oligomerization of the chemokine [8]. Proteoglycan structure can vary significantly, even between tissues, and it has been hypothesized that these structural differences may account for tissue localization of specific chemokines and, as a result, specific leukocyte populations [11]. The chemokine interleukin-8 (IL-8) has an established role in mucosal disease [12], and intestinal epithelia are primary sources of IL-8 [13]. Mechanistic in vitro studies suggest that epithelial-derived IL-8 serves as a recruitment signal for PMN. For example, apical colonization of epithelia with pathogenic bacteria induces the basolateral release of IL-8 and imprints the extracellular matrix with chemokine [14]. This insoluble, matrix-associated chemokine serves as a haptotactic recruitment signal for PMN to the basolateral epithelial surface [14, 15]. Less is known, however, about the subsequent steps in PMN transmigration (i.e., after initial PMN-epithelial contact). PMN adhesive interactions to an as yet unidentified epithelial ligand are mediated primarily by CD11b/18, whereas transmigration through the epithelial paracellular space is predominantly CD47-dependent [16]. This epithelial paracellular pathway provides a unique environment for leukocyte movement. Unlike the thin, flat substrate of the vascular endothelium, transmigration through columnar epithelium mandates that

Correspondence: Sean P. Colgan, Ph.D., Brigham and Women’s Hospital, Center for Experimental Therapeutics and Reperfusion Injury, Thorn 704, 75 Francis Street, Boston, MA 02115. E-mail: [email protected] Received January 12, 2000; revised March 31, 2000; accepted April 10, 2000.

Journal of Leukocyte Biology Volume 68, August 2000 251

PMN move greater than two cell lengths through the paracellular space. It is possible, therefore, that epithelia orchestrate PMN movement through the paracellular space by providing chemokine in a surface-expressed fashion. At present, this pathway has not been studied. We demonstrate here that epithelial cells pre-exposed to cellular hypoxia express functional IL-8 in a surface-expressed manner. Such induction of surface chemokine required the parallel activation of both IL-8 release and the surface up-regulation of perlecan. Thus, under defined conditions, the epithelial cell surface may provide a haptotactic substrate for PMN movement.

Reverse transcriptase-polymerase chain reaction (RT-PCR) analysis of mRNA levels was performed using DNAse-treated total RNA as previously described [22] using primers specific for IL-8 (5’-ATG ACT TCC AAG CTG GCC GTG GCT-3’ and antisense primer 5’-TCT CAG CCC TCT TCA AAA ACT TCT C-3’, 289-bp fragment), epican (5’-GCA GAA TGT GGA CAT GAA GA-3’, and antisense primer 5’-ATG CTA AAA AAG ATT CGC AAT G-3’, 150-bp fragment), perlecan (5’-CTG CCC CTC GTA GGC ACC-3’ and antisense primer 5’-CAC TCC CAT CCA ACC TGC-3’, 231-bp fragment), syndecan-1 (5’-GAA CCA AAT CTG GAA GCC AA-3’ and antisense primer 5’-GAC ACA GGC ACG ACT GCT T-3’, 182-bp fragment), or control ␤-actin [22] (5’-ATG ACT TCC AAG CTG GCC GTG GCT-3’ and antisense primer 5’-TCT CAG CCC TCT TCA AAA ACT TCT C-3’, 661-bp fragment). Each primer set was amplified using 25 cycles of 94°C for 1 min, 60°C for 2 min, 72°C for 4 min, and a final extension of 72°C for 7 min. The PCR reactions were then visualized on a 1.5% agarose gel containing 5 ␮g/mL of ethidium bromide.

MATERIALS AND METHODS

Chemokine assessment in biotinylated plasma membrane fractions

Cell culture T84 intestinal epithelial cells (passages 67– 85) were grown and maintained as confluent monolayers on collagen-coated permeable supports as previously described in detail [17]. Monolayers were grown on 0.33-, 5-, or 45-cm2 ring-supported polycarbonate filters (Costar, Cambridge, MA), as indicated, and utilized 6 –12 days after plating as described previously [18]. Epithelial cultures were exposed to hypoxia as previously described [15]. Briefly, growth media was replaced with fresh media equilibrated with hypoxic gas mixture and cells were placed in the hypoxic chamber (Coy Laboratory Products, Ann Arbor, MI). Oxygen concentrations were as indicated in torr (normoxia equal to pO2 of 150 torr) with the balance made up of nitrogen, carbon dioxide (constant pCO2 35 torr), and water vapor from the humidified chamber. These culture conditions are non-toxic to epithelia [15].

Quantitation of IL-8 Supernatants from normoxic and hypoxic epithelial monolayers were quantitated for IL-8 secretion with the use of a capture sandwich enzyme-linked immunosorbent assay (ELISA) and an IL-8 standard curve as described before [15]. Values are reported as IL-8 per 106 cells in a 1-mL volume of conditioned supernatant. Detection of surface-associated IL-8 or surface-associated heparan sulfate was determined using intact monolayers of normoxic or hypoxic T84 cells or by flow cytometry of cells that were lifted from the permeable support. Briefly, confluent T84 cells were grown to electrical confluence on permeable supports, incubated in normoxia or hypoxia as indicated. Analysis of surface chemokine or proteoglycan on intact cell monolayers was performed by ELISA on insertgrown cells (0.33 cm2) as described previously [19] (values are reported as IL-8 per 106 cells relative to a standard curve). For assessment by flow cytometry, intact monolayers (5 cm2) were transferred to a solution of modified Hanks’ balanced salt solution (HBSS; without Ca2⫹ or Mg2⫹) containing 5 mM ethyleneglycol-bis(␤-aminoethyl ether)-N, N, N’, N’-tetraacetate (EGTA, Sigma Chemical) at 37°C. Monolayers were agitated for 2 h at 37°C and vigorously washed to remove cells, as described elsewhere [14]. Cells elicited in this fashion were cooled to 4°C, blocked with 100 ␮L media containing 10% bovine calf serum (BCS) for 1 h at 4°C, followed by addition of rabbit anti-human IL-8 polyclonal antibody (20 ␮g/mL in media, Endogen, Boston, MA) or control rabbit serum (Sigma, 1:500 dilution in media) for 2 h at 4°C. Cells were extensively washed with cold HBSS followed by addition of fluorescein-conjugated goat anti-rabbit secondary Ab (1:1000 in media, Cappel, West Chester, PA) and analyzed by flow cytometry. Control cells to determine background were incubated with only the secondary antibody. Cells were analyzed for surface expression of IL-8 or CD11b, as indicated, by flow cytometry as described before [20] with a FACScan flow cytometer (Becton Dickinson Immunocytometry Systems, Mountain View, CA) and CellQuest™ software.

Epithelial cells were grown to confluence on 45-cm2 permeable supports, and exposed to hypoxia or normoxia as indicated. Plasma membranes were isolated using nitrogen cavitation (200 psi, 8 min, 4°C) as previously described [16]. Plasma membrane fractions were labeled with biotin (1 mM in HBSS) as previously described [16] and re-pelleted by ultracentrifugation to remove excess biotin. Recombinant human IL-8 (100 ng/mL) was directly biotinylated [1 mM NHS-biotin (Pierce Chemical, Rockford, IL) in HBSS] and excess biotin was removed by multiple washes on a 5-kDa cut-off membrane filter (Amicon, Beverly, MA). Fractions were pre-cleared with 50 ␮L pre-equilibrated protein-G Sepharose (Pharmacia, Uppsala Sweden). Immunoprecipitation of IL-8 was performed with rabbit polyclonal anti-IL-8 followed by addition of 50 ␮L pre-equilibrated protein-G Sepharose and overnight incubation. Washed immunoprecipitates were boiled in non-reducing sample buffer [2.5% sodium dodecyl sulfate (SDS), 0.38 M Tris pH 6.8, 20% glycerol, and 0.1% bromophenol blue], resolved by non-reducing SDS-polyacrylamide gel electrophoresis (PAGE; 18% polyacrylamide gel), transferred to nitrocellulose, and blocked overnight in blocking buffer. Biotinylated proteins were labeled with streptavidin-peroxidase (Pierce Chemical) and visualized by enhanced chemiluminescence (ECL; Amersham, Arlington Heights, IL).

Antisense oligonucleotide treatment of cells Electrically confluent T84 cells grown as monolayers on permeable supports were washed three times in warm T84 cell media. IL-8 phosphorothioate antisense oligonucleotide (40 base oligonucleotide) derived from the 5’ untranslated sequence of exon 1 [23], purchased from Oncogen Sciences (Cambridge, MA), were diluted to 200 nM in serum-free media containing N-[1(2,3-dioleyloxy)propyl]-N, N, N-trimethylammonium chloride (DOTMA, Lipofectin solution at 10 ␮g/mL, GIBCO) and applied to cells (150 ␮L) to the basolateral surface only. Serum-containing media were used in the opposing well (i.e., apical surface). Cells were incubated in normoxia or hypoxia, as indicated, for 24 h followed by addition of 100 ␮L serum-containing medium to the basolateral surface. Cells were allowed to incubate for an additional 24 h, then used as described above to assess IL-8 secretion or PMN transmigration. Controls consisted of media only or media containing DOTMA without antisense oligonucleotides.

Heparinase treatment of intact epithelia Monolayers of the human intestinal epithelial cell line T84 were grown as inverted monolayers on permeable supports. Confluent monolayers were exposed to hypoxia (pO2 20 torr, 48 h), EGTA-elicited (as described above), and incubated in the presence or absence of heparinase III (1 U/mL, Sigma) for 2 h at 37°C. Monolayers were extensively washed at 4°C, and processed for flow cytometry as described above.

Transcriptional analysis

Immunoprecipitation of biotinylated heparan sulfate

The transcriptional profile of epithelial cells exposed to ambient hypoxia was assessed in RNA derived from control or hypoxic epithelia (T84 cells at 6 or 18 h hypoxia) using quantitative genechip expression arrays (Affymetrix) [21].

T84 cells were grown to confluence on 5-cm2 polycarbonate rings, exposed to experimental hypoxia conditions, as indicated, washed with HBSS followed by labeling of extracellular cell surface proteins with biotin (NHS-Biotin, 1 mM,

252

Journal of Leukocyte Biology Volume 68, August 2000

http://www.jleukbio.org

Pierce), as described previously [24]. Unbound biotin was quenched with NH4Cl (50 mM) in HBSS. Labeled T84 cells were lysed with lysing buffer (150 mM NaCl, 25 mM Tris, 1 mM MgCl2, 1% Triton X-100, 1% Nonidet P-40, 5 mM EDTA, 5 ␮g/mL chymostatin, 2 ␮g/mL aprotinin, and 1.25 mM PMSF, all from Sigma). Cell debris was removed by centrifugation (10,000 g, 5 min). Lysates were pre-cleared with 50 ␮L pre-equilibrated protein G-Sepharose (Pharmacia) for 2 h. Immunoprecipitation of heparan sulfate was performed by addition of anti-heparan sulfate monoclonal antibody (clone A7L6, rat IgG2a from Upstate Biotechnology, Lake Placid, NY) for 2 h followed by 50 ␮L pre-equilibrated protein G-Sepharose overnight on an end-over-end rotator. Washed immunoprecipitates were boiled in non-reducing sample buffer (2.5% SDS, 0.38 M Tris, pH 6.8, 20% glycerol, and 0.1% bromophenol blue), separated by SDS-PAGE (10% linear gel) under non-reducing conditions and transferred to nitrocellulose through the use of standard protocols. Biotinylated proteins were labeled with streptavidin-peroxidase and visualized by enhanced chemiluminescence (ECL; Amersham). Resulting heparan sulfate bands were quantified from scanned images using NIH Image software (Bethesda, MD).

PMN transepithelial migration PMN transmigration into and across confluent intestinal epithelial monolayers was examined as previously described in detail [25, 26]. Briefly, human PMN were isolated from normal human volunteers by a gelatin sedimentation technique [27] and suspended in modified HBSS (without Ca2⫹ and Mg2⫹, with 10 mM HEPES, pH 7.4, Sigma) at a concentration of 5 ⫻ 107/mL. T84 monolayers, plated in the inverted position to allow PMN interaction with the physiologically relevant basolateral surface, were washed free of media with HBSS (containing Ca2⫹ and Mg2⫹). Transmigration assays were performed by the addition of PMN to the upper chambers after chemoattractant (1 ␮M fMLP) was added to the opposing (lower) chambers. At time 0, 1 ⫻ 106 PMN were added and transmigration was allowed to proceed for 2 h at 37°C. In subsets of experiments, PMN were pre-incubated with anti-IL-8 R mAb (anti-CXCR-1, clone 42705.111, subclass IgG2a from R & D Sysytems, Minneapolis, MN) at indicated concentrations for 30 min at 4°C before addition to epithelia. All experiments were performed at 37°C. PMN migration into epithelial monolayers was quantified by assaying for the PMN azurophilic granule marker myeloperoxidase (MPO) as described previously [26]. After each transmigration assay, nonadherent/loosely adherent PMN were extensively washed from the surface of the monolayer and PMN cell equivalents, estimated from a standard curve, were assessed as the number of PMN associated with the monolayer. Previous studies have clearly demonstrated that the morphological location of monolayer-associated PMN under these conditions is within the epithelial paracellular space and not adherent to the permeable support or the apical membrane surface [25, 28].

PMN CD11b/18 expression The influence of epithelia on PMN CD11b/18 was assessed by flow cytometry during co-incubation of PMN with intestinal epithelial cells. Briefly, 2.5 ⫻ 106 PMN were co-incubated with 1.25 ⫻ 107 T84 cells (pre-exposed to hypoxia or hypoxia with added antisense oligonucleotides) for 20 min at 37°C. Cells were immediately washed in HBSS, and incubated with anti-CD11b monoclonal antibody (clone 44a, subclass IgGsa obtained from American Type Tissue Collection) followed by FITC-conjugated secondary antibody. CD11b/18 was analyzed on PMN from co-incubations by gating flow cytometer on PMN.

Data presentation ELISA data were compared by analysis of variance (ANOVA) or by Student’s t test. Values are expressed as the mean and SEM of n experiments. Flow cytometry data were expressed as mean fluorescence intensity (MFI) and analyzed by the Kolmogorov-Smirnov (K-S) two-sample test.

RESULTS Induction of soluble and surface IL-8 by hypoxia Epithelia are demonstrated sources of bioactive chemokines [13]. Here, we explored the possibility that hypoxia induces epithelial surface IL-8. T84 cells grown on permeable supports

were exposed to ambient oxygen tension (pO2 150 torr, 72 h) or hypoxia (pO2 20 torr, 12–72 h). Cell culture supernatants were harvested, pooled, and assayed for IL-8 by capture ELISA, and washed monolayers were examined for expression of IL-8 by cell ELISA. As shown in Figure 1A, and consistent with previous data [15], hypoxia induced a time-dependent increase in soluble IL-8 production (P ⬍ 0.001 by ANOVA), with maximal production at 48 h. Similarly, as shown in Figure 1B, monolayers grown in this fashion revealed a parallel increase in surface-bound (monolayer-associated) IL-8 (P ⬍ 0.01 by ANOVA) and mRNA induction by hypoxia (maximal at 12 h hypoxia, see Fig. 1C). It has been previously demonstrated that soluble chemokines bind to cell matrix proteins, and that such bound chemokine can serve as a recruitment signal for PMN [14, 15]. Thus, it is possible that the cell-associated IL-8 demonstrated in Figure 1B could reflect matrix-associated IL-8. To circumvent this potentially confounding issue, cells pre-exposed to hypoxia were elicited from permeable supports using EGTA and examined for surface-bound chemokine by flow cytometry. As shown in Figure 2, hypoxia induced a time-dependent induction (P ⬍ 0.01 by ANOVA) of surface-bound IL-8 examined under these conditions, with maximal expression at 48 h of hypoxia (MFI of 130.5 vs. 8.9 for 48 h hypoxia and control normoxia, respectively, P ⬍ 0.01). It is important to note that control normoxic cells expressed significant cellsurface IL-8 (MFI of 8.9 vs. 2.2 for normoxia and background control, respectively, P ⬍ 0.05). As further verification for hypoxia-elicited induction of surface expressed chemokine, plasma membrane fractions from control normoxic cells, and cells pre-exposed to hypoxia (24 or 48 h) were probed for IL-8 through the use of immunoprecipitation of biotinylated protein. As shown in Figure 2B, a clear induction of IL-8 is evident in plasma membrane fractions from hypoxic cells (densitometry values of 31.5 ⫾ 8.8 and 63.2 ⫾ 14.5 relative densitometry units for 24 and 48 h hypoxia, respectively, compared to 0.6 ⫾ 2.2 for controls, P ⬍ 0.05, n ⫽ 3 experiments). Similar analysis of biotinylated recombinant IL-8 (100 ng total protein) treated in the same manner revealed a corresponding band at a similar molecular weight. These data indicate the likelihood that a fraction of hypoxiaelicited IL-8 remains tightly associated with the epithelial plasma membrane.

Influence of IL-8 antisense oligonucleotides on cell-surface chemokine To better define the role of hypoxia in induction of IL-8 surface-bound chemokine, IL-8 mRNA was blocked in hypoxic cells with antisense oligonucleotide probes. As we have demonstrated previously [15], the conditions for loading oligonucleotides do not influence cell viability (judged morphologically and assessed by transepithelial resistance, a sensitive measure of epithelial viability, data not shown). As shown in Figure 3, hypoxic epithelia incubated with IL-8 antisense oligonucleotides for 48 h result in significantly decreased surface IL-8 (MFI of 620.5 vs. 12.8 for 48 h hypoxia and 48 h hypoxia pre-loaded with IL-8 anti-sense oligonucleotides, respectively, P ⬍ 0.001). As a control, pre-loading epithelia with a scrambled oligonucleotide did not influence surface IL-8

Furuta et al. Bioactive surface IL-8 on hypoxic intestinal epithelia

253

Fig. 1. Hypoxia induces soluble and monolayer-associated IL-8, and IL-8 mRNA. Monolayers of the human intestinal epithelial cell line T84 were grown on polycarbonate permeable supports. Confluent monolayers were exposed to hypoxia (pO2 20 torr, 0 – 48 h). (A) Soluble supernatants were collected from the basolateral surface and assayed for IL-8 by capture ELISA. (B) Washed monolayers were assessed for monolayer-associated IL-8 by ELISA. Results represent the mean ⫾ SEM for 10 –12 individual monolayers in each condition. Values are reported as IL-8 per 106 cells relative to a standard curve. (C) RT-PCR was used to determine IL-8 message levels in response to hypoxia. After exposure to indicated periods of hypoxia, total RNA was isolated, DNAse I treated, and amplified by RT-PCR using IL-8 specific primers. As indicated, lanes represent (from left to right); normoxia, 2-, 6-, 12-, and 24-h exposure to hypoxia, respectively. In the bottom panel, corresponding ␤-actin mRNA expression are shown to demonstrate equivalent loading. Data shown are representative of three experiments.

expression (data not shown). No differences were observed between control normoxia cells and 48 h hypoxia pre-loaded with IL-8 anti-sense oligonucleotides (MFI of 9.3 vs. 12.8, respectively, P ⫽ not significant). These data indicate that IL-8 antisense oligonucleotides prevent hypoxia-induced surface IL-8.

Role of heparan sulfate in cell-surface IL-8 Chemokines such as IL-8 bind with high specificity to the polysaccharide components of cell-surface and extracellularmatrix proteoglycans, particularly heparan sulfate proteoglycans [7]. Thus, we determined whether hypoxia-induced cellsurface IL-8 was diminished by enzymatic degradation of heparan sulfate on intact cells. As shown in Figure 4, heparinase III significantly decreased hypoxia-induced IL-8 expression (MFI of 54.6 vs. 280.5 in the presence and absence of heparinase III treatment, respectively, P ⬍ 0.025). Similar treatment of normoxic cells (Fig. 4, CTL) also decreased surface IL-8 staining (MFI of 3.4 vs. 3.9 for normoxic control treated with heparinase and 2° Ab only, respectively, P ⫽ not significant), suggesting that baseline surface expression of IL-8 in normoxic cells is likely mediated by heparan sulfate. These 254

Journal of Leukocyte Biology Volume 68, August 2000

data suggest that IL-8 binds to a surface component consistent with heparan sulfate.

Hypoxia induces epithelial proteoglycans; role for intracellular cAMP Increased cell-surface IL-8 expression may be only in part explained by hypoxia-induced IL-8. For instance, conditions other than hypoxia which induce soluble IL-8 (e.g., phorbol myristic acetate, 10 ng/mL) [15] do not result in increased surface IL-8 (data not shown), suggesting additional factors in this observation. It is possible, therefore, that hypoxia could influence expression of surface proteoglycans, providing a dual mechanism for increased surface IL-8 accumulation (i.e., parallel induction of both IL-8 and proteoglycan). A transcriptional profiling approach was adopted to identify induction of proteoglycans that might contribute to surface-expressed chemokine. Microarray analysis [29] was utilized to broadly screen hypoxia-regulated genes in RNA derived from intestinal epithelia (T84 cells). This approach identified a time-dependent induction of the epican (3.2- and 5.4-fold increase over control normoxia at 6 and 18 h hypoxia, respectively) and perlecan (1.6- and 3.3-fold increase at 6 and 18 h hypoxia, respectively), http://www.jleukbio.org

Fig. 3. Pre-loading epithelia with IL-8 antisense oligonucleotides (ASO) blocks hypoxia-induced surface IL-8. Monolayers of T84 cells were grown on polycarbonate permeable supports. Confluent monolayers were pre-loaded with IL-8 ASO (200 nM) or mock treated with DOTMA followed by exposure to hypoxia (pO2 20 torr, 48 h, as indicated). Surface IL-8 was assessed from EGTA-elicited epithelial cells by flow cytometry. Also indicated is the secondary antibody only control (2° only) and the normoxic epithelial control (CTL). Results are representative of three experiments.

Fig. 2. Evaluation of epithelial membrane-associated IL-8. Monolayers of T84 cells were grown on polycarbonate permeable supports. Confluent monolayers were exposed to hypoxia (pO2 20 torr, 0 – 48 h, as indicated). (A) Surface IL-8 was assessed from EGTA-elicited epithelial cells by flow cytometry. Also indicated is the secondary antibody only control (2° only). (B) Plasma membrane preparations were generated from epithelial cells pre-exposed to indicated hypoxia. Membranes were biotinylated, re-pelleted, and IL-8 was immunoprecipitated from solubilized fractions. Washed immunoprecipitates were boiled in non-reducing sample buffer, resolved by non-reducing SDS-PAGE (18% polyacrylamide gel), transferred to nitrocellulose, and blocked overnight in blocking buffer. Biotinylated proteins were probed with streptavidin-peroxidase and visualized by enhanced chemiluminescence (ECL). Also indicated is a biotinylated IL-8 standard for reference. Results are representative of four experiments (A) and three experiments (B).

but no influence on syndecan (0.9- and 1.1-fold increase at 6 and 18 h hypoxia, respectively). RT-PCR analysis was employed to verify these microarray results (Fig. 5A) and revealed a time-dependent induction of epican and perlecan mRNA expression, but no obvious change in syndecan expression. Immunoprecipitation of perlecan from surface biotinylated preparations was performed on T84 cells exposed to hypoxia (0 – 48 h), conditions that result in increased surface IL-8 expression (Fig. 1 and 2). As shown in Figure 5B, increased expression of two proteins consistent with perlecan (50 and 90 kDa) [30] were observed with increasing time of hypoxia. Both the ⬃50- and ⬃90-kDa proteins were present in control normoxic cells, but significantly induced by hypoxia at 36 – 48 h (determined by densitometry, increase of 3.7 ⫾ 0.9 and 4.2 ⫾ 1.1 for the 50- and 90-kDa proteins, respectively, P ⬍ 0.02 for both). The ⬃68-kDa protein associated with 24and 36-h hypoxic cells was inconsistently observed. We and others have previously defined a role for intracellular cAMP in regulation of gene expression by hypoxia [22,

31–33]. This cAMP influence maps to genes that bear a cAMP response element (CRE) [22, 34]. Our analysis revealed that the cloned perlecan gene [30] bears a previously unappreciated CRE (TGACGTGG). Similar analysis of epican did not reveal such a CREB binding site (data not shown). Decreased intracellular cAMP is commonly associated with cellular hypoxia [35]. Moreover, previous studies have determined that elevation or maintenance of cAMP levels in hypoxic cells can prevent induction of CRE-bearing gene products [22, 31, 36]. Thus, we determined whether co-treatment of epithelia with hypoxia and the adenylate cyclase agonist forskolin reversed the influence of hypoxia on induction of heparan sulfate. As shown in Figure 6, ELISA analysis of surface perlecan re-

Fig. 4. Heparinase treatment of epithelia decreases surface expression of IL-8 elicited by hypoxia. Monolayers of the human intestinal epithelial cell line T84 were grown on polycarbonate permeable supports. Confluent monolayers were exposed to hypoxia (pO2 20 torr, 48 h), EGTA-elicited (as described above), and incubated in the presence or absence of heparinase III (1 U/mL) for 2 h at 37°C. Monolayers were extensively washed at 4°C, and assessed for surface-expressed IL-8 by flow cytometry. Also indicated is the secondary antibody-only control (2° only) and the normoxic epithelial control (exposed to similar conditions of heparinase, CTL). Results are representative of three experiments.

Furuta et al. Bioactive surface IL-8 on hypoxic intestinal epithelia

255

Fig. 5. Hypoxia induces mRNA and surface expression of proteoglycans. T84 monolayers were cultured to confluency on 5-cm2 permeable supports and exposed to indicated periods of hypoxia (pO2 20 torr). (A) Total RNA was isolated and examined for perlecan, epican, and syndecan transcript by RTPCR. ␤-Actin transcript was used as an internal control. (B) Cells were cooled to 4°C and labeled (apical and basolateral) with biotin followed by immunoprecipitation of heparan sulfate using clone A7L6 mAb. Immunoprecipitates were subjected to SDS-PAGE on a 10% polyacrylamide gel followed by Western blotting, incubation with streptavidinperoxidase, and development by enhanced chemiluminescence. Prominent 50- and 90-kDa bands are observed (open arrows). The ⬃68-kDa protein associated with 24 and 36 h hypoxic cells was inconsistently observed. Results are representative of three experiments.

vealed that increasing concentrations of forskolin resulted in decreased hypoxia-elicited heparan sulfate expression (64 ⫾ 13% decrease with 10 ␮M forskolin, P ⬍ 0.025). Similar to previous results with rat glomerular epithelia [37], forskolin suppressed T84 cell surface expression of perlecan in normoxic cells (45 ⫾ 7% decrease with 10 ␮M forskolin, P ⬍ 0.05). These data indicate that hypoxia induces epithelial perlecan in a cAMP-regulated fashion.

Bioactivity of surface-expressed IL-8 We next defined whether hypoxia-induced, epithelial surface IL-8 was functional. To do this, two experimental approaches were used. First, we determined whether hypoxia influenced PMN accumulation within epithelial monolayers, and whether anti-IL-8 receptor antibodies would influence such responses. As shown in Figure 7A, and consistent with previous studies [15], exposure of T84 epithelial monolayers to hypoxia and subsequent assessment of PMN accumulation within washed monolayers revealed significantly increased PMN migration into epithelial monolayers (4.9 ⫾ 0.88 ⫻ 104 for normoxic Fig. 6. Hypoxia-induced surface expression of perlecan is diminished by the cAMP agonist forskolin. Monolayers of the human intestinal epithelial cell line T84 were grown on polycarbonate permeable supports. Confluent monolayers were exposed to normoxia (pO2 150 torr, 48 h) or hypoxia (pO2 20 torr, 48 h) in the presence or absence of indicated concentrations of the adenylate cyclase agonist forskolin. Washed monolayers were assessed for monolayerassociated perlecan by ELISA. Results represent the mean ⫾ SEM for 8 –10 individual monolayers in each condition.

256

Journal of Leukocyte Biology Volume 68, August 2000

controls vs. 13 ⫾ 1.20 ⫻ 104 PMN for hypoxia, P ⬍ 0.001). Previous morphological studies with this model system have demonstrated that PMN accumulate within the paracellular space of epithelia [25, 38]. Pre-incubation of PMN with antiIL-8 receptor antibodies resulted in a concentration-dependent inhibition of PMN accumulation in monolayers pre-exposed to hypoxia (ANOVA, P ⬍ 0.01) but not normoxia (ANOVA, P ⫽ not significant). As a second analysis, we determined whether co-incubations of PMN and epithelia bearing surface IL-8 would influence PMN CD11b/18, because IL-8 is known to up-regulate this molecule on PMN [39]. Epithelia were pre-exposed to conditions that increase surface IL-8 (48 h hypoxia) or to similar conditions in the presence of antisense IL-8 (to block increased surface IL-8, see Fig. 3). Cells were lifted from permeable support membranes (see Materials and Methods) and incubated with PMN at a 5:1 epithelia-to-PMN ratio. CD11b/18 expression on PMN was examined by flow cytometry (gating on PMN). As can be seen in Figure 7B, CD11b/18 expression was induced by more than 10-fold by incubation with epithelia pre-exposed to hypoxia (MFI of 11.9 vs. 160.7 for basal and epithelial-exposed PMN, respectively, P ⬍ 0.01). These results represent a shift in magnitude approximating that induced by 10 ng of recombinant human IL-8 (MFI 145.6 , n ⫽ 4, P ⬍ 0.01 compared to unactivated PMN). Moreover, this shift was significantly diminished when PMN were incubated with cells pre-exposed to the combination hypoxia and IL-8 antisense oligonucleotides (MFI of 58.9 vs. 160.7 in the presence and absence of IL-8 antisense oligonucleotides, respectively, P ⬍ 0.025), indicating that induction of PMN CD11b/18 is, at least in part, dependent on surface IL-8. Experiments addressing whether epithelia and/or PMN release soluble IL-8 during such co-culture revealed no detectable IL-8 in co-culture supernatants (examined by ELISA, lower limit of detection ⬍10 pg/ mL, data not shown). Taken together, these experiments indicate that hypoxia-elicited, epithelial-associated IL-8 moduhttp://www.jleukbio.org

Fig. 7. Functional assessment of hypoxia-induced surface IL-8. (A) Monolayers of the T84 cells were grown as inverted monolayers (i.e., basolateral surface upward) on permeable supports. Confluent monolayers were exposed to normoxia or hypoxia (pO2 20 torr, 48 h) and reoxygenated with buffer containing PMN (1 ⫻ 106/monolayer) pre-incubated with indicated concentrations of anti-IL-8 R mAb (CXCR-1). PMN were allowed to migrate for 2 h at 37°C. Transmigration was quantitated by assaying for monolayer-associated (the number of PMN that remain firmly associated with washed monolayers) PMN myeloperoxidase. Results represent the mean ⫾ SEM for seven to nine individual monolayers in each condition. (B) Monolayers of T84 cells were grown on polycarbonate permeable supports. Confluent monolayers were pre-loaded with IL-8 ASO (200 ng/mL) or mock treated with DOTMA followed by exposure to hypoxia (pO2 20 torr, 48 h). Cells were EGTA-elicited and co-incubated with PMN (5:1 epithelia-to-PMN ratio) for 15 min at 37°C. Washed co-incubations were assessed for PMN CD11b/18 (Mac-1) by gating on PMN with flow cytometry. Also indicated is the secondary antibody-only control (2° only) and baseline PMN Mac-1 expression (basal). Results are representative of three experiments.

lates PMN-epithelial interactions and that surface-associated chemokine may activate PMN in this setting.

DISCUSSION Chemokine and growth factor binding to extracellular matrices provides a mechanism for recruitment and activation of leuko-

cytes at tissue sites. Many cell types also express surface proteoglycans, although less is understood regarding their role in binding chemokines and whether such cell surface chemokine is functional in the setting of complex tissue architechture. We report here that epithelial hypoxia activates parallel induction of IL-8 and proteoglycan mRNA and surface expression. Extensions of these initial findings reveal that hypoxiaelicited induction of surface-tethered chemokine functionally recruits and activates PMN localized within epithelial paracellular space. Leukocyte recruitment to tissue sites is highly ordered. After extravasation from the vascular space, leukocytes migrating within mucosal tissue respond to locally generated signals for successful transit to the epithelium. These signals may exist in soluble forms (e.g., fMLP) or may present themselves bound to matrix proteoglycans (e.g., chemokines). Given its high negative charge, heparan sulfate readily interacts with proteins and such interactions are increasingly implicated in a variety of physiological processes [10]. Previous studies, for instance, have indicated that chemokine bioactivity in vitro may not be predictive in vivo [11], suggesting that additional molecules contribute to chemokine function. A likely mechanism is the selective association of chemokines with tissues specialized to recruit specific leukocyte subpopulations. We have previously demonstrated that epithelial hypoxia activates basolateral release of IL-8, and that IL-8 binds to epithelial extracellular matrix (i.e., is detected in acellular monolayers) [15]. The findings of chemokine binding to proteoglycans are not unique. A number of studies have directly demonstrated chemokine binding to glycosaminoglycan subpopulations on numerous cell types [7, 9]. In some cases, proteoglycan-bound chemokine may show enhanced bioactivity relative to the soluble forms of the molecule [7], and in one published finding, the chemokine interferon-␥-inducible protein-10 (IP-10) was demonstrated to use proteoglycans as a signal-transducing receptor [40]. Similarly, nuclear magnetic resonance analysis of IL-8-IL-8 receptor complexes revealed that the heparin-binding residues of IL-8 are exposed, thus allowing for the distinct possibility that surface proteoglycans can present chemokine in a form that binds to the receptor [41]. Our studies indicate that epithelial surface-bound IL-8 recruits PMN to the basolateral surface of intact epithelia and that such surface-tethered chemokine provides a pathway for induction of PMN CD11b/18 expression. We do not know, however, whether chemokine binding to surface proteoglycan enhances chemokine activity. Accurate estimates of chemokine concentrations expressed on the cell surface are difficult to obtain with this model, limiting the direct comparison of PMN CD11b/18 activation by soluble IL-8. Thus, at the present time, we do not know whether IL-8 binding to epithelial surface heparan sulfate enhances IL-8 bioactivity. Previous studies indicate that hypoxia induces chemokine generation in endothelia [42] and in epithelia [15]. At present the mechanism(s) of such activation are not certain, but likely involves activation mediated by the transcription factors cAMP response element binding protein (CREB) [22] and nuclear factor-␬B [22, 42]. Indeed, our analysis of the cloned IL-8 gene [23] revealed that it too bears a CRE (TTTCGTCA) 150 base pairs upstream from the TATA signal in the human IL-8 gene.

Furuta et al. Bioactive surface IL-8 on hypoxic intestinal epithelia

257

Consistent with this observation, it was recently revealed that cellular hypoxia induces activation of a number of genes bearing CREB binding sites at or near the promoter [22, 34], including E-selectin [32], tumor necrosis factor ␣ [22, 31], COX-2 [43], and MHC class II [22, 31]. A similar hypoxiainduced activation pattern was observed with perlecan, which also bears a CRE (TGACGTGG) upstream of the transcription start site [30]. Moreover, perlecan has been demonstrated to be cAMP responsive [37]. Of note, transcriptional analysis also revealed that epican is prominently induced by hypoxia (Fig. 5), although mechanisms of such induction remain unclear. Taken together, these studies indicate the likelihood that hypoxia activates a cAMP-regulated IL-8 and perlecan expression in epithelia. A strong correlation between reperfusion injury and the acute inflammatory response exists [3]. The direct role of PMN has been implicated in tissue damage resulting from reperfusion injury, although mechanisms remain only partially understood [3]. In the intestine for instance, PMN accumulation at the level of the epithelium has been shown to play a central role in mucosal injury during intestinal reperfusion injury [44 – 46]. In disorders such as Crohn’s disease, significant evidence exists that ischemia may substantially contribute to tissue damage [47– 49]. Thus, it seems likely that intestinal ischemia and PMN-epithelial interactions may be related events. Here, we demonstrate that hypoxia-induced surface IL-8 functionally activates PMN CD11b/18 expression. This observation has significance for a number of reasons. First, CD11b/18 has been extensively studied as an important adhesion molecule for PMN adhesion to, and transmigration across intestinal epithelia [50], and that hypoxia enhances PMN transmigration. Second, although we have previously demonstrated that PMN transmigration across epithelia pre-exposed to hypoxia is CD11b/18-dependent [15], the ligand for CD11b/18 on epithelia remains elusive. Third, the proteoglycan heparin (a glycosaminoglycan structurally related to heparan sulfate) is a demonstrated ligand for PMN CD11b/18 [51], thus it is tantalizing to speculate that heparan sulfate may also serve as a surface ligand for PMN CD11b/18. Using monoclonal antibodies directed against perlecan, we have not observed inhibition of PMN transmigration (data not shown). However, this approach is limited by the fact that these monoclonals may not bind to the perlecan epitope necessary to block function. In conclusion, epithelial exposure to hypoxia activates the cAMP-regulated induction of both heparan sulfate and IL-8, resulting in the expression of a surface-bound form of chemokine that functionally activates PMN migration and CD11b/18. Thus, these studies contribute to the growing body of evidence that cAMP-responsive genes may play an important role in mediating crosstalk between hypoxia and inflammatory events.

ACKNOWLEDGMENTS This work was supported by National Institutes of Health Grants DK-50189, HL-60569, DK-02564, DK-02682, project 3 of PO-1 DE 13499, and by a grant from the Crohn’s and Colitis Foundation of America. 258

Journal of Leukocyte Biology Volume 68, August 2000

REFERENCES 1. Yardley, J. H., Donowitz, M. (1977) Colo-rectal biopsies in inflammatory bowel disease. In The Gastrointestinal Tract (J. H. Yardley and B. C. Morson, eds.) Baltimore, MD: Williams and Wilkins, 57–94. 2. Yardley, J. H. (1986) Pathology of idiopathic inflammatory bowel disease and relevance of specific cell findings: an overview. In Recent Developments in the Therapy of Inflammatory Bowel Disease (J. H. Yardley, ed.) Baltimore, MD: Johns Hopkins, 3–9. 3. Welbourne, C. R. B., Goldman, G., Valeri, C. R., Shepro, D., Hechtman, H. B. (1991) Pathophysiology of ischaemia reperfusion injury: central role of the neutrophil. Br. J. Surg. 78, 651– 655. 4. Hawker, P. C., McKay, J. S., Turnberg, L. A. (1980) Electrolyte transport across colonic mucosa from patients with inflammatory bowel disease. Gastroenterology 79, 508 –511. 5. Taylor, C. T., Colgan, S. P. (1999) Therapeutic targets for hypoxia-elicited pathways. Pharmacol. Res. 16, 1498 –1505. 6. Baggiolini, M. (1998) Chemokines and leukocyte traffic. Nature 392, 565–568. 7. Witt, D. P., Lander, A. D. (1994) Differential binding of chemokines to glycosaminoglycan subpopulations. Curr. Biol. 4, 394 – 400. 8. Amara, A., Lorthioir, O., Valenzuela, A., Magerus, A., Thelen, M., Montes, M., Virelizier, J. L., Delepierre, M., Baleux, F., Lortat-Jacob, H., Arenzana-Seisdedos, F. (1999) Stromal cell-derived factor-1␣ associates with heparan sulfates through the first beta-strand of the chemokine. J. Biol. Chem. 274, 23916 –23925. 9. Kuschert, G. S., Hoogewerf, A. J., Proudfoot, A. E., Chung, C. W., Cooke, R. M., Hubbard, R. E., Wells, T. N., Sanderson, P. N. (1998) Identification of a glycosaminoglycan binding surface on human interleukin-8. Biochemistry 37, 11193–11201. 10. Lander, A. D. (1998) Proteoglycans: master regulators of molecular encounter? Matrix Biol. 17, 465– 472. 11. Miller, M. D., Krangel, M. S. (1991) Biology and biochemistry of the chemokines: a family of chemotactic and inflammatory cytokines. Annu. Rev. Immunol. 9, 617– 648. 12. MacDermott, R. P., Sanderson, I. R., Reinecker, H. C. (1998) The central role of chemokines (chemotactic cytokines) in the immunopathogenesis of ulcerative colitis and Crohn’s disease. Inflamm. Bowel Dis. 4, 54 – 67. 13. Kagnoff, M. F., Eckmann, L. (1997) Epithelial cells as sensors for microbial infection. J. Clin. Invest. 100, 6 –10. 14. McCormick, B. A., Hofman, P. M., Kim, J., Carnes, D., Miller, S. I., Madara, J. L. (1995) Surface attachment of Salmonella typhimurium to intestinal epithelia imprints the subepithelial matrix with gradients chemotactic for neutrophils. J. Cell Biol. 131, 1599 –1608. 15. Colgan, S. P., Dzus, A. L., Parkos, C. A. (1996) Epithelial exposure to hypoxia modulates neutrophil transepithelial migration. J. Exp. Med. 184, 1003–1015. 16. Parkos, C. A., Colgan, S. P., Liang, A., Nusrat, A., Bacarra, A. E., Carnes, D. K., Madara, J. L. (1996) CD 47 mediates post-adhesive events required for neutrophil migration across polarized intestinal epithelia. J. Cell Biol. 132, 437– 450. 17. Dharmsathaphorn, K., Madara, J. L. (1990) Established intestinal cell lines as model systems for electrolyte transport studies. Meth. Enzymol. 192, 354 –389. 18. Madara, J. L., Colgan, S. P., Nusrat, A., Delp, C., Parkos, C. A. (1992) A simple approach to measurement of electrical parameters of cultured epithelial monolayers: use in assessing neutrophil epithelial interactions. J. Tissue Culture Res. 14, 209 –216. 19. Parkos, C. A., Colgan, S. P., Bacarra, A. E., Nusrat, A., Delp-Archer, C., Carlson, S., Su, D. H. C., Madara, J. L. (1995) Model human intestinal epithelial (T84) possess basolateral ligands for CD11b/18 mediated neutrophil adhesion. Am. J. Physiol. 268, C472–C479. 20. Colgan, S. P., Parkos, C. A., McGuirk, D., Brady, H. R., Papayianni, A. A., Frendl, G., Madara, J. L. (1995) Receptors involved in carbohydrate binding modulate intestinal epithelial-neutrophil interactions. J. Biol. Chem. 270, 10531–10539. 21. Lockhart, D. J., Dong, H., Byrne, M. C., Follettie, M. T., Gallo, M. V., Chee, M. S., Mittmann, M., Wang, C., Kobayashi, M., Horton, H., Brown, E. L. (1996) Expression monitoring by hybridization to high-density oligonucleotide arrays [see comments]. Nat. Biotechnol. 14, 1675–1680. 22. Taylor, C. T., Fueki, N., Agah, A., Hershberg, R. M., Colgan, S. P. (1999) Critical role of cAMP response element binding protein expression in hypoxia-elicited induction of epithelial TNF-␣. J. Biol. Chem. 274, 19447–19450. 23. Mukaida, N., Shiroo, M., Matsushima, K. (1989) Genomic structure of the human monocyte-derived neutrophil chemotactic factor IL-8. J. Immunol. 143, 1366 –1371.

http://www.jleukbio.org

24. LeBivic, A., Francisco, X. R., Rodriguez-Boulan, E. (1989) Vectorial targeting of apical and basolateral plasma membrane proteins in a human adenocarcinoma epithelial cell line. Proc. Natl. Acad. Sci. USA 86, 9313–9317. 25. Parkos, C. A., Delp, C., Arnaout, M. A., Madara, J. L. (1991) Neutrophil migration across a cultured intestinal epithelium: Dependence on a CD11b/CD18-mediated event and enhanced efficiency in the physiologic direction. J. Clin. Invest. 88, 1605–1612. 26. Colgan, S. P., Parkos, C. A., Delp, C., Arnaout, M. A., Madara, J. L. (1993) Neutrophil migration across cultured intestinal epithelial monolayers is modulated by epithelial exposure to interferon-gamma in a highly polarized fashion. J. Cell. Biol. 120, 785–795. 27. Henson, P., Oades, Z. G. (1975) Stimulation of human neutrophils by soluble and insoluble immunoglobulin aggregates. J. Clin. Invest. 56, 1053–1061. 28. Nash, S., Parkos, C. A., Nusrat, A., Delp, C., Madara, J. L. (1991) In vitro model of intestinal crypt abscess: a novel neutrophil-derived secretagogue activity. J. Clin. Invest. 87, 1474 –1477. 29. Zweiger, G. (1999) Knowledge discovery in gene-expression-microarray data: mining the information output of the genome. Trends Biotechnol. 17, 429 – 436. 30. Cohen, I. R., Grassel, S., Murdoch, A. D., Iozzo, R. V. (1993) Structural characterization of the complete human perlecan gene and its promoter. Proc. Natl. Acad. Sci. USA 90, 10404 –10408. 31. Taylor, C. T., Dzus, A. L., Colgan, S. P. (1998) Autocrine regulation of intestinal epithelial permeability induced by hypoxia: Role for basolateral release of tumor necrosis factor-␣ (TNF-␣). Gastroenterology 114, 657– 668. 32. Zu¨nd, G., Nelson, D. P., Neufeld, E. J., Dzus, A. L., Bischoff, J., Mayer, J. E., Colgan, S. P. (1996) Hypoxia enhances stimulus-dependent induction of E-selectin on aortic endothelial cells. Proc. Natl. Acad. Sci. USA 93, 7075–7080. 33. Pinsky, D. J., Oz, M. C., Liao, H., Morris, S., Brett, J., Morales, A., Karakurum, M., Campagne, M. M. V. L., Nowygrod, R., Stern, D. M. (1993) Restoration of the cAMP second messenger pathway enhances cardiac preservation for transplantation in a heterotopic rat model. J. Clin. Invest. 92, 2994 –3002. 34. Beitner-Johnson, D., Millhorn, D. E. (1998) Hypoxia induces phosphorylation of the cyclic AMP response element-binding protein by a novel signaling mechanism. J. Biol. Chem. 273, 19834 –19839. 35. Tretyakov, A. V., Farber, H. W. (1995) Endothelial cell tolerance to hypoxia: Potential role of purine nucleotide phosphates. J. Clin. Invest. 95, 738 –744. 36. Taylor, C. T., Lisco, S. J., Awtrey, C. S., Colgan, S. P. (1998) Hypoxia inhibits cyclic nucleotide-stimulated epithelial ion transport: role for nucleotide cyclases as oxygen sensors. J. Pharmacol. Exp. Ther. 284, 568 –575.

37. Ko, C. W., Bhandari, B., Yee, J., Terhune, W. C., Maldonado, R., Kasinath, B. S. (1996) Cyclic AMP regulates basement membrane heparan sulfate proteoglycan, perlecan, metabolism in rat glomerular epithelial cells. Mol. Cell. Biochem. 162, 65–73. 38. Nash, S., Stafford, J., Madara, J. L. (1987) Effects of polymorphonuclear leukocyte transmigration on barrier function of cultured intestinal epithelial monolayers. J. Clin. Invest. 80, 1104 –1113. 39. Baggiolini, M., DeWald, B., Moser, B. (1994) Interleukin-8 and related chemotactic cytokines- CXC and CC chemokines. Adv. Immunol. 55, 97–179. 40. Luster, A. D., Greenberg, S. M., Leder, P. (1995) The IP-10 chemokine binds to a specific cell surface heparan sulfate site shared with platelet factor 4 and inhibits endothelial cell proliferation. J. Exp. Med. 182, 219 –231. 41. Skelton, N. J., Quan, C., Reilly, D., Lowman, H. (1999) Structure of a CXC chemokine-receptor fragment in complex with interleukin-8. Structure Fold Des. 7, 157–168. 42. Karakurum, M., Shreeniwas, R., Chen, J., Pinsky, D., Yan, S.-D., Anderson, M., Sunouchi, K., Major, J., Hamilton, T., Kuwabara, K., Rot, A., Nowygrod, R., Stern, D. (1994) Hypoxic induction of interleukin-8 gene expression in human endothelial cells. J. Clin. Invest. 93, 1564 –1570. 43. Blume, E. D., Taylor, C. T., Lennon, P. F., Stahl, G. L., Colgan, S. P. (1998) Activated endothelial cells elicit paracrine induction of epithelial chloride secretion: 6-keto-PGF1␣ is an epithelial secretagogue. J. Clin. Invest. 102, 1161–1172. 44. Haglund, U. H., Bulkley, G. B., Granger, D. N. (1987) On the pathophysiology of intestinal ischemia injury. Acta Chir. Scand. 153, 321–324. 45. Kurtel, H., Tso, P., Granger, D. N. (1992) Granulocyte accumulation in postischemic intestine: role of leukocyte adhesion glycoprotein CD11/ CD18. Am. J. Physiol. 262, G878 –G882. 46. Parks, D. A., Granger, D. N. (1986) Contributions of ischemia and reperfusion to mucosal lesion formation. Am. J. Physiol. 250, G749 – G753. 47. Wakefield, A. J., Dhillon, A. P., Rowles, P. M., Sawyer, A. M., Pitilo, R. M., Lewis, A. A. M., Pounder, R. E. (1989) Pathogenesis of Crohn’s disease: Multifocal gastrointestinal infarction. Lancet 2, 1057–1062. 48. Osborne, M. J., Hudson, M., Piasecki, C., Dhillon, A. P., Pounder, R. E., Wakefield, A. J. (1993) Crohn’s disease and anastomotic recurrence: microvascular ischaemia and anastomotic healing in an animal model. Br. J. Surg. 80, 226 –229. 49. Thompson, N. P., Wakefield, A. J., Pounder, R. E. (1995) Inherited disorders of coagulation appear to protect against inflammatory bowel disease. Gastroenterology 108, 1011–1015. 50. Parkos, C. A. (1997) Molecular events in neutrophil transepithelial migration. BioEssays 19, 865– 873. 51. Diamond, M. S., Alone, M., Parkos, C. A., Quinn, M., Springer, T. A. (1995) Heparin is an adhesive ligand for the neutrophil ␤2 integrin Mac-1 (CD11b/18). J. Cell Biol. 130, 1473–1482.

Furuta et al. Bioactive surface IL-8 on hypoxic intestinal epithelia

259