Hypoxia-Inducible Factor 1–Dependent Induction of Intestinal Trefoil Factor Protects Barrier Function during Hypoxia

Epithelial cells are anatomically positioned to provide a barrier to antigenic flux and luminal bacteria 1. Mucosal organs, such as the intestine, rely on an extensive underlying vasculature, and therefore are primary sites for conditions of diminished blood flow and resultant tissue hypoxia 2. Under such conditions, maintenance of barrier function depends on structures within the intercellular tight junction at the boundary of the apical and basolateral membrane domain. At present, it is not known if protective mechanisms exist to support functional epithelial barrier properties during episodes of diminished blood flow and resultant hypoxia.

Both adaptive (e.g., erythropoietin [EPO], glycolytic enzymes) and proinflammatory (e.g., TNF-α, IL-8) responses have been attributed to conditions of diminished tissue oxygen delivery (hypoxia), and it is now clear that responses to hypoxia include transcriptionally regulated gene expression 2,3. One such transcriptional pathway relies on the activation of hypoxia-inducible factor (HIF)-1, a member of the rapidly growing per-ARNT-sim family of basic helix-loop-helix (bHLH) transcription factors 4,5,6. Functional HIF-1 exists as an αβ heterodimer, the activation of which is dependent on stabilization of an O₂-dependent degradation pathway 7,_. Binding of HIF-1 to DNA consensus domains (5′-RCGTG-3′) results in the transcriptional induction of HIF-1–bearing gene promoters 3. HIF-1 is widely expressed and recent studies indicate that consensus HIF-1–binding sequences exist in a number of genes 3. Here we demonstrate that hypoxia induces the epithelial-specific, barrier-protective protein intestinal trefoil factor (ITF) through an HIF-1–dependent mechanism.

T84 cells and a subclone of Caco-2 B Be cells with high epithelial resistance were grown as monolayers on polycarbonate permeable supports as described previously 8,9. The KB cell line is a human oral epithelial cell line derived from native human mucosa 10. Human microvascular endothelial cells (MVECs) are primary culture endothelial cells (Cascade Biologics, Inc.) derived from the human dermis and were cultured as described previously 11. Madin-Darby canine kidney (MDCK) cells were cultured and passaged as described previously 12. Confluent cells were exposed to hypoxia in a humidified hypoxic cell chamber (Coy Laboratory Products, Inc.) as described previously 13. Standard hypoxic conditions (based on previous work; references 13,14,15) were pO₂ = 20 Torr, pCO₂ = 35 Torr, with the balance made up of nitrogen and water vapor. Normoxic controls were cells exposed to the same experimental protocols under conditions of atmospheric oxygen concentrations (pO₂ = 147 Torr and pCO₂ = 35 Torr within a tissue culture incubator). Paracellular permeability was assessed as described previously 11 using FITC-labeled dextran (40 kD).

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 gene chip expression arrays (Affymetrix, Inc.; reference 16). Reverse transcription (RT)-PCR analysis of mRNA levels was performed using DNase-treated total RNA as described previously 14 using primers specific for human ITF 17 resulting in a 108-bp fragment or β-actin 14 resulting in a 661-bp fragment. In subsets of experiments, the transcription inhibitor actinomycin D (10 ng/ml final concentration) was used to determine the prevalence of newly synthesized RNA in this response.

After experimental treatment of epithelial cells, cell culture supernatants (for examination of soluble ITF) were prepared as described previously 11. Nuclear extracts (for analysis of HIF-1α) were isolated from confluent monolayers of epithelia on 100-mm petri dishes as described before 18. Samples (25 μg/lane) were resolved by reducing SDS-PAGE, transferred to nitrocellulose, and blocked overnight in blocking buffer (250 mM NaCl, 0.02% Tween 20, 5% goat serum, 3% BSA). For Western blotting, affinity-purified anti–HIF-1α (generated by immunizing rabbits with an antigenic peptide encoded in the HIF-1α protein structure encoding amino acids 527–541 [MVNEFKLELVEKLFA] of HIF-1α) or anti-ITF 19 was added for 3 h, blots were washed, and species-matched peroxidase-conjugated secondary Ab is added, as described previously 14. Labeled bands from washed blots were detected by enhanced chemiluminescence (Amersham Pharmacia Biotech).

Nuclear extracts of cells exposed to indicated experimental conditions were obtained as described elsewhere. Synthetic oligonucleotides were synthesized (Genosys Biotechnologies, Inc.) and used as probes in electrophoretic mobility shift assays (EMSAs). The hypoxic response element (HRE) motif lies at −129 to −122 relative to the transcription start site in the ITF promoter 17. Oligonucleotide probes for EMSA were digoxigenin-labeled, incubated with nuclear lysates, and resolved on a 6% nondenaturing polyacrylamide gel as described previously (Boehringer; reference 14). DNA–protein complexes were transblotted to nylon membrane, probed with antidigoxigenin-peroxidase, and developed by enhanced chemiluminescence. Controls consisted of free probe alone, excess unlabeled probe, or a species-matched irrelevant Ab (anti-cAMP response element–binding protein [CREB]; Transduction Labs, Inc.).

HIF-1α depletion in epithelial cells was accomplished by using antisense oligonucleotide loading as described previously 20 using phosphorothioate derivatives of antisense (GCC GGC GCC CTC CAT) or control sense (ATG GAG GGC GCC GGC) oligonucleotides. T84 epithelial cells were washed in serum free media, then in media containing 20 μg/ml effectene transfection reagent (QIAGEN) with 2 μg/ml HIF-1α antisense or sense oligonucleotide. Cells were incubated for 4 h at 37°C and then replaced with serum containing growth media. Treated cells were exposed to hypoxia or normoxia for indicated periods of time. As indicated, ITF mRNA was quantified by RT-PCR as described above (see Transcriptional Analysis).

Stable cell lines overexpressing HIF-1α (plasmid provided by H.F. Bunn, Harvard Medical School; reference 7) were generated in the high resistance Caco-2 B Be cell line as described previously 8. Primary screening for HIF-1 expression was by RT-PCR analysis. Clones with the highest expression underwent secondary screens based on electrophysiologic characteristics (high resistance monolayers) and Western blot analysis of HIF-1α.

The generation of ITF null mice has been described previously 21. Intestinal permeability was examined in 6–10-wk-old wild-type Bl6129 (Taconic, Inc.) or ITF null mice using an FITC-labeled dextran method as described previously 22. Mice were gavaged with 60 mg/100 g body weight of FITC-dextran (4,000 kD at 80 mg/ml; Sigma-Aldrich) and exposed to ambient hypoxia (8% O₂, 92% N₂) or ambient room air for various times (n = 4–6 per condition). In some experiments, mice were administered either recombinant human ITF (30 mg/kg, 75% gastric lavage, 25% enema) or PBS 1 h before hypoxia. Cardiac puncture was performed and serum analysis of FITC concentration performed. In subsets of experiments, tissues were collected from wild-type and ITF null mice after hypoxia and fixed in 10% buffered formalin. 5-μm sections were cut, stained with hematoxylin and eosin, and histologically characterized. Colonic tissue was harvested and homogenized and RNA extracted with Trizol as described above. Another section of colonic mucosal scrapings was harvested and processed for Western blot analysis as described above. This protocol was in accordance with National Institutes of Health guidelines for use of live animals and was approved by the Institutional Animal Care and Use Committee at Brigham and Women's Hospital.

ITF bioactivity and paracellular permeability data were compared by one- or two-factor analysis of variance or by Student's t test where appropriate. Values are expressed as the mean and SEM from at least three separate experiments.

Previous reports indicate that colonic epithelial barrier function is maintained during even severe conditions of hypoxia 13,14. In contrast, other cell types (e.g., vascular endothelia) appear to markedly lose barrier qualities during ambient hypoxia 23. As shown in Table, comparison of six barrier cell types revealed that colonic epithelial cells (T84 and Caco-2 cells), as opposed to other epithelia (KB, A549, and MDCK) or MVECs, appear to be uniquely resistant to increased permeability induced by hypoxia, suggesting that intestinal epithelial cells possess mechanisms that sustain barrier function under such conditions.

Transcriptional profiling has been used to identify potential factors contributing to maintenance of barrier during hypoxia 24,25. Microarray analysis 26 was adopted to broadly screen hypoxia-regulated genes in RNA derived from intestinal epithelia (T84 cells), and identified a time-dependent induction of the intestinal-specific, barrier-protective ITF gene (2.8- and 4.2-fold increase over control normoxia at 6- and 18-h hypoxia, respectively). Subsequently, RT-PCR analysis was employed to corroborate results of microarray screening (Fig. 1 a). This analysis of Caco-2 cells demonstrated a time-dependent induction of ITF mRNA expression, and was evident by 4-h exposure to hypoxia. Similar results were seen in T84 cells (data not shown). Epithelial exposure to the RNA synthesis inhibitor actinomycin D blocked this hypoxia-elicited ITF induction (Fig. 1 b), indicating that this response reflects new transcriptional activity. To confirm these mRNA observations at the protein level, Western blot analysis of soluble supernatants confirmed an increase of secreted ITF protein by 48 h (Fig. 1 b).

Prompted by these results, a search of the cloned ITF gene identified a previously unappreciated HIF-1α binding site at positions −129 to −122 relative to the transcription start site 17 (DNA consensus motif 5′-TACGTGGG-3′; reference 6). Consequently, we assessed the induction of HIF-1α and ITF by hypoxia. First, we determined whether conditions of hypoxia induce HIF-1α in intestinal epithelia (Fig. 2 a). Western blot analysis of nuclear lysates derived from hypoxic intestinal epithelia Caco-2 cells demonstrated abundant HIF-1α in nuclei. These data are consistent with previous studies suggesting that HIF-1α is expressed in most cell types during periods of hypoxia 3. Next, we examined the binding of HIF-1α to the HRE consensus in the ITF promoter by EMSA (Fig. 2 b). As can be seen, nuclear extracts derived from hypoxic, but not normoxic, Caco-2 cells bind to the HIF-1α consensus of the ITF promoter. Pretreatment of epithelia with 100 μM CoCl₂ for 4 h, conditions that promote nuclear accumulation of HIF-1α 5, resulted in a similar band shift. Addition of anti–HIF-1α mAb, but not isotype-matched mAb, resulted in a diminution of shifted signal, but no obvious supershift consistent with Ab interference with DNA–protein interaction. Incubation with 100-fold excess unlabeled primer resulted in complete loss of signal. Collectively, these results indicate that hypoxia elicits nuclear accumulation of HIF-1α, which binds to the consensus HRE located within the ITF promoter.

As proof of principle for this concept, intestinal epithelial cell lines overexpressing HIF-1α were generated by stable transfection (Fig. 3 a). As shown in Fig. 3 b, overexpression of HIF-1α resulted in increased basal expression of ITF and enhanced hypoxia-elicited induction of ITF. Next, antisense oligonucleotides directed against HIF-1α were used to block HIF-1α expression (Fig. 3 c) and the influence of HIF-1α depletion on ITF mRNA induction by hypoxia was assessed (Fig. 3 d). As can be seen, the directed loss of HIF-1α resulted in a nearly complete loss of ITF mRNA expression at 4- and 24-h periods of hypoxia. Control oligonucleotides generated to the sense strand of HIF-1α did not influence ITF hypoxia inducibility. Taken together, these results indicate that hypoxia induces a time-dependent HIF-1α nuclear accumulation and parallel ITF induction.

Given the findings described above, studies were undertaken to determine whether ITF played a functional role in maintenance of the epithelial integrity during hypoxia. For this purpose, Caco-2 monolayers were exposed to hypoxia in the presence or absence of anti-ITF (Fig. 4 a). Increasing concentrations of anti-ITF sera (titer ∼1:100,000), but not isotype-matched antisera, resulted in a dose-dependent increase in epithelial permeability in response to hypoxia (no Ab versus anti–human [h] ITF at 1:100 dilution; P < 0.001). These results demonstrate that ITF participates in the maintenance of barrier function during hypoxia. Further evidence in support of ITF as a barrier-protective agent were provided by addition of recombinant hITF to hypoxia-sensitive cells. Monolayers of endothelial cells (which respond to hypoxia with increase paracellular permeability, Table) were used to define this principle. As shown in Fig. 4 b, addition of hITF to confluent endothelial cells during conditions of hypoxia resulted in a concentration-dependent protection against barrier disruption (no hITF versus 1 mg/ml hITF; P < 0.01). These data provide direct evidence for a key role of ITF protein in barrier protection and suggest that ITF's functional action may be independent of the cell type, as illustrated here by the vascular endothelia.

To define the role of hypoxia-elicited ITF on mucosal barrier in vivo, intestinal permeability was examined in wild-type and ITF null mice 21 during 4-h exposure to ambient hypoxia (8% O₂). Under such conditions, increases in both ITF mRNA (Fig. 5 a) and protein (Fig. 5 b) were induced in tissues derived from wild-type animals subjected to hypoxia. As shown in Fig. 5 c, upon exposure to hypoxia, ITF null mice exhibited a significantly greater increase in intestinal permeability to FITC-dextran (4 kD) when compared with wild-type mice (ITF knockout versus wild-type exposed to hypoxia; P < 0.01). Interestingly, no histologically evident lesions were observed in the colonic mucosa with 4-h exposure to hypoxia (representative section from wild-type and ITF null animals are shown in Fig. 5 d). Longer periods of time in hypoxia did not significantly increase intestinal permeability beyond the 4-h time point. For example, intestinal permeability of wild-type animals subjected to hypoxia for 18 h (3.3 ± 1.2-fold increase in permeability, n = 4 animals) were not different than animals exposed to 4-h hypoxia (4.4 ± 1.3-fold increase, n = 19 animals). Moreover, these longer time periods resulted in an unexpected death in ITF null (four out of six animals tested) but not wild-type animals (zero out of six animals tested, data not shown). Interestingly, increased intestinal permeability was also evident in ITF null mice not exposed to hypoxia as compared with wild-type (P < 0.05 for ITF null compared with wild-type exposed to normoxia), consistent with the presumption that ITF may contribute to the maintenance of intestinal barrier under physiological conditions.

To further define these principles, wild-type and ITF null mice were administered recombinant ITF protein (30 mg/kg in PBS with 75% intragastric and 25% rectal infusion) before hypoxia. These studies revealed that administration of ITF to normoxic ITF null animals resulted in a significant decrease in permeability (550 ± 40 vs. 320 ± 90 ng/ml in the absence and presence of ITF administration, respectively; n = 4 mice per condition, P < 0.04). ITF addback to ITF null mice in hypoxia resulted in a trend toward resolution of increased permeability (1,310 ± 310 and 970 ± 110 ng/ml in the absence and presence of ITF administration; n = 6 mice per condition), however, these results were not statistically significant (P > 0.05). Similar to these results, ITF administration to wild-type mice in normoxia and hypoxia trended toward resolution of increased permeability but were not statistically significant (data not shown).

These studies provide the first molecular link between hypoxia and intestinal epithelial barrier protection. Here, we identify a previously unappreciated HIF-1α regulated pathway for induction of ITF. ITF is a small, protease-resistant peptide that is abundant throughout the mammalian intestinal tract 27,28. Increased ITF expression has been observed in proximity to sites of injury in the gastrointestinal tract, including peptic ulcers and active inflammatory bowel disease 27,28,29. ITF can promote epithelial barrier function, protecting against injury, and facilitating repair after damage occurs irrespective of the nature of the initial wound. However, specific transcriptional pathways for ITF induction are not well understood 30.

The hypoxia-dependent production of ITF represents a novel, innate protective mechanism that may guard the immunologic components of the lamina propria from exposure to pathogenic luminal bacteria, antigens, and toxins during episodes of diminished oxygen delivery. Accumulating evidence suggests a variety of physiological roles for ITF 28, however, molecular mechanisms regulating ITF expression have not been studied in detail. Rather, the signaling pathways activated by the addition of exogenous ITF have been more closely examined and include the inactivation of extracellular signal regulatory kinase, regulated expression of α- and β-catenin, and transactivation of the epidermal growth factor receptor 31,32,33. Recent work has delineated that ITF can elicit distinct signaling pathways for epithelial restitution and inhibition of apoptosis, the latter of which involves activation of mitogen-activated protein kinase pathways independent of epidermal growth factor receptor phosphorylation 33. Important for the present studies, these same extracellular signal regulatory kinase pathways were recently demonstrated to critically function in the development of intestinal epithelial barrier function through regulated expression of claudin-2 and organization of membrane proteins within the tight junction 34. Thus, it is likely that ITF-specific, extracellular signaling pathways regulate epithelial barrier under these defined conditions.

At the tissue and cellular level, an array of genes pivotal to survival in low oxygen states are activated 4,35,36,37. Original studies of hypoxia-induced EPO production concentrated on the regulation of gene transcription. Sequences from the EPO promoter were isolated and identified as a heterodimer with independently regulated subunits termed HIF-1, a member of the rapidly growing per-ARNT-sim family of basic helix-loop-helix transcription factors 5,6. HIF-1 exists as an αβ heterodimer, the activation of which is dependent on stabilization of an O₂-dependent degradation domain of the α subunit by the ubiquitin–proteasome pathway 7. Although not certain, HIF-1 appears to reside in the cytoplasm of normoxic cells, and like a number of other transcription factors (e.g., nuclear factor κB, β-catenin), HIF-1 translocates to the nucleus to form a functional complex 38,39. Binding of HIF-1 to consensus domains in the EPO enhancer results in the transcriptional induction of HIF-1–bearing gene promoters 4. Subsequently, it was determined that HIF-1 is widely expressed and that consensus HIF-1–binding sequences exist in a number of genes other than EPO and are termed HRE 4. Of note on this accord, we did not precisely map the ITF HRE. Attempts to define this site in exact terms were hampered by the complexity of the flanking region around the HIF-1 consensus sequence (data not shown). For example, the immediate region of the HIF-1 consensus site also contains consensus binding sites for the transcription factors myeloid zinc finger-1 (MZF1, consensus 5′-AGGGGGA-3′), GATA-1 (consensus 5′-GGGGATTCTG-3′), GATA-2 (consensus 5′-CAGGATAAGG-3′), Sp1 (consensus 5′-GTGGCAGGGT-3′), MyoD (consensus 3′-GCTGTCCACCGG-5′), and upstream stimulatory factor (USF, consensus 5′-CCACCTGT-3′). Important in this regard, at least three of these transcription factors (GATA-1, GATA-2, and MyoD) have been recently implicated in either induction or repression of genes in hypoxia 40,41,42,43. Therefore, it is possible that regulation of the ITF gene at this site could be an interplay of positive and negative signaling pathways, and given this complexity, more work will be necessary to define the exact details of this HRE.

Our results suggest that ITF may represent a potential novel therapeutic approach for a variety of diseases associated with mucosal hypoxia and altered epithelial barrier function, including inflammatory bowel disease, necrotizing enterocolitis, and ischemic colitis 44. In addition, these results suggest that a receptor/binding site for ITF may be present at sites other than the gastrointestinal tract, such as the endothelium. Thus, the protective effect of ITF on the hypoxia-induced decline in endothelial permeability suggests that ITF may have even broader therapeutic relevance in conditions of vascular leakiness such as septic shock. Although it is unknown whether ITF would function alone in these responses, it is possible, as previously proposed, that ITF acts in concert with and complementary to other protective growth factors and cytokines resident at sites of tissue damage 29. Efforts to better understand ITF signaling pathways and to identify other hypoxia-elicited protective elements could provide future focus for the development of novel treatments.

We thank Eric Black, Kristin Synnestvedt, and Robyn Carey for their superb technical assistance. The authors also thank I.K. Ezedi and I. Rosenberg for production of recombinant ITF used in these studies.

This work was supported by National Institutes of Health grants DK02564 (to G.T. Furuta), DK02682 (to C.T. Taylor), DK02503 (to J. Turner), DK43351 and DK46776 (to D.K. Podolsky), and DK50189 and HL60569 (to S.P. Colgan).