Anti-inflammatory actions of adrenomedullin through fine tuning of HIF stabilization

Abstract

In intact mucosal tissues, epithelial cells are anatomically positioned in proximity to a number of subepithelial cell types, including endothelia. A number of recent studies have suggested that imbalances between energy supply and demand can result in “inflammatory hypoxia.” Given these associations, we hypothesized that endothelial-derived, hypoxia-inducible mediators might influence epithelial function. Guided by cDNA microarray analysis of human microvascular endothelial cells (HMEC-1 line) subjected to hypoxia (pO₂ 20 torr, 8 h), we identified adrenomedullin (ADM) as a prominent hypoxia-inducible factor (HIF) that acts on epithelial cells through cell surface receptors. We assessed the functional ability for exogenous ADM to signal in human intestinal Caco2 cells in vitro by demonstrating a dose-dependent induction of Erk1/2phosphorylation. Further analysis revealed that ADM deneddylates cullin-2 (Cul2), whose action has been demonstrated to control the activity of HIF. Caco2 cells stably expressing a hypoxic response element (HRE)-driven luciferase promoter confirmed that ADM activates the HIF signaling pathway. Extensions of these studies revealed an increase in canonical HIF-1-dependent genes following stimulation with ADM. To define physiological relevance, we investigated the effect of ADM in a DSS model of murine colitis. Administration of ADM resulted in reduced inflammatory indices and less severe histological inflammation compared to vehicle controls. Analysis of tissue and serum cytokines showed a marked and significant inhibition of colitis-associated TNF-α, IL-1β, and KC. Analysis of circulating ADM demonstrated an increase in serum ADM in murine models of colitis. Taken together, these results identify ADM as an endogenously generated vascular mediator that functions as a mucosal protective factor through fine tuning of HIF activity.—MacManus, C.F., Campbell, E.L., Keely, S., Burgess, A., Kominsky, D.J., Colgan, S.P. Anti-inflammatory actions of adrenomedullin through fine tuning of HIF stabilization.

At mucosal surfaces such as the gastrointestinal tract, a myriad of cell types interact to coordinate physiological and pathophysiological processes, including endothelial/epithelial barrier function, oxygen delivery, motility, and immune responses. The lumen of the colon itself is known to be subjected to moderate resting levels of low oxygen tension, or hypoxia (Hpx), and the epithelium is the most notable area of oxygen deprivation (1–3). The normal architecture of the gut permits sufficient oxygen delivery to the underlying subepithelium and submucosa, owing to the rich endothelial vasculature that not only permits oxygen delivery but is also required for the carriage of absorbed nutrients. Indeed, ultrastructural studies have revealed the intimate positioning of the gut epithelia and underlying endothelial capillary networks (4). On inflammatory insult, Hpx extends much more profoundly within the tissue, reaching beyond the submucosa and colonic muscularis. The signaling consequences of this shift in oxygen demand and availability results in the activation of a number of transcription factors, most notably hypoxia-inducible factor 1 (HIF-1), and NFκB (2, 5). HIF is itself regulated by a well-described post-translational mechanism involving oxygen-dependent degradation of HIF by an E3 ubiquitin ligase centered around a complex containing a Cullin family member (Cul2) and the Von Hippel Lindau (pVHL) tumor suppressor protein, which tag HIF for proteasomal degradation (6). Fine regulation of this process is achieved through the covalent bonding of a small ubiquitin-like protein called NEDD8 to Cul2, which is required for the E3 activity of the signalosome (7). Owing to the deep penetration of Hpx during colitis (2), paracrine signaling might be brought into play between cell types within the tissue. Changes in vascular tone and permeability might suggest the ability of the endothelium to signal out toward the epithelium during Hpx and that vasoactive factors themselves might interact with nonvascular cell types. We therefore sought to investigate the propensity for a Hpx-responsive, endothelium-derived peptide to signal to epithelial cells in an anti-inflammatory context in vitro and the translation of these findings in an in vivo model of dextran sodium sulfate (DSS)-induced colitis.

MATERIALS AND METHODS

Cell culture

Human HMEC-1 microvascular endothelial cells and human Caco2 colonic epithelial cells were obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA). HMEC-1 cells were cultured in molecular, cellular, and developmental biology (MCDB)-131 medium, supplemented with heat-inactivated fetal bovine serum, penicillin, streptomycin, l-glutamine, epidermal growth factor, and hydrocortisone, as described previously (8). Caco2 cells were cultured using Dulbecco's minimum essential medium containing 10% fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin (Invitrogen, Carlsbad, CA, USA). Cells were cultured at 37°C in an atmosphere of 95% air and 5% CO₂ in a humidified incubator. For hypoxic exposure, cell monolayers were subjected to indicated periods of Hpx (pO₂ 20 torr) as described previously (9).

Cell treatments

Cells were treated with ADM (Phoenix Pharmaceuticals, Burlingame, CA, USA) and ADM binding partner/complement factor H (AMBP/CFH; Quidel Corporation, San Diego, CA, USA) or TNF (R&D Systems, Minneapolis, MN, USA) at described concentrations for either 30 min for deneddylation assays, or for 12 h for real-time PCR analysis. ADM/AMBP solutions were allowed to incubate for 15 min prior to cell treatments.

Western blot analysis

Cells were lysed in radioimmunoprecipitation assay (RIPA) buffer [10 mM Tris-HCl, pH 7.4; 150 mM NaCl; 1 mM EDTA; 1% (v/v) Triton X-100; 0.1% SDS; and 1 protease inhibitor tablet/10 ml; Roche Diagnostics, Inc., Indianapolis, IN, USA], and the lysates were cleared by centrifugation at 15,000 g for 20 min at 4°C. For immunoblotting, cleared protein was boiled in Laemmli's SDS sample buffer, resolved by electrophoresis on a 10% SDS-PAGE gel, and electroblotted onto polyvinylidene difluoride (PVDF) membranes (Millipore; Billerica, MA, USA). PVDF membranes were incubated in blocking buffer [Tris-buffered saline (TBS) and 5% nonfat dry milk] for 1 h at room temperature. Membranes were probed at 4°C overnight with the following primary antibodies: rabbit anti-calcitonin receptor-like receptor (CRLR; Genetex, Irvine, CA, USA), 1:500; rabbit anti-receptor activity-modifying protein 2 (RAMP2; ProteinTech, Chicago, IL, USA), 1:800; rabbit anti-Cul2 (Novus Biologicals, Littleton, CO, USA), 1:250; rabbit anti-p44/42 and rabbit anti-phospho p44/42 (Thr202/Tyr204; Cell Signaling, Beverley MA, USA), 1:1000; mouse anti-β-actin (Abcam, Cambridge, MA, USA), 1:20,000, and subsequently with a 1:10,000 dilution of horseradish peroxidase-linked anti-rabbit or mouse IgG (MP Biomedicals, Solon, OH, USA). Antibody staining was detected using LumiGlo chemiluminescence detection system (KPL, Gaithersburg, MD, USA).

Biotinylation of cell surface proteins

Confluent epithelial cells exposed to indicated experimental conditions were surface-labeled with biotin, as described previously (10). Briefly, confluent Caco2 cell monolayers were exposed to a time course of hypoxic or normoxic conditions. Monolayers were then washed once with HBSS with 10 mM HEPES (pH 7.4) and exposed to 1 mM sulfo-NHS-biotin (Pierce Chemical Co., Rockford, IL, USA) in HBSS for 30 min at 4°C on a rocking platform. Monolayers were then washed ×3 in HBSS, followed by incubation in 50 mM NH₄Cl in HBSS for 10 min at 4°C to quench residual biotin. Cells were again washed ×3 in HBSS and lysed in lysis buffer [25 mM Tris-HCl, pH 7.4; 150 mM NaCl; 2 mM EDTA; 1 mM MgCl₂; 1% T_x-100; and 1% octylphenoxypolyethoxyethanol (Igepal)]. Debris was pelleted by centrifugation at 15,000 g for 15 min. Immunoprecipitation was performed using μMACS protein A MicroBeads (Miltenyi Biotec, Auburn, CA, USA), in accordance with the manufacturer's instructions. In brief, an immunocomplex consisting of cell lysate and 3 μg CRLR antibody (Genetex) was incubated with protein A microbeads for 1 h over ice. Washed immunoprecipitates were boiled in nonreducing Laemmli's SDS sample buffer, and protein was analyzed by Western blotting, using an HRP-linked avidin as a detection reagent (Pierce).

Microarray analysis

HMEC-1 cells were incubated either in normoxic room air in a humidified incubator, or in a Hpx chamber in preequilibrated culture medium. Following 8 h incubation, cells were washed with ice-cold PBS and RNA was collected with Trizol and analyzed using an Affymetrix GeneChip Array (Affymetrix, Inc., Santa Clara, CA, USA).

Real-time PCR analysis

RNA was harvested from cultured cells with Trizol. cDNA synthesis was performed by using an iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA, USA) according to the manufacturer's instructions. Real-time quantitative PCR (qPCR) was performed to validate the microarray screen and to examine gene expression levels of ADM (sense 5′-CCCACTTATTCCAC-3, antisense 5′-GTACTTGGCAGATC), CRLR (sense 5′-AGGTAAAGATGAAT, antisense 5′-TATTTCAAGAGCCT), and RAMP2 (sense 5′-TGATTAGCAGGCCTTATAGCA, antisense 5′-GGTGAGTCTCAAAGATGATCC). Samples were controlled for β-actin using following primers: sense 5′-GGAGAAAATCTGGCACCACA-3′, antisense 5′-AGAGGCGTACAGGGATAGCA-3′. qPCR master mix contained 1 μM sense and 1μM antisense primers with iQ SYBR Green (Bio-Rad Laboratories).

Murine models of colitis

C57/BL6 mice were obtained from Jackson Laboratories (Bar Harbor, ME, USA). Animal protocols were performed in accordance with the University of Colorado, Denver, Animal Care and Use Committee. To induce colitis, the drinking water of 8-wk old mice was supplemented with 6% dextran sodium sulfate (DSS; MP Biomedicals, Solon, OH, USA), ad libitum. Animals were administered either ADM (250 μg/kg) or saline via intraperitoneal (i.p.) injection at d −1, 1, and 3 of DSS exposure. Mice were sacrificed on d 4 of DSS exposure.

Determination of adrenomedullin and cytokine protein levels

HMEC-1 cells were grown to confluency in 100-mm dishes and incubated under normoxic or hypoxic conditions for 48 h. Cell supernatants were subsequently collected and centrifuged at 15,000 g for 15 min at 4°C to sediment debris. Secreted ADM levels were assayed by specific enzyme immunoassay (Phoenix Pharmaceuticals) in accordance with manufacturer's instructions. For determination of tissue cytokines, whole colon was extracted from C57/BL6 mice and homogenized by sonication in Tris lysis buffer (150 mM NaCl; 20 mM Tris, pH 7.5; 1 mM EDTA; 1 mM EGTA; and 1% Triton X-100). Protein samples were centrifuged at 15,000 g for 15 min at 4°C to remove debris. Between 20 and 50 μg (25 μl) protein was assayed by electrochemical detection using a Th1/Th2 mouse cytokine multiplexed plate (MesoScale Discovery, Gaithersburg, MD, USA), and normalized to the amount of input protein. Similarly, circulating cytokines were analyzed in serum samples on a Th1/Th2 mouse cytokine multiplexed plate.

RESULTS

Induction of vasoactive peptides in Hpx

In an attempt to identify Hpx-inducible, endothelial-derived mediators that might influence mucosal function, we conducted a microarray using HMEC-1 cells subjected to Hpx (pO₂ 20 torr, 8 h). Analysis of the microarray revealed a number of highly regulated targets of Hpx, including varied regulation of apoptotic, immunomodulary, metabolic, and vasoactive genes (Fig. 1A and Supplemental Table S1). Within these target transcripts, ADM showed the largest induction among the vascular mediators induced by Hpx (see heat map depiction in Fig. 1B; 12.2±2.1-fold; P<0.001). Given that ADM is a secreted vasoactive peptide whose action can influence epithelial cells (11), we pursued this observation. Real-time PCR was used to validate these findings in Hpx and revealed a 6.4 ± 0.5-fold increase at 8 h Hpx (P<0.005; Fig. 1C). To validate these findings at the protein level, we evaluated the influence of Hpx on the production of ADM by ELISA. Figure 1D demonstrates that exposure of HMEC-1 to Hpx over 48 h resulted in an increase in ADM secretion by 4.4 ± 0.6-fold (P<0.05). Together, these results identify ADM as a significant Hpx-induced endothelial target with potential mucosal functions.

Figure 1.

Vasoactive peptides and their receptors are regulated during acute Hpx. A) Gene chip analysis and gene family clustering identified vasoactive peptides as a broadly up-regulated family of genes in HMEC-1 cells. B) Heat map depiction of selected genes from gene chip analysis. Red indicates a >4-fold increase in transcript; blue indicates no significant change in transcript level. C) Real-time qPCR analysis of ADM mRNA expression in HMEC-1 cells following 8 h exposure to Hpx. Data were calculated relative to the β-actin and are expressed as mean ± sd fold change compared with normoxia control (Nmx; n=3). D) EIA assay demonstrating Hpx-induced increase in ADM in supernatants of HMEC-1 cells cultured in Hpx for 48 h (n=3). *P < 0.05, **P < 0.01; Student's t test.

Expression of ADM receptor complex in intestinal epithelial cells

To define the potential of ADM to function as a paracrine factor in the mucosa, we next sought to define the ability of Hpx-induced ADM to signal to intestinal epithelial cells. The ADM receptor consists of a transmembrane G-protein-coupled receptor (GPCR) CRLR, where specificity for ADM is conferred by the coupling of CRLR to RAMP2. As shown in Fig. 2A, CRLR mRNA analysis in Caco2 intestinal epithelia revealed low expression in normoxia (Nmx) and a robust and significant increase of 3.32 ± 0.5 (P<0.05) and 26.5 ± 04.45-fold (P<0.01) following 4 and 6 h exposure to Hpx, respectively. In Nmx, moderate levels of RAMP2 mRNA were detected by real-time PCR, with no significant induction over 6 h treatment with Hpx (Fig. 2B). Furthermore, Western blot analysis of Caco2 whole cell lysates identified CRLR as rapidly induced under conditions of Hpx, with a striking increase in protein levels as little as 2 h exposure to Hpx, which was sustained through 6 h (Fig. 2B). However, RAMP2 protein levels remained at a constitutive and noninducible level throughout a 6-h time course of Hpx treatment. To ascertain whether cell surface expression of CRLR was evident on Caco2 cells, we performed a combination of biotinylation and immunoprecipitation to demonstrate moderate levels of constitutively expressed protein under normoxic conditions, with a clear Hpx response over 6 h treatment (Fig. 2C, bottom panel).

Figure 2.

Caco2 intestinal epithelial cells posess intact ADM receptor complex. A) Real-time qPCR analysis of CRLR mRNA expression in Caco2 cells following a time-course exposure to Hpx. Data were calculated relative to β-actin and are expressed as mean ± sd fold change compared with Nmx control at each time point (n=3). *P < 0.05, **P < 0.01; Student's t test. B) Real-time qPCR analysis of RAMP2 mRNA expression in Caco2 cells on exposure to Hpx (n=3). C) Western blot analysis demonstrating robust induction of CRLR protein in Caco2 cells following 2, 4, and 6 h exposure to Hpx, with RAMP2 levels remaining steady throughout Hpx treatment. β-Actin served as an internal control. Data are representative of 3 individual experiments. Bottom panel identifies the cell surface inducibility of CRLR protein during Hpx treatment. D) Two-fold serial dilution of ADM/AMBP (from 50/25 to 0.4/0.2 nM, 30 min) Identification of a functional ADM signaling axis in Caco2 cells by Western blot analysis of p44/42 phosphorylation demonstrated a dose-dependent increase in phosphorylation following exogenous ADM/AMBP administration. TNF treatment (log dose response from 1 to 100 ng/ml) was used as a positive control of p44/42 phosphorylation. Data are representative of 3 individual experiments.

To assess the presence of a functional ADM/CRLR/RAMP2 signaling axis in human intestinal epithelial cells, we determined the ability of exogenous ADM to regulate established signal transduction pathways, namely p44/42 activity (12, 13). As shown in Fig. 2D, stimulation of Caco2 cells with ADM and AMBP/CFH over 30 min (14) resulted in a dose-dependent phosphorylation of p44/42, reaching maximal phosphorylation at a concentration of 12/6 nM ADM/AMBP and sustained as high as 50/25 nM ADM/AMBP, respectively. Maximal phosphorylation of p44/42 by ADM/AMBP treatment was observed at a similar magnitude to treatment with TNF, with an observed maximal phosphorylation with 1 ng/ml treatment with TNF, sustained out to 100 ng/ml. Together, these data demonstrate the propensity for endothelium-derived ADM to signal to its cognate receptor on human intestinal epithelial cells. These results reveal that intestinal epithelial cells express the machinery for functional ADM responses, particularly in Hpx, and therefore present a complete endothelial-epithelial ADM signaling axis.

ADM induces HIF signaling in Caco2 cells

We next attempted to define the effect of ADM signaling on intestinal epithelial function. For these purposes, we screened a number of luciferase-based reporter assays for transcription factors known to be involved in the inflammatory response. Of these, the hypoxic response element (HRE)-promoter construct was most significantly influenced by ADM (Fig. 3A). Indeed, Caco2 cells transfected with HRE and exposed to ADM/AMBP revealed a significant increase in activity at concentrations as low as 3/1.5 nM (P<0.025) over a 12 h period under normoxic conditions. In an effort to identify the mechanism underpinning the normoxic regulation of HIF activity by ADM, we focused on the E3 ligase machinery responsible for the proteasomal degradation of HIF. The E3 SCF ubiquitin ligase specific to HIFα-family members consists of SKP1, CUL2, and the F-box domain of VHL and results in polyubiquitination of HIFα (15). Full activity of this complex is conferred on the binding of the small ubiquitin-like protein NEDD8 to Cul2. Deneddylation of Cul2 results in the failure of proteasomal breakdown of HIF-1α and therefore increased potential for HIF signaling in the absence of a hypoxic stimulus.

Figure 3.

Exogenous ADM/ADMBP1 modulates HIF activity in Caco2 cells. A) Luciferase promoter assays for a transiently stably transfected HRE-luc promoter vector in Caco2 cells. Cells were treated for 12 h with stated concentrations of ADM/ADMBP under Nmx conditions, and data were normalized to control (untreated condition). Values are means ± sd (n=3). B) Caco2 cells were treated with stated concentrations of ADM/ADMBP for 30 min and analyzed by Western blot for Cul2. Slower-migrating (top) bands represent neddylated forms of the cullin proteins. Densitometric analysis reveals a dose-dependent decrease in Cul2^Nedd:Cul2 ratio. Data are representative of 3 individual experiments. C) Real-time qPCR analysis of canonical HIF target genes enolase, phosphoglycerate kinase-1 (PGK-1), and hexokinase treated with 6/3 nM of ADM/ADMBP over 12 h under Nmx conditions (n=3). *P < 0.05, **P < 0.01, ***P < 0.005; Student's t test.

Based on this premise, we determined whether ADM might influence the neddylation status of Cul2. To do this, we performed Western blot analysis of Cul2 in Caco2 cells exposed to exogenous ADM/AMBP. As shown in Fig. 3B, a 30-min treatment with increasing concentrations of ADM/AMBP resulted in a dose-dependent Cul2 deneddylation (i.e., loss of Cul2^Nedd8). Densitometric analysis revealed a >80% decrease in Cul2^Nedd8 at 25 nM ADM. Such observations would be consistent with the observed increase in HRE activity.

To extend this observation, we determined whether HIF-dependent genes might be induced by ADM. Figure 3C demonstrates the ability of low exogenous administration of ADM/AMBP to induce expression of classically defined HIF-1 targets, namely enolase, phosphoglycerate kinase-1 (PGK-1), and hexokinase. Cumulatively, these data demonstrate that exogenous ADM signals to epithelia for the activation of HIF via a mechanism involving deneddylation of Cul2.

ADM is protective in a DSS model of colitis

A number of studies have indicated that stabilization of endogenous HIF-1α is protective in mucosal inflammation (2, 16, 17). Given our findings that ADM stabilizes HIF in vitro, we examined whether ADM might provide protection in a mucosal inflammation model in vivo, DSS colitis. For these purposes, C57/BL6 mice were administered 6% DSS in drinking water ad libitum with ADM (250 μg/kg, i.p.), or sham injection every other day (d −1, 1, and 3 of DSS exposure; Fig. 4B, arrows). As an initial screen, we purified colonic epithelium and carried out Western blot analysis to determine the presence of CRLR and RAMP2. Figure 4A demonstrates the presence of both ADM-receptor constituent proteins in 4 different individuals. On induction of colitis, all mice subjected to DSS showed a significant loss in body weight when compared with controls (Fig. 4B). In support of our hypothesis that ADM is protective in this model, mice treated with ADM showed a significant attenuation in weight loss compared to controls (P<0.05). Histological examination of the colon revealed marked protection from DSS damage by ADM (Fig. 4C). H&E staining of the colon from vehicle-treated mice revealed a complete loss of epithelial integrity and crypt architecture, absence of goblet cells, immune cell infiltration, and hypertrophy of the mucosa muscularis (Fig. 4C). This pathology was largely absent in mice administered ADM (Fig. 4C). The therapeutic benefit of ADM administration was more marked at further gross examination of the colons from DSS-exposed mice, whereby colon shortening was significant in sham-treated mice but not in mice treated with ADM (Fig. 4D), representing a 25 ± 6 decrease in colon length in vehicle-treated mice and a 5 ± 2% shortening in ADM-treated mice relative to control (P<0.01). Gross analysis of the colons also revealed less stool consistency in the colons of sham-treated mice and bloody stools present in all mice treated with DSS (Fig. 4D). The induction of endogenous ADM during DSS colitis did not correlate with disease protection, as we were unable to demonstrate a significant correlation in circulating ADM levels and disease severity in DSS-treated animals (Supplemental Fig. S1).

Figure 4.

ADM exhibits therapeutic potential in a DSS model of murine colitis. A) Western blot demonstrating the presence of the ADM receptor, comprising CRLR and RAMP2, in murine colonic epithelial cells. B) Weight-loss curve following induction of colitis (d 0) in C57/BL6 mice. ADM (250 μg/kg, i.p.) was administered at d −1, 1, and 3 (arrows). Weight loss was determined mean ± sd change in weight normalized to starting weight at d −1 (n=7). *P < 0.05; ANOVA. C) H&E staining of the colon reveals a loss of epithelial integrity, loss of crypt architecture, absence of goblet cells, and immune cell infiltration, which is rescued by ADM administration. D) Colon length (mm) as a marker of colitic disease in C57/BL6 mice (n=7). NS, nonsignificant. *P < 0.05; Student's t test.

Exogenous administration of ADM reduces the expression of proinflammatory cytokines in DSS colitis

To gain insight into underlying mechanisms of therapeutic benefit for ADM, we screened cytokines produced at the site of inflammation as well as serum cytokines derived from animals administered DSS in the presence and absence of exogenous ADM. This analysis revealed that exogenous ADM decreased DSS-induced colonic and serum levels of IL-1β by 53 ± 15% (P<0.05) and 34 ± 7% (P<0.01), respectively (Fig. 5A, B). Similarly, tissue and serum levels of KC were attenuated by 84 ± 7% (P<0.01) and 73.2 ± 3% (P<0.05), respectively (Fig. 5C, D). Although colon TNF showed a similar trend, we could not show a significant decrease in local cytokine following ADM administration. Circulating TNF, however, decreased significantly following ADM administration (32±5% decrease, P<0.05). Similar decreases were observed for circulating levels of IL-4 and IL-12 following ADM administration (Supplemental Fig. S2). Tissue or circulating levels of IL-2, IL-10, and IFN-γ did not differ significantly in response to ADM treatment. Taken together, these findings identify vasoactive ADM as a potent anti-inflammatory factor in mucosal inflammation.

Figure 5.

Exogenous administration of ADM results in decreased expression of proinflammatory cytokines in DSS-induced colitis. Protein was harvested from colon tissue lysates (A, C, E) or serum (B, D, F) collected from control animals, and animals were treated with DSS with or without ADM treatment. Tissue and serum levels of IL-1β (A, B), KC (C, D), and TNF (E, F) were assayed using mesoscale multiarray detection (n=3). NS, nonsignificant. *P < 0.05, **P < 0.01, ***P < 0.005; Student's t test.

DISCUSSION

Owing to the intimate positioning of the colonic vasculature to the epithelium (4), we hypothesized that a functional signaling axis might exist between these juxtaposed cell types during inflammation. As ADM has been described previously as a canonical HIF-responsive gene (18, 19), and that deep tissue Hpx persists in experimental models of colitis (2), we postulated that such vasoactive peptides might function as mucosal paracrine signaling partners. We therefore investigated the potential role for Hpx-induced vasoactive peptides for their activity of anti-inflammatory signaling mechanisms on the colonic epithelium. In this current study, we describe a novel signaling mechanism that contributes to the resolution of the inflammatory response through the central action of ADM on the neddylation status of Cul2.

Initial studies were guided by a microarray screen that identified ADM as the highest induced gene in HMEC-1 cells subjected to acute (8 h) Hpx. ADM is a 52-aa vasoactive natriuretic peptide belonging to the calcitonin gene-related family of peptides identified in human pheochromoctyoma by its ability to increase cyclic AMP (cAMP) levels in platelets and as a potent vasodilator (20). ADM is a widely distributed peptide throughout the body and has been demonstrated to be secreted by a variety of cell lines, including endothelial cells, vascular smooth muscle cells (VSMCs), cardiac myocytes, fibroblasts, and leukocytes (21). ADM has been demonstrated to mediate its effects as a ligand to its cognate GPCRs, consisting of CRLR and RAMP2 (22). CRLR is one of an emerging family of receptors whose association with the RAMP family of proteins is known to act as a pharmacological switch, effectively modulating ligand binding properties (22). The ADM receptor has been characterized as predominantly mediating its effects through a cholera toxin (CTX)-sensitive G_s signaling pathway (22), whose activity increases intracellular accumulation of cAMP through increased activity of adenylate cyclase (AC). Accumulating evidence does, however, suggest that the ADM receptor promiscuously binds small G proteins, as is evidenced with the ability for both CTX and pertussis toxin (PTX) to inhibit ADM-induced cellular responses, suggesting the recruitment of both G_i and G_s to the receptor, respectively (23). Further signaling analysis has revealed the activation of a number of kinase pathways downstream of ADM signaling, namely protein kinase A (PKA), Src, PI3K/Akt, and Erk mitogen-activated protein kinases (MAPKs) (13, 24–26). ADM is well documented as being up-regulated in cells and tissues in a hypoxic setting, and associated with hypoxic foci in solid tumors (19, 27). Indeed, sequence analysis of the promoter region of the ADM gene has demonstrated a number of functional HIF consensus (ACGTG) regions (18). Further to the ability for Hpx to induce ADM in vivo and in vitro, exogenous administration of ADM has been demonstrated to exhibit therapeutic potential in a number of models of inflammatory disease, including colitis (28, 29) and arthritis (30).

To define the potential for a signaling axis between the epithelial and endothelial cell compartments, we investigated whether the human colonic epithelial cell line Caco2 expressed the ADM receptor, consisting of CRLR and RAMP2. Although basal levels of CRLR protein are detectable in normoxic Caco2 cells, and ADM competently signals in normoxic Caco2 cells, strong regulation of the receptor is evident under hypoxic conditions. This finding is consistent with previous observations in endothelial cells, whereby an HRE has been identified in the promoter region of CRLR, allowing for rapid up-regulation of CRLR message levels in Hpx (31). Furthermore, it has been suggested that CRLR undergoes rapid post-transcriptional regulation in human VSMCs under hypoxic conditions (32). This condition might suggest the ability for ADM signaling to be induced rapidly and amplified effectively at both the signal and receptor level in response to an acute hypoxic insult, representing a first-line response of tissue protection. During normoxic conditions, CRLR mRNA and surface proteins were detectable at constitutive levels of expression, but were rapidly and markedly induced in Hpx. RAMP2 was detectable at moderate resting levels during Nmx, as identified by qPCR detection at a low C_T. Futhermore, we could demonstrate a resting level of RAMP2 in Caco2 whole-cell lysates with no evident hypoxic inducibility. Although RAMP2 was not inducible in Caco2 cells, we postulate that the resting levels of CRLR represent a limiting factor for the formation of a functional ADM receptor on epithelial cells.

We demonstrate here that ADM deneddylates Cul2 in intestinal epithelial cells. Neddylation/deneddylation responses are highly conserved and appear to be universal in all cell types examined (15). Deneddylation reactions on Cullin targets via CSN-associated proteolysis is increasingly implicated as a central point for Cullin-mediated E3 ubiquitylation (7). Notably, other pathways for deneddylation have been reported. For example, the identification of the Nedd8-specific proteases NEDP1 and DEN1 have provided new insight into this emerging field. NEDP1/DEN1 appear to contain isopeptidase activity that can deneddylate Cullin targets directly (33, 34). Whether ADM might influence NEDP1/DEN1 activity is not known but should prove interesting. Notably, ADM has a number of pharmacological features resembling those of adenosine (e.g., liberated during Hpx, signals through GPCR, functionally anti-inflammatory; ref. 35), and our recent studies revealed that adenosine functions to inhibit NF-κB through the deneddylation of Cul-1 (36).

As HIF has been previously implicated as a potential mechanism for anti-inflammatory signaling in mucosal disease, the ability to regulate its activity is an attractive target for the development of novel therapeutics. One such current strategy is the use of inhibitors of a family of HIF prolyl hydroxylases (PHD inhibitors; refs. 37, 38). PHDs represent a regulatory step at the post-translational level and depend on the presence of oxygen, Fe(II), and 2-oxoglutarate as substrates to hydroxylate HIF-1α, leading to its subsequent degradation. The use of pharmacological compounds to inhibit PHD activity, and therefore allow stabilization of HIF, has proven to be protective in a number of colitis models (16, 17, 39) and most recently in Clostridium difficile-induced intestinal injury (40). Here, we report a positive feedback mechanism of HIF activation where an initial stimulus of acute Hpx initiates the expression of a soluble factor from the endothelial cells, whereby paracrine signaling can act on epithelial cells whose receptor profile has been primed for activity. Indeed, several models of colitis and inflammatory bowel disease demonstrate an increase in circulating levels of ADM. However, it is clear that in the context of a chronic, or “runaway,” inflammatory response, this described feedback mechanism is overwhelmed; therefore, supplementation of exogenous ADM might be required for a therapeutic effect. This condition is evident by the observation that circulating levels of ADM show a trend toward an increase in the our DSS model of colitis (Supplemental Fig. S2A) and that no significant correlation exists between severity of disease (as assayed by weight loss) and levels of circulating ADM (Supplemental Fig. S2B). Overall, our findings here that ADM deneddylates Cul1 provide new insight into the potential use of ADM/ADM receptor analogs as anti-inflammatory-based therapies for mucosal disorders.