Neurotensin Promotes the Development of Colitis and Intestinal Angiogenesis via Hif-1α–miR-210 Signaling

Hypoxia plays a central role in the progression of inflammatory bowel disease (IBD), in which intestinal mucosal tissue damage results during both acute and chronic inflammation (1–3). Hypoxia-inducible factor (HIF)-1 is a key transcription factor in the response to hypoxia and is central in maintaining homeostasis during colitis (4, 5). Importantly, several proinflammatory cytokines are able to activate HIF-1 in various cell types, including intestinal epithelial cells (6, 7). We have recently reported that the neuropeptide neurotensin (NT) via its high-affinity receptor neurotensin receptor 1 (NTR1) activates HIF-1α transcriptional activity and promotes intestinal angiogenesis by increasing colonic expression of vascular endothelial growth factor α (8). NT/NTR1 interactions play an important role in the development and progress of experimental colitis and IBD (9–11), whereas colonic expression of both NT and NTR1 is upregulated in both conditions (9, 11). Several studies also indicate that NTR1 signaling is implicated in the pathophysiology of colitis via mechanisms that involve regulation of expression of inflammation-associated genes (9, 10, 12, 13). A subgroup of these genes codes for micro-RNAs (miRNAs), representing short (19–25 nt), ssRNA molecules, acting primarily as negative transcriptional regulators by binding to the 3′ untranslated regions of transcripts (14, 15). Our recent results indicate that exposure of human colonic NCM460 epithelial cells overexpressing NTR1 (NCM460-NTR1) to NT altered expression of several miRNAs (16, 17). Interestingly, expression of miR-210, a known downstream target of HIF-1α (18–20), showed the highest increase in response to NT treatment, suggesting that a NTR1-dependent miR-210 signaling pathway may play a role in signaling pathways related to hypoxia during colitis.

Based on these considerations, we tested the hypothesis that NT-driven HIF-1α-miR-210 interactions may regulate progress of colitis and colitis-associated angiogenesis. To address this hypothesis, we used NCM460-NTR1 colonic epithelial and colon cancer adenocarcinoma HCT-116 cells as well NTR1-deficient mice, intestinal epithelial cell–specific HIF-1α–overexpressing mice (21), as well as mice with specific intestinal epithelial disruption of HIF-1α (22). The effect of intracolonic administration of locked nucleic acid (LNA)-anti–miRNA-210 in the development of experimental colitis was also examined.

We used the following reagents: NT (Phoenix-Biotechnology, San Antonio, TX), HIF-1α inhibitor PX-478 (MedKoo, Chapel Hill, NC), LNA-anti–miR-210 (Exiqon, Woburn, MA), lipofectamine 2000, lipofectamine RNAimax, and OptiMEM (Life Technologies, Grand Island, NY); anti–HIF-1α mAb LS-B145/10218 (Lifespan Biosciences, Seattle, WA); and anti–β-actin polyclonal Ab GTX30632 (GeneTex, Irvine, CA).

Total RNAs from colon tissues from patients with active ulcerative colitis (UC), active Crohn’s disease (CD), and normal subjects (n = 12 per group) were purchased from OriGene (Rockville, MD). Biopsies were obtained through Institutional Review Board protocols with documented patient consent, all from accredited United States medical institutions (www.origene.com). cDNA conversion of RNA samples was performed as described below, and levels of miR-210 were determined by quantitative RT-PCR analysis.

The human neurotensin receptor 1 (NTSR1) gene was isolated from pCR2.1 (23) with Eco RV and inserted into lentiviral backbone CMV-IRES-GFP-PGK-Puro at the Eco RV site, 5′ to IRES. Lentiviral particles expressing NTRS1 were generated, as we previously described (16). After transduction, NCM460-NTR1 cells were maintained, as described below.

Human colonic epithelial cells (NCM460-NTR1) and colonic cancer HCT-116 cells were maintained in M3D medium (Incell, San Antonio, TX) and McCoy5a (American Type Culture Collection, Manassas, VA), respectively, supplemented with 10% (vol/vol) heat-inactivated FBS, 1% l-glutamine, 10 U/ml penicillin, and 100 μg/ml streptomycin at 37°C in air supplemented with 5% CO₂. For HIF-1α silencing, NCM460-NTR1 cells were seeded (6-well plates, 3 × 10⁵ cells/well) and, 24 h later, were transfected with small interfering RNA against HIF-1α or small interfering RNA-A control (Santa Cruz Biotechnology, Santa Cruz, CA) using lipofectamine RNAiMAX (Thermo Fisher Scientific, Waltham, MA), and, 48 h posttransfection, were treated as indicated with NT. For miR-210 silencing, HCT-116 or NCM460-NTR1 cells were transfected with antisense miR-210, using lipofectamine RNAiMAX. Cells transfected with antisense control miRNA served as controls. Where indicated, NCM460-NTR1 cells were treated with NT (10⁻⁷ M, 6 h) ± 18 h of pretreatment with PX-478 (40 × 10⁻⁶ M) or their vehicles (1% BSA in PBS and 0.9% NaCl in water, respectively).

The NTR1 knockout (KO) mice Ntsr1^tmDgen (designated NTR1-KO) were from The Jackson Laboratory and bred in our facility. Animals were backcrossed onto C57BL6/J background for five generations. We performed one additional backcross before intercrossing animals to generate littermate controls used in this study. The intestinal epithelium-specific HIF-1α–overexpressing mice (designated HIF-1α-OE), the intestinal epithelium-specific HIF-1α KO mice (designated HIF-1α-KO) HIF-1α^ΔIE, and littermate controls were previously described (22) and were on C57BL/6 background. HIF-1α-OE and HIF-1α-KO colitis experiments were performed by the group of Y. Shah at the University of Michigan, and mouse colon cDNA and tissue samples were analyzed at University of California. All animals were maintained in standard cages in a light- and temperature-controlled room and were allowed standard chow and water ad libitum.

Animal studies were approved by the institutional animal care and use committee. The 2,4,6-trinitrobenzene sulfonic acid (TNBS) colitis was induced to 8- to 12-wk-old mice by 100 μl intracolonic enema of 250 mg/kg TNBS (Fluka, Ronkonkoma, NY) (24). Control groups were injected with 100 μl 30% ethanol intracolonically. Mice were returned to their cages and sacrificed 48 h postcolitis by carbon dioxide. Colon tissues were isolated and dissected for further analysis.

Total RNA from mouse colons was isolated using standard TRIzol reagent protocol (Life Technologies, Carlsbad, CA), and cDNA was prepared using miRCURY LNA Universal RT microRNA PCR cDNA kit (Exiqon). Quantitative RT-PCR (qRT-PCR) for micro-RNAs was performed using micro-RNA–specific primers (Exiqon) and miRCURY LNA Universal RT microRNA PCR SYBR Green master mix (Exiqon). qRT-PCR for mRNAs of interest was performed using specific primers obtained (Applied Biosystems), according to the manufacturer's instructions.

Colon tissues were fixed in 4% paraformaldehyde and embedded in paraffin. Tissue sections were prepared by the Translational Pathology Core Laboratory (University of California, Los Angeles). Sections were blocked and incubated with a rabbit polyclonal von Willebrand factor (vWF) Ab (Millipore, Billerica, MA) overnight at 4°C. After washing, sections were incubated with donkey anti-rabbit IgG and slides, stained with an ABC kit for color development (Santa Cruz), photographed under the microscope, and computerized. Image analysis of vWF-stained cells was performed using the Scion Image Software, as we described (8). For histological scoring, sections were stained with H&E, photographed at multiple locations, and analyzed by scoring specimens on a 0–10 scale for the following colitis parameters: mucosal integrity (0–6 scale), mucosal neutrophil infiltration (0–3 scale), and edema (0–1 scale) (8).

For the localization of miR-210, paraffin-embedded sections of mouse colon were used for in situ hybridization. Custom miRCURY LNA microRNA detection probes for miR-210 were designed and labeled with digoxigenin at both 3′ and 5′ ends, and used according to the manufacturer’s instructions (Exiqon).

A modification of our recent published procedure was used (17). Briefly, C57BL/6 male mice (8–12 wk old) received intracolonically vehicle or 10 mg/kg LNA-anti–miR-210 (Exiqon) 48 and 24 h before induction of TNBS colitis (250 mg/kg). All intracolonic enemas were performed using a 3.5-cm-long polyethylene cannula (Intramedic PE-20 tubing; BD Biosciences, Parsippany, NJ) attached to an insulin syringe and introduced into the colon reaching ∼3 cm from the anus. Animals were sacrificed after 2 d, and colon tissues were collected and dissected. Briefly, colons were cut 4 cm to 2 cm away from the anus and separated in three parts for RNA, protein isolation, and immunohistochemistry. The efficiency of miR-210 knockdown after LNA-anti–miR-210 treatment was evaluated by quantitative PCR in mouse colon samples.

Anti–HIF-1α mAb LS-B145/10218 (Lifespan Biosciences, Seattle, WA) or nonspecific human IgG (Sigma-Aldrich) was coupled (2 μg Ab/30 μl beads, according to the manufacturer’s instructions) with Dynal magnetic beads (Dynal Biotech, Carlsbad, CA), pretreated with BSA (1% in PBS for 30 min at 4°C). Equal protein samples of cell lysates from NCM460-NTR1 cells treated ± NT (10⁻⁷ M, 6 h) were brought to equal volumes and were subsequently incubated with anti–HIF-1α–coupled beads (0.5 M NaCl, 30 μl beads) or nonspecific IgG-coupled beads for 16 h at 4°C with gentle shaking. Ab-bound material was eluted from the Ab with 2% SDS Laemmli sample buffer and kept as SDS Elute.

Equal volumes of SDS Elutes were subjected to NaDodSO₄-PAGE, according to Laemmli, and transferred to polyvinylidene difluoride membranes in 25 mmol/L Tris, 192 mmol/L glycine. Equal volume of cell lysates incubated with the beads was also subjected to NaDodSO₄-PAGE to serve as control. Membranes were blocked (PBS, 10% nonfat dry milk, 0.05% Tween 20) and probed with anti-PHD2 polyclonal Ab (ab4561; Abcam, Cambridge, MA) and anti–β-actin polyclonal Ab (GTX3063; GeneTex, Irvine, CA), followed by corresponding HRP-labeled secondary Abs (1:1000). Blots were developed with ECL reagent (PerkinElmer Life Sciences, Waltham, MA). Western blot bands were quantified using image analyzer LAS-4000 mini (Fujifilm). Data are represented by cropped images from the original membranes. Loading controls (β-actin) are derived from the same original membranes.

Quantitative results were expressed as means with SEs of the means. The results were analyzed using Prism professional statistics software (GraphPad, San Diego, CA). A Student t test was used for intergroup comparisons.

NTR1 signaling in human colonic epithelial cells stimulates expression of proinflammatory genes (25, 26) and alters expression of several miRNAs, including miR-210 (16). We previously showed that NT exposure of NCM460-NTR1 and HCT-116 cells [that express high levels of endogenous NTR1 (12)] stimulates increased miR-210 expression (16). Therefore, both cell lines were used in this part of the study. We first used LNA-anti–miR-210 to examine whether NT-induced cytokine expression in NCM460-NTR1 and in colonic adenocarcinoma HCT-116 cells is miR-210 dependent. We show that NT significantly induced TNF-α and IL-6 expression in both HCT-116 cells (Fig. 1Aa versus Fig. 1Ab, p = 0.0001 and Fig. 1Ae versus Fig. 1Af, 0.0003) and NCM460-NTR1 cells (Fig. 1Ba versus Fig. 1Bb, p = 0.0001 and Fig. 1Ba versus Fig. 1Bf, 0.0004). Transfection of HCT-116 cells with LNA-anti–miR-210 reduced NT-induced TNF-α and IL-6 expression to levels comparable to untreated cells (Fig. 1Ad versus Fig. 1Ab versus Fig. 1Aa). Transfection of NCM460-NTR1 cells with LNA-anti–miR-210 also attenuated NT-induced TNF-α and IL-6 expression as compared with NT-treated controls (Fig. 1Bb versus Fig. 1Bd, p = 0.0127 and Fig. 1Bf versus Fig. 1Bh, 0.0134). However, these levels were still significantly higher in NT-treated compared with control NCM460-NTR1 cells pretreated with LNA-anti–miR-210 (Fig. 1Bc versus Fig. 1Bd, p = 0.0137 and Fig. 1Bg versus Fig. 1Bh, p = 0.0149). Pretreatment of NCM460-NTR1 cells with LNA-scrambled had no effect on NT-induced cytokine expression (Supplemental Fig. 1Ab, 1Bb versus Supplemental Fig. 1Ad, 1Bd). Moreover, NT treatment of NCM460 cells that normally express very low levels of NTR1 had no effect on miR-210 levels of expression (Supplemental Fig. 1Ca versus Supplemental Fig. 1Cb). These results suggest that in human colonocytes, induction of proinflammatory cytokine expression in response to NT stimulation is at least in part miR-210 dependent.

Several studies indicate that miR-210 gene transcription is primarily regulated by HIF-1 transcriptional activity (18–20) via a highly conserved HIF-responsive element region within the miR-210 gene promoter (27). We have recently reported that NT exposure of NCM460-NTR1 cells does not promote HIF-1α levels of expression, but rather causes HIF-1α protein stabilization and a subsequent increase in its transcriptional activity (8). HIF-1α protein levels are primarily dependent on prolyl hydroxylation by prolyl hydroxylase 2 (PHD2) and subsequent proteolytic degradation (28). Downregulation of PHD2 activity is a potential means by which NT signaling results in HIF-1α accumulation at the protein level. NT treatment of NCM460-NTR1 cells did not alter PHD2 mRNA levels (data not shown), but caused a decrease in the amount of PHD2 coimmunoprecipitated with HIF-1α (Supplemental Fig. 2A), suggesting that NT attenuates PHD2/HIF-1α interaction, reducing the rate of HIF-1α proteolytic degradation.

We next studied whether NT-induced upregulation of miR-210 in human colonocytes depends on increase of HIF-1α levels and transcriptional activity. We blocked HIF-1α activity with either pharmacological inhibition or small interfering RNA and confirmed inhibition by Western blot analysis (Supplemental Fig. 2B). NT treatment significantly increased miR-210 levels of expression (Fig. 2Aa versus Fig. 2Ab, p = 0.0057;). In contrast, pretreatment with PX-478 significantly reduced both basal (Fig. 2Aa versus Fig. 2Ac, p = 0.0018) as well as NT-induced miR-210 expression (Fig. 2Ab versus Fig. 2Ad, p = 0.0026). However, PX-478 did not completely reverse NT-induced miR-210 expression (Fig. 2Ac versus Fig. 2Ad, p = 0.0309). We attribute this effect on incomplete inhibition of HIF-1α by the pharmacologic inhibitor PX-478 (Supplemental Fig. 2B). In a separate set of experiments, NT treatment of NCM460-NTR1 cells again caused a significant increase in miR-210 levels of expression (Fig. 2Ba versus Fig. 2Bb, p = 0.0001) that was abolished by pretreatment with small interfering HIF-1α (Fig. 2Ba versus Fig. 2Bc versus Fig. 2Bd). These data taken together with our recent results (8) suggest that, in human colonocytes, NT/NTR1 signaling decreases HIF-1α/PHD2 interaction, stabilizes HIF-1α, and increases its transcriptional activity, leading to increased miR-210 expression.

To investigate the role of HIF-1α transcriptional activity in colitis-induced miR-210 upregulation in vivo, we employed experimental colitis in transgenic mouse models. We used intracolonic enema of TNBS in wild type (WT) or mice that either overexpress (HIF-1α-OE) or lack HIF-1α expression (HIF-1α-KO) in their intestinal epithelium. Histologic analysis showed that TNBS administration caused histologic changes consistent with colitis in all three mouse models compared with ethanol-injected mice (Supplemental Fig. 3). We found no difference in the total histologic score following TNBS administration between the different mouse groups (d versus e versus f; Supplemental Fig. 3). However, compared with control littermates, HIF-1α-OE mice had increased colonic miR-210 expression both at baseline (Fig. 3Aa versus Fig. 3Ab, p = 0.0001) and following TNBS exposure (Fig. 3Ad versus Fig. 3Ae, p = 0.001). Interestingly, there was a slightly higher, but not statistically significant increase in miR-210 expression between control- and TNBS-treated HIF-1α-OE mice (Fig. 3Ab versus Fig. 3Ae). In contrast, HIF-1α-KO mice showed reduced miR-210 expression at baseline (Fig. 3Aa versus Fig. 3Ac, p = 0.0048), whereas TNBS treatment did not increase colonic expression of miR-210 (Fig. 3Ac versus Fig. 3Af). These data suggest that in the mouse colon HIF-1α is capable of driving miR-210 expression.

We next examined localization of miR-210 in the colonic mucosa by in situ hybridization. We found that miR-210 expression was not readily evident in control mouse colon tissue sections (Fig. 3Bb), but was highly expressed in TNBS-exposed colon. miR-210 was expressed in both the epithelial and subepithelial colonic mucosa areas, but was primarily present in epithelial versus subepithelial areas (Fig. 3Bc). HIF-1α-KO mouse colon exposed to TNBS showed a much more modest increase in miR-210 expression, primarily confined in the colonic submucosa (Fig. 3Bd). As expected, miR-210 expression levels were very high in HIF-1α-OE mice exposed to TNBS, primarily in colonic epithelial cells (Fig. 3Be). Thus, miR-210 expression in colonic epithelial cells is largely dependent on HIF-1α transcriptional activity.

To evaluate the functional importance of an NT/NTR1/HIF-1α/miR-210 signaling axis in vivo, we silenced miR-210 expression in mouse colon by intracolonic enema of LNA-anti–miR-210 (Supplemental Fig. 4A, 4B) (17). Colonic miR-210 expression was significantly reduced in LNA-anti–miR-210–treated mice (Supplemental Fig. 4Ca versus Supplemental Fig. 4Cb, p = 0.0112), whereas the corresponding levels of two arbitrarily chosen miRNAs, let7a and let7b, remained unaffected (Supplemental Fig. 4Cc versus Supplemental Fig. 4Cd and Supplemental Fig. 4Ce versus Supplemental Fig. 4Cf), suggesting that colonic miR-210 in vivo silencing is relatively specific.

NT stimulates expression of ILs 6 and 8 (IL-6 and IL-8) and TNF-α in colonocytes in vitro (29) and during colitis in vivo (10). To evaluate the role of miR-210 during colitis, we assessed mRNA expression of these cytokines in WT mice that received intracolonic enema of LNA-anti–miR-210 and subsequently subjected to TNBS colitis. As expected, TNBS increased TNF-α, IL-6, and the murine analog of IL-8, Cxcl1, compared with untreated controls (Fig. 4Aa versus Fig. 4Ab, p = 0.0005; Fig. 4Ad versus Fig. 4Ae, p = 0.0023; and Fig. 4Ag versus Fig. 4Ah, p = 0.0007). Importantly, pretreatment of mice with LNA-anti–miR-210 significantly attenuated mRNA expression of these cytokines following TNBS treatment (Fig. 4Ab versus Fig. 4Ac, p = 0.0349; Fig. 4Ae versus Fig. 4Af, p = 0.045; and Fig. 4Ah versus Fig. 4Ai, p = 0.0392) to levels comparable to, or lower than those of untreated controls (Fig. 4Aa versus Fig. 4Ac, p = 0.009; Fig. 4Ad versus Fig. 4Af, NS; and Fig. 4Ag versus Fig. 4Ai, p = 0.0095). In contrast, intracolonic enema of LNA-scrambled control in combination with TNBS treatment did not affect expression of these cytokines compared with TNBS alone (Supplemental Fig. 4D). Histological scoring of colitis of mice treated with TNBS revealed that mice pretreated with LNA-anti–miR-210 exhibited significantly lower colitis score associated with improved mucosal integrity and neutrophil infiltration scores (p = 0.0001, 0.0001, and 0.0004, respectively; Fig. 4B, 4C), compared with mice exposed to LNA-scrambled control. These results suggest a proinflammatory role for miR-210 in experimental colitis.

We have previously shown that NTR1 promotes intestinal angiogenesis in vivo, whereas NT induces the expression of vascular endothelial growth factor α in NCM460-NTR1 cells in vitro (8). To assess angiogenesis during colitis in the presence or absence of HIF-1α, we used endothelial cell staining with vWF in colon tissue sections collected from HIF-1α-OE, HIF-1α-KO, and control littermates following intracolonic exposure to TNBS or control. As expected, angiogenesis was significantly increased in TNBS-treated WT mice as compared with controls (Fig. 5Aa versus Fig. 5Ad, p = 0.019). Importantly, this increase was exacerbated in TNBS-treated HIF-1α-OE mice compared with untreated WT and HIF-1α-OE mice (Fig. 5Aa versus Fig. 5Ae, p = 0.0004 and Fig. 5Ab versus Fig. 5Ae, p = 0.0058). A comparison between TNBS-treated WT and HIF-1α-OE mouse colon also indicated a significant increase in vWF-positive cells in HIF-1α-OE mice (Fig. 5Ad versus Fig. 5Ae, p = 0.0264). In contrast, TNBS-treated HIF-1α-KO mice showed significantly lower angiogenesis compared with TNBS-treated HIF-1α-OE mice (Fig. 5Ae versus Fig. 5Af, p = 0.0422), whereas they showed no significant difference when compared with healthy HIF-1α-KO mice (Fig. 5Ac versus Fig. 5Af). These data indicate that HIF-1α transcriptional activity is required for colitis-induced colonic angiogenesis.

miR-210 is also known to target and inhibit the expression of ephrin-A3 (EFNA3), a receptor protein-tyrosine kinase that blocks angiogenesis (30). Using qRT-PCR analysis, we found that EFNA3 mRNA expression is significantly decreased in colons of TNBS-treated mice compared with control littermates (Fig. 6Aa versus Fig. 6Ab, p = 0.004). Importantly, EFNA3 mRNA expression levels were restored in colons of LNA-anti–miR-210–treated mice (Fig. 6Ab versus Fig. 6Ac, p = 0.0381). Moreover, endothelial cell staining showed significantly reduced angiogenesis in TNBS-treated mice pretreated with intracolonic enema of LNA-anti–miR-210 (p = 0.0081; Fig. 6B, 6C). Collectively, these results suggest that miR-210 exerts its proinflammatory effects, at least in part, by regulating angiogenesis.

To directly assess the relationship of miR-210 expression with NTR1 signaling, we compared the expression of miR-210 in the colon of NTR1-KO and control littermates. TNBS treatment of control mice caused a significant increase in the colonic miR-210 expression (Fig. 7Aa versus Fig. 7Ab, p = 0.0109) that was significantly reduced in TNBS-treated NTR1-KO mice (Fig. 7Ab versus Fig. 7Ad, p = 0.0194), suggesting that increased miR-210 expression during colitis is NTR1 dependent.

To confirm and extend our animal model findings in IBD tissues, we assessed the expression levels of miR-210 in colon specimens obtained from UC and CD patients. We found that miR-210 mRNA expression is significantly increased in colon tissue samples from UC, but not CD patients compared with control samples (Fig. 7Ba versus Fig. 7Bb, p = 0.0027 and Fig. 7Bb versus Fig. 7Bc, p = 0.0009). Importantly, all UC colon samples showed higher miR-210 expression than the mean level of miR-210 expression detected in control samples (Fig. 7B). In contrast, CD patients’ colonic samples showed no significant changes in colonic miR-210 levels compared with controls (Fig. 7B). The molecular mechanism of NT-induced intestinal inflammation and angiogenesis via miR-210 expression is summarized in Fig. 8.

NT and NTR1 expression is increased in both human and animal intestine during experimental colitis and IBD (9, 11). NT/NTR1 proinflammatory signaling in colonic epithelial cells (23, 26, 31, 32) is linked to the development and progress of colitis (9, 10, 33), by mechanisms related, at least in part, to increased angiogenesis, via a NT-regulated HIF-1α–dependent pathway at the colonocyte level (8). We have recently presented evidence supporting the ability of NT to differentially regulate expression of miRNAs in human colonocytes (16), with miR-210 showing the highest upregulation in response to NT (16). We now show increased expression of miR-210 in colonic biopsies of UC patients as well as in the colon of mice with experimental colitis (Fig. 7A, 7B), and present evidence that increased miR-210 expression during colitis is NTR1 dependent (Fig. 7A). Most importantly, we show that intracolonic silencing of miR-210 attenuates experimental colitis (Fig. 4B, 4C), inhibits miR-210–driven angiogenesis (Fig. 5B, 5C), and reduces inflammatory cytokine expression (Figs. 1A, 1B, 4A). Lastly, using two genetically modified HIF-1α mouse models, we present direct in vivo evidence suggesting the importance of epithelial HIF-1α activity in the regulation of miR-210 expression in the colon (Figs. 3, 5).

Our results demonstrate increased expression of miR-210 in human (UC) and experimental colitis. miR-210 is frequently referred to as “hypoxamir” because its expression is increased in a variety of cell types during hypoxia in response to HIF-1α transcriptional activity (34) and has been the focus of several laboratories, because hypoxia is a common hallmark of inflammation (35). Expression of miR-210 is decreased in the gastric mucosa during chronic inflammation with Helicobacter pylori (36), but increased in murine macrophages treated with LPS, leading to diminished LPS-induced proinflammatory cytokine expression, suggesting an anti-inflammatory role for miR-210 (37). In acute kidney tissue injury, circulating levels of miR-210 are significantly upregulated and correlated with patient mortality, suggesting a proinflammatory role for this miRNA (38). miR-210 was also shown to have a proinflammatory role by promoting inflammatory cytokine expression in CD-4⁺ T cells from psoriasis vulgaris patients (39). Together, our results from the intestine and intestinal epithelial cells compared with results from other organs (stomach or skin) and different cells (macrophages or T cells) indicate that regulation of miR-210 expression is tissue and cell type specific and it differs during acute or chronic inflammation. Interestingly, miR-155, which is also highly upregulated in response to NT signaling (16), targets HIF-1α expression and provides a negative-feedback loop for the resolution of HIF-1α activity in cells exposed to prolonged hypoxia, leading to oscillatory behavior of HIF-1α–dependent transcription (40). Studies examining the role of miR-155 in the context of HIF-1α and NT are certainly warranted.

NT signaling causes the stabilization, nuclear translocation, and transcriptional activity of HIF-1α (8), an important transcription factor for miR-210 expression (18–20). During colitis, NT and NTR1 expression is increased (9, 11), affecting HIF-1α transcriptional activity (8). In the current study, we found that colonic miR-210 expression is increased in WT mice with experimental colitis (Fig. 7A), as well as in mice overexpressing HIF-1α in intestinal epithelial cells both at baseline and during colitis (Fig. 3Ab, 3Ae). Consistent with these findings, HIF-1α epithelial cell–specific deficient mice had reduced miR-210 levels at baseline and during colitis (Fig. 3Ac, 3Af). Using NTR1-deficient mice, we also present direct evidence that the regulation of miR-210 during colitis is NTR1 dependent (Fig. 7A). In support of our in vivo findings, NT treatment of colonic epithelial cells resulted in increased miR-210 expression in a HIF-1α–dependent manner (Fig. 2). The mechanism of NT-mediated HIF-1α activation (8) involves, at least in part, a decrease in PHD2-HIF-1α interaction (Supplemental Fig. 2A). De novo synthesized cytoplasmic HIF-1α is rapidly hydroxylated by PHDs (primarily PHD2), a recognition signal for binding of von Hippel-Lindau ubiquitin E3 ligase complex and subsequent ubiquitination and degradation of HIF-1α (41). Attenuation of PHD2/HIF-1α interaction leads to accumulation of HIF-1α at the protein level, its nuclear translocation, and subsequent miR-210 upregulation (Fig. 2), through HIF-1α transcriptional activity (42–44).

Using in vitro silencing with LNA-anti–miR-210 in two different colonic epithelial cell lines, we found that miR-210 reduces NT-induced TNF-α and IL-6 expression (Fig. 1), both NF-κB– and MAPK-driven, proinflammatory cytokines (45–48). Consistent with these results, signaling of NTR1 to both NF-κB– and MAPK-driven pathways has also been reported (25, 32), whereas miR-210 is also linked to NF-κB and MAPK activation (49, 50). Importantly, in vivo intracolonic silencing of miR-210 reduced colonic levels of both TNF-α and IL-6 in mice with TNBS-induced colitis (Fig. 4A) and decreased histologic colitis scores (Fig. 4B, 4C), confirming our cellular findings and supporting a proinflammatory role for this miRNA in colitis pathophysiology (Fig. 8).

Histologic colitis scores of transgenic mice overexpressing or lacking HIF-1α did not show any difference on the degree of colitis compared with WT, TNBS-exposed mice (Supplemental Fig. 3), suggesting that in the acute 2-d colitis model used in our study, HIF-1α expression alone is not sufficient to affect colitis phenotype. This is in contrast to a previous mouse TNBS colitis study showing that decreased epithelial cell expression of HIF-1α correlated with more severe clinical symptoms, whereas increased epithelial cell expression of HIF-1α was protective (51). Differences in the TNBS treatment duration and dose may account for the colitis phenotype differences between our study and the study by Karhausen et al. (51). Karhausen et al. (51) used a chronic TNBS colitis model in which mice were presensitized with TNBS and were then treated with intracolonic TNBS enema (2.5 mg/mouse of 20 g, 7 d), whereas in our study an acute model of TNBS-induced experimental colitis (5 mg/mouse of 20 g, 2 d) was used.

We show that NT-induced HIF-1α promotes neovascularization through the upregulation of miR-210 that in turn negatively regulates the expression of genes that inhibit angiogenesis. One such miR-210 target is the angiogenesis blocker EFNA3 (30). Indeed, our results show attenuated EFNA3 expression during TNBS colitis that was reversed in LNA-anti–miR-210–pretreated mice (Fig. 6). As expected, TNBS-induced colonic angiogenesis was attenuated in HIF-1α-KO mice and increased in HIF-1α-OE mice (Fig. 5). These results suggest that NT, at least in part, promotes colonic angiogenesis via a HIF-1α-miR-210-EFNA3 pathway (Fig. 8).

To our knowledge, this is the first report indicating HIF-1α as the main transcription factor responsible for NT-induced miR-210 upregulation. Moreover, our studies using a significant number of transgenic mice elucidate an important mechanism by which NT exerts its proinflammatory and angiogenic effects by promoting miR-210 expression in intestinal epithelial cells. Our results point to miR-210 as a potential novel target treatment of colonic inflammation, whereas the relative expression levels of miR-210 and/or its downstream target EFNA3 may serve as potential biomarkers for UC diagnosis.

We thank Dr. Sean P. Colgan (University of Colorado) for guidance and for providing mouse tissue samples used for preliminary experiments, which led to this study, and the Translational Pathology Core Laboratory at the University of California at Los Angeles for processing colonic tissue sections.

The authors have no financial conflicts of interest.