DMOG Ameliorates IFN-γ-Induced Intestinal Barrier Dysfunction by Suppressing PHD2-Dependent HIF-1α Degradation

Abstract

Hypoxia-inducible factor 1α (HIF-1α) has been well established as a protective factor for intestinal barrier function in intestinal epithelial cells. Recently, a study found that increased HIF-1α-induced intestinal barrier dysfunction. We proposed that lymphocyte-derived interferon-gamma (IFN-γ) might be responsible for the intestinal barrier dysfunction caused by increased HIF-1α. HT-29 cell monolayers were grown in the presence or absence of IFN-γ under hypoxia. Then, the transepithelial electrical resistance was measured, and HIF-1α-modulated intestinal barrier protective factors were quantified by polymerase chain reaction (PCR). PCR, western blotting, and chromatin immunoprecipitation of HIF-1α were performed. Dimethyloxalyglycine (DMOG), an inhibitor of prolyl-hydroxylases (PHDs) that stabilizes HIF-1α during normoxia, and RNA interference of PHDs were also used to identify the signal pathway between IFN-γ and HIF-1α. We demonstrated that IFN-γ caused barrier dysfunction in hypoxic HT-29 cell monolayers via suppressing HIF-1α and HIF-1α-modulated intestinal barrier protective factors. We found that IFN-γ decreased HIF-1α protein expression instead of affecting HIF-1α transcription or transcriptional activity. Study also showed that DMOG reversed the IFN-γ-induced decrease in HIF-1α protein expression. Further, we found that PHD2 is the major regulator of IFN-γ-induced HIF-1α degradation by PHD inhibition and RNA interference. We conclude that IFN-γ caused barrier dysfunction by promoting PHD-, especially PHD2-, dependent HIF-1α degradation, and DMOG or PHD2 inhibition reversed this HIF-1α suppression and ameliorated barrier dysfunction. Combined with other studies demonstrating HIF-1α activation in lymphocytes promotes IFN-γ secretion, these findings suggest a mechanism by which increased HIF-1α-induced intestinal barrier dysfunction.

Introduction

A monolayer of epithelium lines the entire gastrointestinal tract and forms an essential barrier to the outside world (Turner 2009). Intestinal barrier dysfunction and the resultant invasion of luminal bacteria and other antigens into the systemic circulation leads to gut inflammation and secondary disruption of the epithelial barrier, inducing or exacerbating systemic inflammatory response syndrome and resulting in 50%–80% of all deaths in surgical intensive care units (Taylor and Colgan 2007; Marchiando and others 2010). To date, the mechanism of intestinal barrier dysfunction remains incompletely understood.

Hypoxia plays a role in many physiological and pathological processes and in a variety of clinical conditions, such as inflammatory bowel disease (Colgan and Eltzschig 2012), celiac disease (Vannay and others 2010), and ischemia/reperfusion (Conde and others 2012). Recent studies have suggested that hypoxia-regulated pathways are highly associated with the loss of epithelial function because the intestinal mucosa experiences profound fluctuations in blood flow and metabolism after irritation (Furuta and others 2001). Notably, hypoxia-inducible factor 1 (HIF-1), which plays a key role in the cellular adaptation to hypoxia, triggers the transcription of many genes that enable intestinal epithelial cells to act as an effective barrier (Colgan and Taylor 2010). HIF-1 is a heterodimeric complex composed of the HIF-1α subunit and the constitutively expressed HIF-1β subunit. During normoxia, HIF-1α is hydroxylated by a family of prolyl-hydroxylases (PHDs), including PHD1, PHD2, and PHD3, which enables the von Hippel-Lindau factor (VHL) to bind, resulting in ubiquitination and rapid degradation by proteasomes (Brahimi-Horn and Pouyssegur 2009). However, during hypoxia, HIF-1α is stabilized and translocates to the nucleus. Then, HIF-1α binds to the specific sequence motif called the hypoxia response element (HRE) in the promoter region of hypoxia response genes, which encode a range of intestinal barrier protective factors in intestinal epithelial cells that increase mucin production (mucin-3) (Louis and others 2006), modify mucins intestinal trefoil factor (ITF) (Furuta and others 2001), modify nucleotide metabolism (5′-nucleotidase ecto; nt5e; CD73) (Synnestvedt and others 2002; Colgan and others 2006), and regulate nucleotide signaling through CD55 (Louis and others 2005).

Nevertheless, the effect of constitutively active HIF-1α on the intestinal barrier has been controversial. As an extension of the original studies identifying the induction of HIF expression within the intestinal mucosa, Shah and others generated transgenic mice that constitutively express HIF-1α by disrupting VHL. They found that increased HIF-1α expression correlated with more severe clinical symptoms and an increase in histological damage in a dextran sulfate sodium (DSS)-induced colitis model (Shah and others 2008). Feinman and others (2010) also demonstrated that partial HIF-1α deficiency attenuated hemorrhagic shock-induced increases in intestinal permeability and bacterial translocation.

HIF-1α activation, which was thought to be protective for the intestinal barrier commonly, causes epithelial barrier dysfunction in an unclear way. In fact, hypoxia occurs not only in intestinal epithelial cells but also in intestinal lymphocytes. More recently, there have been studies showing that hypoxia can activate lymphocytes, leading to interferon-gamma (IFN-γ) secretion (Shibahara and others 2005; Roman and others 2010). IFN-γ is a pleiotropic cytokine that exerts a number of biological activities involved in host defense and immunomodulation (Beaurepaire and others 2009). The effect of IFN-γ on hypoxia-induced, HIF-1α-dependent transcription in intestinal epithelial cells remains to be determined. However, there is evidence that IFN-γ can influence HIF-1α transcriptional activity in glioblastoma cells (Hiroi and others 2009). Thus, we hypothesized that lymphocyte-derived IFN-γ might be responsible for the intestinal barrier dysfunction caused by increased HIF-1α.

In this study, we examined the effect of exogenous IFN-γ on intestinal barrier function and the expression of HIF-1α and HIF-1α-induced intestinal barrier protective factors in hypoxic or hypoxic-like HT-29 cell monolayers, using the PHD inhibitor dimethyloxalyglycine (DMOG), which stabilized HIF-1α in vitro and in vivo (Milkiewicz and others 2004; Cummins and others 2006; Elvidge and others 2006). Our results demonstrated that IFN-γ led to barrier dysfunction in hypoxic HT-29 cells by repressing HIF-1α and HIF-1α-regulated intestinal barrier protective factors, and DMOG attenuated the barrier dysfunction caused by IFN-γ by inhibiting PHD2-dependent HIF-1α degradation.

Materials and Methods

Materials

HT-29 (human colorectal adenocarcinoma) cells were purchased from the Institute of Biochemistry and Cell Biology of Chinese Academy of Sciences [The same cell lines are used in Xu and others (2012)]. IFN-γ was purchased from Invitrogen, PHD inhibiter DMOG was purchased from Sigma-Aldrich. Anti-HIF-1α antibody was purchased from Novus Biologicals, Inc. (NB100-105). Anti-PHD1, anti-PHD2, and anti-PHD3 antibodies were purchased from Santa Cruz Biotechnology, Inc. Anti-tubulin antibody was purchased from Goodhere Biotechnology.

Cell culture and treatments

HT-29 was maintained in McCOY's 5A medium (HyClone; Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (Gibco, Life Technologies) and antibiotics at 37°C with 95% air and 5% CO₂. Cells were grouped according to culturing conditions as normoxia (Nx), hypoxia (Hx), DMOG (DMOG), and IFN-γ (IFN) treatment. DMOG or IFN-γ treatment was achieved by culturing the cells in the medium containing DMOG (100 μg/mL) or IFN-γ (20 ng/mL). Hypoxia was achieved by culturing the cells in sealed hypoxic chambers (Thermo Fisher Scientific, Inc.) after the chambers had been flushed with a gas mixture composed of 2% O₂, 5% CO_2, and 93% N₂.

Transepithelial electrical resistance measurement

Cells were split at confluence at a ratio of 1:10 using 0.05% trypsin/0.02% EDTA (Invitrogen) and then grown on Millipore Transwell in 24-well plates (0.3 μm pore size; 0.33 cm²). The electrical resistance of HT-29 intestinal monolayers was measured by using an epithelial voltohmmeter (Millicell-ERS II; Millipore). Both apical and basolateral sides of the monolayers were bathed with medium. All treatment (Hx, IFN, DMOG, etc.) were achieved after 2 days from seeding. The transepithelial electrical resistance (TER) of monolayers were reported after subtraction of the background filter resistance with the result multiplied with the monolayer area of 0.33 cm² leading to TER as Ω×cm².

Quantitative polymerase chain reaction

Total RNA was extracted from cultured HT-29 cells using Trizol (RNAiso Plus; TaKaRa) following the manufacturer's instructions. The cDNA was synthesized using the PrimeScript RT reagent Kit (TaKaRa) with oligo (dT) primer through reverse transcription. Quantitative polymerase chain reaction (PCR) (qPCR) analysis was performed using SYBR Premix EX Taq II (TaKaRa) on Bio-Rad IQ5 (Bio-Rad). The sequences of the primers and lengths of the amplified products are detailed in Table 1. The particular reaction conditions were performed as follows: initial denaturation at 95°C for 1 min, template denaturation at 95°C for 20 s, annealing at 60°C for 30 s, extension at 72°C for 1 min (total of 40 cycles), and final extension at 72°C for 10 min. The cycle threshold (CT) was determined with automatic baseline calculations. CT value more than 30 was considered to be unacceptable. The relative gene expression was calculated using the 2^−ΔΔCT methods. β-actin in identical reactions was used to control the starting template.

Table 1.

The Real-time Polymerase Chain Reaction Primers Used in this Study

Name	Primer sequences	Size of product (bp)
ITF	F: CCAAGCAAACAATCCAGAGCA	79
ITF	R: GCTCAGGACTCGCTTCATGG
Mucin-3	F: CTGCTGCTACTACTTACAAGGTC	75
Mucin-3	R: GCAGGGCAGATAGGCATTCT
CD73	F: CCAGTACCAGGGCACTATCTG	136
CD73	R: TGGCTCGATCAGTCCTTCCA
CD55	F: AGGCCGTACAAGTTTTCCCG	80
CD55	R: CCTTCTCGCCAGGAATTTTCAC
β-actin	F: GCACCACACCTTCTACAATGAGC	159
β-actin	R: TAGCACAGCCTGGATAGCAACG

ITF, intestinal trefoil factor.

Western blot analysis

Western blot analysis was performed as previously described (Sun and others 2012). Nuclear epithelial cell fractions were prepared using the NE-PER extraction kit (Thermoscientific). The protein concentration was determined according to the Bradford method using BCA assay reagent (Beyotime). Samples (20 μg protein) were loaded onto SDS-PAGE gels and the gels were transferred to polyvinylidene difluoride membrane (Millipore) after electrophoresis. Membranes were blocked by 5% bovine serum albumin in TBS-T (50 mM Tris–HCl pH 7.5, 140 mM NaCl, and 0.1% Tween) and then incubated with the following primary antibodies at 4°C: anti-HIF-1α (1:200), anti-PHD1 (1:500), anti-PHD2 (1:500), anti-PHD3 (1:500), and rabbit polyclonal anti-tubulin (1:1000). The membranes were then washed thrice and incubated with horseradish peroxidase-conjugated secondary antibodies. Membrane imaging was performed using the enhanced chemiluminescence detection system (ECL; Boster) according to the manufacturer's instructions.

Chromatin immunoprecipitation assay

Chromatin immunoprecipitation (ChIP) assays were conducted according to the manufacturer's manual for the ChIP Kit (Upstate; Millipore). HT-29 cells were fixed with 1% paraformaldehyde. Chromatin from lysed HT-29 cells were sheared by using an F550 microtip cell solicitor (Fisher Scientific). After centrifugation, supernatants containing sheared chromatin were incubated with an anti-HIF-1α antibody (1:200; Novus Biologics) or control IgG (OE). Protein A Sepharose was then added, incubated overnight, and immune complexes were eluted. Complexes were next treated with protein K, extracted with phenol/chloroform, then with chloroform. DNA was precipitated, washed, dried, resuspended in water, and analyzed by PCR (33 cycles). The primers used in this analysis spanned 165 bp around the HIF-1 binding site within the HRE (Forward primer, 5′-CCACTTCCTCCTCAC CTTCC-3′ and Reverse primer, 5′-TTC TCC CTG AAA TCC CAT CTT-3′) (Rosenberger and others 2007).

Transient transfection assay

The inhibition of PHD1, PHD2, or PHD3 function, cells were transiently transfected with short hairpin RNA (shRNA) plasmids, which aimed to interfere with PHD1, PHD2, or PHD3 expression, purchased from Genecopoeia, Inc., according to the manufacturer's instructions. Briefly, the RNA interference sequence for human PHD1, PHD2, or PHD3 are detailed in Table 2.

Table 2.

The RNA Interference Sequence for Human PHD1, PHD2, or PHD3

Name	Gene	Location	Target sequence
si-PHD1-1	EGLN2	561	gcggccgctgcagagtgaa
si-PHD1-2	EGLN2	851	gcatgcggtactacggcat
si-PHD1-3	EGLN2	1043	gctgtcgaagcattggtgc
si-PHD1-4	EGLN2	1211	ggcgctgcatcacctgtat
si-PHD2-1	EGLN1	83	cctagtgctcctaatcata
si-PHD2-2	EGLN1	262	caagcaccacatgtgttta
si-PHD2-3	EGLN1	408	ggagttccagaatcgtgtt
si-PHD2-4	EGLN1	830	atgagtcctcctagtaaag
si-PHD3-1	EGLN3	766	cctgcatctactatctgaa
si-PHD3-2	EGLN3	817	tcctgcggatatttccaga
si-PHD3-3	EGLN3	928	cttacgcaaccagatatgc
si-PHD3-4	EGLN3	155	gagttgcgagaaactttcc

PHD, prolyl-hydroxylase.

Statistical analysis

All of the experiments were independently replicated at least three separate times. Data were analyzed by SPSS software (version 11.0 for windows) using Student's t-test or ANOVA and are showed as the mean±SD.

Results

IFN-γ mediates barrier dysfunction in hypoxic HT-29 cell monolayers

To determine whether IFN-γ modulates intestinal barrier function in hypoxic HT-29 monolayers, the TER was measured under different conditions: normoxia (Nx), 20 ng/mL IFN-γ (IFN), 2% O₂ hypoxia (Hx), and IFN-γ with simultaneous hypoxia (Hx+IFN). The TER in the four groups steadily rose by the first 2 days (Fig. 1). All of the treatments were performed from this time. For the normoxia-treated cells (Nx), very little difference in the TERs was seen. However, there were varying decreases in the TER in the Hx, IFN, and Hx+IFN conditions. Forty-eight hours after treatment, there were significant differences among the four groups studied (P<0.05). The TER for the Hx+IFN group (10.467±0.746) was the lowest, significantly lower compared with the Nx (15.822±0.565), IFN (13.111±0.550), and Hx (14.600±0.551) groups at 48 h after treatment (P<0.05). Hypoxia plus IFN-γ-induced a significant decrease in the TER compared with normoxic treatment (5.356±0.446) that exceeded the decreases caused by hypoxia or IFN-γ individually [(2.711±0.550) or (1.222±0.416)] (P<0.05). These results suggest that both hypoxia and IFN-γ disrupted intestinal barrier function, but hypoxia combined with IFN-γ amplified the disruption caused by either treatment. There is synergy between hypoxia and IFN-γ in the loss of barrier function.

FIG. 1.

Transepithelial electrical resistance (TER) measurements in the normoxia, interferon-gamma (IFN-γ), hypoxia, and hypoxia plus IFN-γ groups. The TERs were recorded for the normoxia (Nx), IFN-γ (IFN, 20 ng/mL), hypoxia (Hx, 2% O₂), and IFN-γ plus hypoxia (Hx+IFN)-treated HT-29 cell monolayers. All of the treatments were performed 2 days after cell seeding. The readings 48 h after treatment were significantly different among the four groups (P<0.05). The TERs of the cells exposed to IFN-γ plus hypoxia (Hx+IFN) were significantly lower than the cells treated with normoxia (Nx), IFN-γ (IFN), or hypoxia (Hx) individually. The readings were performed in triplicate (n=5).

HIF-1α is involved in the IFN-γ-induced loss of intestinal barrier function

IFN-γ suppressed the expression of protective factors

HIF-1α was thought to protect the intestinal barrier during hypoxia by activating several intestinal barrier protective factors, including ITF, mucin-3, CD73, and CD55, rather than modulating the tight junction proteins occludin and claudin (Colgan and Taylor 2010). However, the mechanism for the protective factors involved in intestinal barrier function is not well understood. In this study, we investigated whether the loss of intestinal barrier function caused by simultaneous IFN-γ and hypoxia treatment was due to IFN-γ-mediated suppression of HIF-1α-regulated expression of protective factors (Fig. 2). qPCR showed that the mRNA expression of ITF, mucin-3, CD73, and CD55 was significantly increased after hypoxia treatment, which is consistent with previous reports (Furuta and others 2001; Colgan and Taylor 2010; Colgan and Eltzchig 2012). We found that IFN-γ did not influence the mRNA levels of ITF, mucin-3, CD73, and CD73 in the absence of hypoxia. However, the mRNA expression of ITF, mucin-3, CD73, and CD55 in the Hx+IFN group was significantly lower than the Hx group, indicating that IFN-γ suppressed the expression of these HIF-1α-induced protective factors. These results suggest that the downregulation of intestinal barrier protective factors, ITF, mucin-3, CD73, and CD55, may account for the barrier dysfunction caused by IFN-γ in hypoxic HT-29 cells.

FIG. 2.

Effect of IFN-γ on hypoxia-inducible factor 1α (HIF-1α)-regulated intestinal barrier protective factors in hypoxic HT-29 cells. Confluent HT-29 monolayers were exposed to normoxia (Nx, 12 h), IFN-γ (Nx, 20 ng/mL IFN-γ with normoxia), hypoxia (Hx, 2% O₂, 12 h), or 20 ng/mL IFN-γ with hypoxia (Hx+IFN, 12 h). The total RNA was isolated, and the mRNA levels of the intestinal barrier protective factors intestinal trefoil factor (ITF), mucin-3, CD73, and CD55 were determined by quantitative polymerase chain reaction using semiquantitative analysis. β-actin was chosen as an internal control. As shown, there was no significant difference between the mRNA expression of ITF, mucin-3, CD73, and CD55 in the Nx group and Nx+IFN group. The mRNA expression of ITF, mucin-3, CD73, and CD55 in the Hx group significantly increased compared with the Nx group, but the mRNA expression of ITF, mucin-3, CD73, and CD55 in the Hx+IFN group was significantly lower than the Hx group (P<0.05). Each column represents the mean±SD (n=3, *P<0.05).

IFN-γ regulated intestinal barrier protective factors via modulation of HIF-1α

Next, we tried to determine how IFN-γ affected these HIF-1α-induced intestinal barrier protective factors (Fig. 3). HIF-1α mRNA expression was determined by reverse transcription-PCR (Fig. 3A). The results confirmed that HIF-1α transcription also increased during hypoxia (3.245±0.154) compared with normoxia (1.024±0.087) (P<0.05). However, there was no significant difference in HIF-1α mRNA expression between the hypoxia (3.245±0.154) and IFN-γ plus hypoxia (3.019±0.127) treatment groups (P>0.05). Western blot analysis (Fig. 3C) was also performed to examine the changes in HIF-1α protein expression. The abundance of nuclear HIF-1a protein during hypoxia was significantly decreased by adding IFN-γ (8.966±0.527) compared with the hypoxia only group (15.754±0.764). To detect the ability of HIF-1α, as a transcriptional factor, combined with the HRE, to regulate the transcription of downstream factors, ChIP assays that used primers amplify a sequence within the promoter of VASP gene (Forward primer, 5′-CCACTTCC TCCTCACCTTCC-3′ and Reverse primer, 5′-TTC TCC CTG AAA TCC CAT CTT-3′) were performed. However, the ChIP assays (Fig. 3E) showed that there was no significant difference between the Hx (0.874±0.091) and Hx+IFN groups (0.848±0.075) using an anti-HIF-1α antibody immunoprecipitation (P>0.05), indicating that IFN-γ did not affect the affinity of HIF-1α and HRE (transcriptional activity of HIF-1α). Therefore, these results suggest that IFN-γ regulates the expression of intestinal barrier protective factors via modulation of HIF-1α protein specifically, possibly by blocking HIF-1α protein translation or promoting HIF-1α protein degradation.

FIG. 3.

Effect of IFN-γ on HIF-1α in hypoxic HT-29 cells. HT-29 cells were exposed to normoxia (Nx, 12 h), hypoxia (Hx, 2% O₂, 12 h), or 20 ng/mL IFN-γ with hypoxia (Hx+IFN, 12 h). (A) HIF-1α mRNA expression was determined by reverse transcription-PCR using semiquantitative analysis. β-actin was chosen as an internal control. (B) Densitometry analysis of the PCR results indicated there was no statistically significant difference between the Hx and Hx+IFN groups. (C) Western blot analysis of HIF-1α. Tubulin was chosen as an internal control. (D) Quantitative analyses of the band intensities in (C) showed that HIF-1α protein expression in the Hx group was lower than the Hx+IFN group (P<0.05). (E) PCR analysis of the chromatin immunoprecipitation (ChIP) assays. ChIP was performed using the chromatin from HT-29 cells in the Hx and Hx+IFN groups with or without anti-HIF-1α antibody (Ab+ or Ab−, respectively). Sheared DNA without the immunoprecipitations from each group was also purified as an internal control. Purified DNA was analyzed by PCR using a primer for the hypoxia response element (HRE). (F) Densitometry analysis of the PCR results from the ChIP assays showed that there was no statistically significant difference between the Hx and Hx+IFN groups using the anti-HIF-1α antibody. Immunoprecipitation showed that IFN-γ did not affect the affinity of HIF-1α and HRE. Each column represents the mean±SD (n=3, *P<0.05).

The PHD inhibitor DMOG ameliorates barrier function via repressing the HIF-1α degradation induced by IFN-γ

DMOG was used as a cell-permeable inhibitor of PHDs, including PHD1, PHD2, and PHD3 (Cummins and others 2008). It has been previously shown to activate HIF-1α expression in vitro and in vivo (Cummins and others 2006; Elvidge and others 2006). DMOG was used to replace hypoxia treatment to stabilize HIF-1α without the harmful effects of hypoxia and to investigate the mechanism of hypoxic barrier dysfunction caused by IFN-γ.

Initially, the TERs were measured for the control, IFN-γ (IFN, 20 ng/mL), DMOG (DMOG, 100 μg/mL), and IFN-γ plus DMOG (DMOG+IFN)-treated HT-29 cell monolayers (Fig. 4). Identical to previous TER measurements, the TERs for the four groups rose steadily within the first 2 days. All of the treatments were performed at this time. The readings at 48 h after treatment were significantly different among the four groups (P<0.05). DMOG treatment resulted in a stable TER, and the control group had a slight decrease in TER. Similar results were obtained for the cells exposed to IFN-γ and DMOG (DMOG+IFN), which is different from the cells exposed to IFN-γ only, suggesting that DMOG improved barrier function, which could not be reversed by IFN-γ.

FIG. 4.

TER measurements for the control, IFN-γ, dimethyloxalyglycine (DMOG), and DMOG plus IFN-γ groups. The TERs were recorded for the control, IFN-γ (IFN, 20 ng/mL), DMOG (DMOG, 100 μg/mL), and IFN-γ plus DMOG (DMOG+IFN)-treated HT-29 cell monolayers. All of the treatments were performed 2 days after cell seeding. The readings 48 h after treatment were significantly different among the four groups (P<0.05). TERs for the cells treated with DMOG (DMOG) or IFN-γ and DMOG (DMOG+IFN) were higher compared with the control or IFN-γ only (IFN) cells. The readings were performed in triplicate (n=5).

Given the importance of HIF-1α in adapting to hypoxia and protecting intestinal barrier function (Ratcliffe 2007), western blot assays were performed (Fig. 5). We found that DMOG treatment resulted in a stable increase in functionally active intestinal epithelial HIF-1α protein (8.782±0.243) compared with the control group (1.009±0.076) (P<0.05). Interestingly, unlike hypoxia-induced HIF-1α stabilization, DMOG-induced HIF-1α protein expression could not be blocked by adding IFN-γ. There was no statistically significant difference in HIF-1α protein expression between the DMOG (8.782±0.243) and DMOG+IFN groups (8.457±0.152) (P>0.05). The decrease of HIF-1α protein expression caused by IFN-γ under hypoxia was not recurrent in HT-29 cells administered with DMOG and IFN-γ simultaneously, which possibly suppresses HIF-1α protein translation or promotes HIF-1α protein degradation. Because DMOG inhibits PHDs, DMOG activates HIF-1α by reducing its degradation. All of these results suggest that IFN-γ suppresses HIF-1α protein expression by promoting HIF-1α protein degradation via a PHD-dependent pathway, and DMOG blocked this decrease in HIF-1α protein expression by inhibiting PHDs.

FIG. 5.

Effect of IFN-γ on HIF-1α in DMOG-treated HT-29 cells. HT-29 cells were grown in normal medium as control (Con) and in medium containing IFN-γ (IFN, 20 ng/mL, 12 h), DMOG (DMOG, 100 μg/mL, 12 h), or IFN-γ combined with DMOG (Hx+IFN, 12 h). (A) Western blot analysis of HIF-1α was performed. Tubulin was used as an internal control. (B) Quantitative analyses of the band intensities in (A) indicated that DMOG promoted HIF-1α stability, which was not reversed by IFN-γ. Each column represents the mean±SD (n=3, *P<0.05).

PHD2 is the key factor for IFN-γ-mediated HIF-1α degradation during hypoxia

To further verify our hypothesis that IFN-γ suppressed HIF-1α protein by promoting HIF-1α protein degradation via a PHD-dependent pathway, shRNA plasmids that specifically inhibit PHD1, PHD2, or PHD3 expression were designed. These plasmids and a negative control plasmid (blank plasmid) were transfected into HT-29 cells according to the manufacturer's instructions. Forty-eight hours after transfection, western blot analyses were performed to assess the efficiency of these shRNA plasmids (Fig. 6). Quantitative analyses of the band intensities indicated that si-PHD1-4 (Fig. 6B), si-PHD2-1 (Fig. 6D), and si-PHD3-4 (Fig. 6F) were the most effective.

FIG. 6.

RNA interference (RNAi) of prolyl hydroxylases 1, 2, and 3 (PHD1, PHD2, and PHD3). HT-29 cells were transiently transfected with short hairpin RNA (shRNA) that specifically suppressed PHD1, PHD2, or PHD3. Forty-eight hours after transfection, western blot analyses were performed to assess PHD1 (A), PHD2 (C), and PHD3 (E) expression. Quantitative analyses of the band intensities indicated that si-PHD1-4 (B), si-PHD2-1 (D), and si-PHD3-4 (F) were the most effective. These plasmids were used in subsequent studies. Each column represents the mean±SD (n=3, *P<0.05).

Then, we used si-PHD1-4-, si-PHD2-1-, and si-PHD3-4-transfected HT-29 monolayers (si-PHD1, si-PHD2, and si-PHD3 groups in the graph) to imitate IFN-γ injury models. Normal and transfected HT-29 cells were incubated in the presence or absence of 20 ng/mL IFN-γ. Twelve hours after treatment, western blot analyses were performed to assess HIF-1α expression (Fig. 7A). PHD1, PHD2, and PHD3 inhibition each increased HIF-1α stability. As reported, PHD2 is the major modulator for HIF-1α degradation (Fujita and others 2012), and RNA interference (RNAi) of PHD2 promoted the greatest HIF-1α stability during normoxia. In our study, HIF-1α protein expression in si-PHD2-transfected cells was even higher than DMOG-treated cells, perhaps due to the dose of DMOG used and the efficiency of RNAi. Except for si-PHD2, PHD inhibition did not block IFN-γ-induced HIF-1α suppression. There was no statistically significant difference between the si-PHD2 and si-PHD2+IFN groups (P>0.05), which was seen between the DMOG and DMOG+IFN groups. HIF-1α protein expression in the si-PHD1+IFN and si-PHD3+IFN groups was lower to different degrees than the si-PHD1 and si-PHD3 groups. These results suggest that PHD2 is the key factor in IFN-γ-mediated HIF-1α degradation.

FIG. 7.

Effect of prolyl-hydroxylase (PHD) inhibition on HIF-1α in HT-29 cells with or without IFN-γ treatment. (A) RNAi HT-29 cells were incubated in the presence or absence of 20 ng/mL IFN-γ in the medium. Twelve hours after treatment, western blot analyses were performed to assess HIF-1α protein expression. (B) Quantitative analyses of the band intensities in (A) showed that PHD1, PHD2, and PHD3 inhibitors all increased HIF-1α stability, but DMOG treatment did not. (C) HT-29 cells treated with PHD2-RNAi were exposed to normoxia (Nx, 12 h), hypoxia (Hx, 2% O₂, 12 h), or 20 ng/mL IFN-γ with hypoxia (2% O₂ hypoxia) (Hx+IFN, 12 h). Then, western blot analyses were performed. (D, E) Quantitative analyses of the band intensities in (C) confirmed that PHD2 inhibition suppressed the IFN-γ-induced decrease in HIF-1α protein expression. Each column represents the mean±SD (n=3, *P<0.05).

All previous studies with RNAi used normoxic conditions. To confirm our observation, we utilized si-PHD2-transfected HT-29 cells to repeat the hypoxia plus IFN-γ treatment experiments. First, si-PHD2-transfected HT-29 cells were exposed to normoxia (Nx), hypoxia (Hx, 2% O₂) or 20 ng/mL IFN-γ, and 2% O₂ hypoxia (Hx+IFN). Twelve hours after treatment, western blot analyses were performed (Fig. 7C), which confirmed that si-PHD2 transfection suppressed PHD2 expression and promoted HIF-1α stability during normoxia. The results also confirmed that PHD2 inhibition reversed IFN-γ-induced HIF-1α suppression.

Discussion

The results presented here demonstrated that IFN-γ mediated barrier dysfunction in hypoxic HT-29 cells via repression of HIF-1α and HIF-1α-regulated intestinal barrier protective factors, including ITF, mucin-3, CD55, and CD73. The PHD inhibitor DMOG reversed this repression and attenuated the barrier dysfunction caused by IFN-γ, and PHD2 RNAi reproduced this reversal, indicating that IFN-γ promotes HIF-1α degradation and decreases HIF-1α-regulated intestinal barrier protective factors via the PHD2 signaling pathway. Additionally, IFN-γ expression increased in hypoxic lymphocytes (Shibahara and others 2005). These findings provide impressive support for our hypothesis that hypoxia not only activates HIF-1α in intestinal epithelial cells, playing a protective role, but also activates HIF-1α in intestinal lymphocytes, increasing IFN-γ expression, which has been certified in another study (Roman and others 2010), and suppressing HIF-1α-mediated protection in epithelial cells. These results may explain why overexpression of HIF-1α, an intestinal barrier protective factor, caused more severe intestinal barrier dysfunction in DSS-induced colitis models and some other models (Shah and others 2008; Kannan and others 2011).

Within DMOG treatment, which stabilizes HIF-1α and activates ITF, mucin-3, CD55, and CD73 during normoxia, we found DMOG stably preserved intestinal barrier function compared to the control group. During simultaneous IFN-γ and DMOG treatment, DMOG inhibited IFN-γ-induced HIF-1α degradation and increased the expression of these protective factors. Additionally, intestinal barrier function in IFN-γ plus DMOG treatment group was also stably preserved compared with the IFN-γ-only treatment group. Thus, we can conclude that HIF-1α stabilization and activation of HIF-1α-induced intestinal protective factors maintain intestinal barrier function. In contrast to the above results, IFN-γ caused barrier dysfunction, and IFN-γ suppressed HIF-1α-induced intestinal protective factors. So, we speculate that IFN-γ-induced intestinal barrier dysfunction was caused, at least partially, by inhibition of HIF-1α-induced intestinal protective factors.

Nevertheless, reports have demonstrated that increased HIF-1α expression correlates with more severe clinical symptoms and increased histological damage in animal models (Shah and others 2008). Why overexpression of HIF-1α, an intestinal barrier protective factor, caused more severe intestinal barrier dysfunction is still confusing. According to previous studies from our group and several studies reported by other groups, we hypothesize that lymphocyte-derived IFN-γ, which is also induced by HIF-1α activation, is related to this phenomenon. IFN-γ is a pleiotropic cytokine that exerts a number of biological activities involved in host defense and immunomodulation (Billiau and Matthys 2009). More recently, a study has shown that hypoxia can activate lymphocytes to release IFN-γ and that this increase in IFN-γ expression was HIF-1-dependent (Roman and others 2010). Another study has found that acute intestinal ischemia-reperfusion may activate intestinal intraepithelial lymphocytes (IELs) and increase IELs-derived IFN-γ expression (Shibahara and others 2005). Previous studies from our group also have revealed that IELs play an important role in the loss of intestinal barrier function in a mouse model of total parenteral nutrition (Yang and others 2003), which recently has been proven to be related to intestinal hypoxia (Jose-Cunilleras and others 2012). However, the effect of IFN-γ on HIF-1α-dependent transcription in intestinal epithelial cells remains to be determined. There are reports that IFN-γ modulates HIF-1α transcriptional activity through the IFN-γ transcription factor signal transducer and activator of transcription 1 (STAT1) (Hiroi and others 2009) or silencing of the co-stimulatory factor CBP/P300 in other tissues (Gerber and others 2009). Given the protective effect of HIF-1α in intestinal epithelial cells, we propose that hypoxia not only takes place in intestinal epithelial cells but also in intestinal lymphocytes. Hypoxia activates intestinal lymphocytes to release IFN-γ through an HIF-1α-dependent signaling pathway, which has been certified in other place, and lymphocyte-derived IFN-γ affect HIF-1α-mediated intestinal epithelial cell protection, which has been demonstrated in this study. This proposition may explain why HIF-1α overexpression caused intestinal barrier dysfunction.

We did not demonstrate that IFN-γ affected HIF-1α transcriptional activity through STAT1 or CBP/P300, which has been suggested by other studies that researched the relationship between IFN-γ and HIF-1α in brain or vascular tissues (Gerber and others 2009; Hiroi and others 2009). Alternatively, we found that IFN-γ influenced HIF-1α protein expression during hypoxia in intestinal epithelial cells, which may be related to decreased HIF-1α protein translation or increased HIF-1α protein degradation, instead of HIF-1α transcription or transcriptional activity, as indicated by RT-PCR, western blotting, and ChIP (Fig. 3). This difference may be explained by tissue specificity.

Further, we found that IFN-γ-induced HIF-1α suppression could be reversed by DMOG, which activated HIF-1α by inhibiting the classic PHDs-dependent degradation pathway (Chan and others 2002). Thus, the IFN-γ-induced HIF-1α suppression may be due to increased PHD-dependent HIF-1α degradation. To identify the signaling pathway involved IFN-γ-induced HIF-1α suppression, we designed shRNA plasmids to inhibit PHD1, PHD2, or PHD3 expression and transfected them into HT-29 cells. We confirmed that PHD2 is the main modulator of HIF-1α degradation during normoxia, consistent with a previous report that found that PHD2 was more specific for HIF-1α degradation and that PHD1 and PHD3 were more specific for HIF-2α degradation (Fujita and others 2012). PHD2 inhibition also suppressed the IFN-γ-induced decrease in HIF-1α effectively, but PHD1 and PHD3 inhibition did not reverse the IFN-γ-induced decrease in HIF-1α, suggesting that PHD2 is the key factor in IFN-γ-mediated HIF-1α degradation.

Interestingly, PHDs have been considered to degrade HIF-1α during normoxia. PHDs utilize O₂ as a substrate to hydroxylate HIF-1α, triggering HIF-1α degradation. During hypoxia, HIF-1α hydroxylation by PHDs is suppressed without O₂ (Nakayama 2009). Nevertheless, in our study, we found that IFN-γ promoted HIF-1α degradation through PHD2 in hypoxic conditions in which PHDs were thought to be ineffective. O₂-independent HIF-1α regulation is the focus of the current HIF-1α research (Greer and others 2012). O₂-independent HIF-1α regulation functions not only as an adaptation to hypoxia but also in tumorigenesis and angiogenesis (Semenza 2012). Currently, O₂-independent, HIF-1α-regulated pathways are known to include the small ubiquitin-related modifier (SUMO) (Cheng and others 2007; Ulrich 2007; Chan and others 2011) and hypoxia-associated factor (HAF) signaling pathways (Koh and Powis 2009). Few studies about a PHD2-induced, O₂-independent, HIF-1α-regulated pathway have been reported until now. However, recent studies have found that pyruvate kinase-M2 (PK-M2) can promote HIF-1α protein degradation through PHD2 and PHD3, especially PHD2 (Luo and others 2011; Tennant 2011). PK-M2 is the rate-limiting enzyme for glycolysis and is also regulated by HIF-1α (Kress and others 1998; Kosugi and others 2011). Under hypoxia, PK-M2 accumulates and cell glycometabolism shifts from aerobic to anaerobic metabolism. Accumulated PK-M2 conversely promotes HIF-1α protein degradation through PHD2 and PHD3 to reduce protein translation, conserve energy, and adapt to hypoxia (Luo and others 2011). Does IFN-γ affect PK-M2 via specific pathways? PK-M2 may be the mechanism for IFN-γ-induced HIF-1α degradation under hypoxic conditions.

Notably, another study reported that IFN-γ attenuates HIF activity in intestinal epithelial cells through repression of HIF-1β (Glover and others 2011). In their research they found that IFN-γ represses HIF activity in T84 epithelial cells. They demonstrated that IFN-γ selectively repressed epithelial HIF-1β mRNA expression, but not HIF-1α mRNA, which was partially coincident with our results. However, in their study they did not show the change of HIF-1α protein level, which was critical to HIF activity, induced by IFN-γ. Meanwhile, they mentioned that PHD may be a second potential mechanism for repression of HIF-1 activity. But in their opinion, PHD-3 might be more important.

In summary, we found that IFN-γ facilitated barrier dysfunction by promoting PHDs, especially PHD2, leading to HIF-1α degradation, and DMOG or PHD2 inhibition could reverse this HIF-1α decrease and improve barrier function. This phenomenon may account for how overexpression of the intestinal protective factor HIF-1α can induce more severe intestinal barrier dysfunction. However, this mechanism has only been tested in vitro. Future studies in vivo and in vitro are required, and these studies will identify more components along this pathway and lead to a more complete understanding of this interesting process.

Footnotes

Acknowledgment

This research was supported by the National Natural Science Foundation of China (NSFC 30973113 to H.Y. and NSFC 81020108023 to H.Y.).

Author Disclosure Statement

No competing financial interests exist.

References

Beaurepaire

, Smyth

, McKay

. 2009. Interferon-gamma regulation of intestinal epithelial permeability. J Interferon Cytokine Res, 29(3):133–144.

Billiau

, Matthys

. 2009. Interferon-gamma: a historical perspective. Cytokine Growth Factor Rev, 20(2):97–113.

Brahimi-Horn

, Pouyssegur

. 2009. HIF at a glance. J Cell Sci, 122(Pt 8):1055–1057.

Chan

, Sutphin

, Denko

, Giaccia

. 2002. Role of prolyl hydroxylation in oncogenically stabilized hypoxia-inducible factor-1alpha. J Biol Chem, 277(42):40112–40117.

Chan

, Tsai

, Wu

, Li

, Dai

, Sun

, Chan

, Chang

. 2011. Sumoylation of hypoxia-inducible factor-1alpha ameliorates failure of brain stem cardiovascular regulation in experimental brain death. PLoS One, 6(3):e17375

Cheng

, Kang

, Zhang

, Yeh

. 2007. SUMO-specific protease 1 is essential for stabilization of HIF1alpha during hypoxia. Cell, 131(3):584–595.

Colgan

, Eltzschig

. 2012. Adenosine and hypoxia-inducible factor signaling in intestinal injury and recovery. Annu Rev Physiol, 74:153–175.

Colgan

, Eltzschig

, Eckle

, Thompson

. 2006. Physiological roles for ecto-5′-nucleotidase (CD73). Purinergic Signal, 2(2):351–360.

Colgan

, Taylor

. 2010. Hypoxia: an alarm signal during intestinal inflammation. Nat Rev Gastroenterol Hepatol, 7(5):281–287.

10.

Conde

, Alegre

, Blanco-Sanchez

, Saenz-Morales

, Aguado-Fraile

, Ponte

, Ramos

, Saiz

, Jimenez

, Ordonez

, Lopez-Cabrera

, del Peso

, de Landazuri

, Liano

, Selgas

, Sanchez-Tomero

, Garcia-Bermejo

. 2012. Hypoxia inducible factor 1-alpha (HIF-1 alpha) is induced during reperfusion after renal ischemia and is critical for proximal tubule cell survival. PLoS One, 7(3):e33258

11.

Cummins

, Berra

, Comerford

, Ginouves

, Fitzgerald

, Seeballuck

, Godson

, Nielsen

, Moynagh

, Pouyssegur

, Taylor

. 2006. Prolyl hydroxylase-1 negatively regulates IkappaB kinase-beta, giving insight into hypoxia-induced NFkappaB activity. Proc Natl Acad Sci USA, 103(48):18154–18159.

12.

Cummins

, Seeballuck

, Keely

, Mangan

, Callanan

, Fallon

, Taylor

. 2008. The hydroxylase inhibitor dimethyloxalylglycine is protective in a murine model of colitis. Gastroenterology, 134(1):156–165.

13.

Elvidge

, Glenny

, Appelhoff

, Ratcliffe

, Ragoussis

, Gleadle

. 2006. Concordant regulation of gene expression by hypoxia and 2-oxoglutarate-dependent dioxygenase inhibition: the role of HIF-1alpha, HIF-2alpha, and other pathways. J Biol Chem, 281(22):15215–15226.

14.

Feinman

, Deitch

, Watkins

, Abungu

, Colorado

, Kannan

, Sheth

, Caputo

, Lu

, Ramanathan

, Attan

, Badami

, Doucet

, Barlos

, Bosch-Marce

, Semenza

, Xu

. 2010. HIF-1 mediates pathogenic inflammatory responses to intestinal ischemia-reperfusion injury. Am J Physiol Gastrointest Liver Physiol, 299(4):G833–G843.

15.

Fujita

, Markova

, Anderson

, Chiba

, Toyama

, Shapiro

, Risbud

. 2012. Expression of prolyl hydroxylases (PHDs) is selectively controlled by HIF-1 and HIF-2 proteins in nucleus pulposus cells of the intervertebral disc: distinct roles of PHD2 and PHD3 proteins in controlling HIF-1alpha activity in hypoxia. J Biol Chem, 287(20):16975–16986.

16.

Furuta

, Turner

, Taylor

, Hershberg

, Comerford

, Narravula

, Podolsky

, Colgan

. 2001. Hypoxia-inducible factor 1-dependent induction of intestinal trefoil factor protects barrier function during hypoxia. J Exp Med, 193(9):1027–1034.

17.

Gerber

, Yatsula

, Maier

, Sadler

, Whittaker

, Pober

. 2009. Interferon-gamma induces prolyl hydroxylase (PHD)3 through a STAT1-dependent mechanism in human endothelial cells. Arterioscler Thromb Vasc Biol, 29(9):1363–1369.

18.

Greer

, Metcalf

, Wang

, Ohh

. 2012. The updated biology of hypoxia-inducible factor. EMBO J, 31(11):2448–2460.

19.

Glover

, Irizarry

, Scully

, Campbell

., Bowers

, Aherne

, Kominsky

, MacManus

, Colgan

. 2011. IFN-γ attenuates hypoxia-inducible factor (HIF) activity in intestinal epithelial cells through transcriptional repression of HIF-1β. J Immunol, 186(3):1790–1798.

20.

Hiroi

, Mori

, Sakaeda

, Shimada

, Ohmori

. 2009. STAT1 represses hypoxia-inducible factor-1-mediated transcription. Biochem Biophys Res Commun, 387(4):806–810.

21.

Jose-Cunilleras

, Viu

, Corradini

, Armengou

, Cesarini

, Monreal

. 2012. Energy expenditure of critically ill neonatal foals. Equine Vet J Suppl (41):48–51.

22.

Kannan

, Colorado

, Reino

, Palange

, Lu

, Qin

, Abungu

, Watkins

, Caputo

, Xu

, Semenza

, Deitch

, Feinman

. 2011. Hypoxia-inducible factor plays a gut-injurious role in intestinal ischemia reperfusion injury. Am J Physiol Gastrointest Liver Physiol, 300(5):G853–G861.

23.

Koh

, Powis

. 2009. HAF: the new player in oxygen-independent HIF-1alpha degradation. Cell Cycle, 8(9):1359–1366.

24.

Kosugi

, Ahmad

, Alam

, Uchida

, Kufe

. 2011. MUC1-C oncoprotein regulates glycolysis and pyruvate kinase M2 activity in cancer cells. PLoS One, 6(11):e28234

25.

Kress

, Stein

, Maurer

, Weber

, Reichert

, Buchmann

, Huppert

, Schwarz

. 1998. Expression of hypoxia-inducible genes in tumor cells. J Cancer Res Clin Oncol, 124(6):315–320.

26.

Louis

, Hamilton

, Canny

, Shekels

, Ho

, Colgan

. 2006. Selective induction of mucin-3 by hypoxia in intestinal epithelia. J Cell Biochem, 99(6):1616–1627.

27.

Louis

, Hamilton

, Kong

, Colgan

. 2005. HIF-dependent induction of apical CD55 coordinates epithelial clearance of neutrophils. FASEB J, 19(8):950–959.

28.

Luo

, Hu

, Chang

, Zhong

, Knabel

, O'Meally

, Cole

, Pandey

, Semenza

. 2011. Pyruvate kinase M2 is a PHD3-stimulated coactivator for hypoxia-inducible factor 1. Cell, 145(5):732–744.

29.

Marchiando

, Graham

, Turner

. 2010. Epithelial barriers in homeostasis and disease. Annu Rev Pathol, 5:119–144.

30.

Milkiewicz

, Pugh

, Egginton

. 2004. Inhibition of endogenous HIF inactivation induces angiogenesis in ischaemic skeletal muscles of mice. J Physiol, 560(Pt 1):21–26.

31.

Nakayama

. 2009. Cellular signal transduction of the hypoxia response. J Biochem, 146(6):757–765.

32.

Ratcliffe

. 2007. HIF-1 and HIF-2: working alone or together in hypoxia?. J Clin Invest, 117(4):862–865.

33.

Roman

, Rangasamy

, Guo

, Sugunan

, Meednu

, Packirisamy

, Shimoda

, Golding

, Semenza

, Georas

. 2010. T-cell activation under hypoxic conditions enhances IFN-gamma secretion. Am J Respir Cell Mol Biol, 42(1):123–128.

34.

Rosenberger

, Khoury

, Kong

, Weissmuller

, Robinson

, Colgan

. 2007. Identification of vasodilator-stimulated phosphoprotein (VASP) as an HIF-regulated tissue permeability factor during hypoxia. FASEB J, 21(10):2613–2621.

35.

Semenza

. 2012. Hypoxia-inducible factors: mediators of cancer progression and targets for cancer therapy. Trends Pharmacol Sci, 33(4):207–214.

36.

Shah

, Ito

, Morimura

, Chen

, Yim

, Haase

, Gonzalez

. 2008. Hypoxia-inducible factor augments experimental colitis through an MIF-dependent inflammatory signaling cascade. Gastroenterology, 134(7):2036–2048, 2048.e1–3.

37.

Shibahara

, Miyazaki

, Sato

, Matsui

, Yanaka

, Nakahara

, Tanaka

. 2005. Alteration of intestinal epithelial function by intraepithelial lymphocyte homing. J Gastroenterol, 40(9):878–886.

38.

Sun

, Wang

, Xiao

, Liang

, Yang

. 2012. Angiotensin II induces apoptosis in intestinal epithelial cells through the AT2 receptor, GATA-6 and the Bax pathway. Biochem Biophys Res Commun, 424(4):663–668.

39.

Synnestvedt

, Furuta

, Comerford

, Louis

, Karhausen

, Eltzschig

, Hansen

, Thompson

, Colgan

. 2002. Ecto-5′-nucleotidase (CD73) regulation by hypoxia-inducible factor-1 mediates permeability changes in intestinal epithelia. J Clin Invest, 110(7):993–1002.

40.

Taylor

, Colgan

. 2007. Hypoxia and gastrointestinal disease. J Mol Med (Berl), 85(12):1295–1300.

41.

Tennant

. 2011. PK-M2 Makes Cells Sweeter on HIF1. Cell, 145(5):647–649.

42.

Turner

. 2009. Intestinal mucosal barrier function in health and disease. Nat Rev Immunol, 9(11):799–809.

43.

Ulrich

. 2007. SUMO teams up with ubiquitin to manage hypoxia. Cell, 131(3):446–447.

44.

Vannay

, Sziksz

, Prokai

, Veres

, Molnar

, Szakal

, Onody

, Korponay-Szabo

, Szabo

, Tulassay

, Arato

, Szebeni

. 2010. Increased expression of hypoxia-inducible factor 1alpha in coeliac disease. Pediatr Res, 68(2):118–122.

45.

, Yao

, Xu

, Ying

, Ma

, Li

, Jin

. 2012. A physical model for the size-dependent cellular uptake of nanoparticles modified with cationic surfactants. Int J Nanomedicine, 7:3547–3554.

46.

Yang

, Fan

, Teitelbaum

. 2003. Intraepithelial lymphocyte-derived interferon-gamma evokes enterocyte apoptosis with parenteral nutrition in mice. Am J Physiol Gastrointest Liver Physiol, 284(4):G629–G637.