Lipopolysaccharide induces CXCL2/macrophage inflammatory protein-2 gene expression in enterocytes via NF-κB activation: independence from endogenous TNF-α and platelet-activating factor

Abstract

CXCL2 (macrophage inflammatory protein-2 (MIP-2)), a critical chemokine for neutrophils, has been shown to be produced in the rat intestine in response to platelet-activating factor (PAF) and to mediate intestinal inflammation and injury. The intestinal epithelium, constantly exposed to bacterial products, is the first line of defence against micro-organisms. It has been reported that enterocytes produce proinflammatory mediators, including tumour necrosis factor (TNF) and PAF, and we showed that lipopolysaccharide (LPS) and TNF activate nuclear factor (NF)-κB in enterocytes. However, it remains elusive whether enterocytes release CXCL2 in response to LPS and TNF via a NF-κB-dependent pathway and whether this involves the endogenous production of TNF and PAF. In this study, we found that TNF and LPS markedly induced CXCL2 gene expression in IEC-6 cells, TNF within 30 min, peaking at 45 min, while LPS more slowly, peaking after 2 hr. TNF- and LPS- induced CXCL2 gene expression and protein release were completely blocked by pyrrolidine dithiocarbamate (PDTC) and helenalin, two potent NF-κB inhibitors. NEMO-binding domain peptide, a specific inhibitor of inhibitor protein κB kinase (IKK) activation, a major upstream kinase mediating NF-κB activation, significantly blocked CXCL2 gene expression and protein release induced by LPS. WEB2170 (PAF antagonist) and anti-TNF antibodies had no effect on LPS-induced CXCL2 expression. In conclusion, CXCL2 gene is expressed in enterocytes in response to both TNF and LPS. LPS-induced CXCL2 expression is dependent on NF-κB activation via the IKK pathway. The effect of LPS is independent of endogenous TNF and PAF.

Introduction

Intestinal epithelial cells are the first line of defence against bacterial invasion by intestinal bacteria, and enterocytes produce many pro-inflammatory cytokines in response to bacterial products such as lipopolysaccharide (LPS), a component of the Gram-negative bacterial membrane.¹ Nuclear factor-κB (NF-κB) has been found to be activated in intestinal epithelial cells of inflamed intestinal mucosa of patients with inflammatory bowel disease (IBD)² and in the rat intestine in an experimental model of acute bowel injury induced by platelet-activating factor (PAF),³ a potent endogenous phospholipid. NF-κB is a central transcription factor regulating the transcription of many pro-inflammatory cytokines, chemokines and adhesion molecules. Because it is constitutively present in the cytoplasm in an inactive state, bound to an inhibitory protein κB (IκB), NF-κB can be rapidly activated by a variety of pro-inflammatory signals. Upon activation, it dissociates from IκB and is translocated into the nucleus, where it binds to 10-bp sequences in the regulatory regions of several inflammatory genes and up-regulates their transcription.

The activation of NF-κB by pro-inflammatory stimuli requires the activity of the IκB-kinase (IKK) complex, which is the convergence point of all upstream signalling pathways that activate NF-κB.⁴ The IKK complex consists of two catalytic subunits named IKKα and IKKβ and a regulatory protein named NEMO (NF-κB essential modulator)⁴. Accumulated genetic evidence has established that pro-inflammatory NF-κB activation requires both the NEMO subunit and IKKβ that directly phosphorylates the IκB proteins. This phosphorylation leads to the subsequent ubiquitination and targeting of IκBs to the 26S proteasome where they are rapidly degraded. IκB degradation then liberates NF-κB to migrate to the nucleus and regulate gene expression.

We previously found that both LPS and tumour necrosis factor-α (TNF-α) activate NF-κB in IEC-6 cells, a non-transformed small intestinal crypt cell line⁵ and the effect of TNF is more rapid than that of LPS, already strong at 10 min following incubation.⁵ In mice with colitis, treatment with antisense phosphorothioate oligonucleotides to the p65 subunit of NF-κB has been shown to reduce the severity of the inflammation.⁶ A major role of the epithelial cells in regulating the inflammatory response by NF-κB was further confirmed by experiments on transgenic mice with specific ablation of IKKβ in their intestinal epithelial cells, preventing the activation of the canonical NF-κB pathway in the enterocytes. These mice are protected against systemic inflammation following intestinal ischemia-reperfusion.⁷

Neutrophil infiltration is a feature of IBD⁸ and necrotizing enterocolitis.⁹ In a rat model of acute bowel injury induced by PAF, we found that neutrophils mediate the injury, since neutrophil depletion prevents PAF-induced bowel injury.¹⁰ Furthermore, we found that CXCL2 (macrophage inflammatory protein-2; MIP-2), a critical chemokine for neutrophils¹¹ plays a major role in neutrophil recruitment in this model, because blocking CXCL2 in vivo decreased PAF-induced intestinal neutrophil sequestration and alleviates PAF-induced bowel injury.¹² CXCL2 is a C-X-C chemokine with potent chemotactic properties, secreted by rat macrophages and epithelial cells in response to inflammatory stimuli, such as LPS.¹³ It has been shown to control mucosal lymphocytes and neutrophil migration.¹¹ Binding sites for NF-κB transcription factor have been identified within the regulatory region of the CXCL2 gene.¹⁴ While the PI3K pathway has been shown to mediate the chemotactic properties of TNF¹⁵, the role of the NF-κB pathway in the induction of CXCL2 gene expression by TNF and LPS has not been investigated.

TNF is a central mediator in endotoxic shock.¹⁶ The intestine is a major organ producing endogenous TNF¹⁷ and TNF mRNA has been localized in small intestine Paneth cells.¹⁸ TNF induces the production of C-X-C chemokines such as IL-8 in HT-29 cells, a colonic carcinoma cell line.¹⁹ However, whether TNF induces CXCL2 in enterocytes has not been shown.

In a model of LPS-induced shock and bowel injury, we have previously found that: (a) LPS administration causes the production of TNF and PAF in the intestine^17,20 and PAF induces CXCL2¹² and neutrophil influx in the intestine, resulting in injury; and (b) endogenously produced PAF and TNF mediate the NF-κB activation in the intestine.²¹ Other investigators have found that LPS stimulates intestinal epithelial cells to produce TNF^22,23 and PAF.²⁴ Thus, it is possible that LPS may directly stimulate enterocytes to express CXCL2 via the activation of NF-κB, and this effect may be mediated via endogenous production of TNF and PAF.

In the present study, we first investigated whether LPS and TNF directly induce CXCL2 gene expression and protein release in IEC-6 cells, a non-transformed small intestinal crypt cell line. Next, we examined the role of NF-κB in LPS and TNF induced CXCL2 gene expression and protein release by using the following NF-κB inhibitors: (1) pyrrolidine dithiocarbamate (PDTC), a metal chelator and antioxidant which inhibits IκB-ubiquitin ligase activity;²⁵ (2) helenalin, a naturally occurring cell-permeable pseudoguainolide sesquiterpenoid lactone that inhibits NF-κB DNA binding activity by selectively alkylating the p65 subunit of NF-κB;²⁶ and (3) NEMO-binding domain (NBD) peptide, a cell-permeable peptide, which, by binding to NEMO, blocks its association with the IKK complex and inhibits NF-κB activation.²⁷ We also examined whether PDTC inhibits NF-κB activation in IEC-6 cells and whether it blocks the processing of its inhibitory IκB proteins. Lastly, we investigated the role of endogenously produced PAF and TNF on LPS-induced CXCL2 expression.

Materials and methods

Materials

IEC-6 cells, a non-transformed rat small intestinal crypt cell line, were obtained from the American Type Tissue Culture Collection (ATCC, Manassas, VA). All cell culture media were purchased from Invitrogen Corporation (Carlsbad, CA). LPS (from Salmonella typhosa) was obtained from Sigma Chemical Co. (St. Louis, MO), and rat TNF and anti-TNF antibodies were purchased from R & D Systems (Minneapolis, MN). WEB2170, a specific PAF receptor antagonist, was kindly provided by Dr H. Heuer, Boerhinger-Ingelheim, Mainz, Germany. Antibodies against p50 (sc-114X), p65 (sc-109X), IκBα (C-21, sc-371) and IκBβ (C-20, sc-945) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). RNA stat-60 was purchased from Tel-Test (Friendswood, TX). Rat CXCL2 and reduced glyceraldehyde phosphate dehydrogenase (GADPH) primers and probes for real-time polymerase chain reaction (PCR) were purchased from Applied Biosystems (Foster City, CA). (Rn01525859 and Rn 99999916, respectively). Rat CXCL2 enzyme-linked immunosorbent assay (ELISA) kit was obtained from Biosource International (Camarillo, CA).

Cell culture and treatment

IEC-6 cells were cultured (1 × 10⁷ cells/10-cm tissue culture dish) for 3–4 days in Dulbecco's modified Eagle's minimal essential medium (DMEM) with high glucose containing 10% (v/v) heat-inactivated fetal bovine serum, 0·1 IU/ml insulin, 100 IU/ml penicillin and 100 µg/ml streptomycin until 95% confluence. Cells were incubated at 37° for different time-periods in DMEM containing 10% fetal bovine serum and the following treatments: (a) LPS (0·2–5 µg/ml); (b) TNF (2–50 ng/ml); (c) Culture medium alone; (d) any of these treatments and cycloheximide (CHX) (50 µg/ml), an inhibitor of protein synthesis. Prior to these treatments, some cells were pretreated with the following: (a) PDTC (200 µm) for 30 min; (b) helenalin (50 µm) for 30 min; (c) NBD (200 µm) for 60 min; (d) mutant peptide (200 µm) for 60 min (both NBD peptide and mutant peptide were dissolved in undiluted dimethylsulphoxide (DMSO); (e) anti-TNF antibodies (0·08 µg/ml) for 30 min; (f) control antibodies (goat immunoglobulin G (IgG)) (0·08 µg/ml) for 30 min; (g) WEB2170 (1–10 µg/ml) for 30–60 min; (k) control medium alone.

Preparation of nuclear extracts/electrophoretic mobility shift assay (EMSA)

Cells were washed twice with 10 ml of 4° phosphate-buffered saline (PBS) and scraped with a rubber policeman in 500 µl of buffer A (10 mm HEPES, pH 7·9, 1·5 mm MgCl₂, 10 mm KCl, 0·5 mm dithiothreitol (DTT)). Cell suspensions were incubated for 15 min on ice, then briefly vortexed following the addition of 25 µl of Nonidet P-40. Nuclei were collected by centrifugation at 7500 g for 1 min at 4° then washed with 200 µl of buffer A. Nuclear extracts were obtained by high-salt extraction (incubation in 150 µl of buffer B (20 mm HEPES, pH 7·9, 1·5 mm MgCl₂, 0·42 m NaCl, 0·2 mm ethylenediaminetetra-acetic acid (EDTA), glycerol 25%, 0·5 mm DTT, 0·5 mm phenylmethylsulphonylfluoride (PMSF)). Nuclear debris were pelleted by centrifugation, 150 µl of buffer C (20 mm HEPES, pH 7·9, 50 mm KCl, 0·2 mm EDTA, 0·5 mm DTT, 0·5 mm PMSF) were added and the nuclear extracts were stored at −80°. Protein concentration was measured by Bradford's method. NF-κB was identified by EMSA using a kit from Promega (Madison, WI), as previously described²¹ after gel loading of equivalent amounts of nuclear extracts. Supershift experiments for p50 and p65, the two subunits identified in our previous studies⁵ were performed with anti-p50 (sc-114X) and anti-p65 (sc-109X) antibodies from Santa Cruz Biothechnology.

Preparation of whole cell extracts and western blot analysis

After treatment, cells were washed twice with 10 ml of PBS 4°. Whole cell extracts were prepared as followed: 500 µl of lysis buffer (10 mm Tris.Cl, pH 7·6, 150 mm NaCl, 5 mm EDTA, 0·5 mm DTT, 0·5 mm PMSF, 5 µg/ml each of pepstatin, leupeptin and aprotinin) were added to the cell plates and cells were scraped with a rubber policeman. The mixture was sonicated on ice for 30 s and cellular debris were removed by centrifugation at 10 000 g for 5 min at 4°. Equal amounts of cell extracts were diluted 1 : 1 with Laemmli buffer, boiled for 5 min and electrophoresed on sodium dodecyl sulphate (SDS)/15% polyacrylamide gels. The proteins were transferred to a nitrocellulose membrane, using 80 mA of current overnight at 4°. Nitrocellulose blots were blocked with 5% dry milk in PBS for 30 min, then incubated with antibodies (Santa Cruz Biotechnology) against IκBα (C-21, sc-371) and IκBβ (C-20, sc-945) for 2 hr. The membrane was washed with 0·05% Tween-20 in PBS for 30 min, incubated in 1 : 800 anti-rabbit IgG antibody or 1 : 2000 anti-mouse IgG antibody, conjugated with horseradish peroxidase (Amersham, Arlington Heights, IL), and detected using an ECL kit (Amersham).

Analysis of mRNA by reverse transcription (RT)–PCR

Total cellular RNA was extracted using RNA Stat-60, a monophase solution containing phenol and guanidinum thiocyanate then stored at −80°. RNA samples were then purified using RNeasy (Qiagen, Valencia, CA). The integrity of each sample was verified by electrophoresis in a 1% denatured agarose gel. The relative quantities of mRNA for each gene of interest were assessed by RT–PCR. cDNA was synthesized using 2 µg of total RNA, 50 mm Tris-HCl, 75 mm KCl, 2·0 mm MgCl₂, 10 mm DTT, 1·25 mm of each deoxynucleotide-triphosphate (dNTP), 7·5 pm random hexamer (pdN6), 1 U/µl RNasin inhibitor, and 10 U/µl Moloney murine leukaemia virus (M-MLV) reverse transcriptase (Invitrogen, Carlsbad, CA). The reaction was carried out for 1 hr at 37°, followed by heating to 95° for 10 min.

cDNA was amplified on a Thermal Cycler Model 9600 (Perkin-Elmer, Boston, MA) in a total volume of 50 µl containing 2 µl of cDNA, 10 mm Tris-HCl, 50 mm KCl, 2 mm MgCl₂, 0·1 mm of each dNTP, 0·1 U/µl of Taq DNA polymerase, and 1·0 µm of each primer. The primers used were: CXCL2 primer 1: 5′-TCC TCA ATG CTG TAC TGG TCC-3′, primer 2: 5′-ATG TTC TTC CTT TCC AGG TC-3′; rat GAPDH primer 1: 5′-ATT CTA CCC ACG GCA AGT TCA ATG G-3′, primer 2: 5′-AGG GGC GGA GAT GAT GAC CC-3′. The amplification was carried out by 32 cycles of denaturation at 94° for 45 s, an annealing step at 60° for 45 s, and an extension step at 72° for 1 min. In the final cycle, the 72° step was extended to 10 min. The PCR products were detected after electrophoresis on a 1·5% agarose gel. The size of the corresponding PCR products for CXCL2 and GAPDH are 299 bp and 224 bp, respectively.

For real-time PCR, RNA templates (12·5 ng) were reverse transcribed and amplified. Primers and probes for rat CXCL2 and GAPDH were purchased from Applied Biosystems (see Materials section for details). The threshold cycle (CT value) was defined for each PCR reaction, which is the cycle number at which the reporter fluorescence generated by the cleavage of the sequence specific probes passes above a fixed baseline. The number of CXCL2 copies present in each sample was normalized to GAPDH copies. Results were expressed as ratio to baseline. Non-amplification controls (NAC) and no template controls (NTC) were run in parallel and their CT values equalled 40.

Analysis of CXCL2 protein release from IEC-6 cells by ELISA

Cells (10⁶) were treated in 1 ml of culture medium at 37° as described above. At the indicated time following incubation, the culture medium was collected and CXCL2 released in the culture medium was measured using a commercially available rat CXCL2 ELISA kit (Biosource International). All standards, controls and samples were run in duplicate. The absorbance of each well was read at 450 nm with a MR600 microplate reader (Thermomax-Molecular Devices; Fisher Scientific Instrument, Nepean, ON, Canada) and background absorbency values of blank wells were subtracted from the obtained values. CXCL2 measurements were expressed as picograms of CXCL2 per 10⁶ cells. Four samples per group were analysed.

Results

LPS and TNF induce CXCL2 gene expression and protein release in IEC-6 cells, which is inhibited by PDTC

Unstimulated IEC-6 cells do not express CXCL2 (Fig. 1a). This is consistent with what we observed in the normal intestine in vivo.¹² Following incubation of IEC-6 cells with LPS (5 µg/ml), CXCL2 gene was slowly but markedly expressed, the expression becoming significant at 1 hr (1204 ± 31-fold increase compared to baseline (P < 0·001)), peaking at 2 hr (13 288 ± 1422-fold increase (P < 0·001)) and remaining significantly increased at 24 hr (133·8 ± 10-fold increase (P < 0·001)) following LPS (Fig. 1a, b). CXCL2 gene induction was seen with the lowest dose of LPS used (0·2 µg/ml) and was dose dependent (Fig. 1c). However, only the difference in expression between 0·2 µg/ml and 5 µg/ml was statistically significant (P < 0·001) (Fig. 1c). This correlates with our previous findings, that LPS-induced NF-κB activation is dose dependent at similar concentrations.⁵ In parallel, an increase in CXCL2 protein release (2·11 ± 0·3-fold increase compared to baseline (P < 0·01)) could be detected 2 hr following LPS incubation, increasing further at 4 hr (3 ± 0·3-fold increase (P < 0·001)) (Fig. 2). Increasing LPS concentration to 10 and 20 µg/ml did not significantly increase CXCL2 induction at 4 hr (data not shown). In contrast, TNF (10 ng/ml) rapidly induced CXCL2 in IEC-6 cells within 30 min (17·6 ± 5·64 -fold increase compared to baseline (P < 0·01)), peaking after 45 min (859 ± 153-fold increase (P < 0·01)) and remaining high at 2 hr (336 ±14-fold increase (P < 0·001)) (Fig. 3a, b). There was a statistically significant dose-dependent increase in CXCL2 gene expression (P < 0·001) with increasing concentrations of TNF up to 10 ng/ml, but no further increase was noted with higher doses (50 ng/ml) (Fig. 3c). In parallel, an increase in CXCL2 protein could be already detected at 1 hr following TNF incubation (1·63 ± 0·14; P < 0·01)) and remained present at 2 and 4 hr ((1·2 ± 0·09-fold (P < 0·01) and 2·16 ± 0·37-fold (P < 0·01) increase, respectively) (Fig. 4a). TNF-induced CXCL2 protein release is dose dependent at the doses tested (2–50 ng/ml) (Fig. 4b).

Figure 1.

Time-course and dose–response of LPS-induced CXCL2 gene expression in IEC-6 cells and its inhibition by PDTC. IEC-6 cells were incubated with culture medium alone, LPS alone (5 µg/ml) (row 1) or PDTC (200 µm) for 30 min, followed by LPS (5 µg/ml) (row 2) for various time periods (a, b). In (c), IEC-6 cells were incubated with increasing concentration of LPS (0·2, 1 and 5 µg/ml) for 1 hr. RNA was extracted, RT–PCR performed using primer specific for CXCL2 (upper panel) and GAPDH (lower panel). PCR products were run on a 1·5% agarose gel and stained with SYBR green (a). CXCL2 mRNA was quantified by real-time PCR and results (normalized to GAPDH) were expressed as the mean value of the ratio to baseline ± SEM (four samples per point) (b, c).

Figure 2.

Time-course and dose–response of LPS-induced CXCL2 protein release in IEC-6 cells. IEC-6 cells were incubated with culture medium alone or 5 µg/ml of LPS (or control culture medium) for 1–4 hr. CXCL2 protein released per 10⁶ cells was measured by ELISA. Results represent the mean ± SEM of four samples with duplicated determination for each sample.

Figure 3.

Time-course of TNF-induced CXCL2 gene expression in IEC-6 cells and its inhibition by PDTC. IEC-6 cells were incubated with culture medium alone, TNF alone (10 ng/ml) (left side) or PDTC (200 µm) for 30 min, followed by TNF (10 ng/ml) (right side) for various time periods (30 min, 45 min, 1 hr and 2 hr) (a and b). In (c), IEC-6 cells were incubated with increasing concentration of TNF (2, 10 and 50 ng/ml) for 1 hr. Cellular RNA was extracted, RT–PCR performed using primer specific for CXCL2 and GAPDH. PCR products were run on a 1·5% agarose gel and stained with SYBR green I (a). CXCL2 mRNA was quantified by real-time PCR and results (normalized to GAPDH) were expressed as the mean value of the ratio to baseline ± SEM (four samples per time point) (b).

Figure 4.

Time-course and dose–response of TNF-induced CXCL2 protein release in IEC-6 cells. IEC-6 cells were incubated with 10 ng/ml of TNF (or control culture medium) for 1–4 hr (a) or with increasing doses of TNF (2–50 ng/ml) for 4 hr (b). CXCL2 protein released per 10⁶ cells was measured by ELISA. Results represent the mean ± SEM of four samples with duplicated determination for each sample.

We then investigated whether LPS-induced CXCL2 gene expression was inhibited by PDTC, a potent NF-κB inhibitor. The dose of PDTC was selected based on a previous study which showed that PDTC (100 µm and 300 µm) inhibits IL-1β-induced IL-8 production in intestinal epithelial cells, without having toxic effects on cell respiration.²⁸ Preincubation with PDTC (200 µm) nearly completely blocked the induction of the CXCL2 gene by both LPS (Fig. 1a and Fig. 8a) and TNF (Fig. 3a and Fig. 9a).

Figure 8.

LPS-induced CXCL2 mRNA expression and protein release is dependent on NF-κB activation: IEC-6 cells were incubated with LPS alone (5 µg/ml), culture medium, PDTC 200 µm (or helenalin 50 µm) for 30 min, followed by LPS (5 µg/ml) for 2 hr or NBD peptide (or control peptide) 200 µm for 1 hr, followed by LPS. (a) RNA was extracted, CXCL2 mRNA was quantified by real-time PCR and results (normalized to GAPDH) were expressed as the mean value of the ratio to baseline ± SEM (four samples per group). (b) CXCL2 protein released per 10⁶ cells was measured by ELISA. Results represent the mean ± SEM of four samples with duplicated determination for each sample.

Figure 9.

TNF-induced CXCL2 mRNA expression and protein release in IEC-6 cells: effect of PDTC and helenalin. Cells were treated with TNF (10 ng/ml), culture medium, PDTC (200 µm) or helenalin (50 µm) for 30 min followed by TNF (10 ng/ml). (a) RNA was extracted, CXCL2 mRNA was quantified by real-time PCR and results (normalized to GAPDH) were expressed as the mean value of the ratio to baseline ± SEM (four samples per group). (b) CXCL2 protein released per 10⁶ cells was measured by ELISA. Results represent the mean ± SEM of four samples with duplicated determination for each sample.

PDTC blocked TNF- and LPS-induced NF-κB activation in IEC-6 cells

We previously showed that, in IEC-6 cells, LPS activated NF-κB at low dose in a dose-dependent manner (0·2 < 1 < 5 µg/ml) with an effect peaking at 1 hr while TNF-induced activation was much more rapid and already strong within 10 min.⁵ Here, we assessed the effect of PDTC, a potent NF-κB inhibitor, on both LPS- and TNF-induced NF-κB. Incubation of IEC-6 cells with LPS or TNF for 1 hr markedly activated NF-κB (Fig. 5) and the complex contained both p50 and p65. Preincubation with PDTC for 30 min completely blocked both LPS and TNF-induced NF-κB nuclear translocation (Fig. 5).

Figure 5.

PDTC attenuates LPS- and TNF-induced NF-κB activation in IEC-6 cells. IEC-6 cells were incubated with PDTC alone (200 µm), control medium, LPS (5 µg/ml), PDTC for 30 min, followed by LPS (5 µg/ml), TNF (10 ng/ml) or PDTC for 30 min, followed by TNF (10 ng/ml) for 1 hr. The NF-κB activity of the cell nuclear extracts was assessed by EMSA. Supershift experiments were done: –, no antibody added; p50, anti-p50 antibody; p65, anti-p65 antibody added. Similar results were obtained in three independent experiments.

PDTC blocks TNF-induced IκBα and IκBβ degradation in IEC-6 cells

In IEC-6 cells, as we had previously shown⁵ TNF induced a rapid and transient degradation of IκBα which was significantly reduced after 10 min, was nearly undetectable at 30 min, being resynthesized and reappearing at 2 hr. TNF-induced IκBβ degradation was slower, detectable after 20 min and increasing up to 1–2 hr (Fig. 6). Preincubation with PDTC completely blocked TNF-induced IκBα and IκBβ degradation (Fig. 6).

Figure 6.

PDTC blocks TNF-induced IκBα and IκBβ degradation in IEC-6 cells: cells were treated with control medium with or without PDTC (200 µm) for 30 min, followed by TNF (10 ng/ml) for different time periods (10, 20, 30 min, 1 hr and 2 hr). Cells were then collected and cell lysates analysed by western blot for IκBα and IκBβ proteins.

PDTC inhibits LPS-induced IκBα and IκBβ turn-over in IEC-6 cells

While we previously showed that LPS activated NF-κB in IEC-6 cells⁵ we did not see any changes in the level of IκBα and IκBβ by western blot.⁵ However, coincubation of the cells with CHX, an inhibitor of protein synthesis⁵ revealed an increased IκBα and IκBβ turn-over in IEC-6 cells following LPS treatment, which was not present in cells treated with CHX alone.⁵ Preincubation of the cells with PDTC completely blocked LPS-induced increase in IκBα and IκBβ turnover (Fig. 7).

Figure 7.

PDTC inhibits LPS-induced IκBα and IκBβ turn-over in IEC-6 cells. Cells were treated with control medium or PDTC (200 µm) for 30 min, followed by LPS (5 µg/ml) for different time periods (1 hr, 2 hr and 4 hr). Cells were then collected and cell lysates analysed by western blot for IκBα and IκBβ proteins.

LPS-induced CXCL2 mRNA and protein release is dependent on NF-κB activation

The induction by LPS of CXCL2 mRNA (2467·7 ± 336-fold increase compared to baseline) was totally blocked by PDTC 200 µm (1·86 ± 0·61-fold increase (P < 0·001)) and by helenalin 50 µm (0·65 ± 0·31-fold increase (P < 0·001)) (Fig. 8a).

Also, LPS-induced CXCL2 gene expression was significantly decreased by NBD peptide (185·88 ± 57·3-fold increase compared to the mutant peptide (2434·68 ± 330-fold increase (P < 0·01) (Fig. 8a). In parallel, LPS-induced CXCL2 protein release (567 ± 16 pg/ml, a 5·78-fold increase compared to baseline (98·4 ± 19 pg/ml)) was nearly completely blocked by PDTC 200 µm (103·1 ± 14 pg/ml) and was significantly decreased by NBD (257·8 ± 10·64 pg/ml) compared to mutant peptide (419·1 ± 26·75 pg/ml; Fig. 8b).

TNF-induced CXCL2 mRNA and protein release: effects of NF-κB inhibitors

The induction by TNF of CXCL2 mRNA (6280·7 ± 1085)-fold increase compared to baseline) was totally blocked by PDTC 200 µm (2·27 ± 1·14-fold increase (P < 0·001)) and by helenalin 50 µm (2·02 ± 0·74-fold increase (P < 0·001) (Fig. 9a). In parallel, TNF-induced CXCL2 protein to 279(± 14) pg/ml (a 2·84-fold increase compared to baseline (98·4 ± 19 pg/ml)) was nearly completely blocked by PDTC 200 µm (91·79 ± 2 pg/ml; Fig. 9b).

Anti-TNF antibodies do not block LPS-induced CXCL2 transcription in IEC-6 cells

As we had previously found that TNF mediates LPS-induced NF-κB activation in the intestine in vivo, we sought to investigate whether TNF mediates LPS-induced CXCL2 transcription in vitro in enterocytes. The neutralization dose₅₀ (ND₅₀) of anti-rat TNF had been previously determined to be approximately 0·02–0·08 µg/ml in the presence of 0·025 ng/ml of rTNF (R & D systems) in L-929 cell line. Pre-treatment for 30 min with anti-TNF antibodies (0·08 µg/ml) did not decrease LPS-induced CXCL2 transcription after 2 hr, compared to control antibodies (0·08 µg/ml) (Fig. 10).

Figure 10.

Anti-TNF antibodies do not block LPS-induced CXCL2 transcription in IEC-6 cells. Cells were treated with anti-TNF antibodies (0·08 µg/ml) or control antibodies (0·08 µg/ml) for 30 min, then incubated with LPS (5 µg/ml) or culture medium for 2 hr. Cellular RNA was extracted and RT–PCR performed using primer specific for CXCL2 and GAPDH. PCR products were run on a 1·5% agarose gel and stained with SYBR green I. Similar results were obtained in two independent experiments with three samples per group.

WEB2170 does not block LPS-induced CXCL2 transcription in IEC-6 cells

We had previously found that PAF mediates LPS-induced NF-κB activation in the intestine in vivo.²¹ Here we examined whether PAF mediates LPS-induced CXCL2 gene expression in vitro in enterocytes. Pre-treatment for 30–60 min with WEB2170 (1–5−10 µg/ml), a PAF receptor antagonist, did not decrease CXCL2 gene expression induced by LPS after 2 hr. This observation concurs with our previous study,⁵ which showed that LPS-induced NF-κB activation in enterocytes is independent of PAF (Fig. 11).

Figure 11.

WEB2170 does not block LPS-induced CXCL2 transcription in IEC-6 cells. Following pretreatment with either WEB2170 (1, 5 and 10 µg/ml) or culture medium for 30 and 60 min, cells were treated with LPS for 2 hr. Some cells were treated with WEB2170 alone. Cellular RNA was extracted and RT–PCR performed using primer specific for CXCL2 and GAPDH. PCR products were run on a 1·5% agarose gel and stained with SYBR green I. Similar results were obtained in two independent experiments.

Discussion

There is increasing evidence that the intestinal mucosa plays a primary role in the defence against bacterial invasion and contributes to the inflammatory response. The intestinal lumen is filled with more than 400 species of bacteria²⁹ and therefore the mucosa needs to tolerate commensal bacteria while being able to develop an inflammatory response against pathogenic bacteria. In an in vivo model of acute bowel injury induced by PAF, we showed that luminal bacteria are required for PAF to induce intestinal injury: because (a) germ-free rats are protected against bowel injury unless primed with LPS³⁰ and (b) endotoxin-resistant mice are protected against PAF-induced bowel injury.³¹ Necrotizing enterocolitis might be caused by the lack of tolerance to commensal bacteria during the colonization process, and IBD is thought to be caused by an inappropriate response of the host to its own resident luminal bacteria.²⁹ In a rat model of PAF-induced bowel injury, we have found that: (a) neutrophil influx into the intestine is increased and blocking neutrophil influx by blocking neutrophil–endothelial adhesion prevents neutrophil adhesion and subsequent bowel injury;¹⁰ (b) intestinal NF-κB is activated³ and the activated NF-κB is mostly immunolocalized to the epithelial cells of the villi and to some inflammatory cells of the lamina propria;³ and (c) CXCL2, a major chemokine for neutrophils, is induced and mediates PAF-induced bowel injury and intestinal neutrophil recruitment.¹² A previous investigation³² and our own study (data not shown) have found that LPS induces CXCL2 in the intestine in vivo.³² These observations suggest that NF-κB activation and CXCL2 induction lead to neutrophil influx, which in turn mediates bowel injury. Although it is difficult to discern the role of intestinal epithelial cells in vivo, a major role of the epithelial cells in the inflammatory response via NF-κB activation was demonstrated in a model of intestinal ischemia–reperfusion induced by a 30 min clamping of the superior mesenteric artery. In this model, the ablation in the enterocytes of IκB kinase, the enzyme responsible for the activation of the canonical NF-κB pathway, prevented the systemic inflammatory response.⁷ In the present study, we investigated the role of enterocytes in intestinal inflammation and injury by directly challenging the epithelial cells with inflammatory stimuli such as LPS and TNF and examined the production of CXCL2, the major chemokine responsible for the pathogenesis of bowel injury in our in vivo model.¹² IEC-6 cells were selected for the study because these cells are non-transformed rat small intestinal crypt cell line. Indeed, the small intestine is the major target organ of injury in the in vivo model³³ and TNF production has been localized to the small intestinal Paneth cells.¹⁸

Enterocytes have been shown to produce many pro-inflammatory cytokines in response to bacterial products such as LPS.¹ We previously showed that intestinal epithelial cells respond to LPS and TNF by activating NF-κB⁵ and inducing inducible nitric oxide synthase gene expression.⁵ In RAW264.7 macrophages, NF-κB and c-Jun binding sites are essential for LPS-induced CXCL-2 gene expression.³⁴ In this study, LPS induced CXCL-2 at very low concentration (100 ng/ml) within 30 min,³⁴ which was blocked by the coexpression of a mutant IκBα protein.³⁴ While a previous study showed that LPS induces CXCL2 in IEC-6 cells,¹³ whether these cells express CXCL2 gene in response to TNF and whether LPS and TNF-induces CXCL2 gene expression in IEC-6 cells via NF-κB activation remain unclear. Reports on CXCL2 expression by enterocytes are limited. In this study, we showed that TNF induces CXCL2 gene expression within 30 min, the effect peaking at 45 min and remaining significantly increased at 2 hr. Following TNF incubation, CXCL2 protein release increased at 1 hr, remaining elevated at 2 and 4 hr. In contrast, LPS induced CXCL2 expression more slowly, the induction becoming significant at 1 hr, peaking at 2 hr and remaining significant at 24 hr. Also, following LPS incubation, an increase in CXCL2 protein release could be detected only at 2 hr, increasing further at 4 hr.

To investigate the role of NF-κB on TNF and LPS-induced CXCL2 gene expression in IEC-6 cells, we examined the effect of 3 potent NF-κB inhibitors with different mechanisms of action: (1) PDTC, a metal chelator and antioxidant, which has been widely used to inhibit NF-κB, both in vivo³⁵ and in vitro;²⁸ PDTC blocks the ubiquitination of IκB;²⁵ (2) helenalin, a naturally occurring cell-permeable pseudoguainolide sesquiterpenoid lactone that inhibits NF-κB DNA binding activity by selectively alkylating the p65 subunit of NF-κB;²⁶ and (3) NBD peptide, a cell-permeable peptide, which by binding to NEMO, blocks its association with the IKK complex and inhibits NF-κB activation.²⁷ We found that PDTC blocked both the nuclear translocation of NF-κB and the IκBα and IκBβ degradation and turn-over induced by TNF and LPS. These findings are consistent with a previous study showing that PDTC blocked NF-κB activation by inhibiting the IκB-ubiquitin ligase activity, and this was independent of the antioxidant activity of PDTC.²⁵ We found that, in IEC-6 cells, both PDTC and helenalin blocked TNF- and LPS-induced CXCL2 gene expression. Furthermore, the specific disruption of the NEMO–IKK interaction with NBD peptide significantly blocked CXCL2 gene expression and protein release induced by LPS. These data suggest that the IKK pathway is involved in CXCL2 induction by LPS. In IEC-6 cells treated with TNF, both the NBD and control mutant peptides decreased CXCL2 mRNA and protein release (data not shown). As the NBD peptide is highly specific for the IKK complex and the mutant peptide has no effect on IKK activity²⁷ this effect of both peptides most likely reflects non-specific sensitization to toxicity in response to TNF but not LPS (Fig. 8a, b) in IEC-6 cells.

Because (a) LPS stimulates TNF production in the intestine,¹⁷ (b) the action of LPS on NF-κB activation and CXCL2 induction is much slower than TNF, and (c) enterocytes have been shown to produce TNF, and endogenously produced TNF by IEC-6 cells²³ has been shown to mediate the modulatory effect of LPS on cell growth in an autocrine/paracrine way,²³ it is also possible that endogenous TNF may mediate LPS-induced CXCL2 gene expression. In this study, anti-TNF antibodies did not decrease LPS-induced CXCL2 gene expression, and therefore a mediating role of TNF in an autocrine loop seems unlikely.

PAF is produced by intestinal epithelial cells²⁴ and these cells express PAF-receptors.³⁶ PAF has been found to have potent chemotactic properties on neutrophils via the production of endogenous leukotriene B4.³⁷ PAF is a potent activator of mitogen-activated protein kinase,³⁸ signal transducer and activator of transcription-3³⁹ and phosphoinositol-3 kinase.⁴⁰In vivo, PAF has been shown to be produced in a model of endotoxic shock²⁰ and to mediate LPS-induced intestinal injury.²⁰ PAF was also found to mediate LPS-induced NF-κB activation.²¹ It is therefore possible that PAF could play a role in mediating LPS-induced CXCL2 gene expression in enterocytes in vitro in an autocrine loop. However, in this study, we found that PAF did not mediate CXCL2 gene expression by LPS in enterocytes. This is consistent with our previous findings that PAF does not activate NF-κB in IEC-6 cells, and that WEB2170 does not inhibit LPS-induced NF-κB activation in these cells.⁵

In summary, we found that CXCL2 expression is induced and CXCL2 protein is released in enterocytes in response to LPS and TNF. These effects were inhibited by PDTC and helenalin, two NF-κB inhibitors. The specific disruption of the NEMO–IKK interaction inhibits LPS-induced CXCL-2 suggesting that the IKK pathway is involved. We did not identify an autocrine loop by either endogenous production of TNF or PAF to mediate LPS-induced NF-κB activation at the level of the enterocytes. However, in an in vivo setting, it is likely that the production of PAF and TNF by other cells, such as macrophages and dendritic cells, may amplify the inflammatory response in the disease process of septic shock and necrotizing enterocolitis. It is probable that, in infection or sepsis, the local production of CXCL2 by intestinal epithelial cells via the NF-κB pathway contributes to the mucosal neutrophil recruitment to amplify the inflammatory process, resulting in intestinal injury.

Abbreviations

CHX

cycloheximide

DTT

dithiothreitol

EDTA

ethylenediaminetetra-acetic acid

EMSA

electrophoretic mobility shift assay

IBD

inflammatory bowel disease

IκB

inhibitor protein κB

IKK

IκB kinase

LPS

lipopolysaccharide

MIP-2

macrophage inflammatory protein-2 (CXCL2)

NBD

NEMO-binding domain

NF-κB

nuclear factor-κB

NEMO

NF-κB essential modulator

PAF

platelet-activating factor

PDTC

pyrrolidine dithiocarbamate

PMSF

phenylmethylsulphonylfluoride

TNF

tumour necrosis factor-α