We have shown that ObRb, the leptin receptor, is overexpressed in colorectal cancer cells, and that this may influence the patients' outcome. We investigated colonocytes as leptin targets and characterized their pivotal role in antitumor immune response. Cytokine and chemokine mRNAs in HT29 cells were measured by targeted arrays. In vitro, normal colonocytes and human colon cancer cells (HT29, Caco-2, SW480, and HCT116) were used to investigate ObRb transduction system and cytokine releases. Animal colonocytes and CD8 splenocytes and human HT29, HCT116, and CD8+ cells from blood donors were used to investigate the lymphocyte response to the colonocytes when stimulated by leptin. Leptin-induced cytokine releases in the normal colonic mucosa and tumor growth and cytokine releases within tumors in vivo were measured in male rats and nude mice, respectively. Statistical analysis was done by Fisher's exact and Mann-Whitney U tests. Various cytokines and their receptors were produced in normal and tumoral colonocytes in response to leptin by increasing nuclear factor-κB activation. Interleukin-8 (IL-8) was the main cytokine produced in vitro. The levels of IL-8 and its receptor, CXCR1, were higher in tumors than in homologous normal mucosa. Systemic leptin enhanced the proinflammatory cytokines in normal colonocytes and in HT29 xenografted tumor colonocytes. Colonocyte-derived products after leptin treatment stimulated perforin and granzyme B expressions in normal CD8+ T cells in vitro. Leptin triggers an inflammatory response in tumor tissue by directly stimulating colonocytes, which can recruit T cytotoxic cells in the tumor microenvironment. [Cancer Res 2008;68(22):9423–32]
- Cytotoxic cell
- leptin receptor
Leptin is produced by fat tissue and the stomach in human adults ( 1). It is involved in energy balance and regulation of food uptake and nutrient absorption in the enterocytes ( 2) and enhances mitotic and apoptotic cells in the colonic mucosa. It targets ObRb, its main receptor on colonocytes ( 3– 5), and may induce the expression of nuclear factor-κB (NF-κB), activation of which leads to both cell proliferation ( 5) and production of proinflammatory cytokines and chemokines ( 6, 7). Leptin also targets the lymphocytes, within which it may stimulate development, differentiation, proliferation, activation, and cytotoxicity ( 6, 8– 10). However, the effect of leptin on the colonic mucosa is complex. In animals, exogenous leptin has been shown to promote experimental colitis by targeting colonocytes, PBMCs, and lymphocyte cells ( 9, 11), but its role in colon carcinogenesis is not yet fully understood, and contradictory data have been reported ( 12– 14). Mice with recessive ob mutations have nonfunctional ObRb and display increased susceptibility to spontaneous tumors ( 15), which could be associated with decreased systemic or intestinal immune function ( 7, 11). It has been suggested that leptin may be protumorigenic in wild-type animals because of its proliferative effect on the intestine ( 16). However, we have shown that in spite of this effect, it fails to exercise a protumorigenic role ( 3, 4) and that overexpression of the leptin receptor in colorectal cancer is associated with a better prognosis ( 17), suggesting that colonocytes could be a source of the proimmune response in colorectal cancer. In the present study, we describe mechanisms by which leptin triggers immune response of colonocytes.
Materials and Methods
Normal and Tumoral Human Tissues
Tumoral and normal colonic tissues from 32 patients (13 males, 19 females; mean age, 70.3 ± 9.5 y) were obtained from our tissue bank for the present study. They were selected on the basis of both quantitative reverse transcription-PCR (RT-PCR) and immunohistochemistry as having leptin receptor (ObRb) overexpression in tumors, as compared with normal homologous colonic tissues. All the tissue samples (32 normal and 32 colon cancer tissues) were characterized for cytokine and chemokine expression ( 17). For the current study, we focused on the expression of interleukin (IL)-8 and CXCR1 by using the following antibodies and dilutions: mouse anti-human IL-8 (clone BMS136, 1:50; Bender MedSystems) and mouse anti-human CXCR1 antibody (clone 501, 1:500; Biosource).
Human Colonic Cells and Cell Line Treatments for Cytokine Releases in Response to Leptin
Normal colonic epithelial cells from three individuals were freshly isolated and used as primary colon culture cells for dose-response assays of the cytokine release in response to recombinant human leptin (PeproTech, Inc.). Briefly, colon segments were made into a loop, inverted, and incubated in saline with 5 mmol/L EDTA at 4°C while shaking gently to dislodge the colon crypts. The isolated crypts were then collected and incubated for 30 min at 37°C in PBS without calcium containing 10% trypsin-EDTA (Gibco-Life Technologies). The cells were used for fresh culture or RNA extraction. The human colon cancer cell lines HCT116, Caco-2, SW480, HT-29, and Jurkat T cells [American Type Culture Collection (ATCC)] were cultured in Eagle's MEM (Gibco-Life Technologies) supplemented with 10% decomplemented fetal bovine serum (FBS; Eurobio) and 1% glutamine (Gibco-Life Technologies) in a humid atmosphere with 5% CO2 at 37°C, and then used for experiments once the cultures reached confluency. Normal human colon cells were maintained for 12 h in DMEM Glutamax without FBS before being used for cytokine release assays. We used polymyxin B sulfate (40 μg/mL; Sigma Aldrich) to abolish the effects of lipopolysaccharide (LPS) on the release of the cytokines ( 18). Escherichia coli 055:B5 LPS (Sigma Aldrich) was used as a control LPS. The monoclonal antibody (mAb) against the leptin receptor (9F8) that acts as an antagonist ( 19) was a kind gift from Prof. Richard J.M. Ross (University of Sheffield, Sheffield, United Kingdom).
Quantitative RT-PCR Analysis of the ObRb Leptin Receptor, Chemokines, and Cytokines
Total mRNA was prepared from tumoral and homologous normal specimens; cDNA was synthesized in reverse transcriptase samples; and appropriate primers (Supplementary Fig. S1) were chosen to detect expression of the genes using quantitative RT-PCR, as described elsewhere ( 20– 22). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH), β2-microglobulin, or β-actin genes were considered as housekeeping genes.
IL-8 Protein Measurement
Cytokine secretion into the supernatants of normal and tumor colon cell cultures was quantified by using the DuoSet ELISA Development kit (R&D Systems). Colorimetric results were read in a MRX Dynatech Microplate reader at a wavelength of 450 nm in 96-well high-binding Stripwell Costar EIA microplates (Costar). Substrate Reagent Pack (R&D Systems) was used for all streptavidin-horseradish peroxidase (HRP) reactions. Each sample was assayed in triplicate. The sensitivity limit of the assay was 3.5 pg/mL for IL-8.
The human gene arrays from Superarray system were used for gene expression analysis in HT-29 cells in basal and leptin-treated conditions. RNAs were extracted, and synthetic cDNAs were hybridized with human inflammatory cytokine, gene array membranes printed with cDNA fragments specific for 96 cytokine and receptor genes associated with the inflammatory response, and 72 genes encoding interleukins and interleukin receptors (GEArray Q Series HS-015.2 and HS-014 from SuperArray Bioscience Corp.). Results were expressed as the ratio of stimulated to control cells. For probe synthesis and biotin labeling, we used the GEArray AmpoLabeling-LPR Kit (SuperArray Bioscience) and biotin-16-dUTP (Roche Diagnostics) according to the manufacturer's instructions. Biotinylated and amplified cDNA probes were hybridized and alkaline phosphatase-streptavidin chemiluminescent detection was used (SuperArray Detection Kit). Membranes were used to expose X-ray films (Hyperfilm, Amersham Biosciences) and images were acquired with a desk scanner using 200 dpi. For data acquisition and analysis, we used ScanAnalyze software version 2.50 and GEArray Analyzer software, respectively. The mean signals for two negative controls [areas without spotted gene sequences (blanks) or areas spotted with nonhuman sequences (pUC18)] were calculated and subtracted from the raw results. We normalized these subtracted data by calculating their ratio to the averaged signals of three positive housekeeping genes. Each experiment was done at least twice to ensure the reproducibility of results.
Cytokine and Chemokine Measurements by RNase Protection Assay
RNA was extracted from the lysates using the chloroform-isopropanol method. mRNA was assayed using the Non-Rad RiboQuant Multiprobe RNase Protection Assay system (BD Biosciences) according to the manufacturer's instructions. Antisense RNA probes were generated using templates from the human cytokine probe sets 1 and 5 (hCK-1 and hCK-5) and the rat cytokine probe set 1 (rCK-1). With the exception of biotin-16-UTP (Roche Diagnostics), all reagents were supplied by the manufacturer of the RiboQuant Non-Rad In Vitro Transcription and RNase Protection Assay Kit (BD Biosciences Pharmingen). For chemiluminescent probe detection, we used the BD RiboQuant Non-Rad Detection kit (BD Biosciences Pharmingen). Samples were transferred onto membranes and treated with Luminol/Enhancer reagents; signals were detected using ECL Hyperfilm (Amersham Biosciences). Nucleotide lengths versus migration distances were compared with standards on a logarithmic grid.
Transfection and NF-κB Luciferase Reporter Gene Assays
Cells with impaired NF-κB activation were generated by transfecting them with pC-SR-IκB (kindly provided by Dr. H. Ashktorab) using Lipofectamine 2000 reagent (Invitrogen Life Technologies) according to the manufacturer's protocol. Normal or SR-IκB–positive cells were transiently transformed with pNF-κB-Luc carrying the luciferase reporter gene (Clontech BD Biosciences) using the calcium-phosphate method as described by the manufacturer (Mercury Pathway Profiling System, Clontech). Twelve hours after transfection, the cells were starved, cultured for 16 h in a medium lacking FBS, and stimulated with leptin (100 nmol/L) for up to 24 h. Luciferase activity was determined (Luciferase Assay System, Promega) using a luminometer (Junior; Berthold) as follows: 100 μL of assay buffer containing luciferin were injected into each culture, and the light emission was measured for 10 s. The results were normalized in terms of those obtained for the Renilla luciferase reference plasmid, and the values expressed as the relative luminometric intensity.
NF-κB p65 Activation
Cells were treated with leptin (100 nmol/L) or 9F8 mAb (10 μg/mL), harvested, washed with PBS, and resuspended in 1 mL hypotonic lysis buffer (Active Motif); then 0.1 mL of 10% NP40 was added and the sample centrifuged. The nuclear pellet was resuspended in 0.2 mL complete lysis buffer, incubated on ice for 30 min, and then centrifuged; the supernatant containing the nuclear extracts was snap frozen. The protein concentration of the nuclear extract was determined by a MicroBCA protein assay (Pierce). Nuclear extract was tested for NF-κB activation with the NF-κB p65 TransAM transcription factor assay kit (Active Motif). The active form of NF-κB p65 was measured by colorimetric reaction at 450 nm. Results were given as mean absorbance ± SD (at 450 nm, n = 4).
Western blotting. Cytoplasmic protein extracts from cells that were incubated for different lengths of time in the presence or absence of leptin were then subjected to SDS-PAGE, and the bands transferred onto Hybond polyvinylidene difluoride membranes (Amersham, Inc.). Membranes were probed overnight at 4°C with 1 μg/mL of polyclonal antibody to human ABIN-2 (Imgenex). HRP-conjugated goat anti-rabbit IgG (diluted 1:1,000; Santa Cruz Biotechnology) and chemiluminescence detection system (Amersham) were used.
Semiquantitative RT-PCR. Total RNA was extracted before and after exposure to 100 nmol/L leptin for 16 h as described above; cDNA synthesis was done; and templates were screened for the presence of ABIN-2 by PCR. Reaction products were visualized by ethidium bromide staining.
Male rats (220–260 g; Iffa-Credo, Saint Germain sur l'Arbresle, France) were handled according to the European Community guidelines for care and use of laboratory animals and were given i.p. injections of recombinant murine leptin (R&D Systems Europe Ltd.) diluted in saline at a dose of 2.5 mg/kg (n = 6) or saline alone (n = 6) as a control to analyze cytokine production by normal colonic mucosa in vivo. They were killed 16 h later and samples of colon mucosa were fixed in formalin; histopathologic examinations were done on representative H&E-stained sections ( 23). The colons were then reinverted and incubated in saline with 5 mmol/L EDTA to obtain freshly isolated cells for mRNA extraction as previously described ( 11), and the supernatants were collected and stored at −20°C until use.
For the analysis of in vivo cytokine production by tumor colonocytes, exponentially growing HT-29 cells (107) were harvested and s.c. transferred into nude mice after the functional activity of their leptin receptors was verified (IL-8 synthesis and/or NF-κB activation). Treatment with leptin (800 mg/kg/d for 4 wk) or its vehicle (four mice per group) began on day 0 as previously described ( 3), and body weight and tumor size were measured up to the time of killing (day 28). The tumors were surgically removed and weighed and then used for histopathologic analysis, estimation of the Ki-67 proliferative index ( 3), and IL-8 and CXCR1 measurements using RNase Protection Assay and RT-PCR. The levels of cytokines were normalized using the ratios of tissue necrosis (without necrosis/total tissue area) as assessed by the histopathologic analysis done before RNA extraction.
Tissue-associated myeloperoxidase activity was determined using a standard enzymatic assay ( 24) after tissues were removed, homogenized, centrifuged, and pellets resuspended in the buffer. The pellets were resolubilized by sonication in hexadecyltrimethylammonium bromide and collected in phosphate buffer and centrifuged; myeloperoxidase activity in the supernatant was assayed by adding 0.05% hydrogen peroxide. The change in absorbance at 460 nm was measured spectrophotometrically over 3 min, and 1 unit of myeloperoxidase activity was defined as that degrading 1 μmol of hydrogen peroxide per minute.
Colonocyte and Immune Cell Interaction
To study the effect of leptin-stimulated colonocytes on immune cells, we used both normal and tumor colon culture cells, which were treated with leptin (10, 50, and 100 nmol/L) or vehicle for 24 h; supernatants were harvested and used to simulate immune cells.
Human tumor cells. The HT-29 cells, at confluence, were incubated in new fresh media without or with recombinant human leptin (OB-300-27, PeproTech) for 24 h. Supernatants were then gently collected and aliquoted or transferred directly to human peripheral blood mononuclear cell (PBMC) or enriched CD8+ T-cell culture flasks.
Preparation of CD8+ enriched T cells. PBMCs were obtained by Ficoll-Hypaque (Pharmacia) gradient centrifugation from EDTA-anticoagulated whole blood samples (n = 6 healthy blood donors). One part of these cells was labeled with human whole blood CD8 MicroBeads (130-090-878 Miltenyi Biotec; 50 μL/mL of sample) according to the manufacturer's protocol. Labeled suspensions were separated positively by Posseld2 program in an autoMACS Separator unit (Miltenyi Biotec). Positive fractions containing CD8+ cells (>96% purity verified by fluorescence-activated cell sorting) were resuspended (5 × 105 cells per flask) in DMEM Glutamax I supplemented with 5% human heat-inactivated AB serum (Valley Biomedical), 100 units/mL penicillin, 100 μg/mL streptomycin, and 50 μg/mL gentamicin (Sigma). After 24 h, the CD8+ enriched T-cell flasks and the flasks containing nonseparated cells were incubated for 48 h with control or leptin-treated HT-29 culture supernatants by simple transfer (v:v) of these supernatants on monolayer human immune cultured cells.
Preparation of normal mouse colonocytes. Colon segments from 48 male adult BALB/c mice were removed and colonocytes isolated following the Roediger and Truelove method ( 25). Cell viability was determined by trypan blue and cells were distributed (1.2 × 106 per flask) in 40 culture flasks (Nalge Nunc International) and incubated for 24 h; the media were discarded and cell layers were washed. New fresh media with or without recombinant mouse leptin (498-OB-01M, R&D Systems Europe) were added (10, 50, and 100 nmol/L) and incubation was continued for another 24 h. The supernatants were then collected and transferred (1v:v) directly to mouse splenocytes or mouse CD8+ enriched T-cell culture flasks.
Preparation of normal mouse splenocytes and CD8+ T cells. Twenty normal spleens were obtained from the same mice and a single-cell suspension was prepared by squeezing spleens through a sterile 70-μm nylon cell strainer (Falcon) with the rough end of a 2-mL syringe plunger. Pooled suspensions were centrifuged; the pellets were resuspended in erythrocyte lysis buffer (Sigma; 5 mL/spleen) for 3 min; and cells were counted (65 ± 5.7 million splenocytes per spleen). CD8+ T cells were purified; samples of splenocytes were collected in MACS labeling buffer (Miltenyi Biotec) containing anti-CD8 mAb–coated microbeads (Miltenyi Biotec), and MACS-positive cells were isolated in an autoMACS machine (Miltenyi Biotec). Pooled CD8+ T cells were cultured (106 per flask). Culture flasks containing nonseparated splenocytes and those with CD8+ enriched cells were incubated for 24 h (37°C, 5% CO2), and then supernatants of mouse control- or leptin-treated colonocytes were added and incubation was continued for another 48 h. Remaining monolayer splenocytes in culture flasks after supernatant removal were harvested and resuspended in radioimmunoprecipitation assay cell lysis buffer II solution for 10 min on ice before testing for capacity of cell lysis. Supernatants were centrifuged (13,000 rpm, 10 min) and prepared for ELISA and Western blotting.
Granzyme B and perforin ELISA tests. The concentrations of granzyme B and human perforin were measured in 96-well ELISA plates (Nalgo Nunc) according to the manufacturer's protocol by using ELISA Ready-SET-Go kit (eBiosciences) and human perforin ELISA kit (Cell Sciences), respectively.
Granzyme B and perforin Western blotting. Cell lysates from mouse CD8+ cells or human CD8+ enriched PBMCs were homogenized in a Protease Inhibitor Cocktail (Pierce Biotechnology) and total proteins were separated by SDS-PAGE. Protein bands were transferred onto polyvinylidene difluoride membranes, probed with mouse mAb IgG1 against granzyme B (2C5, Santa Cruz Biotechnology) diluted 1:300, β-actin mouse mAb IgG1 (Ac-15, Abcam) diluted 1:500, and mouse mAb IgG2a perforin 1 (Santa Cruz Biotechnology); all reactions were visualized with goat anti-mouse IgG-HRP (Santa Cruz Biotechnology) by using chemiluminescence ECL detection system (Pierce Biotechnology).
Cytotoxicity test. The cytotoxic activity of CD8+ enriched splenocytes was determined in a standard 51Cr release assay. These effector cells were harvested after incubation with supernatant collected from control- or leptin-treated colonocyte flasks and incubated for 4 h with 51Cr-labeled (Amersham) P815 [mastocytoma cells of BALB/c mice (MHC-H2d); ATCC TIB-64, LGC Standards] cells as syngenic target cells and EL4 [lymphoma cells of C57BL/6 mice (MHC-H2b); ATCC TIB-39, LGC Standards] as control target cells for specificity. Varying numbers of effector cells were added to 10,000 51Cr-labeled target cells in appropriate effector/target cell ratios. Supernatants were removed and radioactivity was counted, to determine the percentage of specific lysis, as 100 × [(cpm released from target cells in the presence of effector cells − cpm released from target cells alone) / (cpm released from target cells lysed with 10% Triton X-100 − cpm released from target cells alone)], where cpm is counts per minute. Maximal cell lysis was determined in control target cell samples (n = 3) after Triton X-100 was added and radioactivity measured.
The correlation between parameters was analyzed using a correlation t test. Values were considered statistically significant at P < 0.05. Quantitative variables are expressed as means (± SD) or medians (extremes), as appropriate; categorical variables were expressed as numbers (%). Data were analyzed using Fisher's exact test and Mann-Whitney nonparametric U test for categorical and quantitative parameters, respectively, with P < 0.05 considered as significant.
Leptin-induced proinflammatory cytokine expression in colorectal tissues. Of the 96 genes encoding for cytokine and cytokine receptors and the 72 genes encoding for interleukins and interleukin receptors, the expression of several genes (i.e., IL-1, IL-1β, IL-8, CXCR1, IL-8 Rb, LT-β, TARC, MCP-1, MIP-3α, RANTES, and MCP-3) was found to be significantly higher in the leptin-treated cells than in control cells ( Fig. 1A and B ) when a trend toward down-regulation was observed for the IL-10, TGF-β and IL-23 genes ( Fig. 1A and B). Figure 1C shows the chemokines and cytokines for which quantitative RT-PCR revealed significantly higher levels of mRNA in tumoral than in normal mucosa. In addition, a significant close relationship was found between the mRNAs of IL-8 and its receptor, CXCR1, with the analysis focused on normal tissues only, tumoral tissues only, or the tumor/normal ratio mRNA tissue levels (R2 = 0.361; P = 0.0001; n = 32 for tumor/normal ratio). IL-8 and its receptor were expressed in normal colonocytes and heterogeneously in the tumoral cells as assessed by immunohistochemistry (Supplementary Fig. S2). Among the leptin-activated genes, IL-8 seemed to have underwent the greatest stimulation. Further investigations were therefore done on IL-8 production in normal and tumoral colonocytes using in vitro and in vivo models.
IL-8 expression in colonocytes according to ObRb activation, in vitro. Leptin stimulated IL-8 gene expression in a time- and dose-dependent manner as assessed by ELISA protein testing ( Fig. 2A ) and mRNA (RPA) levels ( Fig. 2B) in normal human colonic cells and in HT-29, SW 480, Caco-2, and HCT116 cells. In isolated normal colonocytes, levels of IL-8 protein increased significantly in response to 10 nmol/L leptin [from 480 ± 50 pg/mL (n = 3) to 860 ± 116 pg IL-8/mL (n = 5)]. No further increase was observed after adding 100 nmol/L leptin (940 ± 114 pg IL-8/mL; not significant). Incubating HT-29 cells with leptin enhanced IL-8 secretion to levels significantly higher than those observed with normal colonocytes. This effect was time dependent up to 48 h. Leptin-induced IL-8 was not abolished by polymyxin B, whereas polymyxin B completely inhibited the induction of IL-8 production by E. coli LPS (ref. 18; Fig. 2C). This implies that the leptin-induced effect could not be due to nonspecific effects of microbial flora LPS. The specificity of the leptin effects on colonocytes was confirmed using 9F8 mAb, which targets ObRb and inhibits the effects due to leptin ( Fig. 2D; ref. 26).
The effect of leptin on normal colonic mucosa and colon cancer cells in vivo. A single i.p. injection of either recombinant murine leptin or saline resulted in the absence of colon injury as checked by macroscopic and microscopic examinations, and histopathologic analysis did not show significant stigmata of inflammation in the mucosa. However, myeloperoxidase activity in the colonic mucosa of leptin-treated animals was significantly higher than that in the controls ( Fig. 3A ). Although IL-8 is not produced in the rat, this finding was consistent with the fact that the levels of IL-1β, IL-6, and tumor necrosis factor-α (TNFα) mRNAs found in both the left and right colonic mucosa were higher in the leptin-treated rats than in the controls ( Fig. 3B).
During the tumorogenicity experiment, leptin- and vehicle-treated mice had similar body weight curves, and tumor volume remained unchanged. The leptin receptor remained expressed in these tumor cells, as confirmed by immunohistochemistry and Western blotting in both groups of mice from baseline and at day 28. The mean Ki-67 proliferative index was significantly lower in the leptin-treated mice (35%) than in the controls (55%; P = 0.005) at day 28 ( Fig. 3C), whereas the levels of IL-8 and CXCR1 were higher in tumor cells from leptin-treated than in those from saline-treated mice ( Fig. 3D).
Mechanisms. The peak of NF-κB activation ( Fig. 4A ) appeared 4 h after leptin treatment in normal and tumor colon cells ( Fig. 4B); the SR-IκB super-repressor gene, which inhibits constitutive NF-κB activity, suppressed the leptin-induced production of IL-8 ( Fig. 4C) in these cells. The specific suppression by 9F8 mAb of the effects of leptin on NF-κB nuclear translocation in HT-29 cells was also verified (data not shown). ABIN-2 was found to be lower in leptin-treated HT-29 confluent cells than in untreated confluent cells by both the semiquantitative RT-PCR and Western blotting techniques ( Fig. 4D).
Effect of colonocytes on immune cells in vitro. Supernatants from colon cancer cell cultures (HT29 and HTC116 cells) significantly enhanced perforin expression in human CD8+ enriched cells when colon cells were preincubated in the presence of leptin (10, 50, and 100 nmol/L). Similarly, supernatants from normal mouse leptin-treated colonocytes enhanced granzyme B expression in normal mouse splenocytes and CD8+ cells and increased cytotoxic activity ( Fig. 5 ). Furthermore, in these conditions, the percentage specific lysis of P816 cells used as syngenic targets was significantly (P < 0.01) higher than that of EL4 control target cells or control ( Fig. 5D). All these effects were suppressed if the colonocytes were treated with leptin in the presence of 9F8 mAb, whereas pretreatment of human T cells with 9F8 mAb did not suppress the cytotoxic activity of T cells (data not shown).
We show that leptin increases the expression of several cytokines in normal and tumoral colonocytes by targeting ObRb, its main receptor. Among them, IL-8, a potent proinflammatory cytokine ( 27– 29), is shown to be related to the level of NF-κB activity ( 7), which was significantly higher in tumoral than in normal colon cells. Further, we showed enhancement of CXCR1, the receptor for IL-8, and down-regulation of ABIN-2 ( 30) in leptin-stimulated colonocytes. These effects could be inhibited by a specific mAb that targets ObRb or by in situ inhibition of the NF-κB transcript. In vivo, leptin is shown to initiate proinflammatory response in the normal colon mucosa and to enhance cytokine release from colon tumor cells. Therefore, we hypothesized that leptin stimulates colonocytes to produce substances that are able to stimulate normal human and mouse T cytotoxic cells. We suggest that leptin can trigger colonocytes to recruit and stimulate T lymphocytes with possible role in antitumor response in colorectal cancer.
Inflammation in tumors was shown to favor progression ( 31), and aberrant regulation of NF-κB signaling pathway in cancer development has been extensively studied. It is now believed that overactivation of NF-κB in promoting cell survival is an important obstacle in many cancer therapies, and several chemical inhibitors of NF-κB have been developed. However, up to now, their total relevance could not be shown in clinical studies. Indeed, NF-κB machinery regulates the expression of hundreds of genes that are involved in different cellular processes, including not only inflammation but also stress responses and cellular immunity. If mutations in the core components of NF-κB in some tumors require direct targeting of NF-κB, one could suggest that in the majority of tumor cell systems its inhibition per se may not result necessarily in favorable final consequence. This is at least the case with leptin-induced NF-κB activity as documented in the present study and in the literature. Leptin down-regulates ABIN-2, a cytoplasmic protein that inhibits NF-κB ( 30), and still increases the NF-κB activity in tumor cells, resulting in increased level of cytokine production that, in turn, may start immune response. In addition, in rat receiving leptin, we did not observe tumor growth in vivo in spite of the augmentation of colon cell proliferation ( 3, 16) and NF-κB activation in the colonocytes ( 5).
Besides inflammation, the presence of effector memory CD8+ lymphocytes within tumors is associated with better survival ( 32), and we have recently shown that tumor phenotype (i.e., microsatellite instability) influences shifting from proinflammatory to efficient immune response in the tumor microenvironment ( 33). In a series of patients undergoing surgery for colorectal cancer ( 17), we showed a significant relationship between overexpression of the leptin receptor in tumor cells, higher infiltration of T lymphocytes and cytotoxic cell activity in tumors at baseline, and better patient outcomes during follow-up. We found that lymphocyte infiltration within tumors was associated with the levels of those cytokines and chemokines that have been reported by others as markers of antitumor immune response (i.e., perforin and granzyme B; ref. 32). We hypothesized that colonocytes might influence antitumor immune response by involving colonocytes and immune cells in a “cell-cell” interaction. First, normal and tumoral colonocytes were both shown to be sources of proinflammatory cytokines with higher levels of IL-8 and its receptor. CXCR1, in tumoral colonocytes than in normal cells, suggesting that the capacity of colon cells to produce cytokines may vary in proportions that depend on the leptin receptor expressions. This is consistent with the fact that db/db rats, which have a mutated nonfunctional leptin receptor, did not display increased inflammation in the colonic mucosa ( 11), whereas in nude mice, in which IL-8 cannot normally be detected, HT29 xenografted cells overproduced both IL-8 and its receptor, CXCR1, in vivo, in response to leptin. Second, although leptin is able to recruit leukocytes and lymphocytes by direct targeting ( 7– 9, 11), we show in the present study that leptin-induced cytokine enhancement in the normal colonic mucosa occurs before leukocytes and/or lymphocytes infiltrate the mucosa with no difference between right-sided and left-sided normal mucosa. These findings are also consistent with the kinetics of inflammatory cell infiltration in the colonic mucosa in response to leptin in the ulcerative colitis ( 12, 34, 35), and suggest that overexpression of leptin receptor in tumors with MSI phenotype is likely not associated with right-sided colorectal cancer, which is usually the case in this subgroup of colorectal cancer. Thus, we suggest that leptin targets colonocytes, triggering the production of various proinflammatory and proimmune cytokines. Because expressions of both IL-8 and its receptor, CXCR1, are found to be significantly associated in normal tissues, in tumor tissues, and in HT29 xenografted cells in nude mice, IL-8 may be considered as an amplificatory mediator in the leptin-induced inflammatory response of colonocytes.
Third, colonocytes can interact with immune cells via a paracrine pathway. We therefore used a coculture cell model including colon cells and immune cells to clarify this point. Because lymphocytes can also be stimulated by leptin by direct targeting of ObRb ( 36), we first stimulated colon cells with leptin, and then harvested the media from these culture cell flasks. Supernatants have been found to activate perforin gene ( 37) in Jurkat T lymphocytes (data not shown) when cell cultures were incubated in the presence of leptin. More interestingly, we could clearly show that human cytotoxic cells were stimulated by these culture media. A direct effect of leptin on immune cells seems to be unlikely because pretreating them with 9F8 mAb, a specific leptin receptor antagonist, did not suppress this effect. By contrast, pretreating colon culture cells with this antagonist did suppress the effect of the culture media on T cells, indicating that the leptin receptor in the colonocytes is required for this effect to occur. Whether normal colonocytes can stimulate normal immune cells was investigated in mice, and levels of perforin and granzyme B expressions, two known markers of T cytotoxic activity, were found to be related to the leptin concentrations in the colon culture cell media. As a final pathophysiologic response, the cytotoxic activity of mouse splenocytes was documented when they were restimulated by normal colonocytes. These cells seem to activate CD8+ cells in vitro in proportions depending on the leptin concentration when pretreated with leptin. This effect is mediated by the natural killer–dependent pathway because stimulated colonocyte-splenocyte–induced lysis was observed with target P815 but not with EL4 cells. Thus, we suggest that the density of leptin receptor on tumor colon cells influences cytokine and chemokine releases from these cells, which in turn can recruit and stimulate immune cytotoxic cells ( Fig. 6 ). This is consistent with tumors with MSI phenotype that are known to be associated with improved patient outcome, likely due to higher lymphocyte recruitment in the tumor microenvironment ( 31, 33), and express higher number of leptin receptor as we have recently reported ( 17). Because IL-8 favors T-cell homing and lymphocyte migration ( 38), and human T cells are known to be activated by a perforin-dependent mechanism ( 39), the colonocytes could be the link between leptin receptor expression and the mucosal cytotoxic immune response in tumors ( 8). This is the first study to describe pathways by which factors involved in the nutriment uptake may stimulate the tumor immune response in colon and rectal cancers. Whether this effect has a systemic effect requires further investigations.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: Ligue Nationale Contre le Cancer and Association Charles Debray.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. Hamid Zarkesh Isfahani and Prof. Richard J.M. Ross (Head of the Endocrinology and Reproduction Section; University of Sheffield, Sheffield, United Kingdom), Dr. John M. Carethers (Division of Gastroenterology, Department of Medicine, University of California, San Diego, CA), Dr. Hassan Ashktorab (Howard University Cancer Center, Washington, DC), and Dr. Farhad Heshmati (Department of Blood Transfusion Cochin Hospital, Paris, France) for their kind gifts, and Dr. Mohammad Yaghoubi, François Berrehar, and Feriel Bouabbas for their management and technical assistance.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Current address for M. Abolhassani: BCG Department, Pasteur Institute, Tehran, Iran.
- Received March 19, 2008.
- Revision received August 8, 2008.
- Accepted August 19, 2008.
- ©2008 American Association for Cancer Research.