Main

The intestinal epithelium is continuously exposed at the epithelial cell's apical surface to luminal factors, including nutrients and bacterial products. The epithelial cell is also in contact at its basolateral surface with cytokines secreted by cells in the intestinal mucosa. Although the primary role of the epithelium is the absorption of nutrients, it also acts as a protective barrier for the host against infectious agents, toxins, and antigens within the intestinal lumen(1, 2). Mucosal immune responses play a central role in epithelial defensive mechanisms that are mediated, in part, by secretory IgA. In addition, intestinal epithelial cells secrete IL-1, IL-6, tumor necrosis factor-α, IL-10(2), and chemotactic cytokines (chemokines), including IL-8(3–6).

IL-8 is an α-chemokine with a molecular weight of 8-10 kD. IL-8 possesses potent neutrophil chemoattractant and activating properties(7, 8). It is secreted by granulocytes, endothelial cells, fibroblasts, T cells, keratinocytes, monocytes/macrophages(7), and by intestinal epithelial cells(3–6). IL-8 is increased in the mucosa of patients with both active ulcerative colitis and active Crohn's disease(5, 9).

Intestinal epithelial cells may, by secreting chemokines, participate in the attraction of leukocytes into the intestinal mucosa(2–6). The question arises, therefore, whether the intestinal milieu regulates this epithelial participation. The short chain fatty acids, acetate, propionate, and butyrate, are the products of bacterial fermentation of carbohydrate(10–12). We hypothesized, therefore, that factors in the lumen, such as short chain fatty acids, may alter the production of IL-8 by epithelial cells. Thus, butyrate may not only be an energy source for colonocytes(13), stimulating epithelial cell proliferation in vivo(14), but it may also carry information on the bacterial ecosystem to the intestinal epithelium. The concentration of butyrate varies from 5 mM(15) to 24 mM(16) in the normal large intestine.

We examined whether sodium butyrate altered the secretion of IL-8 by epithelial cells stimulated with proinflammatory molecules. Lipopolysaccharide, a component of the outer membrane of Gram-negative bacteria, by itself does not stimulate IL-8 secretion by Caco-2 cells(3). However, it does stimulate IL-1β production from macrophages(17). We, therefore, also examined whether IL-1β could act in synergy with butyrate to alter IL-8 secretion by intestinal epithelial cells. Our studies demonstrate that changes in butyrate concentrations are directly associated with changes in chemokine secretion by epithelial cell line cells stimulated with lipopolysaccharide or IL-1β.

METHODS

Culture of Caco-2 cell or H-4 cell monolayers. Caco-2 cells constitute a cell line derived from human colon carcinoma which differentiates spontaneously in culture(18–20). The H-4 cell line is a nonmalignant human small intestinal epithelial cell clone derived from human fetal small intestine(21). Caco-2 cells (American Type Culture Collection, Rockville, MD, passages 20-40) and H-4 cells (passages 20-36) were plated into 12-well plates (Falcon, Becton Dickinson, Franklin Lakes, NJ) at an initial density of 5 × 104 cells/cm2. Both Caco-2 cells and H-4 cells were grown in DMEM(GIBCO-Life Technologies, Inc., Grand Island, NY) containing FCS (Hyclone Laboratories, Logan, UT), 2 mM glutamine (GIBCO), 100,000 U/L penicillin, and 100 mg/L streptomycin (GIBCO), 0.1 mM minimal essential medium nonessential amino acid solution (GIBCO), and 10 mM HEPES buffer solution (GIBCO) in a humid atmosphere at 37°C and 5% CO2.

On d 7 after plating, the media were changed from media containing serum to media without serum (DTS). DTS medium was DMEM (GIBCO) supplemented with 5 mg/L iron-saturated human transferrin (Sigma Chemical Co., St. Louis, MO) and 5 μg/L selenous acid (Na2SeO3; Collaborative Biomedical, Bedford, MA) but contained no serum(22). Caco-2 cells grow and differentiate normally in DTS medium(22). Experimental protocols were initiated on d 14, when cells were confluent and their state of differentiation stable.

Protocol of incubation with butyrate and stimulation with IL-1β or lipopolysaccharide. IL-8 secretion by intestinal epithelial cells in media containing both sodium butyrate and lipopolysaccharide or IL-1β. Caco-2 cells or H-4 cells were incubated with varying concentrations of butyrate (0-20 mM) (Sigma Chemical Co.) in DTS media (without serum) for 48 h. No changes in the pH of the media were observed with the addition of butyrate. The epithelial cells were stimulated with 100 μg/mL lipopolysaccharide (from Escherichia coli serotype 055:85, Sigma Chemical Co.) or 1 ng/mL IL-1β (R&D Systems, Minneapolis, MN) during the last 24 h. Further experiments were performed to examine IL-8 mRNA accumulation within Caco-2 cells preincubated with 20 mM sodium butyrate for 24 h and subsequently stimulated for 2 h with 1 ng/mL IL-1β or 100 μg/mL lipopolysaccharide.

Dose response to lipopolysaccharide or IL-1β. Caco-2 cells were incubated with and without sodium butyrate (5 mM) for 48 h. After 24 h, cells were stimulated with varying concentrations of lipopolysaccharide (0-100μg/mL) or IL-1β (0-100 ng/mL) for a further 24 h.

Time course of IL-8 secretion. Caco-2 cells were incubated with and without sodium butyrate (20 mM) for 0, 12, 24, 26, and 30 h. After 24 h, cells were stimulated with IL-1β (1 ng/mL), for 2 h and for 6 h. Supernatants, Caco-2 cell proteins and Caco-2 cell RNA from the same wells were collected at the different time points described.

Sequential media exchange experiments. Cells were preincubated with sodium butyrate (0-20 mM) for 24 h, then washed twice with DTS media. Fresh DTS media without butyrate were placed into the wells followed by addition of lipopolysaccharide (100 μg/mL) or IL-1β (1 ng/mL) for a further 24 h. In these sequential media exchange experiments, butyrate and lipopolysaccharide or IL-1β were not present concurrently.

Assay of IL-8 secretion and mRNA accumulation . Measurement of IL-8 secretion. IL-8 levels were measured in cultured supernatants by ELISA (Quantikine™ R&D Systems). The culture supernatants from Caco-2 cells or H-4 cells were collected, centrifuged at 800 rpm, and stored frozen at -20°C. The amount of IL-8 was quantified within each supernatant in duplicates. The maximum variation between duplicates was 2%. Colorimetric results were read on a Titertek Multiscan II 96-well plate reader at a wavelength of 450 nm.

Measurement of total Caco-2 or H-4 cell protein. The cell protein content was measured to express IL-8 production per unit weight of cellular material. Deionized water was placed into wells containing adherent Caco-2 cells or H-4 cells to detach the cells. They were immediately sonicated with an ultrasonic cell disrupter (micro-ultrasonic cell disrupter; Kontes, Vineland, NJ). The amount of total protein was then measured with the bicinchonic acid protein assay (BCA; Pierce, Rockford, IL). Colorimetric results were read on a DU-70 spectrophotometer (Beckman Instrument Inc., Wakefield, MA) at a wavelength of 562 nm and converted into protein concentration using an albumin standard calibration curve. The protein content within each well was measured in duplicates with an experimental variation of less than 5%.

Caco-2 cell mRNA collection. Caco-2 cells were cultured in the presence or absence of butyrate for 24 h followed by lipopolysaccharide or IL-1β for 2 h. Caco-2 cells were homogenized, and the RNA was extracted, precipitated, washed, and collected according to the manufacturer's recommendations as described in the procedures for RNA isolation in the RNAzol kit (Biotecx Laboratories, Inc., Houston, TX).

RNA transfer blot analysis of Caco-2 cell mRNA. Purified total RNA (30 μg) from each well was separated by electrophoresis in a 2% agarose gel containing 1.8% formaldehyde (Mallinckrodt, Paris, KY) and 20% 3-[N-Morpholino]propane sulfonic acid (Sigma Chemical Co.) and transferred to Gene-Screen Plus™ membranes (Dupont NEN, Boston, MA) by capillary action. The membranes were then baked, prehybridized, hybridized with labeled IL-8 cDNA (a gift from Dr. Joost Oppenheim, Laboratory of Molecular Immunoregulation, NCI-FCRPC, Fredrick, MD). Blots were washed in 0.5× SSC and 0.5% SDS (Bio-Rad Laboratories, Hercules, CA) at 65°C. Washed blots were analyzed by autoradiography between intensifying screens at-70°C as previously described(23).

Polarity of IL-8 secretion by Caco-2 cells. Caco-2 cells were grown on 12-mm cell culture inserts with a 0.45-μm pore size (Falcon). The change from normal media to serum-free (DTS) media was made on d 19. Confluency of Caco-2 cell monolayers in the culture inserts was confirmed by measuring transepithelial electrical resistance using a Millicell-ERS system(Millipore Corporation, Bedford, MA) on d 21. Sodium butyrate (0, 2.5, and 5 mM) was then added into both the upper and lower compartments of the inserts for 24 h, and lipopolysaccharide (100 μg/mL) was added into the upper compartments for a further 24 h. Culture supernatants were then collected, and IL-8 was measured (as described above).

Statistics. Statistical significance of the differential effects of sodium butyrate on IL-8 secretion was assessed by regression analysis. Each point entered into the data represented a separate well. Statistical differences between controls and the lowest butyrate concentrations (2.5 mM) were also examined using the t test. p values of less than 0.05 were considered statistically significant.

RESULTS

Butyrate induces Caco-2 cells to secrete IL-8 in response to lipopolysaccharide. Caco-2 cells did not secrete IL-8 when stimulated with lipopolysaccharide (100 μg/mL). This confirmed previous reports showing a lack of response of Caco-2 cells to lipopolysaccharide(3). However, when Caco-2 cells were incubated with sodium butyrate, lipopolysaccharide stimulated IL-8 secretion (Fig. 1). Furthermore, the IL-8 production increased with increasing concentrations of sodium butyrate (p < 0.001;Fig. 2). There was no plateau to this effect even at 100 mM (data not shown). There was a small, but nevertheless statistically significant, increase in IL-8 secretion (p < 0.001) by unstimulated cells incubated with sodium butyrate alone (Figs. 1 and 2). To ensure that there was no direct extracellular interaction between lipopolysaccharide and sodium butyrate, sequential media exchange experiments were performed removing butyrate completely from the culture medium before adding lipopolysaccharide. Sodium butyrate induced lipopolysaccharide responsiveness even after it had been removed before lipopolysaccharide stimulation (data not shown).

Figure 1
figure 1

Sodium butyrate induces the secretion of IL-8 in Caco-2 cells stimulated with lipopolysaccharide (LPS). Only cells treated with sodium butyrate (5 mM) for 24 h responded to lipopolysaccharide. Concentrations greater than 1 μg/mL lipopolysaccharide stimulated IL-8 secretion. The bars represent standard deviations of three different wells for each point.

Full size image
Figure 2
figure 2

The effect of sodium butyrate on lipopolysaccharide(LPS)- or IL-1β-stimulated secretion of IL-8 by Caco-2 cells. Lipopolysaccharide at 100 μg/mL alone did not induce IL-8 secretion by Caco-2 cells, but in the presence of sodium butyrate added 24 h previously, lipopolysaccharide induced IL-8 production (p < 0.001). The magnitude of IL-8 secretion was dependent on the concentration of sodium butyrate (p < 0.001). Sodium butyrate alone induced Caco-2 cells secretion of IL-8 but to a limited extent. Sodium butyrate also enhanced the secretion of IL-8 by Caco-2 cells stimulated with IL-1β (1 ng/mL). The bars represent standard deviations of three different wells for each point. The results are representative of five experiments.

Full size image

The effects of sodium butyrate were compared with equimolar concentrations of sodium bicarbonate, another weak base, to exclude the possibility that the effects of sodium butyrate may have been due to any alteration in the charge of lipopolysaccharide in the media, or of proteins at the cell membrane. Unlike sodium butyrate, sodium bicarbonate did not induce IL-8 production by Caco-2 cells stimulated with lipopolysaccharide (data not shown). Furthermore, alterations in the concentration of glutamine (a recognized epithelial nutrient) had no effect on IL-8 secretion. Thus, the effect of sodium butyrate on lipopolysaccharide induction of IL-8 secretion by Caco-2 cells was due to the butyrate moiety itself.

Butyrate enhances IL-1β stimulation of IL-8 secretion by Caco-2 cells. The cells of the intestinal epithelium are directly exposed not only to bacterial products from the lumen but also to proinflammatory cytokines from immune cells in the lamina propria that are activated during inflammation. The effect of sodium butyrate on IL-1β stimulation of IL-8 secretion by Caco-2 epithelial cell line cells was, therefore, examined. In contrast to the effects of lipopolysaccharide, IL-1β stimulated IL-8 production by Caco-2 cells in the absence of sodium butyrate as has been previously shown by others(3). However, sodium butyrate markedly enhanced IL-8 secretion by Caco-2 cells when stimulated by IL-1β (p < 0.001; Fig. 2).

The dose-response curve of IL-8 secretion by Caco-2 cells after stimulation with varying concentrations of IL-1β showed a peak IL-8 production at a concentration of 1 ng/mL IL-1β (Fig. 3). This concentration of IL-1β also induced a maximal response in cells incubated with 5 mM sodium butyrate. However, butyrate increased the response of IL-8 secretion to IL-1β at this and every other concentration of IL-1β tested (Fig. 2). Sodium bicarbonate had no effect on IL-8 production by Caco-2 cells stimulated with IL-1β, again demonstrating that butyrate's effects were not due to its being a weak base. In addition, in sequential media exchange experiments, sodium butyrate enhanced the responsiveness of cells to IL-1β stimulation, even though it was not present concurrently with IL-1β in the medium (data not shown). Butyrate did not, therefore, interact with IL-1β in the media, but acted instead by upregulating IL-8 synthesis within the cell.

Figure 3
figure 3

Sodium butyrate enhances the secretion of IL-8 by Caco-2 cells. IL-8 secretion increased with increasing doses of IL-1β(0-100 ng/mL) for 24 h with a peak IL-8 secretion at 1 ng/mL IL-1β. At every concentration of IL-1β, 5 mM sodium butyrate enhanced the secretion of IL-8 by Caco-2 cells. The bars represent standard deviations of three different wells for each point. The results are representative of two experiments.

Full size image

Similar effects on the secretion of another chemokine, such as GRO-α, by Caco-2 epithelial cells were observed when cells were stimulated with IL-1β and sodium butyrate. IL-1β stimulated GRO-α secretion by Caco-2 cells in the absence of sodium butyrate. This secretion was further enhanced when cells were preincubated with sodium butyrate and then stimulated with IL-1β (Table 1).

Table 1 Sodium butyrate enhances the secretion of both IL-8 and GRO-α by Caco-2 cells
Full size table

Butyrate enhances Caco-2 cell accumulation of IL-8 mRNA. Treatment with lipopolysaccharide alone did not induce IL-8 mRNA except when cells had been treated with sodium butyrate (Fig. 4). IL-1β stimulation induced IL-8 mRNA in the absence of sodium butyrate. However, the amount of IL-8 mRNA was increased by incubation with sodium butyrate (Fig. 4). This effect was dose-dependent (data not shown). In unstimulated cells, IL-8 mRNA was discernible in cells incubated with 20 mM sodium butyrate (Fig. 4), but could not be detected by Northern blot analysis at concentrations below this.

Figure 4
figure 4

Sodium butyrate (20 mM) enhanced IL-8 mRNA accumulation in Caco-2 cells in response to stimulation with lipopolysaccharide(LPS) (100 μg/mL) or IL-1β (1 ng/mL). Caco-2 cell RNA (30μg) was electrophoresed in each lane and probed with an IL-8 cDNA. The upper panel shows hybridization of IL-8 cDNA to RNA. The membrane was reprobed with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA (lower panel) showing equivalent loading of total RNA. Sodium butyrate enhanced IL-8 accumulation in Caco-2 cells in response to stimulation with lipopolysaccharide or IL-1β. The results are representative of five experiments.

Full size image

IL-8 mRNA accumulation in Caco-2 cells was detected 2 h after stimulation with 1 ng/mL IL-1β (Fig. 5). IL-8 mRNA was greater when the Caco-2 cells had been incubated with 20 mM sodium butyrate for 24 h (Fig. 5). IL-8 mRNA was no longer detectable 6 h after IL-1β stimulation whether or not cells had been incubated in butyrate. The increase in IL-8 mRNA, induced by butyrate, was associated with greater secretion of IL-8 in Caco-2 cells (p < 0.05). Unlike mRNA, IL-8 secretion continued to rise even at 6 h after incubation with IL-1β whether or not cells received butyrate.

Figure 5
figure 5

The time course of IL-8 mRNA accumulation and IL-8 secretion by Caco-2 cells after stimulation with IL-1β. The secretion of IL-8 (upper panel) and accumulation of IL-8 mRNA (lower panel) were examined in the same cells. Sodium butyrate (20 mM) was given 24 h (-24 h) before stimulation with IL-1β (1 ng/mL) at time 0. Higher IL-8 mRNA production resulted in greater secretion of IL-8 protein in Caco-2 cells. IL-8 protein secretion continued after IL-8 mRNA production in Caco-2 cells. Each lane contained equal amounts of RNA.

Full size image

Butyrate induces nonmalignant human intestinal epithelial cells(H-4 Cells) to secrete IL-8 in response to IL-1 β stimulation. Caco-2 cells are derived from intestinal cancer cells and thus, may have an aberrant gene expression which could possibly affect the regulation, synthesis, and secretion of IL-8. H-4 cells are nonmalignant human small intestinal epithelial cells cloned from normal human fetal small intestine(32). As in Caco-2 cells, butyrate enhanced the secretion of IL-8 by H-4 cells after stimulation with IL-1β(Fig. 6). Thus, the effects of butyrate on IL-8 secretion were not limited to epithelial cells that had undergone malignant transformation.

Figure 6
figure 6

Sodium butyrate enhances IL-8 secretion by the nonmalignant human small intestinal epithelial cell line (H-4 cells) in response to IL-1β stimulation. Sodium butyrate induced IL-8 secretion by H-4 cells in response to 1 ng/mL IL-1β stimulation in a dose-dependent manner (p < 0.05, IL-1β vs 5 mM butyrate + IL-1β). The bars represent standard deviations of three different wells for each point. Results were representative of five experiments.

Full size image

Butyrate induces bipolar IL-8 secretion by Caco-2 cells after stimulation with lipopolysaccharide . In vivo, the intestinal epithelium is a polarized cell monolayer which is constantly exposed to bacterial products, such as lipopolysaccharide, at its apical surface. To examine the effect of sodium butyrate on the polarity of IL-8 secretion by Caco-2 cells after lipopolysaccharide stimulation, Caco-2 cells were cultured in cell culture inserts suspended in media with or without butyrate in both compartments. Lipopolysaccharide (100 μg/mL) was then added into the apical compartment of the wells. Lipopolysaccharide alone had no effect on IL-8 secretion. In the presence of 2.5 mM sodium butyrate, IL-8 secretion was secreted from both apical and basal poles in response to lipopolysaccharide stimulation (Fig. 7). This secretion was further increased with 5 mM sodium butyrate (Fig. 7).

Figure 7
figure 7

The effect of sodium butyrate on the polarity of IL-8 secretion by Caco-2 cells stimulated by lipopolysaccharide (LPS). Sodium butyrate induced Caco-2 cells to secrete IL-8 from both its apical and basolateral poles in response to apical lipopolysaccharide (100 μg/mL) stimulation. The bars represent standard deviations of three different wells for each point. Results were representative of four experiments.

Full size image

DISCUSSION

The epithelium serves as a line of defense against an array of substances in the intestine, including resident and pathogenic microorganisms(1). There is now evidence that the intestinal epithelium is able to secrete a wide array of cytokines, including chemokines, which alter responses in the mucosal immune system(2–6). However, the evidence that the production of such molecules is regulated by events in the lumen of the intestine is sparse. It is known that invasive bacteria may induce the release of IL-8(4, 6, 24), but the possibility that epithelial cells respond to variations in the normal environment of the intestinal lumen is less well documented. Normal resident bacteria alter the fucosylated glycoconjugates on the epithelial cell surface(25). These alterations were due to a bacterial metabolite, as yet not identified. Our present data now demonstrate that a normal metabolic product of intestinal bacteria, sodium butyrate, significantly increases the secretion of IL-8 by intestinal epithelial cells. This effect of butyrate on IL-8 secretion is particularly marked when epithelial cells are also stimulated by the proinflammatory cytokine, IL-1β, or by the bacterial product, lipopolysaccharide. Thus, bacterial signals in the form of butyrate are translated into chemokine secretion by epithelial cells.

In the normal human large intestine, butyrate concentrations vary from 5 mM(15) to 24 mM(16); and in ileostomy effluents, the concentration of butyrate ranges between 0.1 and 16 mM(26). If butyrate is a normal constituent of the intestine, the question arises as to why in vivo intestinal epithelial cells are not continuously releasing IL-8. Such secretion would cause a continued state of inflammation in the normal intestine. The answer may lie in the observation that butyrate had little effect on IL-8 secretion by epithelial cells in the absence of a second signal, such as IL-1β or lipopolysaccharide. IL-1β is not secreted in appreciable quantities in the healthy intestine. Lipopolysaccharide concentrations reaching the apical membrane in health would also be low because of the mucous barrier. By contrast, in inflammatory conditions where butyrate concentrations are elevated, such as Crohn's disease(15) or necrotizing enterocolitis(27), there is appreciable IL-1β secretion, loss of the mucosal barrier, and large numbers of Gram-negative bacteria. Our data indicate that a small amount of IL-8 is, however, secreted by Caco-2 cells given high concentrations of butyrate without IL-1β or lipopolysaccharide. In healthy mice, high concentrations of butyrate instilled into the rectum do, in fact, cause inflammatory changes in the colon(28). The effects of butyrate in vivo are, therefore, consistent with its actions in vitro.

Caco-2 cells, although derived from colonic cancer cells, were chosen because they are a well established model for the human intestinal epithelium(18–20). We have used, in addition, a nonmalignant cell line (H-4 cells) derived from human fetal intestine(21) to confirm that our results could be reproduced in intestinal epithelial cells that have not undergone malignant transformation (Fig. 6).

The up-regulatory effects of sodium butyrate on chemokine secretion by intestinal epithelial cells indicate that the effects of butyrate are more complex than has been hitherto thought. Sodium butyrate is a normal metabolite of luminal bacteria. Butyrate not only functions as an energy source(13), but modulates the intestinal epithelial cell's response to inflammatory mediators, resulting in chemokine secretion. The cells examined in our models were both healthy and viable. The lack of response in cells without butyrate was not due to any nutritional deprivation, because Caco-2 cells express brush border enzymes(22) and proliferate normally in the absence of butyrate. Addition of serum to the cells in parallel studies did not alter the effects of butyrate. Furthermore, variations in glutamine concentration (another epithelial cell nutrient) did not alter IL-8 secretion patterns.

Future experiments will be needed to show whether butyrate's effects on IL-8 production are through diverse mechanisms (for example by increasing the expression of the receptors for lipopolysaccharide, including CD14 and the IL-1β receptor) or through a single mechanism (for example by binding a specific nuclear protein to the IL-8 promoter). A single mechanism would appear more likely because, at high concentrations, sodium butyrate alone induces small amounts of IL-8 secretion. One attractive hypothesis is that sodium butyrate, by its known action on c-fos induction in Caco-2 cells(29), results in increased heterodimerization of the c-fos and c-jun nuclear proteins, which could then bind to the AP-1 site of the IL-8 promoter(30) resulting, in turn, in the enhancement of IL-8 transcription. In stimulated cells, this nuclear factor may act in association with NFκB, induced by both lipopolysaccharide(31) and IL-1β(32), in increasing IL-8 synthesis(33).

Lipopolysaccharide induced IL-8 secretion from both basolateral and apical poles of Caco-2 cells (Fig. 7). Apical stimulation can, therefore, induce basolateral secretion of IL-8. This is important if intestinal epithelial cells are to act as a system signaling events in the lumen to the mucosal immune system. IL-8 was also secreted apically. It is more difficult to ascribe a function to IL-8 secreted apically than to that secreted basolaterally, but our data are consistent with other observations. First, IL-8 levels in the lumen are elevated during intestinal inflammation in vivo(34). Second, IL-8 secretion by polarized intestinal epithelial cell lines is greater in the direction of the proinflammatory stimulus(35).

In summary, we have shown that butyrate, a normal bacterial metabolite, has a marked effect on the ability of intestinal epithelial cells to secrete IL-8. Because intestinal epithelial cells may participate in the attraction of leukocytes into the intestinal mucosa, we believe that we have identified a mechanism by which bacteria potentially modulate inflammatory responses within the intestine.