The effect of chronic exposure to transforming growth factor-α (TGF-α) on bradykinin-stimulated acute prostanoid production and ion secretion in monolayers of HCA-7 colony 29 colonic epithelial cells has been studied. Monolayers synthesized prostaglandin E2(PGE2) at a basal rate of 2.10 ± 0.31 pg ⋅ monolayer−1 ⋅ min−1over 24 h. Bradykinin (10−8–10−5M) dose dependently increased acute PGE2 release by three orders of magnitude. This was associated with a rise in cAMP from 1.60 ± 0.14 to 2.90 ± 0.1 pmol/monolayer (P < 0.02) and a dose-dependent increase in short-circuit current (SCC). When monolayers were primed by a 24-h exposure to TGF-α, basal PGE2 release rose to 6.31 ± 0.38 pg ⋅ monolayer−1 ⋅ min−1(TGF-α concn 10 ng/ml; P = 0.001). However, the stimulation of acute prostaglandin release, intracellular cAMP, and increased SCC by bradykinin was significantly reduced by preincubation with TGF-α. Priming with PGE2(10−8–10−6M) over 24 h mimicked the effect of TGF-α on bradykinin-induced changes in cAMP and SCC. These data suggest that enhanced chronic release of prostaglandins in response to stimulation with TGF-α may downregulate acute responses to bradykinin. In vivo, TGF-α could have an important modulatory function in regulating secretion under inflammatory conditions.
- transforming growth factor-α
- electrogenic ion transport
electrogenic chloride secretion in intestinal epithelial cells occurs when Cl− are actively transported from the basolateral to the apical side of epithelial cells, and excessive Cl− secretion in vivo gives rise to secretory diarrhea (3). A number of inflammatory mediators, including bradykinin (BK), eicosanoids such as prostaglandin E2(PGE2), histamine, 5-hydroxytryptamine, and a range of cytokines are known to mediate this secretory response (26).
BK is a biologically active nine-amino acid peptide formed during inflammation from protein kininogen and is known to stimulate electrogenic Cl− secretion in a variety of epithelial cells, including those of the gastrointestinal tract (31), airways, and uterus (11). BK also acts directly on colonic epithelial cells to stimulate Cl− secretion in vitro (1). The effect of BK in directly stimulating electrogenic Cl− secretion has best been characterized for monolayers of the HCA-7 colony 29 colonic epithelial cell line (5, 16). In this cell line, the effect of BK in stimulating Cl− secretion is mediated through cAMP-dependent pathways, pathways thought to be due to eicosanoid production (13), although other mediators such as intracellular Ca2+ also play a part.
Although much attention has focused on agents that stimulate electrogenic Cl− secretion, relatively little is known of endogenous regulatory mechanisms that will attenuate the secretory response. Epidermal growth factor (EGF) has recently been recognized to have an important physiological role in maintaining intestinal homeostasis (28) as well as stimulating proliferation and growth. Growth factor receptors are found throughout the human gastrointestinal tract and are primarily located in the basolateral domain (4, 27, 34).
Transforming growth factor-α (TGF-α) is a homolog of EGF and may be the most important physiological ligand to the EGF receptor in the intestinal mucosa (2, 12, 17). TGF-α has recently been shown to enhance vectorial prostaglandin release from HCA-7 colony 29 cells when applied to the basolateral compartment in which the EGF receptor resides (4). Cyclooxygenase (COX) is the rate-limiting enzyme for the conversion of arachidonic acid (AA) to prostanoids. Two isoforms of COX are recognized: COX-1, which is constitutively expressed, and COX-2, which is induced in pathological conditions by a range of chemicals including cytokines and inflammatory mediators (22). TGF-α induces COX-2 expression in the HCA-7 colony 29 cell line, resulting in increased prostaglandin production (4). However, the effect of TGF-α on BK-mediated acute colonic epithelial ion transport and prostaglandin synthesis is not known. Our original hypothesis was that the induction of COX-2 expression in HCA-7 cells by TGF-α (4) could lead to both enhanced acute PGE2 release and acute secretory response to BK. In this study, we therefore investigated the effect of TGF-α on BK-induced ion transport and prostaglandin release in monolayers of HCA-7 colony 29 cells.
Although TGF-α increased PGE2release over 24 h, it did not enhance acute prostaglandin release or the acute secretory response to BK. In fact, both prostaglandin release and the acute Cl− secretory response were significantly attenuated and BK-stimulated cAMP production was reduced; these changes could be mimicked by exogenously applied PGE2.
These findings suggest that TGF-α may have an important role in limiting the acute secretory response to BK during inflammation via a prostanoid-dependent mechanism(s).
HCA-7 colony 29 (16) cells were grown in DMEM supplemented with 10% FCS, glutamine (0.29 mg/ml), ampicillin (8 μg/ml), penicillin (40 μg/ml), streptomycin (368 μg/ml), and nonessential amino acids in an atmosphere of 5% CO2 at 37°C. Cells were seeded on Snapwell and Transwell filters (polycarbonate membrane; pore size 0.45 μm; surface area 1 cm2; Costar) and formed confluent monolayers within 8–10 days, as assessed with an epithelial volt-ohmmeter (World Precision Instruments).
COX-1 and COX-2 Western blotting.
HCA-7 cells were grown to confluence, lysed (10 mM PBS, 0.1% Triton X-100, 0.1% EDTA, 1 mM phenylmethylsulfonyl fluoride, 0.03% leupeptin; Sigma), and centrifuged for 10 min. Cell lysates were separated by SDS-PAGE with COX-1 and COX-2 standards (Cayman Chemical, Ann Arbor, MI), transferred to nitrocellulose, and blocked with 5% nonfat dried milk in Tris-buffered saline (TBS). Membranes were incubated overnight in 0.1% Tween 20 in TBS with a 1:500 dilution of either COX-1 or COX-2 rabbit polyclonal antibodies (both from Cayman Chemical). The blots were washed, and biotinylated goat anti-rabbit IgG was added. Labeled bands were detected by an avidin-biotin complex (ABC) technique (Vectastain Elite ABC kit; Vector Laboratories, Burlingame, CA).
Assay of basal and BK-induced prostaglandin synthesis.
To study prostaglandin production, cells were grown to confluence on Transwell filters. After a change of the medium, cells were exposed to basolateral TGF-α or control medium for 24 h. The supernatant was collected from the basolateral compartment to measure PGE2 levels. Monolayers were washed twice with fresh medium, then stimulated with basolateral BK (10−6 M), and after 1 min supernatant was collected for the PGE2 assay. Dose-response curves showing the effect of different concentrations of BK and the attenuation of the response after preincubation with the COX inhibitor piroxicam (10−9–10−5M) were constructed. The PGE2 enzyme immunoassay (Amersham) is based on the competition between unlabeled PGE and a fixed quantity of peroxidase-labeled PGE for a limited amount of the PGE-specific antibody.
Briefly, supernatant of HCA-7 cells or PGE2 was incubated with specific anti-PGE2 reagent in a prepacked 96-well plate containing a goat anti-mouse solid phase. Peroxidase-labeled PGE2 was added to each well, and the wells were incubated for 1 h. The peroxidase ligand that was bound to the antibody was immobilized on a precoated microtiter well, and the amount of unlabeled PGE in a sample was determined by interpolation from a standard curve. After the well was washed three times with washing buffer, tetramethylbenzidine-hydrogen peroxide was added and the well was incubated for 30 min. The addition of acid solution stopped the reaction, and the optical density at 450 nm was read on a microtiter plate reader. The detection range at room temperature was 50–6,400 pg/ml. Cross-reactivity with PGE1 was 25% and with prostaglandin F2α(PGF2α), 6-keto-PGF1α, and AA was <0.1%.
Assay of BK-induced cellular cAMP production.
In subsets of experiments BK-induced cAMP production was measured by a direct enzyme immunoassay (Amersham). Briefly, confluent monolayers of HCA-7 cells grown on permeable supports and pretreated for 24 h according to experimental conditions were washed twice with fresh medium. After acute stimulation with BK (10−6 M), membrane inserts were transferred to new wells and the medium was replaced with lysis reagent at a concentration of 0.25% dodecyltrimethylammonium bromide. Cell lysate was collected, and cAMP levels were measured according to the manufacturer’s instructions. The detection range at room temperature was 12.5–3,200 fmol/well.
For electrophysiological studies, cells were treated with either control medium or TGF-α (10 and 100 ng/ml, respectively) for up to 24 h. Filters (area 1 cm2) were placed into an Ussing chamber (World Precision Instruments) bathed in oxygenated (95% O2-5% CO2) Krebs-Henseleit solution (in mM: 117 NaCl, 4.7 KCl, 2.5 CaCl2, 1.0 MgSO4, 24.8 NaHCO3, 1.2 KH2PO4, and 11.1 glucose) and maintained at 37°C. In some experiments, the Cl−-containing salts of the Krebs-Henseleit solution were replaced with (in mM) 117 sodium gluconate, 4.7 potassium gluconate, and 2.5 calcium sulfate. The epithelium was voltage-clamped to 0 mV by continuous application of a short-circuit current (SCC) with a DVC-1000 dual-voltage clamp (World Precision Instruments). Periodic constant-amplitude voltage pulses were used to assess transepithelial resistance. Basal SCC (μA/cm2) and resistance (Ω ⋅ cm2) were measured after the monolayers were allowed to equilibrate for 15 min. The changes in SCC (ΔSCCs) in response to BK, carbachol, and forskolin administered to the basolateral side of the monolayer were recorded. SCC was digitally recorded and analyzed with the Acqknowledge III (BIOPAC Systems) data acquisition system. ΔSCCs were expressed in units of microamperes per square centimeter. In some experiments ΔSCCs over a given time period were integrated and converted into nanoequivalents by using the Faraday relationship (26.8 μA/cm2 = 1 μeq ⋅ cm−2 ⋅ h−1).
BK, carbachol, forskolin, piroxicam, DMEM, and FCS were purchased from Sigma; TGF-α was purchased from R&D Systems.
Data are expressed as means ± SE. The unpaired, two-tailed Studentt-test (adjusted for multiple testing by Bonferroni’s correction) or a one-way ANOVA was used to determine the significance of differences between means.P < 0.05 was accepted as indicating statistical significance. The dose-response curves for BK and EC50 values were obtained by applying data to the curve-fitting program PRISM (Graphpad, San Diego, CA).
Characteristics of HCA-7 cells.
HCA-7 cells expressed COX-1 and COX-2 proteins as shown by Western blotting (Fig.1 A,inset). They synthesized PGE2 at a basal rate of 3.03 ± 0.45 ng ⋅ monolayer−1 ⋅ 24 h−1, equivalent to an average rate of 2.10 ± 0.31 pg ⋅ monolayer−1 ⋅ min−1. HCA-7 epithelial cells exhibited a basal ΔSCC of 1.29 ± 0.21 μA/cm2(n = 20) and a resistance of 138.1 ± 9.9 Ω ⋅ cm2(n = 20).
Effect of BK.
BK applied to the basolateral compartments of monolayers of HCA-7 cells caused a substantial increase in acute PGE2 release (measured 1 min after addition of the agonist) in a dose-dependent fashion of between 2.70 ± 0.45 (BK 10−8 M) and 6.30 ± 0.84 ng/monolayer (BK 10−5) (n = 4–5; Fig.1 A). This was accompanied by a rise in intracellular cAMP from 1.60 ± 0.14 to 2.90 ± 0.1 pmol/monolayer (n = 4;P < 0.02; Fig. 3). BK stimulated ΔSCC in a dose-dependent fashion with an EC50 = 2.43 × 10−7 M (Fig.1 B).
BK (10−6 M)-stimulated prostaglandin release was dose dependently reduced by pretreating cells with piroxicam, a nonspecific COX inhibitor (Fig.2 A). At a concentration that entirely blocked PGE2 release, piroxicam (10−5 M) attenuated the BK (10−6 M)-induced ΔSCC by 69 ± 6% (n = 4,P = 0.01) (Fig.2 B).
Effect of TGF-α preincubation on basal PGE2 release and acute BK-stimulated PGE2 release.
In previous experiments, the basolateral application of TGF-α increased COX expression and stimulated PGE2 production over 24 h to levels from 0.01 to 100 ng/ml, with a submaximal effect at 10 ng/ml (4). In our experiments, TGF-α (10 ng/ml) administered to the basolateral compartments of monolayers of HCA-7 cells similarly increased the basolateral release of PGE2 over 24 h from 3.03 ± 0.45 (control) to 9.09 ± 0.54 ng/monolayer (rate 6.31 ± 0.38 pg ⋅ monolayer−1 ⋅ min−1;n = 4;P = 0.001).
In contrast to its effect on chronic PGE2 release over 24 h, pretreatment of monolayers with TGF-α (1, 10, and 100 ng/ml) for 24 h reduced acute BK-stimulated PGE2release over 1 min (P < 0.02; Fig.4 A). This was accompanied by a reduced production of intracellular cAMP (P < 0.02; Fig.3).
Effect of TGF-α preincubation on subsequent BK-stimulated ion transport.
Preincubation with TGF-α at 10 ng/ml for 24 h did not affect basal SCC (1.67 ± 0.58 μA/cm2,n = 20), although it was associated with a significant increase in resistance from 138.1 ± 9.9 (control) to 169 ± 16.2 Ω ⋅ cm2(n = 20;P < 0.03). As with acute PGE2 release after stimulation with BK, there was also an unexpected reduction in the ΔSCC response to subsequent BK challenge after a preconditioning with TGF-α for 24 h (Fig.4 B). Compared with that for control cells, the maximal ΔSCC response to BK was reduced by 57 ± 6% for cells conditioned by TGF-α at 10 ng/ml (P < 0.01) and by 60 ± 5.5% for cells conditioned by TGF-α at 100 ng/ml (P < 0.01) (Fig.5).
Effect of exogenous PGE2 on responses to BK.
To examine if exogenously administered prostaglandin could mimic the effect of TGF-α, monolayers of HCA-7 cells were preincubated with PGE2 at concentrations (10−8–10−6M) in the range achieved after exposure to TGF-α in the experiments described above. Under these conditions, the BK-induced rise in cAMP was significantly reduced (P < 0.05; Fig. 3). This was accompanied by a significant dose-dependent reduction in ΔSCC after exposure to PGE2for 24 h compared with that for controls. This reduction was similar to the TGF-α-induced downregulation of BK-stimulated ion transport (Fig.5).
Time course of the effect of TGF-α preincubation on BK-stimulated ion transport.
To determine the duration of exposure of HCA-7 cells to TGF-α required to downregulate the secretory response to BK, TGF-α (10 ng/ml) was applied for different time periods (1, 6, 12, and 24 h) in paired experiments before stimulation with BK. The area under the response curve for the period after the addition of BK was measured from 0 to 1 min (time of initial peak response) and is shown as the percent modulation of the response from control monolayers. Exposure to TGF-α at 10 ng/ml caused a reduction of the ΔSCC response to BK compared with that for the control after 1 h of incubation, and this effect was maximal after 12 h of incubation (Fig 6).
Effect of AA on TGF-α-induced downregulation of ion transport.
To investigate if TGF-α acted by depleting substrate for the acute response to BK, TGF-α-treated monolayers were exposed to AA in Ussing chambers for 30 min before stimulation with BK. AA (10−5–10−3M) induced a slow increase in SCC (ΔSCC 2 μA/cm2), which returned to baseline after 5 min. The responses to BK with and without AA in control monolayers showed no difference (19.2 ± 2.36 vs. 19.8 ± 3.25 μA/cm2), and there was no reversal of the downregulatory effect of TGF-α (Fig.7), indicating that the effect of TGF-α was not due to the depletion of endogenous AA.
Contribution of Cl− secretion to ΔSCC.
To confirm previous reports that Cl− secretion was the main contributor to BK- and carbachol-stimulated ΔSCC in this cell line, ion substitution studies were performed. In Cl−-free media, ΔSCCs in response to BK and carbachol stimulation were reduced to 51.3 ± 7 (n = 6;P < 0.02) and 55.4 ± 1.6% (n = 6;P < 0.02) of the response seen in normal Krebs-Henseleit solution, confirming that the secretory response to BK and carbachol was at least in part due to electrogenic Cl− secretion (23).
Specificity of the secretory response.
To investigate whether TGF-α additionally affected the acute secretory response of HCA-7 cells to other secretagogues and to test the secretory capacity of HCA-7 cells, carbachol (10−4 M; a cholinomimetic agent that increases intracellular Ca2+) and forskolin (10−4 M; a direct adenylyl cyclase activator) were added to controls and TGF-α-pretreated monolayers. There was no significant reduction in the carbachol- or forskolin-stimulated response in monolayers treated with TGF-α at 10 ng/ml for 24 h (Fig. 8). Likewise, PGE2 in a dose range from 10−8 to 10−6 M had no significant effect on the carbachol-induced ΔSCC (Fig. 8).
In this study we have investigated the effect of TGF-α on acute secretory responses to BK in colonic epithelial cells. In the intestine, lamina propria cells, including myofibroblasts and mononuclear cells (11, 18, 33, 37), and epithelial cells (5-7, 13, 20) have been previously implicated in the inflammatory response to BK. BK activates cytosolic phospholipase A2 and increases prostaglandin production thought to be primarily mediated by COX-1, although COX-2 may participate in more chronic responses (22). We chose the HCA-7 colony 29 cell line because previous studies had shown that these cells express both COX-1 and COX-2 (4) and demonstrate a good secretory response to BK (5-7, 13, 20). In addition, TGF-α induces COX-2 expression and basolateral release of prostaglandins in this cell line (4).
Studies of HCA-7 monolayers have previously demonstrated that the secretory response to BK is due to both increased cAMP (as a result of increased endogenous prostaglandin production) and intracellular Ca2+ (20). Although an additive or even synergistic effect of BK-stimulated ΔSCC might have been anticipated, TGF-α preconditioning of HCA-7 cells paradoxically downregulated BK-induced acute prostaglandin production, and there was an accompanying reduction in cAMP production and Cl− secretion. This effect would appear to be mediated by prostaglandins because long-term (24 h) exposure of monolayers to exogenous PGE2 at concentrations similar to that produced by TGF-α also significantly reduced BK-stimulated cAMP production and Cl− secretion in Ussing chambers. The mechanism by which TGF-α reduced the secretory responses to BK was further investigated by adding AA before stimulation with BK. In these experiments, BK-stimulated Cl− secretion could not be restored to control levels in TGF-α-conditioned monolayers, indicating that depletion of substrate (AA) for the COX enzyme is not the mechanism by which TGF-α reduces the acute response to BK. Therefore, chronic exposure of HCA-7 cells to prostaglandins may make them partially resistant to further stimulation by BK, and this may form part of a negative-feedback loop to limit the inflammatory and secretory responses. This is not without precedent, as it has recently been reported that prostaglandins exert negative-feedback effects on the expression of COX-2 in macrophages (25) and on cAMP production in airway smooth muscle cells (24) and may be part of an endogenous regulatory mechanism. The inhibition of BK-induced increases in cAMP by EGF in a human epidermoid carcinoma cell line (A431) has also recently been described (19).
This downregulation of secretory responses, along with the enhancement of barrier function implied by the increased resistance we have reported in the present experiments, may contribute to a homeostatic role for TGF-α in modulating secretory and permeability changes that can be induced by inflammation. Some authors have previously reported reduced Cl− secretion in an inflamed colonic mucosa (14). Because TGF-α primarily upregulated COX-2 expression in a colony of HCA-7 cells similar to that used in our experiments (4), it is tempting to speculate that exaggerated PGE2 production as a result of COX-2 induction is part of the negative-feedback system exerting a braking effect on the inflammatory response.
Our experiments do not establish the mechanism by which TGF-α leads to downregulation of the responses to BK, which is the subject of ongoing research. Possibilities include tachyphylaxis as a result of prostaglandin receptor downregulation and alterations in localization, making COX enzymes less accessible to substrate released by BK. Illuminating these mechanisms will require an understanding of the extent to which COX-1 and COX-2 contribute on the one hand to TGF-stimulated prostaglandin synthesis and on the other to the prostaglandin synthesis and secretory responses stimulated by BK.
It is possible that TGF-α could have effects on HCA-7 cells apart from stimulating prostaglandin production to attenuate the secretory response to BK. The implication that an indirect metabolic mechanism is involved is further supported by time course experiments showing a lag of 1 h to the onset, and 12 h to the maximal effect, of TGF-α attenuation of BK-stimulated ion transport. In T84 cells, pretreatment with EGF, which binds to the same ligand as TGF-α, reduced the secretory response to carbachol partly by increasing intracellulard-myo-inositol 3,4,5,6-tetrakisphosphate, which negatively affected Cl− secretion (36). Whether TGF-α reduces BK-stimulated anion secretion by a similar mechanism in HCA-7 cells is not known. Interestingly, in our experiments with HCA-7 cells, TGF-α or prostaglandin preconditioning did not significantly affect the response to basolateral carbachol compared with that for control monolayers (Fig. 8), a finding that may reflect inherent differences in secretory mechanisms between the T84 and HCA-7 cell lines. In addition, we have shown that both COX-1 and COX-2 are constitutively expressed in HCA-7 cells, whereas T84 cells show only low levels of mRNA for COX-1 and none for COX-2 (Beltinger, unpublished observations) and respond only weakly to BK (1).
Although growth factors are generally recognized for their role in proliferation and restitution of epithelia and in maintaining mucosal integrity, they have only recently been recognized to have a role in regulating intestinal ion transport and inflammation. Previous studies have shown that TGF-α and EGF regulate other secretory functions in the gastric mucosa (32) and in pancreatic acinar cells (35). EGF has also been demonstrated to enhance glucose and sodium absorption in the rat ileum (10, 15, 23). As already mentioned, EGF inhibits carbachol-stimulated electrogenic Cl− secretion in T84 cells (36).
Moreover, TGF-α is elevated during gut inflammation (8, 9, 29), and pretreatment with EGF in a rat model of colitis reduced inflammation (30). Increases in prostaglandin synthesis associated with the induction of COX are generally regarded as having proinflammatory effects; our data suggest that they may also play a homeostatic counter role, which may be compromised during treatment with COX inhibitors. Regulatory mechanisms in the presence of an activated COX system would be useful to prevent high-level prostaglandin production (due to inflammatory mediators such as BK), which may in turn lead to the stimulation of intestinal secretion and excessive fluid and electrolyte losses from the body. Finally, a complete understanding of the effect of TGF-α on colonic mucosal secretory responses will need to take into account the ability of immunocytes to stimulate, and myofibroblasts to downregulate, Cl− secretion.
Hitherto, research on growth factors in the gut has emphasized their role in maintaining normal epithelial integrity and stimulating repair processes, as well as their possible involvement in malignant change. Our data are consistent with the notion that growth factors may also have an important homeostatic role in limiting the secretory responses to BK.
HCA-7 colony 29 cells were a kind gift from Dr. Susan Kirkland.
Address for reprint requests and other correspondence: W. A. Stack, Division of Gastroenterology, Univ. Hospital, Queen’s Medical Centre, Nottingham NG7 2UH, UK (E-mail:).
J. Beltinger was supported by a grant from the Swiss National Science Foundation and the Ciba-Geigy-Jubilaeums-Stiftung, Switzerland.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.
- Copyright © 1999 the American Physiological Society