Somatostatin peptides prevent increased human colonic epithelial permeability induced by hypoxia

Abstract

Mesenteric ischemia increases gut permeability and bacterial translocation. In human colon, chemical hypoxia induced by 2,4-dinitrophenol (DNP) activates basolateral intermediate conductance K⁺ (IK) channels (designated KCa3.1 or KCNN4) and increases paracellular shunt conductance/permeability (G_S), but whether this leads to increased macromolecule permeability is unclear. Somatostatin (SOM) inhibits IK channels and prevents hypoxia-induced increases in G_S. Thus, we examined whether octreotide (OCT), a synthetic SOM analog, prevents hypoxia-induced increases G_S in human colon and hypoxia-induced increases in total epithelial conductance (G_T) and permeability to FITC-dextran 4000 (FITC) in rat colon. The effects of serosal SOM and OCT on increases in G_S induced by 100 µM DNP were compared in isolated human colon. The effects of OCT on DNP-induced increases in G_T and transepithelial FITC movement were evaluated in isolated rat distal colon. G_S in DNP-treated human colon was 52% greater than in controls (P = 0.003). G_S was similar when 2 µM SOM was added after or before DNP treatment, in both cases being less (P < 0.05) than with DNP alone. OCT (0.2 µM) was equally effective preventing hypoxia-induced increases in G_S, whether added after or before DNP treatment. In rat distal colon, DNP significantly increased G_T by 18% (P = 0.016) and mucosa-to-serosa FITC movement by 43% (P = 0.01), and 0.2 µM OCT pretreatment completely prevented these changes. We conclude that OCT prevents hypoxia-induced increases in paracellular/macromolecule permeability and speculate that it may limit ischemia-induced gut hyperpermeability during abdominal surgery, thereby reducing bacterial/bacterial toxin translocation and sepsis.

NEW & NOTEWORTHY Somatostatin (SOM, 2 µM) and octreotide (OCT, 0.2 µM, a long-acting synthetic analog of SOM) were equally effective in preventing chemical hypoxia-induced increases in paracellular shunt permeability/conductance in isolated human colon. In rat distal colon, chemical hypoxia significantly increased total epithelial conductance and transepithelial movement of FITC-dextran 4000, changes completely prevented by 0.2 µM OCT. OCT may prevent or limit gut ischemia during abdominal surgery, thereby decreasing the risk of bacterial/bacterial toxin translocation and sepsis.

INTRODUCTION

Bacterial translocation (BT) in the gut occurs when bacteria cross the lamina propria to local mesenteric lymph nodes and from there to other normally sterile extraintestinal sites (1). Mesenteric lymph node cultures reported BT in 59% of patients undergoing laparotomy for intestinal obstruction, none of whom had necrotic bowel (2). Loss of gut barrier function appears to contribute to systemic infection and multiorgan failure in critically ill or injured patients (3), and DNA analysis revealed BT during general or cardiac surgery (4). However, BT triggers cytokine release, which may itself predispose to septic morbidity and multiple organ failure (1, 5). Indeed, bacteria per se are not required to cross the intestinal epithelial barrier: the translocation of inflammatory mediators produced adjacent to the mucosal surface, or other toxic products from the gut, may breach the epithelial barrier and lead to systemic effects (3). Thus, passage of endotoxins and/or antigens as well as viable bacteria from the intestinal lumen to the circulation may cause systemic inflammation and distant organ injury (6), with severe postoperative sepsis having a mortality rate of up to 50% (7, 8). Postoperative sepsis occurred in 4.1% of general surgical cases leading to higher healthcare costs, particularly in patients undergoing liver or pancreatic resection (9), and similar outcomes were reported in another large study of abdominal surgical procedures (10).

Experimental models of mesenteric ischemia, as well as surgical handling of the gut, result in increased levels of inflammatory cytokines (11–13), all capable of disrupting tight junction (TJ) proteins and gut barrier function (14, 15). An early effect of oxidative stress is dissipation of transmembrane cation gradients and release of K⁺ ions, and metabolic stress enhances K⁺ release via K⁺ channels in the plasma membrane of liver, vascular, and other cell types (16). In biliary cells, metabolic stress markedly increased membrane K⁺ permeability, mediated in part by the activation of small conductance (SK) K⁺ channels (16). In the human colon, chemical hypoxia induced by 2,4-dintrophenol (DNP) stimulated the activity of basolateral intermediate conductance K⁺ (IK) channels (designated KCa3.1 or KCNN4) while simultaneously increasing epithelial paracellular shunt conductance (paracellular permeability), and both changes were prevented by the specific IK channels inhibitors clotrimazole and triarylmethane-34 (TRAM-34) (17). Furthermore, the addition of somatostatin (SOM; a naturally occurring gut peptide) to isolated human colonic crypts rapidly inhibited basolateral IK channels in a G-protein-dependent manner (18). In the present study, we hypothesized that SOM and its long-acting synthetic analog octreotide (OCT) might prevent hypoxia-induced increases in paracellular conductance in the human colon via the inhibition of basolateral IK channels. Our aim was to determine whether OCT, already in clinical use, prevents hypoxia-induced increases in both colonic paracellular conductance and mucosa-to-serosa macromolecule transfer. That being the case, it would raise the possibility of using OCT to limit or prevent BT during major abdominal surgery and decrease the likelihood of postoperative sepsis. The human colon was not readily available to evaluate the effect of hypoxia on macromolecule transfer, but basolateral IK channels are present in rat colon (19). We therefore investigated whether hypoxia disrupts rat colonic epithelial barrier function in a similar manner, as judged by increases in the overall electrical conductance of the epithelium and its permeability to FITC-dextran 4000 (FITC), and if so, whether these changes could be prevented by pretreatment with OCT.

MATERIALS AND METHODS

Human Colon

With written informed consent and approval from the Leeds Teaching Hospitals Ethics Committee, tissue was obtained from patients undergoing resection of colonic cancer. Normal looking (nondiseased) 2–3 cm segments of ascending or sigmoid colon were chosen at least 8 cm away from the tumor. The colonic segments were placed immediately into oxygenated Ringer solution containing (in mM/L): 136 Na⁺, 7 K⁺, 1.2 Mg²⁺, 2.0 Ca²⁺, 121 Cl⁻, 1.2 SO₄²⁻, 25 HCO₃⁻, 1.2 H₂PO₄⁻, and 11 glucose at 37°C, and opened as a flat sheet within 5 min while bathed continuously in oxygenated Ringer solution, and the serosal muscle and fat were dissected away from the mucosal epithelium, as previously described in Sandle et al. (20). In some experiments with human colon, G_S was estimated in controls tissues and in a separate set of unpaired tissues treated with 100 µM 2,4-dinitrophenol (DNP; MiliporeSigma, St. Louis, MO). In other experiments, each tissue was divided into two pieces, one acting as a control and the other treated with DNP.

Electrophysiology of Human Colon

Tissues were mounted between two Perspex slides with an exposed area of 0.5 cm² and secured between identical open-top Perspex Ussing half-chambers fitted with rubber rings to ensure a tight seal. Half-chambers accommodated equal volumes of bathing solution on both sides of the mucosa to which were added DNP and nystatin (500 U/mL; Sigma Chemical, St. Louis, MO; see below). Tissues were bathed on the mucosal (apical) side with a high-K⁺, Cl⁻-free solution containing (in mM/L): 114 K⁺ gluconate, 25 KHCO₃⁻, 1.2 KH₂PO₄⁻, 10 Ca²⁺ methane sulfonate, 1.2 MgSO₄, and 11 glucose, and on the serosal (basolateral) side with a high-Na⁺, Cl⁻- free solution containing (in mM/L): 113 Na⁺ gluconate, 25 NaHCO₃⁻, 5.8 K⁺ gluconate, 1.2 KH₂PO₄⁻, 10 Ca²⁺ methane sulfonate, 1.2 MgSO₄, and 11 glucose at 37°C. Under these conditions, following apical membrane permeabilization with nystatin, the K⁺ concentration of the mucosal solution mimicked intracellular K⁺ concentration, whereas the absence of mucosal Cl⁻ prevented a rise in intracellular Cl⁻ concentration and cell swelling. Half-chambers were gassed continuously with 95%O₂-5%CO₂, or with 95%N₂-5%CO₂ after the addition of DNP. A pair of 1 M KCl-agar electrodes were positioned at opposite sides of both half-chambers for the application of a 2-s rectangular current pulse (I = 50 μA). A second pair of 1 M KCl-agar electrodes positioned on either side of the tissue allowed measurement of transepithelial voltage (V_T), and their other ends were linked via calomel electrodes to a single channel voltage clamp (Physiologic Instruments, Model VCC600) connected to a computer running Axoscope software, which recorded real-time data. The current pulse was passed across the tissue every 10 s, and the resulting changes in V_T (ΔV_T) were recorded. In all cases, measurements of V_T, total tissue conductance (G_T = I/ΔV_T), and calculated short-circuit current (I_sc = ΔV_T·G_T) had reached steady-state values after 10 min. Figure 1 shows a representative trace from preliminary experiments in DNP-treated tissues done to establish the speed of response to nystatin and SOM during hypoxia. Nystatin, a polyene antibiotic, combines with membrane cholesterol to form water-filled pores freely permeable to monovalent ions (21–23). Stock solution (0.1 mL) of nystatin dissolved in dimethyl sulfoxide was added to the mucosal chamber (500 U/mL final concentration) (22), which rapidly increased apical membrane conductance (G_a) to an infinitely high value without changing the conductance of the basolateral membrane (G_b) or the paracellular shunt pathway (G_S) (21–23). Postnystatin, V_T increased (Fig. 1), reflecting the steep [K⁺] gradient across the basolateral membrane as the apical membrane was permeabilized. V_T, I_sc, and G_T were monitored at 10-s intervals until steady-state conditions were achieved. The amplitude of the deflections decreased, indicating an increase in G_T which reflected the increase in apical membrane conductance to infinity, and the effect of nystatin was complete after 8 min. Post-SOM, V_T decreased, and the amplitude of the deflections increased, reflecting SOM-induced inhibition of basolateral IK channels and a decrease in basolateral membrane conductance, and the effect of SOM was complete after 16 min The electrical changes in response to nystatin and then SOM were as expected and are consistent with tissue integrity being intact after stripping. Thus, in experiments with control (nonhypoxic) tissues, V_T and G_T were monitored for 8 min after the addition of nystatin (total experimental time 18 min).

Figure 1.

Time course of changes in transepithelial voltage (V_T) and total tissue conductance (G_T, indicated by deflections in V_T in response to 2-s rectangular current pulses of 50 μA) after the mucosal (apical) addition of nystatin and the subsequent serosal (basolateral) addition of somatostatin (SOM) to DNP-treated human colon. DNP, 2,4-dinitrophenol.

Cell hypoxia was achieved using DNP, which uncouples oxidative phosphorylation and rapidly depletes cellular ATP. Before the apical membrane was permeabilized with nystatin, 0.1 mL of a stock solution of DNP dissolved in dimethyl sulfoxide was added to both half-chambers (final concentration of 100 µM), which were then gassed with 95%N₂-5%CO₂. In tissues treated with DNP alone, DNP was present for 3 min after reaching the initial steady-state, following which nystatin was added with monitoring for 8 min (total experimental time 21 min). Nystatin was used in all tissues, and the relationship between G_T and I_sc was linear as described by the equation G_T = 1/E_b· I_sc + G_S, where E_b was the electromotive force across the basolateral membrane, and G_S was the shunt or paracellular conductance (21, 22). G_S was estimated from the y-axis intercept of the linear relationship between G_T and I_SC. The were no differences in the responses to nystatin, DNP, SOM, or OCT between tissues derived from the ascending and sigmoid colon.

Somatostatin and Octreotide

Two sets of experiments were performed to evaluate the effect of serosal 2 μM SOM (MilliporeSigma) on G_S. In one set, 2 μM SOM was applied 3 min after adding 100 µM DNP, followed 16 min later by permeabilization of the apical membrane with nystatin, with monitoring for a further 8 min (total experimental time 37 min). In the other set, tissues were preincubated with 2 μM SOM for 16 min before adding 100 µM DNP, followed by permeabilization of the apical membrane with nystatin (total experimental time 37 min). Additional experiments were done using DNP-treated nystatin-permeabilized tissues to establish the dose-response of OCT (MilliporeSigma). After adding 100 µM DNP and permeabilizing the apical membrane, a stock solution of OCT was applied to the serosal side of tissues to achieve a final concentration of 0.002 µM. When I_sc had stabilized, the serosal concentration of OCT was raised by 0.02 µM, then by 0.2 µM, and finally by 2 µM, allowing I_SC to stabilize between each increase in OCT concentration. Based on the results of those experiments (see  results), 0.2 µM OCT was used in subsequent experiments (total experimental time 37 min) to determine its effect on DNP-induced changes in G_S.

Electrophysiology and FITC Fluxes in Rat Colon

Nonfasting normal male Sprague-Dawley rats (126–150 g) were maintained on standard rat chow. Animals were given food and water ad libitum. The experimental protocols used in these studies were approved by the West Virginia University Institutional Animal Care and Use Committee. Two sets of experiments with rat distal colon stripped of serosal fat and muscle were performed under voltage-clamp conditions at 0 mV in EasyMount Ussing chambers (Physiologic Instruments, San Diego, CA), as previously described (24). In the first set, mounted tissues were stabilized by gassing with 95%O₂-5%CO₂ for 15–20 min, then 250 µg/mL FITC (FITC-dextran, MW 4,000, Sigma) was added to the mucosal chamber, and G_T was monitored every 20 s. After 15 min, 200 µL of serosal solution was removed to measure the basal mucosa-to-serosa FITC flux, following which 100 µM DNP was added to the mucosal and serosal chambers, and the gassing was switched to 95% N₂-5% CO₂. After 15 min, 200 µL of serosal solution was again removed to measure the mucosa-to-serosa FITC flux during the period of hypoxia. In the second set, the protocol differed only in that, after removing 200 µL of serosal solution to measure the basal mucosa-to-serosa FITC flux, OCT (Tocris Bioscience; 0.2 µM final concentration) was added to the serosal chamber followed by sampling for a 15-min FITC flux, after which 100 µM DNP was added to both chambers while gassing with 95%N₂-5%CO₂, followed 15 min later by serosal sampling to determine the FITC flux during hypoxia. FITC-dextran concentrations were determined by measuring fluorescence excitation at 485 nm and emission at 525 nm using a microplate reader (Synergy H4 Hybrid Reader, BioTek Instruments, Winooski, VT).

Statistics

Results from experiments done in human colon are shown as means ± SD and were analyzed using the Mann–Whitney U test for unpaired data and the Wilcoxon Signed Rank test for paired data. Results of the FITC flux experiments done in rat distal colon are shown as means ± SE and were analyzed using Student’s t test for paired data. P < 0.05 indicated a statistically significant difference between two sets of data.

RESULTS

Hypoxia Increases G_S in Human Colon

Results from experiments in unpaired tissues revealed a significant difference (P = 0.002) in paracellular shunt conductance (G_s) between control (3.40 ± 1.10 mS/cm²; n = 10) and DNP-treated tissues (5.53 ± 1.16 mS/cm²; n = 10) (Fig. 2). Representative data from unpaired control and DNP-treated tissues, showing increases in G_T and I_sc following the addition of nystatin, are shown in Fig. 3, A and B, respectively. A similar difference was observed in a smaller subset of paired tissues (n = 6) where one half of the tissue acted as control and the other was DNP-treated (3.51 ± 1.17 mS/cm² in controls and 5.95 ± 1.02 mS/cm² in DNP-treated, P = 0.0031).

Figure 2.

Summary of paracellular shunt conductance (G_S) data in control colon (n = 10), colon treated with 2,4-dinitrophenol (DNP) alone (n = 10), colon treated first with DNP then 2 μM somatostatin (SOM; n = 6), and colon pretreated with 2 μM SOM then DNP (n = 6). *P = 0.0038 indicates a significant difference compared with control colon. **P = 0.011 and ***P = 0.0002 indicate significant differences compared with colon treated with DNP alone. Error bars indicate standard deviation (SD).

Figure 3.

Representative results from control colon (A), 2,4-dinitrophenol (DNP)-treated colon (B), colon pretreated with 2 µM somatostatin (SOM) 10 min before adding DNP (C), and colon treated with 2 µM SOM after adding DNP (D). Best-fit plots show the relationship between G_T and I_sc following mucosal (apical) addition of nystatin. The intercepts of the plots with the y-axes indicate G_S. G_S, paracellular shunt conductance; G_T, total tissue conductance; I_sc, short-circuit current.

SOM Inhibits the Hypoxia-Induced Increase in G_s in Human Colon

Patch-clamp studies in isolated intact human colonic crypts previously showed that 2 μM SOM added to the basolateral membrane inhibited basolateral IK channel activity by 50% after ∼8 min and maximally after ∼20 min (18). In the present study, human colon was treated first with DNP, and 2 μM SOM added to the serosal (basolateral) side 5 min later. After a further 10 min, the apical membrane was permeabilized with nystatin. G_S in DNP-pretreated/SOM-treated tissues (3.66 ± 1.06 mS/cm²; n = 6) was significantly lower than in those treated with DNP alone (5.53 ± 1.16 mS/cm²; n = 10, P = 0.015) (Fig. 2), and Fig. 3D shows a representative plot of G_T versus I_sc in a DNP-pretreated/SOM-treated tissue. In separate experiments, the DNP-SOM sequence was reversed, and tissues pretreated with 2 μM SOM for 10 min before adding DNP, and the apical membrane was then permeabilized with nystatin 5 min later. As shown in Fig. 2 and the representative experiment (Fig. 3C), G_S in SOM-pretreated/DNP-treated tissues (2.85 ± 0.24 mS/cm²; n = 6) was significantly lower than that in tissues treated with DNP alone (5.53 ± 1.16 mS/cm²; n = 10, P = 0.0003), indicating that SOM pretreatment prevented the DNP-induced increase in G_S. Thus, although DNP induced a marked increase in G_S, this effect was completely inhibited by SOM irrespective of whether it was added after or before the onset of DNP-induced chemical hypoxia. Interestingly, G_S in SOM-pretreated/DNP-treated tissues (2.85 ± 0.24 mS/cm²; n = 6) tended to be lower than in controls (3.40 ± 1.10 mS/cm²; n = 10), although this difference was not significant (P = 0.238).

OCT Inhibits the Hypoxia-Induced Increase in G_s in Human Colon

In the present study, 2 μM SOM completely inhibited the DNP-induced increase in G_S. Since 2 μM OCT inhibits basolateral IK channel activity to the same extent as 2 μM SOM (18), we studied the ability of increasing concentrations of OCT (0.002 µM, 0.022 μM, 0.222 μM, and 2.222 μM using appropriate aliquots of a stock solution) to inhibit the DNP-induced increase in G_S. OCT (0.002 μM) was added to the serosal side of nystatin-permeabilized DNP-treated human colon, and after V_T and I_sc had reached steady-state values, serosal OCT concentration was increased successively to 0.022 μM, 0.222 μM, and 2.222 μM, allowing V_T and I_sc to reach steady-state values over 10–15 min between each increment. During these experiments, where the apical membrane was freely permeable to K⁺ ions and a steep K⁺ concentration gradient existed across the basolateral membrane, I_sc reflected K⁺ current across the basolateral membrane, where IK channels constituted the dominant K⁺ conductive pathway (25). Thus, cumulative decreases in I_sc (reflecting dose-dependent inhibition of IK channels by OCT) were plotted against OCT concentration (Fig. 4,) and individual results from 6 tissues are shown in the inset. The results indicate that 0.222 µM OCT and 2.222 μM OCT had identical effects on I_sc and presumably similar inhibitory effects on the DNP-induced increase in G_S.

Figure 4.

Cumulative decreases in I_sc at increasing concentrations of octreotide (OCT) in nystatin-permeabilized 2,4-dinitrophenol (DNP)-treated colon (n = 6). Results from individual tissues are shown in the inset. Data points are means ± SD.

Additional experiments were performed using nystatin-permeabilized DNP-treated human colon to determine whether 0.2 μM (∼0.222 μM) OCT and 2 μM SOM had the same inhibitory effect on the DNP-induced increase in G_S. As shown in Fig. 5, when 0.2 μM OCT was added after DNP, G_S (3.69 ± 0.71 mS/cm²; n = 6) was significantly lower than in tissues treated with DNP alone (5.53 ± 1.16 mS/cm²; n = 10, P = 0.008). Similarly, when DNP was added to tissues pretreated with 0.2 μM OCT, G_S (2.96 ± 0.49 mS/cm²; n = 6) was significantly lower than in tissues treated with DNP alone (5.53 ± 1.16 mS/cm²; n = 10, P = 0.0005). From the data presented in Fig. 2 and Fig. 5, it is clear that 2 μM SOM and 0.2 μM OCT had similar inhibitory effects on DNP-induced increases in G_S.

Figure 5.

Summary of paracellular shunt conductance (G_S) data in control colon (n = 10 as in Fig. 2), colon treated with 2,4-dinitrophenol (DNP) alone (n = 10 as in Fig. 2), colon treated first with DNP then 0.2 μM octreotide (OCT; n = 6), and colon pretreated with 0.2 μM OCT then DNP (n = 6). *P = 0.0038 indicates a significant difference compared with control colon. **P = 0.0075 and ***P = 0.005 indicate significant differences compared with colon treated with DNP alone. Error bars indicate SD.

Estimates of G_T, G_S, and G_C (Cellular Conductance) in Individual Tissues of Human Colon

Figure 2 and Fig. 5 summarize the mean (±SD) G_S estimates in the various experimental groups. Individual values of G_T, G_S, and G_C (calculated as G_{T −} G_S) are shown in Table 1. These values showed wide variability in the control group and the group treated with DNP alone, which may have reflected differences in tissue handling before mounting and different degrees of tension between tissues once mounted. Nevertheless, although there was no significant difference in G_T between controls (7.26 ± 2.06 mS/cm²) and tissues treated with DNP alone (9.37 ± 2.75 mS/cm²), G_S in the DNP group (5.53 ± 1.16 mS/cm²) was greater than in controls (3.40 ± 1.10 mS/cm², P = 0.0038). Interestingly, G_T and G_S were less variable in the four groups of tissues treated with either SOM or OCT. We attribute this to the ability of SOM and OCT to decrease G_S (and thus G_T) irrespective of the hypoxia induced by DNP. In tissues treated with SOM or OCT after the addition of DNP, G_C (1.34 ± 0.98 mS/cm² and 1.35 ± 0.88 mS/cm², respectively) was significantly lower (P = 0.03 and P = 0.02, respectively) than in tissues treated with DNP alone (3.86 ± 1.81 mS/cm²). These differences may reflect inhibition of basolateral IK channels by SOM and OCT, resulting in a decrease in basolateral membrane conductance. There was a similar trend when comparing G_C in tissues treated with SOM or OCT before the addition of DNP with tissues treated with DNP alone, although this was not significant (Table 1).

Table 1.

Values of total conductance (G_T), paracellular (shunt) conductance (G_S), and cellular conductance (G_C) in each tissue in control group, group treated with 2,4-dinitrophenol (DNP) alone, and DNP groups treated with somatostatin (SOM) or octreotide (OCT) either after or before the addition of DNP

Control			100 µM DNP			100 µM DNP + 2 µM SOM			2 µM SOM + 100 µM DNP			100 µM DNP + 0.2 µM OCT			0.2 µM OCT + 100 µM DNP
G T	G S	G C	G T	G S	G C	G T	G S	G C	G T	G S	G C	G T	G S	G C	G T	G S	G C
5.79	1.85	3.94	5.61	3.48	2.13	3.79	1.82	1.97	5.64	2.54	3.10	3.62	2.89	0.73	3.23	2.00	1.23
5.87	1.87	4.00	5.04	4.10	0.94	5.88	2.69	3.19	5.88	2.69	3.19	3.56	2.96	0.60	6.21	2.87	3.34
5.13	2.59	2.54	10.50	5.06	5.40	5.01	3.84	1.17	5.05	2.73	2.32	4.25	3.13	1.12	4.23	2.90	1.33
4.89	2.72	2.17	13.90	5.18	8.69	4.73	4.26	0.47	5.12	2.81	2.31	4.98	4.17	0.81	7.62	3.19	4.43
10.80	3.33	7.44	11.10	5.44	5.62	4.93	4.53	0.40	5.62	3.06	2.56	7.61	4.47	3.14	7.90	3.27	4.63
11.00	3.90	7.10	7.88	5.48	2.40	5.67	4.81	0.86	5.65	3.26	2.39	6.23	4.51	1.72	7.80	3.54	4.26
6.73	3.94	2.79	7.70	5.96	1.74
7.43	4.36	3.07	12.90	6.21	6.69
6.62	4.41	2.21	10.10	6.53	3.60
8.30	5.00	3.30	8.98	7.82	1.16
7.26	3.40	3.86	9.37	5.53	3.84	5.00	3.66	1.34	5.49	2.85	2.65	5.04	3.69	1.35	6.17	2.96	3.20
(2.06)	(1.00)	(1.81)	(2.75)	(1.16)	(2.50)	(0.68)	(1.16)	(0.98)	(0.30)	(0.24)	(0.36)	(1.46)	(0.71)	(0.88)	(1.83)	(0.49)	(1.42)
a	b	c	d	e	f	g	h	i	j	k	l	m	n	o	p	q	r

Individual values with means (SD) in bold. Comparisons of G_T, G_S, and G_C between groups: a vs. g, P = 0.02; b vs. e, P = 0.002; c vs. i, P = 0.005; c vs. o, P = 0.005; d vs. g, P = 0.003; d vs. j, P = 0.04; d vs. m, P = 0.003; d vs. p, P = 0.05; e vs. h, P = 0.015; e vs. k, P = 0.0003; e vs. n, P = 0.008; e vs. q, P = 0.0005; f vs. i, P = 0.03; f vs. o, P = 0.02; all other comparisons not significantly different.

OCT Pretreatment Prevents Hypoxia-Induced Changes in G_T and FITC Fluxes in Rat Colon

Two sets of experiments were done in rat distal colon (both n = 4) to first determine the effect of DNP on G_T and the mucosa-to-serosa FITC flux, and then to determine the effects of pretreatment with OCT on these parameters. Basal G_T values in the two sets of colons were significantly different (4.3 ± 0.20 mS/cm² vs. 6.7 ± 0.80 mS/cm², P = 0.0255), and although short-circuit current (I_sc) was similar in both groups (14.1 ± 0.30 µA/cm² vs. 15.6 ± 1.30 µA/cm², P = 0.3019), the group with the lower basal G_T exhibited a higher V_T (−3.4 ± 0.30 mV vs. −1.8 ± 0.10 mV, P = 0.0054). We attribute the difference in basal G_T between the groups of tissues to the fact that the two batches of animals were obtained from the same source several weeks apart, since they were housed, maintained, and their colons prepared and mounted in identical ways. In the first set of experiments, addition of DNP significantly increased G_T (by 18.6%, n = 4, P = 0.016) and the mucosa-to-serosa FITC flux (by 43%, n = 4, P = 0.011), as shown in Fig. 6, A and B, respectively. In contrast, in the second set of experiments done in colons pretreated with 0.2 μM OCT, these indicators of colonic epithelial permeability were unchanged by DNP (Fig. 6, C and D, respectively). These results show that the hypoxia-induced increase in FITC flux observed in rat distal colon is consistent with the hypoxia-induced increase in G_S seen in human colon, both changes being prevented by pretreatment with OCT.

Figure 6.

Effect of 2,4-dinitrophenol (DNP)-induced hypoxia on total tissue conductance (G_T) and the mucosa-to-serosa flux of FITC (A and B) in rat distal colon in the absence of 0.2 µM octreotide (OCT; n = 4) and (C and D) after pretreatment with 0.2 µM OCT (n = 4). Individual data points are shown, and error bars indicate SEM. *P = 0.0157 and **P = 0.0105 indicate significant differences compared with basal value.

DISCUSSION

We have previously shown that DNP-induced hypoxia rapidly stimulates basolateral IK channel activity in intact human colonic crypts (an effect fully reversible after DNP washout) and significantly increases paracellular shunt conductance (G_S) in isolated sheets of human colon (17). Hypoxia-induced increases in basolateral IK channel activity and G_S were inhibited by the specific IK channel inhibitors clotrimazole and TRAM-34, which suggests that basolateral IK channels have a regulatory effect on G_S (17). SOM and its long-acting synthetic analog OCT also inhibit human colonic basolateral IK channels in a rapid and sustained manner (18). These observations prompted the present study, which was designed to evaluate the effects of SOM and OCT on the increase in overall paracellular permeability (G_S) that occurs during DNP-induced hypoxia. The new findings from this study are that 1) pretreating human colon with 2 μM SOM prevented the increase in G_S produced by DNP; 2) after first inducing tissue hypoxia, subsequent addition of 2 μM SOM largely attenuated the hypoxia-induced increase in G_S; and 3) 0.2 μM OCT had identical effects. These observations suggest that SOM and OCT inhibit basolateral IK channels and initiate intracellular changes that maintain or restore tight junctional (TJ) integrity, thus preventing the deleterious effect of chemical hypoxia on paracellular permeability. We addressed this possibility in additional experiments in rat distal colon, which showed that DNP significantly increased both G_T and the mucosa-to-serosa flux of FITC (consistent with a DNP-induced increase in G_S), whereas these indicators of colonic epithelial permeability were unchanged by DNP in tissues pretreated with 0.2 μM OCT (Fig. 5).

The possible mechanisms underlying our results remain speculative, but it is worth considering 1) how hypoxia might activate basolateral IK channels, 2) the possible link between IK channel activation and the increase in G_S, and 3) how the effects of SOM and OCT are mediated. Hypoxia increases intracellular Ca²⁺ (which may itself stimulate basolateral IK channel activity) and activates phospholipase A₂, thereby increasing the production of arachidonic acid (26), although arachidonic acid has been reported to inhibit Ca²⁺-dependent IK channel activity (27, 28). O₂-dependent prolyl and asparaginyl hydroxylases, which regulate the activity of hypoxia-inducible transcription factors (HIFs) may have a role in acute O₂ sensing (29). However, in cultured HeLa cells, increased levels of HIF-1α protein occur after 4 h of hypoxia (1% O₂) (30), and it is difficult to envisage the involvement of HIFs in the stimulation of colonic basolateral IK channels occurring within minutes of adding DNP (which inhibits ATP production) and is fully reversible following DNP washout (17). Furthermore, a review of O₂-sensitive K⁺ channels provides limited insights into how basolateral IK channels might be activated by hypoxia; previous studies investigated the effects of hypoxia (due to lower O₂ tension) on other K⁺ channel types in nonintestinal cells (29). Thus, in chemoreceptor/neurosecretory cells, voltage-gated Kv channels, Ca²⁺-activated K⁺ channels, and TASK-like channels have been implicated in acute O₂ sensing, but in those cases, hypoxia decreased whole cell K⁺ currents (31, 32), which is inconsistent with the stimulatory effect of hypoxia on colonic basolateral IK channels observed in the present study. Hypoxia also decreased macroscopic K⁺ currents in pulmonary myocytes (33, 34). On the other hand, hypoxia and the resultant decrease in ATP production caused relaxation of vascular smooth muscle cells found in coronary and cerebral vessels, which reflected increased ATP-sensitive K⁺ channel activity (35). Whether hypoxia-sensitive K⁺ channels are inherently O₂-sensitive or an additional O₂ sensor molecule interacts with K⁺ channel subunits remains unclear (29).

With regard to the possible link between IK channel activity and G_S, TJ integrity is a major determinant of G_S and thus paracellular permeability. Thus, mesenteric ischemia in patients undergoing major abdominal surgery, trauma, hemorrhagic shock, or intensive care (36–39) might disrupt TJ proteins and increase G_S, which could enhance the translocation of bacteria or bacterial toxins, leading to multiorgan dysfunction and increased morbidity associated with sepsis (40–43). We are not the first to demonstrate that somatostatin peptides reduce hypoxia-induced damage to gut epithelia. In rats subjected to acute intestinal ischemia-reperfusion, SOM partially prevented intestinal epithelial barrier injury by upregulating Toll-interacting protein, which inhibited myeloid differentiation factor 88/nuclear factor-κB/MLC kinase signaling (44), and pretreatment with OCT significantly reduced the severity of intestinal injury via early induction of heme oxygenase-1 (45, 46). Whether these changes in intracellular signaling had specific effects on IK channel activity is unclear. Moreover, in those studies, ischemia was induced by clamping the superior mesenteric artery for 60 min, whereas our electrical and FITC flux measurements in human and rat colon were taken 5–15 min after inducing acute cell hypoxia with DNP. Nevertheless, our previous results (17, 18) and those from the present study point to basolateral IK channel activation and increases in G_S and permeability to macromolecules (e.g., FITC) being early events during acute hypoxia, which are completely prevented by SOM and OCT. It should be noted that acute changes in TJ permeability can be a normal physiological response. For example, the addition of glucose to the lumen of perfused mouse intestine increases transepithelial conductance and paracellular permeability to small nonelectrolytes, reflecting the contraction of the perijunctional actomyosin ring following myosin light chain (MLC) phosphorylation (47). In contrast, tissue hypoxia disrupts TJ proteins and impairs gut barrier function through the action of inflammatory cytokines (11, 12, 14, 15). DNP-induced hypoxia reflects cellular ATP depletion and leads to loss of F-actin from the apical pole and redistribution of ZO-1, occludin, and cadherin from the TJ complex (48, 49). The junctional localizations of occludin and ZO-1 are regulated by Rho-GTPase and increase with constitutive Rho signaling. Constitutive Rho signaling also protects against hypoxia-induced changes in TJ integrity, whereas inhibition has the opposite effect, suggesting that Rho has an important role in ischemia-induced injury (48). Ischemia is predicted to inhibit MLC phosphorylation, since Rho kinase (ROK), the downstream effector of Rho-GTPase, phosphorylates MLC directly (50). Interestingly, ROK-mediated phosphorylation of MLC is increased by membrane depolarization in LLC-PK1 cells, providing a link between electrogenic solute transport, membrane potential, and TJ integrity (51). In addition, K⁺ channel activity has been implicated in the regulation of epithelial barrier function. For example, in T₈₄ cells, the G-protein agonist mastoparan stimulated basolateral K⁺ channels, increased paracellular conductance fourfold, and altered the distribution of F-actin (52). These changes were inhibited by Ba²⁺ ions, indicating that the mastoparan-induced increase in paracellular conductance reflected basolateral K⁺ channel activation. Mastoparan also sequestrated and decreased the activity of Rho, similar to the changes seen during hypoxia. Irrespective of the intracellular signaling pathways involved, with regard to the observations in the present study, additional work is required to first determine whether changes in colonic barrier function during acute chemical hypoxia and the more gradual ischemia induced by superior mesenteric artery clamping reflect dissociation/migration of individual TJ proteins, which could be investigated using confocal microscopy and TJ-specific antibodies.

Given the apparent link between basolateral IK channel activity and G_S, it is somewhat easier to propose a mechanism to explain how the hypoxia-induced increases in these two factors are prevented or reversed by SOM and OCT. Inhibition of human colonic basolateral IK channels by SOM (and OCT) is G-protein-mediated and may reflect phosphoprotein tyrosine phosphatase-mediated dephosphorylation (18), similar to its inhibitory effect on apical high-conductance K⁺ (BK) channels (53). Basolateral IK channels are highly Ca²⁺-sensitive, and SOM may alter the phosphorylation state of IK channel protein, thereby decreasing its Ca²⁺-sensitivity (18). Little is known about the signaling mechanisms linking basolateral IK channel activity with TJ functionality, although intracellular Ca²⁺ appears to be an important determinant of TJ integrity. In MDCK cells, Ca²⁺ chelation reduced TJ assembly, decreased R_T, changed the sorting of ZO-1, and disrupted the association between the TJ and the cytoskeleton (54). When extracellular Ca²⁺ was raised, TJ reassembly appeared to depend on signaling pathways involving a heterotrimeric G protein, regulated intracellular Ca²⁺ stores, and protein kinase C (55). In addition, Rho-mediated TJ assembly mechanisms were disrupted in ATP-depleted MDCK cells (50), which suggests that Rho signaling regulates epithelial TJ protein functionality (56). Having previously demonstrated that both 2 μM SOM and 2 μM OCT markedly inhibit basolateral IK channels in human colonic crypts (18), we found that 2 μM SOM and 0.2 μM OCT were equally effective in preventing hypoxia-induced increases in G_S (Fig. 2 and Fig. 5). We also found that 0.2 μM OCT and 2 μM OCT exerted the same inhibitory effect on I_SC (the IK channel-mediated K⁺ current across the basolateral membrane) in apically permeabilized hypoxic human colon. Furthermore, pretreatment of rat distal colon with 0.2 μM OCT completely prevented hypoxia-induced increases in G_T and mucosa-to-serosa FITC fluxes, two indicators of overall colonic epithelial permeability. Of the five types of SOM receptor (SSTRs 1–5, SSTR2 having two forms SSTR2A and SSTR2B) which are linked to G-proteins (57), human colonic crypts express SSTR1 and SSTR2 but not SSTR3 or SSTR4 (58). Rat distal colonic epithelial cells express predominantly SSTR2 (59). Basolateral IK channels in human colon are therefore likely to be regulated by SSTR1 and/or SSTR2, since 2 µM SOM rapidly decreased IK channel activity by >80% in nonstimulated colonic crypt cells, an effect prevented by pretreatment with pertussis toxin, which points to a G-protein-dependent mechanism (18). Similar degrees of IK channel inhibition were elicited by 2 µM SOM and 2 µM OCT in crypts stimulated with dibutyryl cAMP (18). Thus, in the present study, it seems likely that the inhibitory effects of SOM and OCT on the electrophysiological and permeability changes induced by chemical hypoxia were mediated by SSTR1 and/or SSTR2 in human colon, and by SSTR2 in rat distal colon.

The results of the present study, done using human and rat colon, point to OCT mitigating ischemia-induced disruption of colonic epithelial barrier function, thereby limiting the translocation of bacteria and other toxic factors in patients undergoing major abdominal surgery. With regard to the small intestine, IK channels are present in Paneth cells (60), and Ca²⁺-sensitive basolateral K⁺ channels with a conductance of 19–28 pS (similar to human colonic IK channels) have been identified in rat duodenal crypt cells (61), but it is unclear whether hypoxia-sensitive basolateral IK channels are present in human small intestinal epithelial cells. Nevertheless, OCT administered subcutaneously is widely used to manage enterocutaneous fistulae, diarrhea secondary to VIPomas, and obscure secretory diarrhea. Systemic absorption is rapid and complete, with 100% bioavailability and peak plasma concentrations of 2–4 µg/L in patients receiving 50–100 µg daily (62, 63). Since OCT has a molecular weight of 1,019, plasma concentrations of 2–4 µg/L equate to 0.002 μM–0.004 μM, 50- to 100-fold lower than the “effective” concentration of OCT (0.2 μM) used in our in vitro experiments. Nevertheless, the mean fasting plasma SOM concentration of 13.3 pg/mL reported in 45 normal subjects (64), which equates to 8.1 pM based on SOM’s molecular weight of 1,638, is at least two orders of magnitude lower than plasma OCT concentrations achieved in clinical settings. Thus, plasma OCT concentrations of 0.002 μM–0.004 μM may still prevent significant ischemic damage to the colonic epithelial barrier in vivo, given that they readily inhibit other aspects of gastrointestinal function. Furthermore, if OCT was to be used during surgery, much higher concentrations could be achieved within the intestinal mucosa by infusing it directly into splanchnic vessels.

Perspectives and Significance

Acute changes in TJ integrity are a normal physiological response during glucose absorption in mammalian small intestine, leading to increased paracellular permeability to small nonelectrolytes. This is associated with contraction of the perijunctional actomyosin ring following myosin light chain (MLC) phosphorylation. On the other hand, tissue hypoxia leads to inflammatory cytokine-mediated disruption of TJ proteins and impairment of gut barrier function. Here, we report that SOM and OCT (a long-acting synthetic analog of SOM) both prevent chemical hypoxia-induced increases in paracellular shunt permeability/conductance in isolated human colon, and in rat distal colon, OCT prevents chemical hypoxia-induced increases in total epithelial conductance and the transepithelial movement of the macromolecule FITC-dextran 4000. The precise cellular mechanisms underlying the effects of tissue hypoxia on TJ functionality and transepithelial macromolecule movement, and the ability of somatostatin peptides to prevent these effects, remain to be established. Further mechanistic studies are required to underpin the possible use of intraoperative OCT to prevent or limit gut ischemia during abdominal surgery, with the aim of decreasing the risk of bacterial/bacterial toxin translocation and sepsis.

DATA AVAILABILITY

Laboratory data may be available upon reasonable request.

GRANTS

This work was supported by Leeds Teaching Hospitals Charity 1170369 (Rays of Hope R5T42) and National Institute of Health NIDDK Grants DK104791 and DK 112085.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

G.I.S. conceived and designed research; I.R., V.M.R., and A.J.N. performed experiments; I.R., V.M.R., A.J.N., and G.I.S. analyzed data; J.P.A.L. and G.I.S. interpreted results of experiments; I.R., V.M.R., and G.I.S. prepared figures; G.I.S. drafted manuscript; J.P.A.L. and G.I.S. edited and revised manuscript; I.R., V.M.R., A.J.N., J.P.A.L., and G.I.S. approved final version of manuscript.