Report

Role of the Enteric Nervous System in the Fluid and Electrolyte Secretion of Rotavirus Diarrhea

See allHide authors and affiliations

Science  21 Jan 2000:
Vol. 287, Issue 5452, pp. 491-495
DOI: 10.1126/science.287.5452.491

Abstract

The mechanism underlying the intestinal fluid loss in rotavirus diarrhea, which often afflicts children in developing countries, is not known. One hypothesis is that the rotavirus evokes intestinal fluid and electrolyte secretion by activation of the nervous system in the intestinal wall, the enteric nervous system (ENS). Four different drugs that inhibit ENS functions were used to obtain experimental evidence for this hypothesis in mice in vitro and in vivo. The involvement of the ENS in rotavirus diarrhea indicates potential sites of action for drugs in the treatment of the disease.

Rotavirus is the major cause of infantile gastroenteritis worldwide and is associated with about 600,000 deaths every year, predominantly in developing countries (1). Although two decades of research have significantly increased our understanding of virus immunology and have led to the development of an oral vaccine, our knowledge of the mechanisms that induce rotavirus diarrhea, nausea, and vomiting remains limited.

Rotavirus infects the mature enterocytes in the mid and upper villous epithelium of the small intestine, ultimately leading to cell death and villus atrophy. A striking observation in both animals and humans is that only a few percent of the mature villus epithelial cells and no crypt cells seem to be infected (2–5). Fluid secretion is usually ascribed to an imbalance between the secretory crypt cells and the mature absorptive villous epithelium. The death of the villus cells leads to a repopulation of the epithelium with immature secretory type cells. Mechanisms proposed to explain the rotavirus-induced intestinal secretion of fluid and electrolytes include villous ischemia (5) and a toxin-like effect by a nonstructural virus protein, NSP4 (6).

A localized systemic response triggered by rotavirus-enterocyte interaction has been proposed previously (4). According to Stephen and collaborators, rotavirus induces ischemia in villi before significant virus replication occurs in epithelial cells. These changes in the microcirculation are believed to be of pathophysiological importance and to be mediated by endogenous, neuroactive, hormonal substances (5).

During the last two decades it has become increasingly apparent that intestinal secretion of fluid and electrolytes evoked by luminal agents is often induced by activation of the nervous system located in the intestinal wall, the enteric nervous system (ENS) (7–9). Here we have investigated the role of the ENS in the fluid secretion evoked by rotavirus in newborn mice.

In Ussing chamber experiments (10), transmural potential difference (PD) was monitored continuously in mice intestines in vitro. The intestinal segment was voltage-clamped at 0 V and the short circuit current (SCC) determined at various intervals, which allowed us to estimate tissue electrical resistance or conductance. The nearly twofold increase in PD in infected (1.89 ± 0.15 mV;n = 27) relative to uninfected (0.90 ± 0.07 mV;n = 21) animals was significant by the Mann-Whitney U test (P < 0.0001).

Two types of Ussing chambers were used to measure the SCC. In one type, the electrical field was homogeneous, so that it was possible to reliably quantify tissue conductance. In the virus-infected tissue, conductance was significantly higher than in the noninfected tissue [28.6 ± 2.7 mS cm−2 (n = 13) versus 18.3 ± 1.1 mS cm−2 (n = 12);P < 0.01]. The corresponding values of PD in these experiments were 1.91 ± 0.15 mV versus 0.79 ± 0.06 mV (P < 0.0001).

Two drugs that specifically abolish nerve action potentials were tested, tetrodotoxin (TTX; blocker of sodium channels in excitable tissues) and lidocaine (local anesthetic). Neither drug significantly affected tissue conductance. Addition of TTX to the serosal and the mucosal compartments in infected and noninfected animals attenuated the PD in a dose-dependent manner (Fig. 1A). The decrease in PD was significantly larger in the infected than in the noninfected animals (P < 0.01).

Figure 1

Results obtained in the Ussing chamber experiments. (A) The effect of increasing concentrations of TTX on the transepithelial PD in control intestinal segments (n = 5) and segments exposed to rotavirus (n = 6). TTX was administered both to the mucosal and serosal compartments. (B) The effect of lidocaine (0.4 and 4 mM) on PD in control (n = 5) and virus-infected (n = 6) intestinal segments. Only the serosal surface was exposed to the drug. (C) The effect of increasing concentrations of mecamylamine on the PD in control (n = 5) and virus-infected (n = 6) intestinal segments. The drug was administered both to the serosal and mucosal compartments. Crosses indicate statistically significant differences compared with control observations (C). Asterisks indicate statistically significant differences between control and virus groups.

In initial series of experiments, lidocaine was applied to both sides of the intestinal wall at three concentrations (0.4, 4, and 40 mM). When the intestinal mucosa was exposed to theophylline at the end of those experiments (to test the viability of the tissue), no change in PD was observed, suggesting that lidocaine had affected enterocyte function. In a second series of experiments lidocaine was therefore only administered to the serosal side at two concentrations (0.4 and 4 mM). Theophylline evoked a significant increase in PD in these experiments. Lidocaine attenuated the PD in infected and noninfected intestines in a dose-dependent manner (Fig. 1B), with a significantly larger decrease seen in infected than in control animals (P< 0.05).

Hexamethonium, which inhibits synaptic transmission by blocking the nicotinic receptor, had no effect (2, 20, and 200 μM) (11). Because hexamethonium is a highly polar compound and therefore not readily soluble in lipids, we investigated the effects of a more lipophilic nicotinic receptor blocker, mecamylamine. Mecamylamine tested at the same concentrations as hexamethonium decreased the PD in infected relative to noninfected intestinal segments without affecting tissue conductance (Fig. 1C).

In perfusion experiments (12) we monitored the PD and net fluid transport (NFT), which was measured with 14C-labeled polyethylene glycol as a nonabsorbable radioactive marker (Fig. 2). In experiments to test the viability of the intestinal segments, PD and NFT remained relatively constant in the noninfected intestines, but changed significantly in the intestines exposed to virus. Because the differences in NFT and PD were not significant at 60 min, the results reported below are based on experiments lasting only 40 min.

Figure 2

Viability of the intestinal preparation perfused in vitro. The PD was monitored continuously for 60 min, and NFT was measured during three consecutive 20-min periods. (A) Changes in NFT during perfusion of infected (n = 7) and noninfected (n = 5) intestinal segments. Negative values indicate net fluid secretion. (B) Changes in the PD during perfusion of control segments (n = 6) and in intestinal segments exposed to rotavirus (n = 7). Both the NFT and PD remained constant in the control series, but decreased in the virus group. For technical reasons NFT was not recorded in one experiment. Asterisks indicate statistically significant differences between control and virus groups.

The PD measured in infected animals was significantly higher than in noninfected mice (Fig. 2B). This was also apparent in a different series of experiments comparing the PD in infected and noninfected animals during the control period (10) (control animals: 2.71 ± 0.10 mV; n = 24; infected animals: 4.24 ± 0.20 mV; n = 27; P < 0.0001). Similarly, in the control period, the amount of fluid secreted in the infected animals (28.8 ± 4.9 μl hour−1cm−1; n =26) was significantly different from the fluid absorption observed in the noninfected animals (1.1 ± 2.6 μl hour−1 cm−1; n = 22,P < 0.0001).

When the serosal surface of the perfused segment was exposed to a TTX concentration of 0.1 μM, there was a significant decrease in the PD of virus-infected segments, whereas the PD of control animals was not affected. Concomitantly, net fluid secretion in the infected intestines was completely inhibited. NFT in the noninfected intestines changed from a small net secretion to net absorption (Fig. 3). The changes caused by TTX were much larger in the infected than in the control intestines (Table 1). Exposure of the intestinal serosa to a 4 mM solution of lidocaine evoked a response similar to that seen with TTX (Fig. 4 and Table 1).

Figure 3

(A) NFT and (B) PD before and after exposure of the intestinal serosa to TTX (10−7 M). Perfusion experiments were performed on control segments (n = 6) and on segments infected with rotavirus (n = 6). Crosses indicate statistically significant differences compared with the corresponding No drug observations. Asterisks indicate statistically significant differences between control and virus groups.

Figure 4

The effect of lidocaine (1.6 × 10−3 M) on (A) NFT and (B) PD. Control experiments were performed on six segments, and rotavirus experiments on eight segments. Crosses indicate statistically significant differences compared with the corresponding No drug observations. Asterisks indicate statistically significant differences between control and virus groups.

Table 1

Differences in transmural epithelial potential difference (PD) and net fluid transport (NFT) between control measurements and measurements made 20 min after exposure of the intestinal mucosa to the drug tested. In the “No drug” experiments, the change in NFT between the first and second 20-min period was determined. The number of observations is indicated in parentheses.

View this table:

Hexamethonium at a concentration of 20 μM significantly decreased the PD in the virus-infected segments but not in control segments (Fig. 5B). Similarly, net fluid secretion was attenuated in the infected segments, whereas in the control segments no significant effect was detected (Fig. 5A). The changes in PD and NFT were larger in the infected than in the control intestines (Table 1).

Figure 5

(A) NFT and (B) PD before and after exposure of the intestinal serosa to hexamethonium (2 × 10−5 M). Perfusion experiments were performed on control segments (n = 5 or 6) and on segments infected with rotavirus (n = 6). For technical reasons, NFT was not determined in one experiment. Crosses indicate statistically significant differences compared with the corresponding No drug observations. Asterisks indicate statistically significant differences between control and virus groups.

A third type of experiment was carried out on 4- to 6-day-old awake animals. Diarrhea was determined by judging the stool according to a scoring system (13) after gently pressing the abdomen. Only lidocaine given intraperitoneally (i.p.) was tested in these experiments, because the effect of hexamethonium or mecamylamine given intravenously (i.v.) lasts for only 1 to 2 hours (14). Lidocaine (25 mg/kg, dissolved in 50 μl of physiological saline) was injected twice daily after giving rotavirus orally. About 48 hours after rotavirus administration the incidence of diarrhea peaked in the control group, with 14 of 15 mice exhibiting diarrhea. In the lidocaine group, 6 of 14 mice showed signs of diarrhea. The difference between the two groups was highly significant (P < 0.005; Fisher's exact probability test).

The observation that all four drugs significantly attenuated the intestinal secretory response to rotavirus strongly suggests that the ENS participates in the virus-induced electrolyte and fluid secretion, as has been shown for bacterial enterotoxins (7–9). The involvement of the ENS may explain how comparatively few virus-infected cells at the villus tips can cause the intestinal crypt cells to augment their secretion of electrolytes and water. The enhanced fluid secretion may serve as a defense mechanism against the potentially harmful mucosal influence. In the case of the fluid secretion caused by cholera toxin and bile salt, it has been shown that the secretory response is accompanied by a motility response that enhances the luminal transport in an aboral direction (15).

We used the TTX data to estimate the extent of ENS involvement in the intestinal fluid and electrolyte secretion evoked by rotavirus, given that TTX probably is the most investigated and most specific blocker of nervous activity among the drugs used in the present study (16). In the Ussing chamber experiments, 67% of the virus effect was mediated by the nervous system. A similar calculation for the effect of TTX on NFT in the perfusion experiments indicates that 85% of the virus response can be ascribed to neuronal involvement. The PD measurements in the perfusion experiments indicate than the virus effect is entirely due to nervous system activity. However, the latter calculation seems less reliable because the PD measurements in the noninfected segments were abnormally high. We conclude from these calculations that at least two-thirds of the secretory response to rotavirus is mediated by the ENS.

Intestinal inflammation has been shown to evoke fluid secretion by activation of the ENS (9). The results of the present study are consistent with these observations. Inflammation is accompanied by an increased tissue concentration of a large number of biologically highly active compounds that alone or together may activate enteric neurons (17, 18). In the case of rotavirus, ENS activation may also occur by other mechanisms. It has been shown that rotavirus can infect primary neurons (19). Furthermore, it was recently suggested that rotavirus exerts its effects through a “toxin-like” protein (NSP4), a nonstructural protein produced by the virus (6), thereby increasing intracellular calcium concentrations. Such calcium increases may trigger the release of amines or peptides from the endocrine cells of the gut to stimulate dendrites or free nerve endings located underneath the epithelial layer, thereby activating secretory nervous reflexes (20).

The replacement of fluid losses in diarrhea with an oral solution of glucose and sodium chloride represented a major therapeutic advance when it was first introduced. The effect is dependent on an intact absorptive capacity of the intestinal epithelium. Combining the oral glucose-salt solution with a drug that attenuates the intestinal secretion of fluid has the potential to enhance this effect. The results of this and earlier studies (8) strongly suggest that in most, if not all, intestinal secretory states, nerve reflexes in the ENS are stimulated to cause intestinal fluid losses. This implies new potential sites of action for drugs in the treatment of diarrhea. The results with nicotinic receptor blockers indicate that the secretory nervous reflexes contain at least one synapse. Synaptic transmission may in principle be influenced by drugs other than the nicotinic receptor blockers reported here. For example, it may be possible to target the system presynaptically (21) or to blockade neurotransmitter receptors on the enterocytes. Finally, intestinal secretion evoked by diarrheal agents that activate ENS via the intestinal endocrine cells can be significantly inhibited by L-type calcium channel blockers, which markedly attenuate the release of amines and peptides from these cells (20).

  • * To whom correspondence should be addressed. E-mail: ove.lundgren{at}fysiologi.gu.se

REFERENCES AND NOTES

View Abstract

Stay Connected to Science

Navigate This Article