Materials and methods

Thirty four healthy volunteers and eight patients with villus atrophy participated in the study. The healthy volunteers had no history of gastrointestinal disease and were free from medication. The protocol was approved by the ethical committee at Sahlgrenska University Hospital, Göteborg, and all participants had given informed consent. The patients were informed about the scientific nature of the study, and they all participated voluntarily. The study was part of a diagnostic “work up” for patients with suspected coeliac disease, and the patients had agreed to refrain from changes in diet until all diagnostic procedures were completed. The diagnosis of villus atrophy was based on duodenal biopsies in triplicate showing signs of subtotal or total villus atrophy, crypt hyperplasia, and increased numbers of inflammatory cells in the lamina propria. The mean age of the volunteers was 25 (range 20–39) years and the mean age of the villus atrophy patients was 39 (range 20–53) years. The experiments started with the intubation procedure at 7 30 am, and continued for about four hours.

Results

There was no significant difference in motor activity between the groups (table 1). Luminal HCO₃ ⁻ concentration at the entry to the test segment was not significantly different in the control and villus atrophy groups (≈ 20 mmol/l), but was about 20% lower in the atropine group (15.2 (0.9) mmol/l; p<0.01 versus control and villus atrophy groups; results are mean (SE)).

Figure 2 illustrates the relation between luminal concentration and absorption of HCO₃ ⁻ in the control group. During perfusion with the HCO₃ ⁻-containing solution (filled triangles), there was a significant linear correlation between HCO₃ ⁻ absorption and HCO₃ ⁻ concentration measured at the proximal aspiration site (r = 0.80, p<0.01). During perfusion with HCO₃ ⁻-free solution (NaCl, open triangles), HCO₃ ⁻ transport was not significantly different from zero.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Duodenal HCO₃ ⁻ absorption as a function of luminal HCO₃ ⁻ concentration at the entry to the test segment, in healthy controls. All data emanate from the first 30 minutes of the migrating motor complex cycle (low motor activity). Filled triangles indicate Krebs/mannitol perfused segments and open triangles saline perfused segments. HCO₃ ⁻ transport is expressed as μmol/min per 10 cm; positive values denote net absorption. The equation of the regression line for the Krebs/mannitol perfused segment is also given in the figure.

Figure 3 shows individual values for HCO₃ ⁻concentration and HCO₃ ⁻ absorption in the atropine group. In this group, there was no longer any significant correlation between HCO₃ ⁻ concentration and HCO₃ ⁻ absorption (r = 0.32, p = 0.28). However, 10 of 13 subjects (solid squares) followed a regression line that was approximately parallel to the one obtained in the controls (which is indicated by the dashed line in fig 3). The three remaining subjects (open squares) closely followed the control regression line. To test if the displacement of the atropine data from the control regression line was statistically significant, we fitted the concentration data in the atropine group into the control regression equation, and then compared measured and expected HCO₃ ⁻ absorption rate under the null hypothesis that the atropine data were randomly distributed around the control line. This calculation showed that the atropine data were significantly displaced to the left of the control population (control distribution: 0.0 (3.6); atropine distribution: +13.8 (3.7) μmol/min per 10 cm; p<0.05).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Relation between luminal HCO₃ ⁻ concentration and net HCO₃ ⁻ absorption in atropine treated subjects. A total of 10 of 13 subjects (solid squares) followed a regression line (solid line) that was almost exactly parallel to the control regression line (dashed line). The three remaining subjects (open squares) were on the control line. The equation of the regression line for the 10 subjects is given in the figure.

In the villus atrophy patients, there was no consistent relation at all between luminal HCO₃ ⁻ concentration and absorption (r = 0.075, p = 0.85; data not shown), and therefore this mode of recalculation was not considered appropriate for this group. However, luminal HCO₃ ⁻ concentration at the proximal aspiration site in the villus atrophy patients was not significantly different from that in the control group (19.8 (1.1)v 21.4 (1.4) mmol/l). To be able to directly compare absolute HCO₃ ⁻ absorption values in all three groups at a similar luminal HCO₃ ⁻concentration, we therefore used actual measured data from the control and villus atrophy groups and mathematically normalised the atropine data to the same luminal concentration (20 mmol/l), from the slope of the (control and atropine) regression lines. Figure 4 shows the results of this calculation. In the control group, the normalised HCO₃ ⁻ absorption rate was 35.4 (3.8) μmol/min per 10 cm, which increased to 48.0 (3.4) μmol/min per 10 cm after atropine administration (p<0.05 vcontrol data). In the villus atrophy group, the normalised HCO₃ ⁻ absorption rate was 15.2 (5.7) μmol/min per 10 cm (p = 0.01 v controls, p<0.001 v atropine). (Actual luminal HCO₃ ⁻ concentration in the control group was 21.4 (1.4) mmol/l, and the use of measured instead of calculated (20 mmol/l) data only marginally changed the results and did not influence the significance values.)

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Calculated net HCO₃ ⁻absorption at a luminal HCO₃ ⁻ concentration of ≈ 20 mmol/l in the three groups. Control data and atropine data were recalculated from the slope of the regression lines in figs 2 and3. Villus atrophy data were measured data (mean luminal HCO₃ ⁻ concentration 19.8 (1.1) mmol/l) . Values are mean (SE). Asterisks above bars indicate a significant difference from control values. A significant difference between villus atrophy and atropine data is indicated separately. Significance levels were: *p<0.05; **p<0.01; ***p<0.001. For further explanations, please see the text.

To analyse further the mechanism behind the apparent stimulatory effect of atropine on HCO₃ ⁻ absorption, we also calculated luminal H⁺ ion concentration and luminal Pco ₂ values at the proximal and distal aspiration sites. Figure 5 summarises the results. At the proximal aspiration site, there was no significant difference in luminal H⁺ ion concentration between the three groups (Kruskal-Wallis test, p = 0.09) (fig 5A). Between proximal and distal aspiration sites, H⁺ concentration increased significantly in the atropine group (p<0.01) but no significant change occurred in the control group (p = 0.09) or in the villus atrophy group (p = 0.48). Consequently, at the distal aspiration site, luminal H⁺concentration in the atropine group was significantly higher and in the villus atrophy group significantly lower than in the control group (p<0.01).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Luminal H⁺ concentration (top panel) and luminal Pco ₂ (bottom panel) at the proximal and distal aspiration sites, in the three groups. Luminal H⁺concentration is expressed in nmol/l and Pco ₂in kPa. Values are mean (SE). Asterisks indicate a significant difference from control values at the same aspiration site. Significance levels are the same as in fig 4.

Figure 5B summarises in a similar fashion luminal Pco ₂ values at the proximal and distal aspiration sites. Luminal Pco ₂ did not differ between the groups, either at the proximal or distal aspiration sites, and no significant change in Pco ₂ between proximal and distal aspiration sites occurred in any of the three groups.

Discussion

Atropine had three major effects on duodenal HCO₃ ⁻ transport: it reduced luminal HCO₃ ⁻ concentration at the proximal aspiration site, it displaced the relation between HCO₃ ⁻ concentration and absorption to the left, and it induced significant acidification of the luminal contents at the distal aspiration site. Net absorption of a compound is the difference between the absorptive and secretory flux, and stimulation of net absorption may therefore be due to stimulation of the absorptive flux, inhibition of the secretory flux, or a combination of the two.

Atropine inhibits the secretory HCO₃ ⁻ flux in both the human proximal duodenum19 and rat jejunum,20 and this mechanism may therefore account for our findings. Theoretically, inhibition of a (normally reabsorbed) secretory HCO₃ ⁻ flux with a magnitude of about 20 μmol/min per 10 cm may account for the apparent leftward displacement of the atropine curve (fig 3). Inhibition of a secretory flux with a relatively high HCO₃ ⁻concentration may also account for the reduction in HCO₃ ⁻ concentration at the proximal aspiration site in the atropine group. The same mechanism would also reduce HCO₃ ⁻ concentration at the distal end of the test segment which, at a constant Pco ₂, would lead to an increase in luminal acidity. Inhibition of a secretory HCO₃ ⁻ flux is therefore an altogether possible explanation for our findings.

There are, however, some observations that are difficult to reconcile with this interpretation. First, the secretory HCO₃ ⁻ flux may be very small at the level of our test segment—that is, even total inhibition of this flux may not suffice to account for the displacement of the saturation curve in the atropine experiments. Isenberg et al 21 have directly compared the secretory HCO₃ ⁻ flux in proximal and mid duodenum in humans and saw a 90% reduction from the bulb to the level of our mixing segment, which had a HCO₃ ⁻ secretion rate of only 3–4 μmol/min per 10 cm. It is always difficult to interpret segmental differences between species, but data from rat jejunum suggest that the basal secretory HCO₃ ⁻ flux is even lower more distally.20 In the present study, we also tried to measure the secretory HCO₃ ⁻ flux by saline perfusion. HCO₃ ⁻ inflow into the mixing segment was 11.2 (6.0) μmol/min, which generated a luminal HCO₃ ⁻ concentration of 1.6 (0.8) mmol/l. However, despite this very low HCO₃ ⁻concentration, net HCO₃ ⁻ transport in the test segment was not significantly different from zero (fig 2). If a secretory flux of any quantitative magnitude had occurred at this level, it must consequently have been totally reabsorbed.

In a saturated system, a reduced intrinsic substrate load will increase uptake of an extrinsic marker—that is, inhibition of a substantial HCO₃ ⁻ secretion by the mucosa itself will be expected to enhance HCO₃ ⁻ uptake from the lumen. However, the system was not saturated, as seen from the strong concentration dependence of HCO₃ ⁻ absorption. It is therefore very difficult to envisage how inhibition of a minor secretory flux can account for the leftward displacement of the HCO₃ ⁻ absorption curve.

A characteristic feature of HCO₃ ⁻ absorption is that it is very hard to saturate. In the jejunum, Turnberget al 6 obtained only partial saturation with luminal concentrations as high as 200 mmol/l. The linear part of their saturation curve was virtually identical with our own, strongly suggesting that they were studying the same transport mechanism. An important question is whether this saturation curve fully accounts for the HCO₃ ⁻ concentration decrease between the infusion site and the proximal site, or if a secretory HCO₃ ⁻ flux has to be included to account for the observed concentration changes. To test this, we used the intercept of the regression line with the HCO₃ ⁻concentration of our perfusate (25 mmol/l) to estimate HCO₃ ⁻ absorption in the mixing segment. The expected mean absorption rate (fig 2) will be 47/2 = 23.5 μmol/min. Mean flow rate through the mixing segment (including the pressure infusion) was 6.26 (0.19) ml/min and there was no significant net water flux (0.00 (0.15) ml/min). One consequently expects a mean luminal HCO₃ ⁻ concentration at the proximal aspiration site of 25 − (23.5/6.26) = 21.2 mmol/l, which agrees almost exactly with the measured mean value (21.4 (1.4) mmol/l, table1). The most likely explanation for the relatively pronounced scatter around this mean value is interindividual differences in the degree of stirring motor activity in the mixing segment, which probably increases HCO₃ ⁻ uptake by improving the contact with the absorbing epithelium.9 At least as judged from these rough calculations, HCO₃ ⁻ absorption thus seems to account fully for the HCO₃ ⁻concentration changes at the proximal aspiration site.

A similar mode of calculation can also be used to estimate HCO₃ ⁻ inflow into the mixing segment. At the proximal aspiration site, the mean HCO₃ ⁻“flow rate” (volume flow × [HCO₃ ⁻]) was 134 μmol/min. If the mixing segment had a similar absorption rate to the test segment—that is, 20–25 μmol/min per 5 cm, one would expect an initial mean HCO₃ ⁻ inflow rate of 155–160 μmol/min. HCO₃ ⁻ was infused at a rate of about 150 μmol/min—that is, only 5–10 μmol/min remain as a possible mean value for the secretory inflow into the mixing segment. Even this value, which agrees reasonably well with both Isenberg′s data21 and our own NaCl experiments (our mean value was 11.2 (6.0) μmol/min), is probably an overestimation, as this mode of calculation assumes optimal stirring conditions in the mixing segment.

In the atropine group, 10 of 13 subjects followed a regression line that was displaced about 20 μmol/min per 10 cm to the left—that is, at a given concentration, HCO₃ ⁻ absorption was 20 μmol/min per 10 cm higher. The equation for this regression line gave an expected HCO₃ ⁻ absorption in the mixing segment of around 35 μmol/min which, with the mode of calculation used above, should lead to a further decrease in mean proximal HCO₃ ⁻ concentration value by about 3.5 mmol/l. The actually measured mean difference between controls and atropine data was 5 mmol/l (table 1). Furthermore, the three experiments in the atropine group which seemed to follow the control regression line (solid squares in fig 3) also had a proximal HCO₃ ⁻ concentration which was virtually identical with the mean value in the control group. Attempts to estimate the amount of HCO₃ ⁻ entering the mixing segment in the atropine group gave negative values (−12.9 (6.6) μmol/min)—that is, in the atropine group, HCO₃ ⁻ absorption capacity in the mixing segment was probably even higher than in the test segment.

To conclude these numeric excursions, our data are theoretically compatible with both inhibition of HCO₃ ⁻secretion and stimulation of HCO₃ ⁻absorption, but increased affinity of the absorptive mechanism accounts for both transport and concentration data in a strikingly consistent manner.

To obtain additional information on the transport mechanisms responsible for HCO₃ ⁻ absorption, we also included a small group of patients with villus atrophy in the study. The idea was that, if HCO₃ ⁻ absorption depends on NHE activity in the apical membrane of villus enterocytes, one would expect villus atrophy to have the opposite effect to atropine—that is, to reduce HCO₃ ⁻ absorption and to decrease H⁺ and CO₂ formation. As a group, the patients did exhibit significantly reduced HCO₃ ⁻ absorption in their test segment, but the interindividual variation was fairly large, with “normal” absorption values occurring in some subjects. The linear correlation between HCO₃ ⁻ absorption and luminal HCO₃ ⁻ concentration was also lost in the patients, which precluded “backward” calculation of the amount of HCO₃ ⁻ entering the mixing segment. Another puzzling finding was that, despite their inability to absorb HCO₃ ⁻ in the test segment, the patients had a HCO₃ ⁻ concentration at the proximal aspiration site that was virtually identical with that in the controls (table 1). However, in the patients, the decrease in HCO₃ ⁻ concentration was almost exactly paralleled by dilution of the volume marker—that is, the mechanism was probably increased inflow of secreted fluid with a low HCO₃ ⁻ concentration into the mixing segment.22

The reason for the much larger variability in the HCO₃ ⁻ transport data in the patients is not altogether clear. Prostaglandins, which are released during inflammation, stimulate HCO₃ ⁻ secretion also in the distal duodenum,21 and it cannot therefore be excluded that a numerically relevant secretory HCO₃ ⁻ flux may have occurred in some of the villus atrophy patients. Other possibilities are interindividual differences in the degree of villus atrophy, or simply shortcomings of our fairly crude methodology. However, it should be stressed that, despite the relatively large scatter, HCO₃ ⁻transport rate in the patient group was significantly different from that in the healthy subjects.

In all three groups, the same solution was infused at the same rate and one therefore also expects differences in HCO₃ ⁻ absorption between the groups to be reflected by changes in luminal acidity and Pco ₂, particularly at the distal aspiration site. Since our hypothesis was that HCO₃ ⁻ is absorbed by Na⁺/H⁺ exchange, which is in turn stimulated by atropine and reduced by villus atrophy, the pattern expected from the model is increased acidity and Pco ₂ at the duodenojejunal junction in the atropine group, and essentially the opposite pattern in the villus atrophy group. The actual findings were slightly different: luminal acidity did show the expected profile, but luminal Pco ₂ was virtually identical in all three groups.

When interpreting changes in luminal Pco ₂, one has to remember that CO₂ is a very lipid-soluble gas that readily diffuses across the duodenal mucosa.23 The development of Pco ₂ gradients therefore depends not only on production rate but also on factors affecting diffusion—that is, thickness and hydrophobicity of unstirred layers.24 Thus, despite a longer transport segment and a higher luminal bicarbonate concentration, Turnberget al 6 had to use coupled buffer systems and reduced perfusion rates to show a HCO₃ ⁻-induced increase in luminal Pco ₂ during triple lumen perfusion of human jejunum. In a better controlled system (perfused rat jejunum in vivo), Hubel25 did see an increase in luminal Pco ₂ with time if HCO₃ ⁻ was present in the lumen. However, the magnitude of this increase was not significantly affected by atropine.8

The most likely explanation for the poor correlation between luminal Pco ₂ and HCO₃ ⁻absorption seems to be that CO₂, being a highly lipid-soluble gas, rapidly diffuses into the mucosa. The segment consequently behaves like a “Pco ₂-stat” system, and changes in luminal HCO₃ concentration are primarily reflected by pH changes. Provided that one accepts the “Pco ₂-stat” model, the data thus suggest that atropine enhances HCO₃ ⁻ absorption by a process that involves increased proton formation from the duodenal villi.

The mechanism of action of atropine remains unknown. It seems quite clear that cholinergic neurones do participate in the control of intestinal chloride secretion2 ,26-30 but in our model, we have also postulated a cholinergic inhibitory effect on the absorbing epithelium. There is actually some evidence against this hypothesis. Both intestinal villi and crypts are innervated by cholinergic neurones31 ,32 and are supplied with muscarinic receptors,33-35 but activation of enteric neurones by electrical field stimulation does not inhibit the absorptive sodium or chloride fluxes.2 Another contradictory finding is that the absorptive response to a meal was not affected by luminal administration of bupivacaine, a local anaesthetic agent.36 One clearly has to keep an open mind about the exact mechanism behind the atropine response.

In summary, our results show that, in the interdigestive state, duodenal net HCO₃ ⁻ absorption is increased by atropine when given in a dose that blocks muscarinic receptors. A detailed analysis of the transport data suggests stimulation of active HCO₃ ⁻ absorption as the most probable mechanism. The results are compatible with interdigestive cholinergic inhibition of villus NHE activity, but the data are far too indirect to allow this conclusion. Mechanisms for neural inhibitory control of the vectorial NHE do, however, exist1 and our results certainly encourage further search for the physiological role of this system. The isoform responsible for this vectorial Na⁺transport is probably NHE3, which is inhibited by high dose amiloride given from the luminal side.37 A good experiment would be to try to block the atropine response with high dose amiloride, preferably in a less complicated animal model.