The active and passive components of glucose absorption in rat jejunum under low and high perfusion stress

Abstract

In order to determine how perfusion design affects the relationship of the apparent ‘active’ and ‘passive’ components of glucose absorption, rat jejunum was perfused with 50 mm glucose under conditions of low and high mechanical stress. Phloretin or cytochalasin B was used to inhibit GLUT2 and phloridzin to inhibit SGLT1. In low stress perfusions, the ratios of the ‘passive’ to the ‘active’ components determined using phloretin and phloridzin were 2.2 and 0.43, respectively. This discrepancy was explained by the fact that phloridzin inhibits not only SGLT1 but also indirectly that part of the GLUT2-mediated component controlled by SGLT1 through the glucose-induced activation and recruitment of GLUT2 to the brush-border membrane. In high stress perfusions, the ratios of the ‘passive’ to the ‘active’ components determined using phloretin and phloridzin were 0.94 and 0.95, respectively; cytochalasin B gave 0.95. The identity of these results was explained by the observation that the passive component is not dependent on the active component, because glucose-induced activation and recruitment of GLUT2 does not occur in high stress perfusions. Simultaneous inhibition of SGLT1 and GLUT2 in high stress perfusions with phloridzin and cytochalasin B inhibited absorption by 92 ± 7 %; non-carrier-mediated transport is therefore minimal. Our data provide support for the view that the term ‘facilitated’ should be used to replace the term ‘passive’ in describing the component now known to be mediated by GLUT2. The study of the mechanism and regulation of this facilitated component depends crucially on the design of the perfusion system.

For almost fifty years, it has been reported that intestinal glucose absorption in vivo comprises two components: an active component, which saturates between 30 and 50 mm glucose, and a passive component, which increases in a broadly linear manner up to concentrations well in excess of 100 mm (Fullerton & Parsons, 1956; Manome & Kuriaki, 1961; Debnam & Levin, 1975; Ilundain et al. 1979; Ugolev et al. 1986; Lostao et al. 1991). At higher concentrations, the passive component is 3-5 times greater than the active component and is therefore likely to be the major pathway by which intestinal glucose absorption occurs during the assimilation of a meal. The active component is mediated by the Na⁺-glucose cotransporter, SGLT1. However, the mechanism, and even the existence, of the passive component has been a matter of debate for over a decade (for a review, see Kellett, 2001). On the one hand, Ferraris & Diamond have proposed that all glucose absorption can be explained solely in terms of the currently known kinetic properties of SGLT1 (Ferraris & Diamond, 1989, 1997; Ferraris et al. 1990). On the other, Pappenheimer & Reiss (1987; see also Pappenheimer, 1993, 1998) have proposed that the passive component of glucose absorption was the result of SGLT1-dependent paracellular solvent drag resulting from the glucose-induced dilatation or ‘opening’ of the tight junctions (Madara & Pappenheimer, 1987).

Recently, we have proposed that the passive component of glucose absorption in rat jejunum in vivo is in fact facilitated by GLUT2 (Corpe et al. 1996; Helliwell et al. 2000a,b; Kellett & Helliwell, 2000). The presence of glucose in the lumen causes the rapid activation and recruitment of the facilitative glucose transporter, GLUT2, to the brush-border membrane; regulation of the facilitated component correlates with the activation of protein kinase C (PKC) βII by glucose transport through SGLT1 and also involves mitogen-activated protein (MAP) kinase signalling pathways. The level of GLUT2 at the brush-border membrane is several times greater in vivo than in vitro, presumably because of the presence of endogenous hormones (as well as glucose and other nutrients) in vivo that activate PKC βII. When jejunum is excised for measurements of glucose uptake in vitro, part of the GLUT2 is lost from the brush-border membrane within minutes. We suggest this is one reason why the passive component was more readily seen in vivo, but less readily in vitro. Of particular note the facilitated component in our experiments was SGLT1 dependent. This was demonstrated by the fact that activation of PKC βII showed a saturation response and had the same K_m as glucose absorption by SGLT1. Moreover, inhibition of SGLT1-mediated absorption with phloridzin resulted in inhibition, not only of the active component, but also of most of the facilitated component, concomitant with the loss of GLUT2 from the brush-border membrane.

In the earlier work cited above, however, it is clear that the passive component in vivo does not show dependence on SGLT1. For example, the original demonstrations of a passive component rested primarily on the fact that when the active component was inhibited by phloridzin, a large passive component remained, which must be independent of SGLT1. What then is the reason for the difference between our findings showing dependence on SGLT1 and those of earlier perfusion studies showing independence? Debnam & Levin (1975) used an in vivo perfusion technique based closely on that of Sheff & Smyth (1955); the technique used a gas lift to recirculate the luminal sugar perfusate and had a pressure head of 25 cm. The precise flow rate was not given, but is likely to have been something of the order of 6-7 ml min⁻¹ in such an apparatus. A characteristic of such a preparation is that the intestine becomes blown up and distended; it also becomes white, possibly indicating that the circulation of blood round the intestine is at least partially restricted. Such a preparation contrasts sharply with the single-pass preparation used in our previous work to demonstrate that the passive component of glucose absorption is dependent on the transport of glucose through SGLT1. The latter preparation has a pressure head of zero and a low flow rate of 0.75 ml min⁻¹ maintained by a peristaltic pump; the jejunum is not distended in any way and remains red. We therefore set out to answer the question of whether the difference in perfusion techniques might be responsible for the difference in the dependence of the passive component of absorption on SGLT1. The answer to this question is vitally important for the experimental design of future studies of the mechanism and regulation of the passive component.

METHODS

Animals

All procedures used conformed to the UK Animals (Scientific Procedures) Act 1986. Male Wistar rats (240-260 g) were fed ad libitum on standard Bantin & Kingman rat and mouse diet with free access to water.

Perfusion of the jejunal loops in vivo

High mechanical stress perfusion

Rats were anaesthetised by an intraperitoneal injection of a mixture of 1.0 ml Hypnorm (Janssen, UK) and 0.4 ml Hypnovel (Roche, UK) per kilogram body weight. The abdomen of an anaesthetized rat was opened by mid-line incision and a 20 cm loop of jejunum starting at a point 5 cm below the ligament of Treitz was defined by two incisions across half the diameter of the intestine. The loop was cleaned by flushing with warm, modified Krebs-Henseleit buffer (containing (mm): 120 NaCl, 4.5 KCl, 1.0 MgSO₄, 1.8 Na₂HPO₄, 0.2 NaH₂PO₄, 1.25 CaCl₂, 25 NaHCO₃) adjusted to pH 7.4 by gassing (19:1, O₂:CO₂) before use. The loop was cannulated, connected to the luminal perfusion circuit and the perfusion commenced immediately with perfusate (50 ml modified Krebs-Henseleit buffer, 37 °C), which was thoroughly oxygenated in its reservoir with O₂:CO₂ (19:1). The perfusate was recirculated with a Minipuls IV peristaltic pump (Gilson, UK) at a rate of 7.0 ml min⁻¹ and segmented with gas at a rate of 2.0 ml min⁻¹ to minimise the effect of unstirred layers. The pump was on the inflow side of the jejunal segment and the pressure head of the perfusate reservoir was 18 cm. The choice of 7.0 ml min⁻¹ for the flow rate approximated to the rate used by Debnam & Levin (1975; E. S. Debnam, personal communication). The jejunum was therefore distended and under mechanical stress. The jejunum also appeared white in colour, suggesting that the circulation of the blood supply was inadequate; this was confirmed by the fact that as soon as the pressure head was released, the jejunum rapidly became pink again. In these two respects, then, the perfusion closely mimicked earlier perfusion systems in which recirculation of perfusate was driven by a gas lift (Debnam & Levin, 1975).

The system had two perfusate reservoirs to permit a paired comparison between a control and an experimental perfusion period for a single loop in which glucose absorption was inhibited during the experimental period by the addition of drug. Thus the first reservoir contained 50 mm glucose, while the second contained 50 mm glucose plus either 0.5 mm phloridzin, 1.0 mm phloretin or 0.2 mm cytochalasin B. Since the drugs are not readily soluble in water, they were first dissolved in a small volume of DMSO, not greater than 200 μl, to give a final maximum concentration of 0.4 % (v/v). The stock solution was then added drop-wise to perfusate at 37 °C with rapid stirring, so that no precipitation of drug occurred. Control experiments showed that vehicle alone at the concentrations used had no significant effect on glucose absorption. Water transport was measured directly from the height of the perfusate level in the narrow luminal reservoir; constant osmolarity was maintained between solutions of different glucose concentration by the addition of mannitol, so as to eliminate effects of differences in osmolarity between samples on water and glucose transport. At the end of the perfusion, whilst still anaesthetised, the rat was killed by exsanguination.

Samples (50 μl) of perfusate from the luminal reservoirs were taken at 5 min intervals for glucose analysis. Glucose concentrations were determined with a COBAS MIRA analyser (Roche, UK) by the GOD-PERID method. Concentrations were converted into amounts of glucose in the luminal perfusate after correction for changes in volume caused by water transport. Dry weights were determined by drying the perfused loop to constant weight at 110 °C in a drying oven overnight. Absorption was measured as the rate of disappearance of glucose from the luminal perfusate under steady-state conditions and expressed in μmol min⁻¹ (g dry wt)⁻¹.

Perfusions to check the viability of the preparation were performed in which jejunum was perfused with 50 mm glucose for the entire length of the control and experimental perfusion periods. After the initial steady-state absorption rate had been achieved in each case these perfusions showed no change in absorption rate confirming that preparations were viable for the whole of the perfusion period.

Low mechanical stress perfusion

Jejunal loops were perfused in vivo with a single pass of perfusate in which a gas-segmented flow system was again used to disrupt the unstirred layer. The jejunum of a rat was cannulated as described above and perfusion commenced immediately. The flow rate of perfusate was controlled by peristaltic pump at 0.75 ml min⁻¹ and that of gas at 0.38 ml min⁻¹; the luminal reservoir was at the same height as the jejunum and the effluent outlet was below the jejunum, so that there was no pressure head. A major difference from the recirculated, high flow rate perfusion was, therefore, that the jejunum was never distended and so was not under mechanical stress; it also remained pink in colour throughout the perfusion, indicating adequate circulation of the blood supply. As above, the system had two perfusate reservoirs to permit a paired comparison between a control and an experimental perfusion period with inhibitors for a single loop. Both reservoirs contained tracer amounts of [³H]inulin (Amersham Life Science) to permit the determination of water transport; constant osmolarity between samples was maintained with mannitol. Control experiments showed that the vehicle in which drugs were dissolved had no significant effect on the rate of glucose absorption in this preparation.

Samples (50 μl) were taken from the perfusate effluent as a function of time and triplicate samples were taken from the reservoirs for analysis of glucose. Concentrations were corrected for the changes in water volume caused by water transport. The rate of glucose transport, termed v and expressed in μmol min⁻¹ (g dry wt)⁻¹, was calculated from the equation:

where Δc is the difference in concentration between the eluate and the perfusate reservoir, F is the flow rate and w is the dry weight of the perfused intestinal loop. The viability of the preparation for the whole of the perfusion period was confirmed in additional perfusions.

Membrane vesicle preparation

Jejuna from two rats were perfused as described above. Perfusions for vesicle preparation were terminated in the steady-state range of glucose absorption. The timing of termination was such as to get as direct a correspondence as possible between rates of transport and extent of trafficking; perfusions for vesicle preparations were therefore terminated at a time point corresponding to half the time period over which the average rate of transport was determined. Every stage of the vesicle preparation was performed at 0-4 °C to prevent changes in trafficking after the intestine had been excised, as described by Helliwell et al. (2000a). Thus immediately after perfusion, each jejunum was flushed with ice-cold buffered mannitol (20 mm imidazole buffer, pH 7.5 containing 250 mm mannitol and 0.1 mm phenylmethanesulphonyl fluoride (PMSF)) whilst maintaining the pressure at the level appropriate to the perfusion. The jejunum was then placed on an ice-cold glass plate and slit longitudinally so that the muscle of the jejunum flattened out on to the cold plate. Mucosal scrapings were taken with an ice-cold glass slide and homogenised immediately at 4 °C in buffered mannitol using a Kinematica Polytron homogeniser (4 × 30 s bursts using the large probe at setting 7). The rest of the preparation and its detailed characterisation for purity were as described by Corpe et al. (1996). Enrichment of sucrase activity in these highly purified preparations ranged from 16- to 20-fold; there was no significant enrichment of Na⁺,K⁺-ATPase activity.

Western blotting

SDS-PAGE and Western blotting were also performed as described previously using ECL (enhanced chemiluminescence) detection (Corpe et al. 1996). Immunoblotting was performed using polyclonal antibodies raised in rabbit to the C-terminal sequence of GLUT2. SGLT1 antibody raised in rabbit was a gift from Professor M. Kasahara (Tokyo) and antibody to the eighteen amino acids at the C-terminus of PKC βII was purchased from Autogen Bioclear (UK). Neutralisation of either GLUT2 antibody, PKC βII antibody or SGLT1 antibody with the corresponding peptide prior to blotting abolished labelling, confirming the specificity of the antibodies. Quantification of Western blots was performed using a Flowgen AlphaImager 1200 analysis system (Alpha Innotech Corporation, CA, USA). The level of GLUT2 determined in vesicle preparations from jejunum was expressed relative to that in control preparations. A liver GLUT2 standard (postnuclear membranes; prepared from rats that were anaesthetised and killed as described above) was routinely used in blotting experiments. The linear range of intensity response in ECL photographs was established using a 20-fold range of the amount of an actin standard (2-40 μg). After background correction, the response was linear (correlation coefficient 0.996) for integrated density values ranging from 13 024 to 465 029. As far as possible, exposures for measurement of GLUT2 relative amounts were such that the intensity values fell within the middle third of the linear response range. The same loading of 15 μg protein was used for all samples. Comparison of relative levels of GLUT2 was made on a protein basis in order to minimise potential complications that might be caused by the trafficking of other proteins in response to the same stimuli that affect GLUT2 trafficking.

Statistical analysis

Values are presented as means ± s.e.m. and were tested for significance using Student's t test.

RESULTS

Low mechanical stress perfusion studies

When rat jejunum was perfused luminally in single-pass mode in vivo with 50 mm glucose, a steady-state rate of glucose absorption was achieved after 10-15 min (Fig. 1). Addition of either phloretin (1 mm) or phloridzin (0.5 mm) rapidly inhibited absorption by 68 ± 2 and 70 ± 2 %, respectively (both P < 0.001). In whole intestine, phloretin and phloridzin are specific for GLUT2 and SGLT1, respectively (see below). Moreover, as noted above, we have shown previously that inhibition of absorption by phloretin in the luminal perfusate reflects the presence of GLUT2 at the brush-border membrane. Accordingly, we can therefore determine that at 50 mm glucose the percentage of absorption occurring as a phloretin-sensitive passive component is ≈68 % and that the ratio of the passive to the phloretin-insensitive, active component is 2.17 ± 0.07 (Fig. 2A). However, conventionally, the passive component has been determined as the phloridzin-insensitive component, in which case the phloridzin-insensitive, passive component accounts for ≈30 % of absorption and the ratio of the passive to active component is 0.43 ± 0.08 (Fig. 2A). The rate of the ‘passive’ component determined in low stress perfusions was 20.7 ± 0.4 μmol min⁻¹ (g dry wt)⁻¹ using phloretin and 9.2 ± 0.5 μmol min⁻¹ (g dry wt)⁻¹ using phloridzin.

Figure 1. Inhibition of brush-border glucose absorption in vivo by phloretin and phloridzin.

Rat jejunum was perfused for 40 min in vivo with 50 mm glucose using the low mechanical stress, single-pass perfusion system as described in Methods. After 40 min, the perfusate reservoir was switched to one containing 50 mm glucose and either 1.0 mm phloretin (•) or 0.5 mm phloridzin (▪). Data are presented as means ±s.e.m.; n = 4 for each perfusion condition.

Figure 2. The relative magnitudes of the passive and active components of glucose absorption determined by different perfusion techniques.

A, rat jejunum was perfused with 50 mm glucose in the absence (control) and presence of either phloretin (1.0 mm) or phloridzin (0.5 mm) using the low mechanical stress, single-pass perfusion technique as described in the legend to Fig. 1. The rate of the apparent ‘active’ component was determined either as a phloretin-insensitive component (□) or as a phloridzin-sensitive component (▪). The rate of the apparent ‘passive’ component was determined as a phloretin-sensitive component (□) or as a phloridzin-insensitive component (▪). The rates were then expressed as a fraction of the control rate. The right hand pair of columns show the ratio of the passive to the active component; □, determinations using phloretin; ▪, determinations using phloridzin. For each perfusion condition n = 4. Data are presented as means ±s.e.m. Different letters above the columns denote a significant difference from the corresponding control (a) and from each other (a, b) (P < 0.001) either for rate or for ratio data. B, rat jejunum was perfused with 50 mm glucose in the absence (control) and presence of either phloretin (1.0 mm) or phloridzin (0.5 mm) using the high mechanical stress technique described in Methods. The rates of the ‘active’ and ‘passive’ components were determined, expressed and presented as in A; n = 4. C, rat jejunum was perfused with 50 mm glucose in the absence (control) and presence of either phloridzin (0.5 mm), cytochalasin B (0.2 mm) or the two together using the high mechanical stress technique described in Methods. The rates of the ‘passive’, phloridzin-insensitive component (pdz), the ‘active’ cytochalasin B-insensitive component (cytB) and the residual rate in the presence of both phloridzin and cytochalasin B (pdz + cytB) are presented as a fraction of the control rate. The rates were determined, expressed and presented as in A; n = 4.

These widely divergent and seemingly contradictory results bring into the question the use of the term ‘passive’ (see Discussion); however, they can be explained by our previous observation that the level and intrinsic activity of GLUT2 at the brush-border membrane is dependent on the transport of glucose by SGLT1 in a single-pass perfusion with low mechanical stress. This fall in level is demonstrated in Fig. 3A, for when glucose transport through SGLT1 was inhibited by phloridzin, the level of GLUT2 was diminished by 29 ± 5.0 % (P < 0.05). A residual amount of GLUT2 that appeared independent of SGLT1 remained at the brush-border membrane. The decrease in level of GLUT2 correlated with a fall in the level of PKC βII by 29 ± 5.2 % (P = 0.01). There was no change in the level of SGLT1 (Fig. 3A).

Figure 3. The effect of phloridzin on the levels of GLUT2, SGLT1 and PKC βII in low and high mechanical stress perfusions with 50 mm glucose.

A, rat jejunum was perfused using the low mechanical stress perfusion technique for 60 min with 50 mm glucose alone (three preparations in lanes 1-3) or for 20 min with 50 mm glucose alone and then for a further 40 min with 50 mm glucose plus 0.5 mm phloridzin (three preparations in lanes 4-6); lane 7 shows a GLUT2 control sample of a post-nuclear membrane liver preparation. After perfusion, brush-border membrane vesicles were prepared; vesicle protein (15 μg) was then separated on 10 % SDS-PAGE gels, transblotted onto PVDF and immunoblotted for GLUT2, SGLT1 and PKC βII. For full details, see Methods. B, rat jejunum was perfused using the high mechanical stress perfusion technique for 30 min with 50 mm glucose alone (three preparations in lanes 1-3) or for 30 min with 50 mm glucose alone and then for a further 40 min with 50 mm glucose plus 0.5 mm phloridzin (three preparations in lanes 4-6). The jejunal segment was excised before releasing the distension pressure to maintain the conditions in vivo. Other details as for A.

High mechanical stress perfusion studies

Jejunum was perfused for 30 min (control) before switching to an otherwise identical perfusate containing inhibitor for a further 40 min perfusion. The average control rate with 50 mm glucose was 42.0 ± 3.7 μmol min⁻¹ (g dry wt)⁻¹ in high stress perfusions (n = 7) compared with 30.2 ± 0.4 μmol min⁻¹ (g dry wt)⁻¹ in low stress perfusions (n = 9, P < 0.001). Figure 2B shows that 0.5 mm phloridzin inhibited glucose absorption by 51.4 ± 4.2 % (P < 0.001) and that the ratio of the phloridzin-insensitive, passive component to the active component was 0.95 ± 0.10, that is, comparable to the value of 1.3 reported by Debnam & Levin (1975). Moreover, 1 mm phloretin inhibited glucose absorption by 48.5 ± 8.1 % (P < 0.001), giving a value for the ratio of the passive to the active component of 0.94 ± 0.17. Thus the phloretin-sensitive passive component was the same magnitude as the phloridzin-insensitive passive component. Experiments in which cytochalasin B (0.2 mm) was used to inhibit the passive component gave a result identical, within experimental error, to that of phloretin (Fig. 2C); the cytochalasin B-sensitive component represented 48.7 ± 7.6 % (P < 0.001) of the total rate and the ratio of the passive to active components was 0.95 (data not shown). Moreover cytochalasin B and phloridzin together inhibited 91.8 ± 6.7 % (P < 0.001) of total absorption. Taken together, these results demonstrate that, when jejunum is perfused with 50 mm glucose in high stress perfusions, carrier-mediated transport accounted within experimental error for total absorption and that the passive component is independent of the active component.

The implication that SGLT1-dependent recruitment of GLUT2 to the brush-border membrane does not occur at 50 mm glucose in high stress perfusions is confirmed by the Western blot in Fig. 3B. Inhibition of SGLT1 with phloridzin had no effect on the levels of either GLUT2 or PKC βII. The level of SGLT1 was also unaltered. Consistent with this view, there was no difference in GLUT2 level between perfusions with 10 mm and 50 mm glucose; nor was there any change in the levels of PKC βII or SGLT1 (Fig. 4).

Figure 4. The effect of glucose on the levels of GLUT2, SGLT1 and PKC βII in high mechanical stress perfusions.

Rat jejunum was perfused using the high mechanical stress perfusion technique for 60 min with either 10 mm (three preparations in lanes 1-3) or 50 mm glucose (three preparations in lanes 4-6); lane 7 is a liver control preparation. Vesicles were prepared and Western blotted as for Fig. 3.

Comparison of brush-border membrane GLUT2 levels at 50 mm glucose showed that levels observed in high stress perfusions were significantly lower (by 36 ± 3.0 %, P < 0.05) than those in the low stress perfusions (Fig. 5). Moreover, the change in GLUT2 level correlated with that of PKC βII, the level of which was also diminished by 31 ± 2 % (P < 0.001) in the recirculated preparation. There was no difference in the level of SGLT1.

Figure 5. Comparison of the levels of GLUT2, SGLT1 and PKC βII in low and high stress perfusions.

Rat jejunum was perfused for 60 min with 50 mm glucose using either the low stress perfusion technique (three preparations in lanes 1-3) or the high stress perfusion technique (three preparations in lanes 4-6); lane 7 is a liver control preparation. Vesicles were prepared and Western blotted as for Fig. 3.

DISCUSSION

The passive component of absorption is independent of the active component in high stress perfusions and absorption is effectively all carrier mediated

Using single-pass, low mechanical stress perfusions, we have obtained evidence that the passive component of intestinal glucose absorption is mediated by the glucose-induced activation and/or recruitment of GLUT2 to the brush-border membrane. The level of GLUT2 correlates with the activation of PKC βII, which in turn is dependent on the transport of glucose through SGLT1 (Helliwell et al. 2000a; Kellett & Helliwell, 2000). Inhibition of SGLT1 with phloridzin therefore diminishes the level of GLUT2 at the brush-border membrane. There appear to be two pools of brush-border membrane GLUT2; GLUT2 appears in Western blots as a tightly spaced doublet and it is the lower band which is involved in rapid trafficking (see for example Fig. 6 of Kellett & Helliwell, 2000). Thus one pool is readily mobilisable and traffics to and from the membrane on a time scale of minutes; the other is relatively static and appears to account for about 20-25 % of the maximum value we have seen either at 100 mm glucose or in the presence of fructose plus 4-βphorbol 12-myristate, 13-acetate (PMA). This relatively static pool therefore seems to represent a base level below which GLUT2 does not fall, unless, perhaps, exceptional circumstances occur. In what follows, we will often state that recruitment of GLUT2 to the brush-border membrane in low stress perfusions is dependent on SGLT1. In doing so, we refer implicitly to the readily mobilisable part of brush-border membrane GLUT2, but remember that this is the majority of, but not all, GLUT2.

This investigation was prompted by the realisation that earlier studies, such as that of Debnam & Levin (1975) using a high stress perfusion system, showed no dependence of the passive component of glucose absorption on SGLT1; in contrast, our studies using a low stress perfusion system had shown that a large part of the passive component was dependent on SGLT1 (Kellett & Helliwell, 2000), so that the ratio of the ‘passive’ to ‘active’ component was about 5 times greater when determined with phloretin than with phloridzin. Using a high stress preparation, which becomes distended and white, we were able to reproduce the results of Debnam & Levin (1975). In both cases the ratio of passive to active components is close to 1 and the former is not dependent on the latter. The independence of the passive and active components in high stress perfusions at 50 mm glucose was demonstrated by the selective inhibition of either the active component with phloridzin or the passive component with phloretin or cytochalasin B. In low stress perfusions, we have also previously used replacement of Na⁺ in the perfusate to abolish the Na⁺ gradient and so inhibit the active component. Thus the assignment of the magnitudes of the active and passive components has been achieved in four separate ways. Under appropriate conditions (when allowance is made for trafficking of GLUT2 in low stress perfusions), each method gives the same answer.

The independence of the passive and active components in high stress perfusions reflected the fact that SGLT1-dependent recruitment of GLUT2 to the brush-border membrane did not occur; thus phloridzin did not diminish the brush-border membrane level of GLUT2 or PKC βII (Fig. 3B). This behaviour contrasted with that in low stress perfusions, where phloridzin diminished the levels of both brush-border membrane GLUT2 and PKC βII at 50 mm glucose (Fig. 3A). Moreover, increasing glucose from 10 to 50 mm did not increase the levels of either GLUT2 or PKC βII at the membrane (Fig. 4). In neither set of experiments was any change in the level of SGLT1 observed.

In low stress perfusions, the rates of the GLUT2- and SGLT1-mediated components were 20.7 and 9.2 μmol min⁻¹ (g dry wt)⁻¹, respectively. Since phloridzin inhibited total absorption by 70 %, we can calculate that about 44 % of the GLUT2 component remains. However, in the presence of phloridzin, the GLUT2 level is 72 % of that in its absence. The relative activity of GLUT2 is therefore about 61 % of that in the presence of 50 mm glucose alone. This calculation is in keeping with our original observation that SGLT1-dependent activation of the intrinsic activity of GLUT2 also occurs in addition to its trafficking. In high stress perfusions, the active component is 21.6 compared with 9.8 μmol min⁻¹ (g dry wt)⁻¹ in low stress perfusions despite the fact that the level of SGLT1 remains constant. Similarly, the facilitated component remains unchanged at ≈21 μmol min⁻¹ (g dry wt)⁻¹, despite the fact that the GLUT2 level is diminished by about 36 %. Thus the intrinsic activities of SGLT1 and GLUT2 appear to be increased. The mechanisms of the changes in intrinsic activity are unknown. However, we have reported that GLUT2 intrinsic activity can be altered as much as ninefold by manipulation of the activities of extracellular signal-regulated kinase (ERK)-, p38- and PI-3kinase-dependent pathways and is specifically activated by stress-activated pathways (Helliwell et al. 2000b). On the other hand, Thorens et al. (1996) have reported that phosphorylation of C-terminal serine/threonine residues diminishes the activity of GLUT2 in pancreatic β-cells to about 52 %. SGLT1 is phosphorylated and activated in response to stress (Ishikawa et al. 1997).

Figure 1 shows that the kinetics of phloretin inhibition in low stress perfusions are slower than those of phloridzin. This difference may reflect the fact that phloretin acts at the endofacial side of facilitative transporters whereas phloridzin most probably acts at the exofacial side of SGLT1. The rapid inhibition by phloridzin demonstrates that inactivation and trafficking away of GLUT2 from the brush-border membrane occur extremely quickly. The half-time (t_1/2) must be less than 3 min in accordance with our previous conclusions from the effects of intestinal excision on GLUT2 levels (Helliwell et al. 2000a); hence only a single phase is seen when perfusate samples are taken every 5 min.

What then is the cause of the lack of GLUT2 trafficking? In a high stress perfusion in vivo, the jejunum is blown up and distended. It is also white, partly because of the distension, and possibly also because the circulation to the jejunum may be restricted to some extent by high pressure; indeed, when the perfusion is ended, the jejunum immediately becomes pink again. Our previous work suggests that endogenous hormones and other nutrients play a role in glucose-induced trafficking (Helliwell et al. 2000a). Any restriction of blood supply in high stress perfusions would therefore be likely to be a factor in the blocking of trafficking. It seems likely that mechanical distension of the jejunum may also be a factor, by preventing essential cytoskeletal rearrangement induced by glucose (Madara & Pappenheimer, 1987).

In both low stress and high stress perfusions, the sum of the active SGLT1-mediated and the facilitated GLUT2-mediated components determined by selective inhibition accounts, within experimental error, for total absorption. Confirmation of this conclusion is provided in Fig. 2C, which shows that phloridzin and cytochalasin B together effectively knock out absorption completely in high stress perfusions. The residual absorption is 8.1 ± 5.0 % of the total absorption. On the time scale that is important immediately after a meal, absorption is therefore carrier mediated and transcellular.

Technical considerations in determination of sugar absorption components in perfusion experiments

The specificity and site of action of inhibitors

An inevitable concern, when using an inhibitor of facilitative transporters such as phloretin or cytochalasin B in perfusion experiments, especially in vivo, is that the inhibitor might reach the basolateral side of the jejunum and inhibit GLUT2-mediated exit. Indeed, when a chow diet is supplemented with large amounts of phloretin (22 mg in a 14 g meal), conjugated phloretin metabolites can be detected in plasma after 4 h, although the concentration of native phloretin is negligible; equivalent doses of phloridzin result in lower plasma levels of phloretin metabolites after 4 h (Crespy et al. 2002). However, perfusion of the entire jejunum and ileum for 30 min with phloretin or phloridzin at concentrations up to 0.15 mm does result in detectable concentrations of metabolites within mucosa (Crespy et al. 2001). The conclusion from these studies is therefore that phloretin (and phloridzin) gets across the intestine at significant levels on a time scale that is much slower than that used in our perfusions and only when its luminal load is very much higher than that used in our experiments. Furthermore, it gets across effectively only as metabolites, which are conjugated and therefore likely to be inactive against facilitative transporters. This conclusion accords well with our observation that four different methods of determining the magnitude of passive and active components of brush-border membrane absorption give the same results, namely the use of phloretin, cytochalasin B, phloridzin and Na⁺ replacement, which act quite differently. Two of those methods target SGLT1, which is exclusively at the brush-border membrane, and one of them, Na⁺ replacement, does not involve inhibitors. Moreover, we have shown that the use of phloretin to inhibit fructose absorption in perfused intestine of diabetic rats gives the same results as the use of glucose to inhibit fructose transport in vesicles, an experiment that also does not involve inhibitors (Corpe et al. 1996).

Furthermore, the fact that all these approaches give the same results also means that phloretin has no effect on SGLT1 in intact intestine. These conclusions are supported by the fact that the phloridzin-sensitive component shows the same kinetics as the phloretin-insensitive component. When studying the hydrolysis of phloridzin to phloretin in fragments of intestine, Diedrich long ago (1968) noted that ‘the fact remains that phloretin is relatively impotent when tested as an inhibitor in intact intestine’. This observation, which was made at low concentrations of glucose with in vitro preparations where SGLT1 transport predominates, has been confirmed by several workers since. We have previously excluded, by HPLC analysis, the possibility that phloretin derived from phloridzin hydrolysis could be a factor in our experiments (Kellett & Helliwell, 2000).

We have already noted that when jejunum is perfused in vitro with 5 mm fructose (Helliwell et al. 2000a,b), no fructose appears on the serosal side, it all being metabolised. The behaviour of fructose absorption and GLUT2 trafficking revealed by the use of phloretin in those studies parallels that for glucose absorption; this supports the conclusion that the inhibitors in the luminal perfusate act at the brush-border membrane in short-term perfusions and that phloretin acts on GLUT2 in glucose perfusions, since fructose is not a substrate for SGLT1.

It is well known that under some circumstances cytochalasins (E, D and presumably B) can disrupt tight junctions and enhance paracellular permeability. However, even with cytochalasin E present on both sides of an intestinal sheet in an Ussing chamber, the effect is slow (≈1 h) and leads anyway to an increase in permeability (Perez et al. 1997; see also experiments with cytochalasin D by Madara et al. 1986). In the present experiments, inhibition is about half-maximal within 5 min and effectively complete at about 15 min (Fig. 2C).

Some inhibitors, especially phloretin, are not readily water soluble and must first be dissolved in organic solvents such as DMSO or ethanol. The final concentrations of solvent in perfusates range from 0.1 to 0.4 % (v/v). Even at the highest concentration, there was no effect of DMSO on glucose absorption in our experiments in vivo. This finding accords with the fact that glucose transport in whole intestine is relatively insensitive to solvents, much less so than in vesicles. For example, it takes as much as 5 % (v/v) ethanol to achieve 27 % inhibition of glucose absorption in vivo (Debnam & Mazzanti, 1987).

The effects of anaesthesia and surgery on glucose absorption

Our conclusion that non-carrier-mediated absorption is minimal accords with several reports in which paracellular flow (passive clearance of mannitol and/or l-glucose) has been measured directly (Fine et al. 1993; Schwartz et al. 1995; Uhing & Kimura, 1995; Lane et al. 1999; Fihn et al. 2000). These studies also raise questions concerning effects of anaesthetic on glucose absorption. For example, in a widely quoted study, Uhing & Kimura (1995) investigated the transepithelial absorption of 3-O-methylglucose at concentrations up to 400 mm, that is, at concentrations far in excess of the 30-50 mm required to saturate SGLT1. Using rats implanted with chronic, in-dwelling catheters in the aorta, portal vein and duodenum, they observed that acute effects of anaesthesia and surgery diminished active but not passive transport processes and concluded that the importance of passive transport processes in the acute, anaesthetised, surgically manipulated rat is overestimated. In our present and previous studies with anaesthetised rats, ‘passive’ non-carrier-mediated absorption was clearly not overestimated; the question therefore is whether active absorption was underestimated. Discussions on this question are complicated, however, by the past use of the terms ‘passive’ and ‘active’ absorption in ways that are no longer sufficiently precise. Thus Uhing & Kimura (1995) defined ‘passive’ as ‘paracellular’ or ‘non-carrier mediated’ and ‘active’ as either ‘the difference between total absorption and passive absorption’ or ‘phloridzin-sensitive 3-O-methylglucose absorption’. Since phloridzin inhibited 100 mm 3-O-methylglucose absorption by 79 % and l-glucose absorption represented 12 % of total absorption, they reported that active transport effectively accounted for all absorption. Uhing & Kimura (1995) concluded, as we do, that non-carrier-mediated transport is minimal; but the assumption that either of the other definitions designates an active process no longer applies. Thus 3-O-methylglucose is transported by both SGLT1 and GLUT2; and, as noted, the phloridzin-sensitive component comprises not only SGLT1-mediated transport but that part of GLUT2-mediated transport regulated by SGLT1.

The inference that both carrier-mediated processes (SGLT1 and GLUT2) are diminished by anaesthesia and surgery in rat is strongly supported by the experiments described in Table 2 of Ugolev et al. (1986). These workers measured rates of glucose absorption in chronic, unanaesthetised rats implanted with jejunal perfusion cannulae and compared them with rates measured in chronic, anaesthetised rats and acute, anaesthetised rats which had just undergone laparotomy (the same as in our experiments). They found that overall rates of glucose absorption in chronic, anaesthetised rats and acute, anaesthetised, laparotomised rats were about 60 and 40 %, respectively, of those observed in chronic, unanaesthetised rats. The corresponding values for fructose absorption were 32 and 25 %. The effects of anaesthesia and surgery are therefore not confined to SGLT1, facilitative transporters are also affected. Of particular note, the rate of glucose absorption in all three types of experimental rat increased significantly from 55 to 110 mm glucose, that is, the same behaviour as in our experiments and those of other workers. Anaesthesia and surgery therefore reduce overall rates, but do not alter the general form of concentration dependence, which shows rates increasing at concentrations above those required to saturate SGLT1.

Evidence that active (SGLT1-mediated) absorption in fact accounts for only about one-third of total absorption at higher concentrations is provided by Fig. 11 of Ugolev et al. (1986); the data show that absorption of 55 mm glucose diminished rapidly by an average of 30 %, when Na⁺ was replaced with mannitol and Na⁺,K⁺-ATPase was simultaneously inhibited with ouabain and strophanthin K in low stress perfusions of chronic, unanaesthetised rats. This compares closely with our observation of a rapid 29 % inhibition when Na⁺ was replaced with choline in acute, low stress perfusions at 50 mm glucose; continued perfusion with choline resulted in a slow loss of GLUT2 from the brush-border membrane and concomitant inhibition of absorption (Kellett & Helliwell, 2000). The magnitude of the Na⁺-dependent component matched those of the phloretin-insensitive and cytochalasin B-insensitive components. Thus in both our work and their work the ratio of the ‘passive’ to ‘active’ components was just over 2 in low stress perfusions. Table 7 of Ugolev et al. (1986) also shows that phloridzin inhibits the rate of absorption of 110 mm glucose by 75 %; this value compares with the 71 % inhibition at 100 mm glucose (Kellett & Helliwell, 2000) and 79 % inhibition at 100 mm 3-O-methylglucose (Uhing & Kimura, 1995). Thus three different laboratories have obtained similar data in rather different ways over a 20 year time scale. As noted earlier, under low stress conditions, phloridzin rapidly inhibits not only SGLT1, but SGLT1-dependent GLUT2-mediated absorption. Ugolev et al. (1986) summarised 10 years research as follows: ‘under physiological conditions the two systems of glucose transport, Na⁺-dependent and Na⁺-independent, function. The first one is less potent, but more resistant to experimental influences.’ Expressed in our terms, the reduction in rates caused by anaesthesia and surgery, if anything, results in underestimation of the GLUT2 contribution.

In unanaesthetised dog, paracellular flow or non-carrier-mediated absorption is also minimal (Lane et al. 1999). However, in that study there is a major component of d-glucose absorption (not commented on) that cannot be mediated by SGLT1 unless the kinetic properties of dog SGLT1 are very different from those of other animal SGLT1 transporters. Table 1 of Lane et al. (1999) records that the percentage of glucose absorbed was unchanged from 50 to 150 mm glucose. All things being equal, that means (to a first approximation) that the absolute rate is increased 3-fold, yet SGLT1 saturates at 30-50 mm glucose; moreover, part of the absorption below 50 mm glucose will be mediated by the component that predominates above 50 mm. This behaviour parallels exactly that seen in anaesthetised rats, which we have now attributed to GLUT2. Finally, in view of a recent report that GLUT2 could not be detected at the brush-border membrane in human intestine (Dyer et al. 2002), it should be noted there are several reports in the literature which show that in human (unanaesthetised) non-carrier-mediated or paracellular transport is minimal compared with carrier-mediated transport. Just as for animals, there has been discussion about the validity of absolute rates of glucose absorption, but none about concentration dependence, which shows that absorption continues to increase well beyond concentrations required to saturate SGLT1 (Holdsworth & Dawson, 1964; for a review, see Pappenheimer, 1993).

Water absorption

The central tenet of the theory of paracellular flow is that osmotic gradients generated by SGLT1-dependent concentration of glucose and Na⁺ in the intracellular spaces induce water flow through tight junctions and cause bulk absorption of glucose by solvent drag. Some experiments, in which direct measurements using mannitol or l-glucose have shown that paracellular or non-carrier-mediated flow of sugars is minimal, have therefore been criticised on the grounds that water flow was minimal. Our experiments are free from that concern. Water transport at 50 mm glucose in high stress perfusions was clearly adequate at an average of 0.37 ± 0.014 ml min⁻¹ (g dry wt)⁻¹. When a combination of phloridzin and cytochalasin B was used to knock out carrier-mediated absorption, leaving a residual absorption of only 8.1 ± 5.0 % of total absorption, water absorption was inhibited ≈76 %. There was thus an association between water absorption and glucose absorption via SGLT1 and GLUT2 together equivalent to ≈374 mol water (mol glucose)⁻¹. This value compares with those of 264 and 424 mol water (mol glucose)⁻¹ reported for isotonic transport via human and rabbit SGLT1, respectively, expressed in oocytes (Zeuthen et al. 2001).

Diet and perfusion design are major determinants of what is observed experimentally

Throughout this and previous studies, we have not seen any change in SGLT1 levels. The diet on which the rats were maintained, Bantin & Kingman rat and mouse diet, upregulates active transport and therefore presumably SGLT1, compared with other diets (Kellett & Barker, 1989). This and the fact that our perfusions are on a very short time scale, so that synthesis is not involved, explain why there are no changes in SGLT1 levels. Nevertheless, we now see there is a new dimension to SGLT1, for it plays a vital role in short-term regulation of glucose absorption. We propose that it is the transport of glucose and Na⁺ through SGLT1 that controls the triggering of GLUT2 activation and trafficking. There is, of course, an established role for SGLT1 in long-term dietary regulation (Ferraris & Diamond, 1989, 1997; Ferraris et al. 1990; Lescale-Matys et al. 1993). There will therefore surely be cases where rapid changes in SGLT1 level are observed when rats have been maintained on diets that perhaps do not up-regulate SGLT1 (see for example up-regulation by glucagon-like peptide 2 (GLP-2); Cheeseman, 1997). Total cellular levels of GLUT2 are also regulated by long-term diet (Corpe et al. 1999). Thus we can envisage that investigation of how long-term diet affects short-term regulation of glucose absorption will become a substantial area of research.

The lesson of the present work is that observation of the facilitated component of glucose absorption depends crucially on the perfusion preparation used. Activation and trafficking of GLUT2 to the brush-border membrane is not only very rapid, but is also sensitive to the activities of a veritable network of intracellular signalling pathways, comprising the ERK MAP kinase and the p38 MAP kinase pathways as well as PKC- and PI-3kinase-dependent pathways. It appears likely that even seemingly small changes in perfusion technique might have significant effects on GLUT2 trafficking and activation. Changing to a different intestinal preparation, modifying an existing preparation or even variations in the manipulation of intestine during surgery all have the potential to exert an effect, possibly a major one. With so many potential variations in diet and perfusion design, it is likely that different workers will see different mechanisms at work: in some instances regulation will occur only by changes in either SGLT1 or GLUT2 activities, in others by changes in transporter levels, and in yet other cases by some combination thereof. It is essential to bear such considerations in mind when studying the facilitated component of glucose absorption.

There is general agreement in the literature that the term ‘active transport’ of glucose means secondary active transport mediated by SGLT1. However, reference to the ‘passive component’ of glucose absorption has been characterised by the almost interchangeable usage of terms such as ‘passive’, ‘diffusive’ and ‘non-carrier mediated’, which, as outlined above, has led in the past to significant confusion. Now that a clearly defined, GLUT2-based mechanism has been identified, we have recommended that the term ‘facilitated’ be used in preference, for it is both accurate and unambiguous (Kellett, 2000).