Effects of Artificial Sweeteners on Insulin Release and Cationic Fluxes in Rat Pancreatic Islets

Cell. Signal. Vol. 10, No. 10, pp. 727–733, 1998 Copyright  1998 Elsevier Science Inc.

ISSN 0898-6568/98 $19.00 PII S0898-6568(98)00017-5

Effects of Artificial Sweeteners on Insulin Release and Cationic Fluxes in Rat Pancreatic Islets Willy J. Malaisse,* Anne Vanonderbergen, Karim Louchami, Hassan Jijakli and Francine Malaisse-Lagae Laboratory of Experimental Medicine, Brussels Free University, 808 Route de Lennik, B-1070 Brussels, Belgium

ABSTRACT. ␤-l-Glucose pentaacetate, but not ␣-d-galactose pentaacetate, was recently reported to taste bitter and to stimulate insulin release. This finding led, in the present study, to the investigation of the effects of both bitter and non-bitter artificial sweeteners on insulin release and cationic fluxes in isolated rat pancreatic islets. Sodium saccharin (1.0–10.0 mM), sodium cyclamate (5.0–10.0 mM), stevioside (1.0 mM) and acesulfame-K (1.0–15.0 mM), all of which display a bitter taste, augmented insulin release from islets incubated in the presence of 7.0 mM d-glucose. In contrast, aspartame (1.0–10.0 mM), which is devoid of bitter taste, failed to affect insulin secretion. A positive secretory response to acesulfame-K was still observed when the extracellular K⫹ concentration was adjusted to the same value as that in control media. No major changes in 86Rb and 45Ca outflow from pre-labelled perifused islets could be attributed to the saccharin, cyclamic or acesulfame anions. It is proposed that the insulinotropic action of some artificial sweeteners and, possibly, that of selected hexose pentaacetate esters may require G-protein-coupled receptors similar to those operative in the recognition of bitter compounds by taste buds. cell signal 10;10:727–733, 1998.  1998 Elsevier Science Inc. KEY WORDS. Artificial sweeteners, Pancreatic islets, Insulin release

INTRODUCTION The ␤-anomer of l-glucose pentaacetate was recently reported to stimulate insulin release, whereas ␤-d-galactose pentaacetate fails to do so [1]. Because these esters enter islet cells and then undergo enzymatic hydrolysis [2], the insulinotropic action of ␤-l-glucose pentaacetate is apparently attributable to its hexose rather than its acetate moiety. Yet, the l-glucose residues of the ester are not metabolised in islet cells (unpublished observation). The idea comes to mind, therefore, that ␤-l-glucose pentaacetate, or possibly l-glucose, may be identified in islet cells by a process similar to that involved in taste recognition [3]. This hypothesis is compatible with the fact that ␤-l-glucose pentaacetate, but not ␤-d-galactose pentaacetate, displays a bitter taste [4]. In the light of these considerations, the present study aims mainly at re-evaluating the possible effect of several artificial sweeteners on insulin release from isolated rat pancreatic islets. MATERIALS AND METHODS Aspartame, sodium saccharin, sodium cyclamate and stevioside were obtained from Sigma (St. Louis, MO) and acesulfame-K from Hoechst (Frankfurt, Germany). *Author to whom all correspondence should be addressed. Received 29 November 1997; and accepted 1 February 1998.

All experiments were conducted in islets isolated by the collagenase procedure [5] from fed female Wistar rats (Proefdierencentrum, Heverlee, Belgium). The methods used to measure insulin output from either incubated [5] or perifused [6, 7] islets and the outflow of 86Rb [8] and 45Ca [9] from pre-labelled islets were previously described in the cited references. All results are expressed as mean values (⫾S.E.M.) together with either the number of individual observations (n) or degree or freedom (d.f.). The statistical significance of differences between mean values was assessed by use of Student’s t-test. RESULTS Control Values for Insulin Output In the absence of any sweetener, the release of insulin, over 90-min incubation, was not significantly different (P ⬎ 0.1) in the absence of d-glucose and at a low concentration (2.8 mM) of the hexose (Table 1, first line). A significant increase (P ⬍ 0.001) in insulin output above basal value was recorded, however, in the presence of 7.0 mM d-glucose. At 20.0 mM d-glucose, the secretory rate was about 16 times as high as basal release. Aspartame Aspartame (1.0–10.0 mM) failed to significantly affect insulin release, whether in the absence or presence of d-glu-

728

W. J. Malaisse et al.

TABLE 1. Effects of four artificial sweeteners on insulin release from islets incubated at increasing concentrations of D-glucose

Sweetener Nilb Aspartame

Saccharin

Cyclamate

Stevioside

mM Nil 1.0 5.0 10.0 Nil 1.0 5.0 10.0 Nil 1.0 5.0 10.0 Nil 0.1 0.5 1.0

Nil

2.7 mM

14.8 ⫾ 2.6 (56) 100.0 ⫾ 15.8 (29)

21.9 ⫾ 3.2 (100) 100.0 ⫾ 10.8 (40)

102.6 ⫾ 11.0 (29) 100.0 ⫾ 14.6 (47)

121.5 ⫾ 12.3 (40) 100.0 ⫾ 9.6 (57)

183.2 ⫾ 22.3 (46)§ 100.0 ⫾ 14.6 (47)

177.8 ⫾ 18.1 (58)k 100.0 ⫾ 20.6 (39)

184.8 ⫾ 19.6 (49)k

131.3 ⫾ 27.7 (37)

D-glucose

a

7.0 mM

20.0 mM

38.2 ⫾ 1.7 (279) 100.0 ⫾ 8.0 (42) 94.6 ⫾ 8.3 (44) 100.3 ⫾ 9.6 (44) 119.0 ⫾ 15.5 (44) 100.0 ⫾ 10.8 (38) 146.0 ⫾ 17.9 (30)* 183.3 ⫾ 21.6 (31)§ 267.9 ⫾ 26.7 (37)k 100.0 ⫾ 10.9 (41) 111.2 ⫾ 16.6 (32) 176.8 ⫾ 21.4 (32)k 236.4 ⫾ 17.4 (41)k 100.0 ⫾ 13.7 (27) 116.0 ⫾ 7.8 (26) 127.5 ⫾ 12.9 (28) 160.3 ⫾ 18.7 (27)†

231.7 ⫾ 10.4 (50) 100.0 ⫾ 4.5 (37) 100.8 ⫾ 4.5 (37) 100.0 ⫾ 4.0 (72) 134.4 ⫾ 4.8 (73)k 100.0 ⫾ 3.5 (50) 116.1 ⫾ 4.6 (52)‡

a Except for the absolute values shown in the first line, all secretory data are expressed relative to the mean control output of insulin found, within the same experiments, at the same concentration of d-glucose in the absence of any sweetener. b Overall mean absolute values (␮U/islet/90 min) for the output of insulin in islets incubated at increasing concentrations of d-glucose in the absence of any sweetener. * P ⬍ 0.05. † P ⬍ 0.02. ‡ P ⬍ 0.01. § P ⬍ 0.005. k P ⬍ 0.001, compared with the mean corresponding control value.

cose (Table 1). Even at 7.0 mM d-glucose, the secretory rate recorded in the presence of 10.0 mM aspartame averaged no more than 40.5 ⫾ 4.6 ␮U/islet/90 min (n ⫽ 44), compared with a control value of 36.6 ⫾ 3.6 ␮U/islet/90 min (n ⫽ 42).

absence of d-glucose and at 20.0 mM d-glucose, respectively. The concentration dependency of the secretory response to the cyclohexylsulphamate in islets exposed to 7.0 mM d-glucose is illustrated in Figure 1 (right panel). Although the increment in insulin output caused by 1.0 mM

Saccharin At a 10.0-mM concentration, sodium saccharin significantly augmented insulin release, whether in the absence or presence of d-glucose (Table 1). The absolute value for the saccharin-induced increment in insulin output averaged 13.6 ⫾ 4.3 (d.f. ⫽ 83; P ⬍ 0.005) and 140.1 ⫾ 19.5 (d.f. ⫽ 129; P ⬍ 0.001) ␮U/islet/90 min in the absence of d-glucose and at 20.0 mM d-glucose, respectively. In islets incubated at 7.0 mM d-glucose, as little as 1.0 mM saccharin was sufficient to cause a sizeable increase in insulin output. Figure 1 (left panel) illustrates the absolute values for the concentration dependency of the saccharin-induced increase in insulin output at 7.0 mM d-glucose. Cyclamate In the absence or presence of d-glucose, sodium cyclamate, when tested at a 10.0-mM concentration, augmented insulin release, such an effect failing to achieve statistical significance only in the presence of 2.7 mM d-glucose (Table 1). The absolute value for the drug-induced increment in insulin output averaged 12.4 ⫾ 3.2 (d.f. ⫽ 86; P ⬍ 0.001) and 32.6 ⫾ 12.5 (d.f. ⫽ 90; P ⬍ 0.02) ␮U/islet/90 min in the

FIGURE 1. Increments (䉭) in insulin output evoked by increas-

ing concentrations of sodium saccharin and sodium cyclamate in islets incubated in the presence of 7.0 mM D-glucose. Mean values (⫾S.E.M.) are derived, in each case, from 30 to 41 individual measurements performed, within the same experiments, both in the absence and in the presence of the artificial sweetener.

Insulinotropic Action of Artificial Sweeteners

729

TABLE 2. Effect of Kⴙ and the acesulfame anion on insulin release from islets incubated at increasing concentrations of

Line

Kⴙ (mM)a

Acesulfame(mM)

1 2 3 4 5 6 7 8 9

Nil 5.0 5.0 6.0 6.0 10.0 15.0 15.0 25.0

Nil Nil 5.0 Nil 1.0 5.0 Nil 10.0 25.0

D-Glucose

Nil

2.7 mM

100.0 ⫾ 14.6 (47)

100.0 ⫾ 11.3 (40)

200.2 ⫾ 16.8 (47)†

152.7 ⫾ 13.3 (40)*

D-glucose

b

7.0 mM 227.8 ⫾ 15.0 (31)† 100.0 ⫾ 5.0 (137) 150.5 ⫾ 6.0 (100)† 92.7 ⫾ 15.4 (8) 133.9 ⫾ 7.7 (28)* 290.9 ⫾ 11.8 (51)† 237.8 ⫾ 17.0 (16)† 296.3 ⫾ 15.0 (36)† 266.0 ⫾ 23.5 (32)†

20.0 mM 100.0 ⫾ 5.1 (30)

96.4 ⫾ 10.4 (16) 95.0 ⫾ 5.3 (32)

The concentration of K⫹ takes into account the cationic contribution of acesulfame-K. b All secretory data are expressed relative to the mean control output of insulin found at the same concentration of d-glucose in islets incubated at a normal K⫹ concentration (5.0 mM); see Table 1 for the absolute values of the mean control data. * P ⬍ 0.005. † P ⬍ 0.001, compared with the mean control value (line 2).

a

cyclamate failed to achieve statistical significance, the results were compatible, as in the case of saccharin, with the absence of any obvious threshold phenomenon. Stevioside In the presence of 7.0 mM d-glucose, stevioside (0.1–1.0 mM) caused a progressive increase in insulin output (Table 1). Such an effect, however, achieved statistical significance (P ⬍ 0.02) only at the highest concentration (1.0 mM) of the diterpene glycoside. Acesulfame-K In the absence of d-glucose or at a low concentration of the hexose (2.7 mM), acesulfame-K (10.0 mM) significantly increased insulin output (Table 2). In absolute terms, the drug-induced increment in insulin output was not significantly different under these two experimental conditions, averaging 16.2 ⫾ 3.8 (d.f. ⫽ 84; P ⬍ 0.001) and 13.9 ⫾ 5.7 (d.f. ⫽ 92; P ⬍ 0.02) in the absence and presence of d-glucose (2.7 mM), respectively. Likewise, in islets exposed to 7.0 mM d-glucose, acesulfame-K (5.0, 10.0 and 25.0 mM) caused a 2- to 3-fold increase in insulin output (Table 2, lines 6, 8 and 9), relative to the control value (Table 2, line 2). This enhancing effect, however, could conceivably be attributable merely to the rise in K⫹ concentration from its normal value of 5.0 mM to 10.0–25.0 mM in the presence of acesulfame-K. Even at non-insulinotropic concentrations of d-glucose, insulin release is indeed enhanced at high extracellular K⫹ concentrations [10, 11]. This issue was further investigated in islets exposed to 7.0 mM d-glucose. At this intermediate concentration of the hexose, the absence of extracellular K⫹ markedly augmented glucose-stimulated insulin release, relative to the value found at a normal K⫹ concentration, in agreement with prior observations [12]. When 5.0 mM acesulfame-K was incorporated in the K⫹-free medium, the rate of insulin release was significantly higher than the control value

found at the same final concentration of K⫹ (5.0 mM), indicating that the acesulfame anion indeed stimulated the islet B cells. Whereas a rise in K⫹ concentration from 5.0 to 6.0 mM failed to significantly affect insulin release, as little as 1.0 mM acesulfame-K was sufficient to increase hormonal output (P ⬍ 0.02) from islets exposed to the same final concentration of K⫹. At a much higher K⫹ concentration (15.0 mM), the acesulfame anion (10.0 mM) still caused a significant increase (P ⬍ 0.03) in insulin output evoked by 7.0 mM d-glucose (Table 2, lines 7 and 8). Such was no longer the case, however, in islets exposed to 20.0 mM d-glucose, in which case a rise in K⫹ concentration from 5.0 to 15.0 mM also failed to augment insulin secretion. Experiments in Perifused Islets The findings so far presented led us to explore the effect of saccharin, cyclamate and acesulfame-K on 86Rb, 45Ca and insulin output from pre-labelled perifused islets. Whether in the absence or presence of d-glucose (7.0 mM), neither saccharin nor cyclamate exerted any marked changes in 86Rb outflow (Fig. 2A, B). If anything, saccharin tended to decrease effluent radioactivity, whereas an opposite trend prevailed in the case of cyclamate. These two artificial sweeteners also failed to provoke any obvious change in 45Ca outflow from islets perifused in the presence of 7.0 mM d-glucose (Fig. 2E, F). Both drugs slowed the progressive decline in insulin output recorded during prolonged exposure of the perifused islets to 7.0 mM d-glucose. For instance, with saccharin, the peak value recorded during the first 5 min of exposure to the sweetener, averaged 126.5 ⫾ 7.4% (n ⫽ 8; P ⬍ 0.01) of the paired nadir value reached shortly before administration of the drug. Likewise, within 2.8 ⫾ 0.5 min of exposure to cyclamate, the output of insulin reached a peak value averaging 142.2 ⫾ 7.5% (n ⫽ 8; P ⬍ 0.001) of the paired nadir value recorded just before circulation of the cyclamic anions. A striking “off” response was recorded after halting the administration of cyclamate, such not being the case in the islets exposed to saccharin.

FIGURE 2. Effects of sodium saccharin (left panels) and sodium cyclamate (right panels), both tested at a 10.0-mM concentration and

administered from minute 46 to minute 70 (vertical dashed lines) on 86Rb and 45Ca fractional outflow rates (FOR) and insulin output from islets perifused either in the absence of D-glucose (A, B) or in the presence of 7.0 mM D-glucose (C–H). Mean values (⫾S.E.M.) were obtained from four to eight individual experiments.

Insulinotropic Action of Artificial Sweeteners

In the absence of d-glucose, acesulfame-K provoked a transient stimulation of 86Rb outflow (Fig. 3B). This increase was also observed in islets exposed throughout the perifusion period to 7.0 mM d-glucose, in which case the sweetener-induced increase in 86Rb efflux became a sustained and rapidly reversible phenomenon. It should be stressed, however, that these changes in effluent radioactivity were also observed, as expected from a prior study [13], when the K⫹ concentration of the perifusate was raised from 5 to 15 mM between minute 46 and minute 70 (Fig. 3A, C). In the latter case, a rapid, sustained and rapidly reversible increase in 45Ca outflow also was recorded. The acesulfame anion apparently antagonised the latter phenomenon. Thus, in the islets exposed to 15 mM K⫹, the integrated value for 45Ca fractional outflow rate between minute 46 and minute 70 averaged, relative to the paired theoretical value calculated by exponential extrapolation of the data recorded between minute 31 and minute 45, 217.0 ⫾ 14.5% (n ⫽ 4) in the absence of the acesulfame anion, as distinct (P ⬍ 0.001) from only 111.4 ⫾ 7.5% (n ⫽ 8) in its presence. The administration of acesulfame-K nevertheless caused a slight, and at least transient, increase in 45Ca efflux. Indeed, the 45Ca fractional outflow rate recorded 3 min after introduction of the artificial sweetener was 0.18 ⫾ 0.06 ⫻ 10⫺2/min as high (n ⫽ 8; P ⬍ 0.05) as the paired reading recorded at the 46th minute of perifusion. Both the increase in K⫹ concentration and the administration of acesulfame-K caused an immediate peak-shaped burst in insulin release, followed by a sustained, but less marked, increase in hormonal ouput (Fig. 3G, H). Over the entire period of stimulation (minute 46 to minute 70), the integrated increment in insulin output above the paired mean control value (minute 42 to minute 45) was twice as high (P ⬍ 0.001) in response to acesulfame-K (⫹0.82 ⫾ 0.08 ␮U/islet/min; n ⫽ 8) as in response to an equimolar concentration of KCl (⫹0.35 ⫾ 0.04 ␮U/islet/min; n ⫽ 8). DISCUSSION The present results unambiguously document that sodium saccharin, sodium cyclamate, stevioside and acesulfame-K stimulate insulin release from rat pancreatic islets, whereas aspartame fails to do so. In regard to acesulfame-K, care was taken to correct for the contribution of the K⫹ ion in the insulinotropic action of this sweetener. Such had not been the case in a prior study [14]. Likewise, the secretory effects of the sodium salts of saccharin and cyclamate cannot be attributed either to the Na⫹ ions or to hyperosmolarity, because a rise in the NaCl content of the incubation medium above its usual value (115 mM) inhibits, rather than augments, glucose-stimulated insulin release [15]. Relatively little information was so far available on the effect of artificial sweeteners on insulin release in vitro. Liang et al. [14] reported on the stimulation of insulin release from isolated rat islets by acesulfame-K (2.5–15.0 mM) but, as heretofore mentioned, did not consider the possible role of the K⫹ ion in their experiments. Niki and Niki [16] indicated that acesulfame-Na (15.0 mM) augments insulin release evoked by 10.0 mM d-glucose. Usami et al. [17] noted

731

that, in the isolated perfused rat pancreas, sodium saccharin (50 ␮M), stevioside (15 ␮M) and sodium cyclamate (440 ␮M) failed to significantly enhance the secretory response evoked by 19.2 mM arginine in the presence of 2.8 mM d-glucose. The concentrations of the three sweeteners used by the Japanese authors were much lower, however, than those tested in the present study. Most of the work relative to the effects of artificial sweeteners on insulin secretion was conducted in vivo to explore the cephalic phase of insulin secretion. In one study [18], the rapid insulin response normally evoked in rats by the oral ingestion of saccharin (about 20 nmol/g body wt.) was shown to be completely abolished in streptozotocin diabetic rats bearing intrahepatic denervated islet isografts, the authors concluding that gustatory and other oral sensory signals act as triggers for neurally mediated insulin release. In our opinion, it is quite remarkable that the four artificial sweeteners displaying insulinotropic action are known to exhibit bitterness or evoke a bitter aftertaste [19], whereas aspartame, which fails to stimulate insulin release, is not reported to taste bitter. A comparable parallelism between insulinotropic action and bitter taste was recently observed for various monosaccharide pentaacetate esters [4]. It is conceivable, therefore, that the insulinotropic action of both artificial sweeteners and hexose pentaacetate esters is accomplished through cellular mechanisms similar to those currently implied in bitter-taste reception by taste buds [3]. This proposal is further supported by the findings (J. Rasschaert, Y. Deng and W. J. Malaisse; unpublished observation) that islet cells are equipped with the G-proteincoupled GUST27 receptor expressed in taste buds [20]. In taste buds, the suggested range of targets for bitter compounds includes channels and receptors, G proteins, phospholipases, phosphodiesterases and Ca2⫹ release channels [3], the main transduction pathways currently under discussion involving either blockage of K⫹ channels or activation of phospholipase C or activation of cAMP breakdown. However, considering the fact that GTP regulatory proteins can act as programmable messengers, alternative modalities of stimulus–secretion coupling could account for the insulinotropic action of bitter-tasting compounds. Saccharin, cyclamate and acesulfame-K all failed to cause any marked decrease in 86Rb efflux from pre-labelled islets, which argues against the blockage of K⫹ channels as a major event in the stimulation of insulin release by these artificial sweeteners. Actually, acesulfame-K increased 86Rb outflow; this effect was reproduced, however, by a rise in extracellular K⫹ concentration and is therefore likely to be attributable to the K⫹ ions that dissociate from acesulfame-K in solution [16]. Likewise, saccharin, cyclamate and acesulfame-K apparently failed to mobilise Ca2⫹ from intracellular stores, as judged from the absence of any sizable increase in 45Ca efflux from pre-labelled islets exposed to these molecules. The acesulfame anion even suppressed the exchange between influent 40Ca2⫹ and effluent 45Ca2⫹ otherwise recorded in response to a rise in extracellular K⫹ concentration and cur-

FIGURE 3. Effect of CKl (left panels) and acesulfame-K (right panels), both tested at a 10.0-mM concentration and administered from

minute 46 to minute 70 (vertical dashed lines) on 86Rb and 45Ca fractional outflow rates (FOR) and insulin output from islets perifused in the absence of D-glucose (A, B) or in the presence of 7.0 mM D-glucose (C–H). Mean values (⫾S.E.M.) were obtained from four to eight individual experiments.

Insulinotropic Action of Artificial Sweeteners

rently ascribed to depolarisation of the plasma membrane with subsequent gating of voltage-sensitive Ca2⫹ channels [7]. These negative findings suggest, as already alluded to, that the insulinotropic action of artificial sweeteners and, by analogy, selected hexose pentaacetate esters may require second messengers other than those regulating cationic fluxes in the islet cells. For instance, preliminary work suggests that ␤-l-glucose pentaacetate may augment cyclic AMP production in islet cells. Further work is obviously required to fully characterise the sequence of cellular events triggered by artificial sweeteners in pancreatic islet B cells. Meanwhile, the present findings draw attention to a possible novel pathway for stimulation of insulin release that may include G-protein-coupled receptors similar to those currently implied in bitter taste reception. This study was supported by a Concerted Research Action of the French Community of Belgium (94/99-183) and a grant from the Belgian Foundation for Scientific Medical Research (3.4513.94). We are grateful to C. Re´mion, for technical assistance, and C. Demesmaeker, for secretarial help.

References 1. Malaisse W. J., Sa´nchez-Soto C., Larrieta M. E., Hiriart M., Jijakli H., Vin˜ambres C., Villanueva-Pen˜acarrillo M. L., Valverde I., Kirk O., Kadiata M. M. and Sener A. (1997) Am. J. Physiol. 273, E1050–1101. 2. Sener A., Welsh N., Malaisse-Lagae F., Kadiata M. M. and Malaisse W. J. (1998) Mol. Gen. Metab., in press. 3. Lindemann B. (1996) Physiol. Rev. 76, 719–766. 4. Malaisse W. J. and Malaisse-Lagae F. (1997) Biochem. Mol. Biol. Int. 43, 1367–1371.

733 5. Malaisse-Lagae F. and Malaisse W. J. (1984) Insulin release by pancreatic islets. In: Methods in Diabetes Research (Vol. 1) (Larner J. and Pohl S. L., Eds), pp. 147–152. Wiley and Sons, New York. 6. Herchuelz A. and Malaisse W. J. (1978) J. Physiol. Lond. 238, 409–424. 7. Jijakli H., Ulusoy S. and Malaisse W. J. (1996) Pharmacol. Res. 34, 105–108. 8. Carpinelli A. and Malaisse W. J. (1970) Mol. Cell. Endocrinol. 17, 103–110. 9. Herchuelz A., Sener A. and Malaisse W. J. (1980) J. Membrane Biol. 57, 1–2. 10. Herchuelz A., Thonnart N., Sener A. and Malaisse W. J. (1980) Endocrinology 107, 491–497. 11. Malaisse W. J. (1997) Cell. Signal. 9, 265–268. 12. Sener A. and Malaisse W. J. (1980) Endocrinology 106, 778– 785. 13. Boschero A. C. and Malaisse W. J. (1979) Am. J. Physiol. 236, E139–E146. 14. Liang Y., Maier V., Steinbach G., Lalic L. and Pfeiffer E. F. (1987) Horm. Metab. Res. 19, 285–289. 15. Malaisse W. J. (1969) Etude de la se´cre´tion insulinique in vitro. Editions Arscia S. A., Bruxelles, 68–69. 16. Niki A. and Niki H. (1994) Similarities between the hexoserecognizing mechanisms of the pancreatic B-cell and the gustatory cell. In: Frontiers of insulin secretion and pancreatic B-cell research (Flatt P. and Lenzen S., Eds), pp. 83–90. Smith-Gordon, London. 17. Usami M., Seino Y., Takai J., Nakahara H., Seino S., Ikeda M. and Imura H. (1980) Horm. Metab. Res. 12, 705–706. 18. Berthoud H. R., Trimble E. R., Siegel E. G., Boreiter D. A. and Jeanrenaud B. (1980) Am. J. Physiol. 238, E336–E340. 19. Schiffman S. S. and Galtin C. A. (1993) Neurosci. Biobehav. Rev. 17, 313–345. 20. Abe K., Kusakabe Y., Tanemura K., Emori Y. and Arai S. (1993) J. Biol. Chem. 268, 12033–12039.