Glucose-induced Cytosolic pH Changes in -Cells and Insulin Secretion Are Not Causally Related STUDIES IN ISLETS LACKING THE NA /H EXCHANGER NHE1*

The contribution of Na(+)/H(+) exchange (achieved by NHE proteins) to the regulation of beta-cell cytosolic pH(c), and the role of pH(c) changes in glucose-induced insulin secretion are disputed and were examined here. Using real-time PCR, we identified plasmalemmal NHE1 and intracellular NHE7 as the two most abundant NHE isoforms in mouse islets. We, therefore, compared insulin secretion, cytosolic free Ca(2+) ([Ca(2+)](c)) and pH(c) in islets from normal mice and mice bearing an inactivating mutation of NHE1 (Slc9A1-swe/swe). The experiments were performed in HCO(-)(3)/CO(2) or HEPES/NaOH buffers. PCR and functional approaches showed that NHE1 mutant islets do not express compensatory pH-regulating mechanisms. NHE1 played a greater role than HCO(-)(3)-dependent mechanisms in the correction of an acidification imposed by a pulse of NH(4)Cl. In contrast, basal pH(c) (in low glucose) and the alkalinization produced by high glucose were independent of NHE1. Dimethylamiloride, a classic...

Normal glucose homeostasis requires precise regulation of insulin secretion, a complex process that pancreatic ␤-cells achieve through changes in their metabolism (1)(2)(3)(4).Current models ascribe the control of insulin secretion by glucose to two hierarchical signaling pathways (5).The triggering pathway involves closure of ATP-sensitive K ϩ channels (K ATP channels), 4 membrane depolarization, Ca 2ϩ influx through voltagedependent Ca 2ϩ channels, and a rise in the cytosolic Ca 2ϩ concentration ([Ca 2ϩ ] c ), which is the indispensable triggering signal for exocytosis of insulin granules (4 -7).Simultaneously, a still incompletely understood amplifying pathway is activated, which augments the amount of released insulin without increasing [Ca 2ϩ ] c further (5, 8 -10).
Glucose induces changes in ␤-cell cytosolic pH (pH c ), the mechanisms and role of which in stimulus-secretion coupling are still debated.Except for two studies reporting no effect of glucose on pH c in mouse ␤-cells (11) or an acidification in rat islets (12), there is a large consensus that high glucose increases pH c in mouse islets (13)(14)(15)(16)(17), rat islets (18,19) and insulin-secreting cell lines (20,21).In contrast, there is no agreement on the mechanisms implicated in this alkalinization.Activation of unidentified HCO 3 Ϫ /Cl Ϫ exchangers (15), Na ϩ /H ϩ exchangers (14,19,20) or both (18) has been suggested using pharmacologically based experiments.
Even more confusing is the issue of a possible role of these pH c changes in insulin secretion.Studies in which changes in ␤-cell pH c were imposed by manipulation of pH and ionic composition of the extracellular medium or by exposure to weak bases or acids have led to the contradictory proposals that alkalinization augments (14,22) or inhibits (23) insulin secretion, and that acidification increases it (24 -26).Other studies, relying on amiloride derivatives to inhibit Na ϩ /H ϩ exchange (27,28) and cause ␤-cell acidification, also concluded that low pH c is beneficial to glucose-induced insulin secretion (19, 24 -26).Glucose-induced priming of insulin secretion (i.e.augmentation of the response to a second stimulation), which is thought to be mediated by the amplifying pathway (29), has been linked to ␤-cell acidification (26,30).These proposals are at odds with the fact that glucose increases islet pH c and with the correlative evidence that only those fuels which increase ␤-cell pH c amplify insulin secretion (31).It is thus unclear whether a Na ϩ /H ϩ exchanger is important for glucose-induced pH c changes in ␤-cells, and whether these changes play a role in the stimulation of insulin secretion, through one or both pathways.These were the questions addressed in the present study.
To date, ten isoforms of Na ϩ /H ϩ exchangers (NHE1 to NHE10), encoded by the Slc9 family of genes, have been identified (32)(33)(34).We recently showed that NHE1 is strongly expressed in rat islets and is present in mouse islets (35).Here, we first compared the expression of NHE isoforms in mouse islets.Having found that NHE1 is, by far, the most abundant among the plasma membrane isoforms, we next studied Slc9A1 swe/swe mutant mice bearing a spontaneous inactivating mutation of NHE1 (36) to test the role of Na ϩ /H ϩ exchange in ␤-cell pH c regulation under basal conditions and glucose stim- ulation.Finally, we measured pH c , [Ca 2ϩ ] c and insulin secretion in control and NHE1 mutant islets perifused with HCO 3 Ϫcontaining and HCO 3 Ϫ -free solutions to assess the possible role of pH c in glucose stimulation of insulin secretion through the triggering and amplifying pathways.
Preparations-Islets were aseptically isolated by collagenase digestion of the pancreas of NHE1 mutant and control mice (37).After hand selection, the islets were cultured for about 18 h at 37 °C in RPMI 1640 medium containing 10 mM glucose, 10% heat-inactivated fetal bovine serum, 100 international units/ml penicillin and 100 g/ml streptomycin.
RNA Extraction and Real-time Quantitative PCR-Total RNA was extracted, quantified, and reverse-transcribed into cDNA as previously described (38) using random hexamers and 200 units of M-MLV Reverse Transcriptase RNase H Ϫ Point Mutant (Promega, Madison, WI).The sense and antisense primers were chosen in the coding region of gene mRNA sequences, and their specificity was checked by BLAST search in the GenBank TM data bank (Table 1).Real-time PCR was performed with the iCycler iQ Real-Time PCR detection system (Bio-Rad) using the fluorescent dye SYBR Green I to monitor DNA amplification.The reaction was performed in duplicate, in a 25-l reaction volume containing cDNA (2-20 ng of total RNA equivalents), 300 nM primers (Table 1), 12.5 l iQ superMIX (Bio-Rad), and water to volume.TATA box-binding protein (TBP) was used as a reporter gene.Under these conditions, PCR efficiencies for amplification of NHE isoforms and TBP were similar.For all NHE isoforms, real-time PCR was performed in parallel in islets and control tissues mRNA (Table 1), with positive amplification in controls.After amplification, the specificity of PCR products was checked by agarose gel electrophoresis and analysis of the melting curve.The threshold cycle (Ct) was determined using iCycler iQ software 3.0.After correction of the Cts for differences in cDNA input in the PCR, the ⌬Ct values were calculated in every sample for each NHE isoform as follows: Ct NHEx Ϫ Ct TBP .The NHEx over TBP mRNA ratios can be estimated by the formula 2 Ϫ⌬Ct (39).
Measurements of Islet [Ca 2ϩ ] c and pH c -For [Ca 2ϩ ] c measurements, islets were loaded with Fura-PE3 (2 M, 2 h at 37 °C) in 1 ml HCO 3 Ϫ or HEPES medium, depending on the medium used during the subsequent experiment.After loading, the islets were transferred into a chamber mounted on the stage of a microscope and maintained at 37 °C.The Fura-PE3 probe was excited at 340 and 380 nm, and emission was captured at 510 nm with a CCD camera.From the ratio of fluorescence (at 340 and 380 nm), [Ca 2ϩ ] c was calculated by comparison with a calibration curve.Details of the method have been reported elsewhere (40).For measurements of pH c , islets were loaded with BCECF (0.5 M, 2 h at 37 °C) in 1 ml of HCO 3 Ϫ or HEPES medium.They were then examined in the same system as that used for Ca 2ϩ , but the probe was excited at 440 and 490 nm and emission was captured at 535 nm.The pH c was calculated from an in situ calibration curve constructed from the ratio values obtained by perifusing loaded islets with solutions of different pH (between 6 and 7.5) containing 10 g/ml nigericin.The medium had the following composition: 136 mM KCl, 4 mM NaCl, 5 mM MgCl 2 , 5 mM glucose, and 20 mM HEPES.
Measurements of Insulin Secretion and Islet or Pancreas Insulin Content-Batches of 20 cultured islets were transferred into chambers of a perifusion system (41).The effluent medium was collected at 42-s or 2-min intervals and saved for measurement of insulin by radioimmunoassay using rat insulin as a standard and ethanol to precipitate bound insulin (42).A small volume of concentrated NaHCO 3 or neutralized HEPES solution was added to all samples of experiments performed in HEPES medium or HCO 3 Ϫ medium, respectively, to achieve an identical final ionic composition.The standard curve was also prepared in a medium containing both HCO 3 Ϫ and HEPES.At the end of the experiments, the islets were recovered from the chambers, counted, and transferred in acid-ethanol for insulin extraction and measurement.The insulin content of the whole pancreas of NHE1 mutant and control mice was determined after extraction by homogenization and sonication of the tissue in acid ethanol.
Morphological Studies-The pancreas of control and NHE1 mutant mice was fixed in formalin for 24 h and embedded in paraffin.Glucagon cells were immunostained as described elsewhere (43) using a mouse monoclonal antibody (GLU-001, NovoBiolabs, Bagsvaerd, Denmark).Stained pancreas slides were digitized through a Zeiss microscope coupled to a DAGE-MTI CCD camera (Michigan City, IN).Fields containing islets (5-11 per slide) were captured through a ϫ10 objective (0.727 ϫ 0.758 m/pixel).Images were analyzed with a Zeiss KS400 imaging system.Islets were delineated by the user, and glucagon cells were identified by their gray level.
Presentation of Results-Results are presented as means Ϯ S.E. for the indicated number of animals or experiments.The statistical significance of differences between control and NHE1 mutant mice was evaluated by unpaired Student's t test.The effect of various drugs on insulin secretion was assessed by analysis of variance followed by a Dunnett's test for multiple comparisons with controls.

Identification of NHE Isoforms Expressed in Mouse Islets-We
used real-time PCR to compare the expression of the 10 known isoforms of NHE in control mouse islets (Fig. 1).The amount of mRNA for each isoform was reported to that of TBP.Among the isoforms present in the plasma membrane (NHE1 to NHE5 and NHE10) (32,34), NHE1 was the most largely expressed.Although primers and conditions of PCR amplification used for the 10 isoforms were different, a cautious comparison indicates that NHE1 expression is about 50-fold higher than NHE5, 450fold higher than NHE2 and 1800-fold higher than NHE4.No NHE3 was detected even after an amplification of 50 cycles.Of the organellar isoforms of NHE (NHE6 to NHE9) (32,33), NHE7 was strongly expressed, ϳ50-fold more than the other three.
Because NHE1 was, by far, the most abundant plasma membrane isoform in normal islets, mice with an inactivating mutation of NHE1 ( 36) could be a useful model to study the role of the Na ϩ /H ϩ exchanger in insulin secretion.As expected, no NHE1 was detected in NHE1 mutant mouse islets, which expressed the other isoforms as in controls, with only two exceptions.The plasma membrane isoforms NHE2 and 4 were more expressed in NHE1 mutant than control islets, but this expression remained very low compared with that of NHE1 in control islets, and lower than that of NHE5 in both types of islets (Fig. 1).
Characteristics of Control and NHE1 Mutant Mice-As previously reported, NHE1 mutant mice (Slc9A1 swe/swe ) exhibited ataxia in the hind limbs and seizures, and grew more slowly than controls (36).Although many died of their neurological problems a few weeks after weaning, we found it possible to Real-time analysis of NHE isoforms was performed after RT-PCR of mouse islets mRNA.The ⌬Ct were calculated as Ct NHEx Ϫ Ct TBP after correction for the total RNA equivalent used (2 or 20 ng).The threshold cycle for TBP using 2 ng of mRNA equivalent was about 28.5 for control and NHE1 mutant.Although primers and conditions of PCR amplification used for the 10 isoforms were different, an approximate comparison between the expression level of NHE1 and other isoforms is given by 2 ⌬Ct(NHEx)-⌬Ct(NHE1) resulting in a difference of ϳ50, ϳ450, and ϳ1800-fold with NHE5, NHE2, and NHE4, respectively.Values are means Ϯ S.E. for four preparations.*, p Ͻ 0.05 between NHE1 mutant and control mice.ND, not detected after 50 cycles of amplification.
prolong the lifespan of the mice by providing food and water close to the floor of the cages.The in vitro experiments using isolated islets were performed with control and NHE1 mutant mice of a mean age of 54 days (range 45-63).Table 2 compares NHE1 mutant and age-matched control mice.The body weight of NHE1 mutant mice was ϳ30% smaller.Their blood glucose was not different, but their plasma insulin was lower.The insulin content of the pancreas was lower, but not different from controls when expressed relative to the mass of pancreas or to body weight, indicating that there is no specific impact of the lack of NHE1 on ␤-cell development.Microscopic examination of the pancreas of NHE1 mutant mice showed that the morphological organization of the islets was normal, with a regular to oval shape and peripheral localization of glucagon cells.However, the average diameter of the islets was smaller (142 Ϯ 7 m, n ϭ 63) than in control mice (193 Ϯ 8 m, n ϭ 76, p Ͻ 0.007), which may explain the 33% lower insulin content of the NHE1 mutant islets that we used for in vitro experiments (Table 2).
Role of NHE1 Exchanger in Islet pH c Regulation-Previous studies, using pharmacological tools, have suggested that pH c regulation in ␤-cells involves unidentified HCO 3 Ϫ /Cl Ϫ and Na ϩ /H ϩ exchangers (14,15,18,44).Basal pH c (in 3 mM glucose) measured during the first 5 min of the experiments was not different in NHE1 mutant islets (7.03 Ϯ 0.01, n ϭ 175) and control islets (7.06 Ϯ 0.01, n ϭ 203) in HCO 3 Ϫ medium, but it was lower (p Ͻ 0.001) in NHE1 mutant islets (6.78 Ϯ 0.01, n ϭ 117) than control islets (7.02 Ϯ 0.01, n ϭ 164) in HEPES medium.It should be noted, however, that basal pH c decreased with time and did so more markedly in HEPES than HCO 3 Ϫ medium.Lower pH c values would thus be obtained at later times of the experiments, without affecting the comparison between NHE1 mutant and control islets.
To ascertain that the lack of NHE1 was not compensated for by other pH c -regulating mechanisms, we compared the ability of control and NHE1 mutant islets to correct an intracellular acidification produced by a prepulse of NH 4 Cl (Fig. 2).The isletswereperifusedwithaHCO 3 Ϫ mediuminwhichbothHCO 3 Ϫdependent mechanisms and Na ϩ /H ϩ exchangers can function, or with a HEPES medium in which only Na ϩ /H ϩ exchangers can operate.In the HCO 3 Ϫ medium, the correction of the acidification was rapid in control islets and delayed in NHE1 mutant islets (Fig. 2A).Dimethylamiloride (DMA), an inhibitor of Na ϩ /H ϩ exchangers (27,28), had no effect on the correction of pH c in NHE1 mutant islets but impaired the correction in control islets, as did the mutation of NHE1 in test islets (Fig. 2B).In contrast, DIDS, an inhibitor of HCO 3 Ϫ /Cl Ϫ exchangers (45,46), blocked the pH c correction in NHE1 mutant islets but had only little effect in control islets (Fig. 2C) because of the operation of Na ϩ /H ϩ exchange.Thus, the simultaneous addition of DIDS and DMA completely prevented pH c correction in control islets (Fig. 2D) as did DIDS alone in NHE1 mutant islets (Fig. 2C).In HEPES medium, when HCO 3 Ϫ -dependent pH regulation is inoperative, the correction of acidification by control islets was slightly slower than in the HCO 3 Ϫ medium, and no correction occurred in NHE1 mutant islets (Fig. 2E).That the correction observed in control islets is mediated by the Na ϩ /H ϩ exchange is shown by its suppression with DMA (Fig. 2F).EIPA and cariporide, two other inhibitors of Na ϩ /H ϩ

TABLE 2
Characteristics of control and NHE1 mutant mice exchange (27) produced similar effects to DMA in controls and were ineffective in NHE1 mutant islets (data not shown).Overall, these results confirm that both Na ϩ /H ϩ and HCO 3 Ϫ -dependent mechanisms contribute to the regulation of pH c in normal ␤-cells (15), show that no mechanism has compensated for the lack of NHE1 in mutant islets, and indicate that Na ϩ /H ϩ exchange is more important than HCO 3 Ϫ -dependent mechanisms for correction of an imposed acidification.
Effects of Na ϩ /H ϩ Exchange Blockers on Insulin Secretion-Because DMA and the genetic loss of NHE1 produced similar effects on pH c , one could argue that the pharmacological approach is adequate to study the impact of Na ϩ /H ϩ exchange and pH c on ␤-cell secretory function.The following experiments show that this is not the case.When control islets were perifused with a HCO 3 Ϫ medium containing 15 mM glucose, DMA produced a small, rapid, and reversible decrease in pH c in ϳ50% (19/36) of the islets (Fig. 3A) and a consistent, delayed, irreversible increase in insulin secretion (Fig. 3B).As anticipated, DMA did not affect pH c in NHE1 mutant islets (Fig. 3A) but, most unexpectedly, increased insulin secretion as in controls (Fig. 3B).In a second series of experiments, we compared the effects of 3 inhibitors of Na ϩ /H ϩ exchange.Although they had virtually no effect on pH c under these conditions (data not shown), DMA and EIPA markedly potentiated glucose-induced insulin secretion (p Ͻ 0.01), while the trend observed with cariporide did not reach statistical significance.Again the effect of the three drugs was similar in control and NHE1 mutant islets (Fig. 3C).Taken together, these results indicate that the increase in insulin secretion produced by these inhibitors is unrelated to a change in pH c but is caused by an action of the drugs on a target other than the Na ϩ /H ϩ exchanger.DMA and related substances cannot reliably be used to evaluate the influence of pH c on insulin secretion.

Influence of pH c on Glucose-induced Insulin Secretion-In HCO 3
Ϫ medium, pH c of control islets decreased slowly when the glucose concentration remained low (3 mM), and increased when the glucose concentration was raised to 15 mM (Fig. 4A) (15).The stimulation by high glucose also produced typical changes in [Ca 2ϩ ] c , characterized by a small initial decrease below basal values, followed by a sharp peak and oscillations during the second phase (Fig. 4B).Glucose also induced a biphasic increase in insulin secretion (Fig. 4C).In HEPES medium, basal pH c decreased more rapidly and no increase occurred upon stimulation by high glucose (Fig. 4A).In contrast, the increases in [Ca 2ϩ ] c and insulin secretion were similar to those in HCO 3 Ϫ medium, with only a small lag in the first phase (Fig. 4, B and C).
When NHE1 mutant islets were perifused with a HCO 3 Ϫ medium, basal pH c slightly decreased in 3 mM glucose and increased in response to 15 mM glucose (Fig. 4D).This shows that NHE1 is not responsible for the alkalinization induced by high glucose.Glucose-induced [Ca 2ϩ ] c and insulin secretion changes in NHE1 mutant islets followed a similar biphasic pattern as in control islets (Fig. 4, E and F).In HEPES medium, basal pH c was lower than in HCO 3 Ϫ medium, and high glucose produced an additional marked decrease to reach a very low pH of 6.3 (Fig. 4D).This decrease in pH c is due to Ca 2ϩ influx as shown by its abrogation by diazoxide or chelation of extracel-lular CaCl 2 with EGTA (data not shown) (15).In HEPES medium, the increase in [Ca 2ϩ ] c produced by glucose was not only delayed, but strongly inhibited, with only small and irreg- ular oscillations during the second phase (Fig. 4E).There was, however, no evidence for desynchronization of these [Ca 2ϩ ] c oscillations between different regions of the islets (data not shown).Paradoxically, glucose-induced insulin secretion by NHE1 mutant islets was similar in HEPES and HCO 3 Ϫ media during the 50 min of stimulation (Fig. 4F).However, when the period of glucose stimulation in HEPES medium was longer, the insulin secretory rate steadily decreased in NHE1 mutant islets (data not shown).Ϫ medium, pH c was similar but [Ca 2ϩ ] c was slightly higher (p Ͻ 0.05) in NHE1 mutant than control islets, which can account for the trend to a higher insulin secretory response.The interpretation of the experiments performed in HEPES medium is complicated by the interference that low pH c can cause with [Ca 2ϩ ] c measurements based on EGTA-derived probes such as Fura-PE3 (47,48).If the measured values are corrected according to Batlle et al. (48), [Ca 2ϩ ] c is increased by 13% in controls (to 266 nM) and by 36% in NHE1 mutant islets (to 253 nM).In the latter case, no difference persists between [Ca 2ϩ ] c in NHE1 mutant islets stimulated by glucose in HCO 3 Ϫ and HEPES media, which is compatible with the similar secretory response despite the marked difference in pH c (Fig. 4, G-I).Overall, the substantial differences in pH c did not have an impact on insulin secretion.
Influence of pH c on the Amplifying Pathway of Insulin Secretion-The amplifying pathway of glucose-induced insulin secretion can be investigated by holding K ATP channels open with diazoxide and depolarizing ␤-cells with 30 mM KCl in the presence of low or high glucose (8).Fig. 5, A, C, and E show the results obtained with control islets perifused with a HCO 3 Ϫ medium.pH c was slightly higher in 15 than 1 mM glucose and barely affected by 30 mM KCl (Fig. 5A).[Ca 2ϩ ] c increased similarly in 1 and 15 mM glucose, following a biphasic pattern with a sharp first peak and a flat second phase (Fig. 5C).Although [Ca 2ϩ ] c was similar at both glucose concentrations, insulin secretion was ϳ3-fold larger at high than low glucose (Fig. 5E), which corresponds to the amplifying action of glu- NHE1 mutant mice established that these islet cells really lack functionally significant Na ϩ /H ϩ exchanger in their plasma membrane.
Under basal conditions (3 mM glucose), the lack of NHE1 did not affect islet pH c in HCO 3 Ϫ medium but resulted in a marked lowering of pH c in HEPES medium.Basal pH c also rapidly decreased in control islets perifused with a HEPES medium, despite the presence of NHE1.This shows that HCO 3 Ϫ -dependent mechanisms are more important than Na ϩ /H ϩ exchange for basal pH c regulation in islet cells.In contrast, NHE1 significantly contributed to normal pH c restoration after an acid load (NH 4 Cl prepulse technique).This contribution was evidenced by the slower increase in pH c in NHE1 mutant than control islets in HCO 3 Ϫ medium, and by the ability of control islets, in contrast to NHE1 mutant islets, to correct pH c in HEPES medium.HCO 3 Ϫ -dependent mechanisms proved less important for this correction of an acid pH c , as shown by the smaller impact of DIDS or HCO 3 Ϫ omission than of NHE1 inhibition by DMA in control islets.However, when NHE1 is lacking or inhibited, these mechanisms ensure a slow pH c correction.
Upon stimulation by high glucose, pH c similarly increased in NHE1 mutant and control islets.These observations conclusively establish that ␤-cell alkalinization by glucose is not mediated by Na ϩ /H ϩ exchange as sometimes proposed (14,19,20,44).The failure of glucose to increase islet pH c in HEPES medium points to the implication of a HCO 3 Ϫ -dependent mechanism, presumably of a HCO 3 Ϫ /Cl Ϫ exchanger because this alkalinization is also inhibited by DIDS (15).However, the exact mechanism, in particular the identity of the implicated exchangers, remains to be determined.Out of the numerous HCO 3 Ϫ /Cl Ϫ exchangers (50), only one Na ϩ -dependent isoform (SLC4A10) has been identified in islet cells (51), but the presence of other isoforms has not been excluded.An intervention of NHE1 was detectable when glucose stimulation occurred in HEPES medium.Under these conditions, high glucose produced a major decrease in pH c in NHE1 mutant islets but only had a marginal acidifying effect in control islets.These results and those of the experiments based on the NH 4 Cl prepulse, therefore, indicate that Na ϩ /H ϩ exchange is mainly involved in a protection of islet cells against marked cytosolic acidification.
NHE1 and Insulin Secretion-Blockers of NHE1, such as amiloride and DMA (27,28) have previously been used to study various facets of ␤-cell function, in addition to pH c regulation.They have been reported to decrease 86 Rb efflux from preloaded islets (an indication of K ϩ channel closure) (19), to increase glucose-induced electrical activity in ␤-cells (24), and to augment insulin secretion (15,19,20,26,30).These effects were attributed to blockade of Na ϩ /H ϩ exchange, leading to cytosolic acidification and, eventually, closure of K ATP channels either directly (19,25) or via improved glucose metabolism (26).However, the decrease in pH c by DMA, which is admittedly significant in HEPES medium, is minimal in HCO 3 Ϫ medium.Moreover, we show here that DMA and EIPA also increased insulin secretion from NHE1 mutant islets without changing pH c .In fact, studies of excised patches of cardiomyocyte (52) or ␤-cell (53) membranes have shown that micromolar concentrations of amiloride and derivatives can directly inhibit K ATP channels by interacting with the pore.Our results do not rule out the possibility that changes in pH c can influence K ATP channels in ␤-cells (54,55), but preclude the use of amiloride derivatives to test the role of Na ϩ /H ϩ exchange itself and of pH c changes in insulin secretion.We therefore used two other approaches: the NHE1 mutant islets and the direct comparison of insulin secretion and pH c changes in HCO 3 Ϫ and HEPES media.
Compared with control islets, glucose-induced insulin secretion was not impaired in NHE1 mutant islets.The Na ϩ /H ϩ exchanger is thus dispensable for stimulus-secretion coupling at least when the experiments are performed in a physiological HCO 3 Ϫ medium.␤-Cell pH c and Insulin Secretion-Most proposals that changes in ␤-cell pH c participate in stimulus-secretion cou- pling are based on experiments testing islets or cell lines in artificial HCO 3 Ϫ -free medium and imposing large pH c changes by drugs (14,19,20,23,26,30), by acidifying or alkalinizing agents (14,22,26,30,56), or by modifications of extracellular pH (22,26,30,56).These large, forced excursions of pH c (above 7.2 and below 6.8) may influence the secretory process, but the underlying mechanisms are unlikely to be physiologically relevant.It has been suggested that the optimal ␤-cell pH c for nutri- ent induced insulin secretion ranges between 6.4 -6.8 (26).Only one of our experiments might suggest that low pH c is favorable for insulin secretion.In NHE1 mutant islets, glucoseinduced insulin secretion was similar in HEPES medium (islet pH c ϳ 6.3) and HCO 3 Ϫ medium (islet pH c ϳ 7.0) although [Ca 2ϩ ] c was significantly lower in HEPES medium, and thus seemingly having a greater efficacy on exocytosis at lower pH c .However, this difference in [Ca 2ϩ ] c is probably apparent only, and attributable to the interference of low pH with the measurement of [Ca 2ϩ ] c (47,48).When this confounding factor was taken into account, no beneficial influence of low pH c on glucose-induced insulin secretion was detectable.␤-Cell pH c admittedly decreases upon stimulation in HEPES medium, and even becomes very low when Na ϩ /H ϩ exchange is pharmacologically inhibited under these conditions.However, this situation is completely artificial, glucose increasing ␤-cell pH c in physiological HCO 3 Ϫ medium.An increase in islet pH c has also been implicated in stimulussecretion coupling and suggested to augment (14,22) or impair (23,30) glucose-induced insulin secretion.Our results show that the rise in pH c that high glucose produces in HCO 3 Ϫ medium is not important for the secretory response.Thus, the absence of such an increase in pH c in HEPES medium was not accompanied by a significant alteration of glucose-induced insulin secretion.Previous observations that all metabolized agents capable of increasing insulin secretion through the amplifying pathway were also causing an increase in islet pH c led us to propose that ␤-cell alkalinization might be implicated in the amplifying process (31).The present results do not support the hypothesis.Thus, the experiments in which [Ca 2ϩ ] c was increased and clamped by depolarizing the islets with KCl in the presence of diazoxide convincingly showed that the amplification of insulin secretion by high glucose is independent of pH c or changes in pH c .
Conclusions-We have shown that NHE1 (plasma membrane isoform) and NHE7 (intracellular isoform) are the most expressed Na ϩ /H ϩ exchangers in mouse islets.NHE1 contributes to correction of an imposed acidification of ␤-cells, but is not implicated in the increase in pH c produced by high glucose.This alkalinization is mediated by HCO 3 Ϫ -dependent mechanisms but is not causally related to insulin secretion.Overall our study provides evidence against the hypotheses that either a decrease or an increase in pH c is important for glucose-induced insulin secretion via the triggering or the amplifying pathways.

FIGURE 1 .
FIGURE 1. Expression of NHE isoforms in control and NHE1 mutant mouse islets.Real-time analysis of NHE isoforms was performed after RT-PCR of mouse islets mRNA.The ⌬Ct were calculated as Ct NHEx Ϫ Ct TBP after correction for the total RNA equivalent used (2 or 20 ng).The threshold cycle for TBP using 2 ng of mRNA equivalent was about 28.5 for control and NHE1 mutant.Although primers and conditions of PCR amplification used for the 10 isoforms were different, an approximate comparison between the expression level of NHE1 and other isoforms is given by 2 ⌬Ct(NHEx)-⌬Ct(NHE1) resulting in a difference of ϳ50, ϳ450, and ϳ1800-fold with NHE5, NHE2, and NHE4, respectively.Values are means Ϯ S.E. for four preparations.*, p Ͻ 0.05 between NHE1 mutant and control mice.ND, not detected after 50 cycles of amplification.

FIGURE 2 .
FIGURE 2. Role of NHE1-exchanger in islet pH c regulation.The NH 4 Cl prepulse technique was used to study how islet cells correct the cytosolic acidification following removal of NH 4 Cl from the extracellular medium.Control islets are shown by thin lines and NHE1 mutant islets by thick lines.The experiments were performed in a HCO 3 Ϫ medium (A-D) or a HEPES medium (E-F) containing 3 mM glucose (G 3), with NH 4 Cl (20 mM) added and removed as indicated.DMA (40 M), or DIDS (200 M) were added for 10 min (B, C, F).In D, both drugs were added together for 10 min before DIDS was omitted while DMA remained present for another 10 min.Values are means Ϯ S.E. for 31-58 islets (from 5 to 10 preparations).

FIGURE 3 .
FIGURE 3. Effects of inhibitors of Na ؉ /H ؉ exchange on pH c and insulin secretion in control and NHE1 mutant mouse islets.pH c (A) and insulin secretion (B) were measured in control islets shown by thin lines and circles, and NHE1 mutant islets shown by thick lines and squares.The glucose concentration in the perifusion medium was kept at 15 mM, and 40 M DMA was added between 15 and 45 min, except in the series shown by open symbols in B. Values are means Ϯ S.E. for 28 -34 islets (from 5 to 6 preparations) in pH c measurements and, for 5 experiments of insulin secretion.C, shows another experimental series, similar to B, in which the effects of 40 M DMA, EIPA, or cariporide on insulin secretion were compared.The data show the difference (⌬) between the average rate of secretion during the last minutes of drug application (35-45 min) and the last minutes preceding drug addition (5-15 min).Values are means Ϯ S.E. for 4-5 experiments.

Fig. 4 ,
G-I show integration of the responses during the last 20 min of the experiments.During glucose stimulation in HCO 3

FIGURE 4 .
FIGURE 4. Roles of NHE1-exchanger in glucose-induced pH c (A, D, G), [Ca 2؉ ] c (B, E, H), and insulin secretion (C, F, I) changes in mouse islets.A-F, control or NHE1 mutant islets were stimulated by an increase in the glucose concentration from 3 to 15 mM (G 3 3 G 15) in a HCO 3 Ϫ medium (thin lines and open circles or squares) or a HEPES medium (thick lines and filled circles or squares).The dotted lines in A and D, correspond to islets maintained in a medium containing 3 mM glucose throughout.G-I, show the integrated responses during the last 20 min in the HCO 3 Ϫ or HEPES media for control and NHE1 mutant islets.Values are means Ϯ S.E. for 27-46 islets (from 5 to 8 preparations) in pH c and [Ca 2ϩ ] c measurements and, for 5 experiments of insulin secretion.