Potential conflict of interest: Nothing to report.
Bicarbonate secretion from cholangiocytes is required for appropriate adjustment of primary canalicular bile along the biliary tract. In human and rat cholangiocytes, bicarbonate secretion is mediated by anion exchanger (AE) 2, an electroneutral Na+-independent Cl−/HCO AE also involved in intracellular pH (pHi) regulation. In Ae2a,b-deficient mice, pHi is increased in lymphocytes and fibroblasts, whereas it is surprisingly normal in cholangiocytes. Here, we analyze the mechanisms for HCO3− secretion in cultured Ae2 and Ae2a,b−/− mouse cholangiocytes by microfluorimetric measurement of pHi changes upon established perfusion maneuvers. Cl− withdrawal by isethionate-based perfusions showed that Ae2 but not Ae2a,b−/− mouse cholangiocytes can display Cl−/HCO exchange, which is therefore entirely mediated by Ae2. Nevertheless, simultaneous withdrawal of Cl− and Na+ revealed that mouse cholangiocytes possess an additional transport activity for HCO3− secretion not observed in control rat cholangiocytes. Propionate-based maneuvers indicated that this supplemental Na+-driven HCO-secreting activity is Cl−-independent, consistent with a Na+-HCO3− cotransport (NBC). NBC activity is greater in Ae2 than Ae2a,b+/+ mouse cholangiocytes, and membrane-depolarization experiments showed that it is electrogenic. Consistent with the potential role of Slc4a4/Nbc1 as the involved transporter, Ae2 mouse cholangiocytes exhibit up-regulated expression of this electrogenic NBC carrier. Whereas Ae2-mediated Cl−/HCO3− exchange in Ae2 mouse cholangiocytes is stimulated by cyclic adenosine monophosphate (cAMP) and acetylcholine, the NBC activity is down-regulated by cAMP and adenosine triphosphate (ATP) in Ae2a,b−/− mouse cholangiocytes. Polarized Ae2 mouse cholangiocytes placed in Ussing chambers show decreased (but not abolished) cAMP-dependent Cl− current and increased ATP-dependent/Ca2+-activated Cl− secretion, which run in parallel with decreased cystic fibrosis transmembrane conductance regulator messenger RNA expression and increased intracellular Ca2+ levels. Conclusion: Bicarbonate secretion in mouse cholangiocytes involves two differentially regulated activities: Ae2-mediated Cl−/HCO3− exchange and Na+-HCO cotransport. (HEPATOLOGY 2009.)
Bile flow is produced through the coordinated secretory function of hepatocytes and cholangiocytes.1 These polarized cells possess different carriers in their apical and/or basolateral plasma membrane that provide vectorial transport of biliary solutes. Primary bile generated by hepatocytes is fluidized and alkalinized by cholangiocytes along the biliary tract. Bicarbonate secretion from cholangiocytes to the duct lumen requires intracellular accumulation of HCO3−, which is partially achieved through catalyzed hydration of CO2 and subsequent H+ extrusion through Na+/H+ exchange. Moreover, HCO loading can be assisted by additional transport activities that may vary between animal species. In rat cholangiocytes, for instance, there is a basolateral HCO3− influx mediated by Na+-HCO cotransport (NBC).2 In human cholangiocytes, however, basolateral influx of HCO3− occurs mainly through Na+-dependent Cl−/HCO exchange.3
In both human4 and rat5 cholangiocytes, the secretion of HCO3− to bile occurs through a Na+-independent electroneutral Cl−/HCO exchange mediated by the anion exchanger (AE) 2 (AE2/SLC4A2), which is an acid loader of the SLC4/AE family generally involved in intracellular pH (pHi) regulation.6 Secretin-stimulated bicarbonate-rich hydrocholeresis has been postulated to involve the synchronized action of AE2 and two additional flux proteins, the cyclic adenosine monophosphate (cAMP)-responsive Cl− channel cystic fibrosis transmembrane conductance regulator (CFTR) and the water channel aquaporin-1.7 Upon secretin/secretin-receptor interaction and subsequent increase in cAMP levels and protein kinase A activation,8–10 intracellular vesicles with all three flux proteins coredistribute to the lumenal membrane of cholangiocytes.7 Accompanying phosphorylation and activation of CFTR11 lead to apical efflux of Cl−, which is ultimately exchanged with HCO3− through AE2.1, 5, 9, 10 Resultant secretion of HCO is followed by osmotically driven efflux of water through aquaporin-1, ending up in the typical bicarbonate-rich hydrocholeresis induced by secretin.7, 12 The effect of this mechanism may be reinforced through the recently described CFTR-associated release of adenosine triphosphate (ATP) from cholangiocytes to bile.13 Autocrine/paracrine stimulation of apical purinergic receptors by ATP is then followed by increased intracellular Ca2+, activation of apical Ca2+-dependent Cl− channels, and efflux of Cl−, which is likewise exchanged with HCO3− through AE2. Bicarbonate-enriching processes elicited by secretin might also involve CFTR-independent effects of cAMP on the AE2-mediated Cl−/HCO exchange, as observed in vitro in human cholangiocytes in the presence of CFTR inhibitors3 and in hepatocyte-derived cells4 that lack CFTR. Acetylcholine (ACh) may further assist biliary HCO3− secretion through Ca2+-dependent Cl− efflux from cholangiocytes and by potentiating the effect of secretin on both intracellular cAMP levels and Cl−/HCO exchange in a calcineurin-dependent manner.13–15
We previously hypothesized that abnormalities of AE2 in primary biliary cirrhosis (PBC) could play a pathogenic role in the disease,16–18 and generated a mouse model targeted for the three major Ae2a, Ae2b1, and Ae2b2 isoforms (Ae2a,b−/− mice).19 These mice showed biochemical, serologic, histopathologic, and immunologic features compatible with PBC.20 pHi assessment in lymphocytes20 and fibroblasts21 of Ae2 mice confirmed the expected intracellular alkalinization, while cholangiocytes isolated from these mice were found to have normal pHi values.20 We therefore decided to analyze the mechanisms involved in HCO3− secretion in cultured Ae2 and Ae2a,b−/− mouse cholangiocytes. Our present data show that wild-type but not Ae2 mouse cholangiocytes exhibit Cl−/HCO3− anion exchange (AE) activity. This indicates that AE activity in normal mouse cholangiocytes is only mediated by Ae2, in agreement with previous data in human and rat cholangiocytes.4, 5 Characteristically, however, mouse cholangiocytes can display an additional transport activity for HCO secretion that is more pronounced in Ae2a,b-deficient cholangiocytes. This additional activity is Cl−-independent and Na+-dependent (consistent with NBC activity), electrogenic, and responds differentially to agonists such as cAMP, ATP, and ACh.
Cultured Cholangiocytes from Normal and Ae2a,b−/− Mice.
Cultured cholangiocytes were obtained from male Ae2 and Ae2a,b−/− mice, starting with isolation of intrahepatic bile duct units, followed by their seeding and culture on collagen monolayer20 (see Supporting Information). Normal rat cholangiocytes (NRCs) were similarly obtained from male Wistar rats. Animal interventions were approved by the Institutional Animal Care and Use Committee.
Determination of HCO3− Transport Activities.
Cultured cholangiocytes were assessed for HCO3− transport activities by microfluorimetry (see Supporting Information). Cells initially perfused with Krebs-Ringer bicarbonate (KRB) medium were monitored for pHi variations upon two types of perfusion maneuvers2, 3, 18 (see detailed buffers in Supplementary Table 1). First, to test Cl−/HCO exchange and Na+ dependency, KRB was changed to KRB without Cl− (substituted with isethionate) or without both Cl− and Na+ (substituted with N-methyl-D-glucamine).3, 18 These Cl−-free based maneuvers are designed to elicit efflux of intracellular Cl− in exchange with extracellular HCO3−, increasing pHi. In the case of AE2, which physiologically functions as a HCO-extruder (i.e., acid-loader), Cl−/HCO3− exchange activity is measured in a reverse mode. The exchange activity (expressed as transmembrane base fluxes JOH−) is calculated from the tangent of the experimental plot of the alkalinization induced by Cl− removal. A second type of maneuver starting with propionate-KRB perfusion5, 18, 22 was aimed to analyze HCO secretion mechanisms in a more physiological manner for AE2. Here, intracellular alkalinization induced by propionate removal may activate acid-loading mechanisms to extrude HCO3− and recover resting pHi. In this case, HCO secretion rates (JOH−) are calculated from the tangent of the experimental plot of the recovery toward resting pHi. The stimulatory effect of either a cAMP/stimulation mixture (cAMP/SM) (100 μM N62-O-dibutyryl-cAMP, 3 μM forskolin, and 100 μM 3-isobutyl-1-methyl-xanthine),5, 18, 22 100 μM ACh13 or 10 μM ATP,18, 23 and the inhibitory effect of 500 μM DIDS (all from Sigma) on HCO3− secretion could be assessed by application of these compounds 5 minutes before and along the perfusion maneuver.
Voltage Clamp Measurement of Short-Circuit Current on Polarized Mouse Cholangiocytes.
Short-circuit current (Isc) was monitored in confluent polarized mouse cholangiocytes as described for NRCs24 (see Supporting Information for details). Briefly, when cells cultured on collagen-coated inserts increased the transepithelial electrical resistance ≥1,000 Ω · cm2, inserts were mounted in U2500 Dual-Channel Self-Contained Ussing Chambers (Warner Instruments, Hamden, CT). Each half-chamber (filled with a NaCl-rich solution at 37°C and continuously bubbled with 5% CO2-95% O2) was connected to a microcomputer-controlled current/voltage clamp amplifier, and Isc (μA/cm2) was determined when transepithelial voltage was clamped to 0 mV. The system was monitored along the experiments in response to current pulses every 6 seconds. Under these conditions, changes in Isc were recorded following simultaneous apical and basolateral stimulation with a particular cAMP-SM (500 μM 8-cpt-cAMP plus 450 μM 3-isobutyl-1-methyl-xanthine) or with 100 μM ATP (all from Sigma), essentially as described.24
Determination of the Intracellular Levels of Ca2+ and cAMP, Immunoblotting, and Real-Time Polymerase Chain Reaction.
These procedures are described in the Supporting Information.
Data are expressed as the mean ± standard error of the mean. Two-group comparisons were performed by using adequate nonparametric (Mann-Whitney) or parametric t tests. Two-tailed P values <0.05 were considered statistically significant.
Normal Mouse Cholangiocytes Display Ae2 Activity.
Ae2 gene silencing in human4 and rat5 cholangiocytes have indicated that HCO secretion is mediated by AE2. In mouse cholangiocytes, however, the mechanisms for HCO3− secretion remain to be fully elucidated. Thus, we assessed HCO secretion activities in cultured Ae2a,b+/+ and Ae2 mouse cholangiocytes, using normal rat cholangiocytes as a control. As shown in Fig. 1A, selective withdrawal of Cl− elicited intracellular alkalinization in control NRCs, indicating the occurrence of Cl−/HCO3− exchange (JOH− = 20.97 ± 0.72 mmol/L/minute). Likewise, Cl− withdrawal induced intracellular alkalinization in normal (Ae2) mouse cholangiocytes, which reveals that these cells also display Cl−/HCO3− exchange activity (JOH− = 10.63 ± 0.95 mmol/L/minute) (Fig. 1B). In Ae2 mouse cholangiocytes, however, no intracellular alkalinization was observed after Cl− withdrawal (Fig. 1C), demonstrating that the Cl−/HCO3− exchange activity displayed by normal mouse cholangiocytes is entirely mediated by Ae2.
Mouse but not Rat Cholangiocytes Possess an Additional Cl−-Independent Na+-Driven HCO3− Secretory Activity that May Maintain Bicarbonate Secretion in the Absence of Ae2.
As mentioned previously, the activity displayed by AE carriers (AE2 activity included) is independent of Na+. Indeed, in the case of control NRCs, we could confirm that the Cl−/HCO exchange activity elicited by simultaneous withdrawal of Cl− and Na+ (JOH− = 20.80 ± 0.97 mmol/L/minute) was similar to that observed in the presence of Na+ (Fig. 1A). In mouse cholangiocytes, however, assessment of Na+ dependency uncovered a surprising effect in these cells. Unlike NRCs, both Ae2a,b+/+ and Ae2 mouse cholangiocytes showed intracellular acidification after simultaneous removal of Cl− and Na+ from the media (Fig. 1B,C), indicating the presence of a Na+-driven acidification activity that involves HCO3− secretion. This Na+-driven activity was more pronounced in Ae2 than normal mouse cholangiocytes (JOH− = 10.92 ± 2.02 versus 1.82 ± 0.95 mmol/L/minute; P < 0.001), which might be partially due to the counterbalancing Ae2-mediated alkalinization that takes place in normal but not Ae2a,b−/− mouse cholangiocytes (see Fig. 1B,C for the first maneuver of Cl− withdrawal).
To further ascertain that the Na+-driven acidification activity selectively observed in mouse cholangiocytes involves secretion of HCO and is Cl−-independent, we turned to propionate-based maneuvers in the presence and absence of Cl− (see Material and Methods). As expected, control NRCs showed a final step of acidification recovery—indicative of HCO3− extrusion—in the presence of Cl− (JOH− = 5.39 ± 0.24 mmol/L/minute) but not in its absence (Fig. 2A), confirming again that HCO secretion only occurs through Cl−/HCO3− exchange in these cells. By contrast, Ae2 mouse cholangiocytes showed no inhibition of the HCO3− extrusion rate despite Cl− withdrawal (JOH− = 2.87 ± 0.33 mmol/L/minute in the presence of Cl− versus 2.83 ± 0.29 mmol/L/minute in the absence of this anion) (Fig. 2B), thus unmasking a Cl−-independent HCO extrusion activity. This observation was even clearer in Ae2a,b−/− mouse cholangiocytes. In spite of their Ae2 deficiency, these cells were able to display HCO extrusion activity regardless of the presence or absence of extracellular Cl− (JOH− = 4.91 ± 1.12 versus 5.89 ± 0.80 mmol/L/minute, respectively; Fig. 2C). In fact, HCO3− extrusion rate values were significantly higher in Ae2 than Ae2a,b+/+ mouse cholangiocytes (P < 0.05 in the presence of Cl− and P < 0.01 in its absence). This Cl−-independent and Na+-driven activity to secrete HCO observed in mouse cholangiocytes is by definition a Na+-HCO3− cotransport (i.e., an NBC activity).25, 26 As shown in Fig. 2D, additional experiments indicated that this activity is sensitive to DIDS (a nonspecific inhibitor of HCO transporters)25 in both Ae2a,b+/+ (JOH− = 2.81 ± 0.23 versus 1.03 ± 0.44 mmol/L/minute in the presence of DIDS; P < 0.001) and Ae2 (JOH− = 6.15 ± 0.86 versus 3.05 ± 0.63 mmol/L/minute; P < 0.05) mouse cholangiocytes.
Na+-Driven Cl−-Independent HCO Secretion Displayed by Mouse Cholangiocytes Is Electrogenic.
Electrogenicity of the NBC activity in mouse cholangiocytes was analyzed using propionate-based maneuvers under depolarizing conditions with 10-fold the physiological extracellular concentration of K+.27 Upon membrane depolarization, an electrogenic-driven HCO3− extrusion would be diminished, whereas an electroneutral transport would remain unchanged.27 Whereas HCO secretion activity in NRCs was maintained (consistent with its complete mediation by electroneutral Ae2 activity) (Fig. 3A), the secretion rate was diminished in Ae2a,b+/+ mouse cholangiocytes (JOH− = 3.09 ± 0.19 versus 1.54 ± 0.16 mmol/L/minute; P < 0.001) (Fig. 3B) and totally abolished in Ae2 mouse cholangiocytes (JOH− = 4.83 ± 0.46 versus 0 mmol/L/minute; P < 0.001) (Fig. 3C), indicating that the Na+-HCO3− cotransport displayed by mouse cholangiocytes is electrogenic.
Slc4a4 Expression Is Up-Regulated in Ae2a,b−/− Mouse Cholangiocytes.
Among NBC carriers, Slc4a4 and Slc4a5 have been reported to be electrogenic,26, 28 whereas Slc4a7 is electroneutral.27, 29 We therefore analyzed the messenger RNA (mRNA) expression of these cotransporters as well as the Na+-dependent Cl−/HCO3− exchanger Slc4a8 in mouse cholangiocytes. Both Ae2 and Ae2a,b−/− mouse cholangiocytes showed expression of Slc4a4 mRNA, which was 1.8-fold up-regulated in Ae2 versus Ae2a,b+/+ mouse cholangiocytes (P < 0.01) (Fig. 3D). Slc4a7 mRNA was similarly detected in mouse cholangiocytes of the two genotypes (Fig. 3D), whereas no mRNA expression of Slc4a5 and Slc4a8 was found in any genotype (not shown). Western blotting confirmed a four-fold increased expression of Slc4a4 at the protein level in the Ae2 mouse cholangiocytes (Fig. 3E).
HCO3− Secretion Is Stimulated by cAMP in Ae2a,b+/+ Cholangiocytes but Inhibited in Ae2a,b−/− Cholangiocytes.
Rise of intracellular cAMP levels is known to stimulate Ae2 activity in human and rat cholangiocytes, resulting in an increase of HCO secretion to bile.1 As shown in Fig. 4A, application of a cAMP/SM to Ae2a,b+/+ mouse cholangiocytes increased the HCO extrusion rate (JOH− = 2.81 ± 0.23 versus 4.10 ± 0.48 mmol/L/minute; P < 0.05), whereas the same application to Ae2a,b−/− mouse cholangiocytes diminished HCO extrusion (JOH− = 6.15 ± 0.86 versus 3.64 ± 0.45 mmol/L/minute; P < 0.001) (Fig. 4B). On the other hand, analysis of intracellular cAMP levels (both baseline and following cAMP stimuli) indicated that Ae2a,b−/− mouse cholangiocytes have raised levels compared with Ae2 mouse cholangiocytes (Fig. 4C).
Acetylcholine Stimulates HCO3− Secretion in Ae2a,b+/+ but not in Ae2a,b−/− Mouse Cholangiocytes.
ACh is also known to regulate HCO3− secretion in the biliary epithelium.13–15 As shown in Fig. 5A, the presence of ACh increased HCO secretion in Ae2a,b+/+ mouse cholangiocytes (JOH− = 3.63 ± 0.48 versus 6.79 ± 0.56 mmol/L/minute; P < 0.001). However, the presence of ACh did not affect HCO secretion in Ae2a,b−/− mouse cholangiocytes (JOH− = 4.15 ± 0.66 versus 4.02 ± 0.65 mmol/L/minute) (Fig. 5B).
ATP Inhibits HCO3− Secretion in Ae2a,b−/− Mouse Cholangiocytes.
ATP has also been reported to play an indirect role in HCO secretion in the biliary epithelium.13, 30 In our experimental setting of perfused cells, however, the presence of 10 μM ATP did not modify the HCO3− secretion rate in Ae2 mouse cholangiocytes (JOH− = 5.53 ± 0.66 versus 4.68 ± 0.52 mmol/L/minute) (Fig. 6A), and interestingly resulted in diminished rate in Ae2a,b−/− mouse cholangiocytes (JOH− = 9.00 ± 0.83 versus 5.99 ± 0.34 mmol/L/minute; P < 0.001) (Fig. 6B).
Ae2a,b−/− Mouse Cholangiocytes Show Altered Cl− Secretion.
Previous studies in Ussing chambers to analyze the Cl− conductance in NRCs with recovered cell polarity indicated that cAMP stimulation leads to apical secretion of Cl− through Cftr activation, while purinergic stimulation by ATP leads to apical Cl− efflux through activation of Ca2+-dependent Cl− channels.24, 30, 31 We therefore decided to directly assess the Cl− conductance in polarized mouse cholangiocytes upon cAMP and ATP stimulations. As shown in Fig. 7A, cAMP stimulation of Ae2 mouse cholangiocytes resulted in increased values of short-circuit current intensity due to Cl− secretion (ΔIsc: 3.45 ± 0.38 μA/cm2), which was similar to that reported for polarized NRCs.24 However, cAMP stimulation of Ae2a,b−/− mouse cholangiocytes resulted in less intense Cl− secretion (ΔIsc: 0.95 ± 0.13 μA/cm2) (Fig. 7A). When we measured the area under the curve (AUC) of Isc, the cAMP response in Ae2 mouse cholangiocytes was found to be significantly lower than that in Ae2a,b+/+ mouse cholangiocytes (AUC 10.63 ± 1.071 versus 2.85 ± 0.22; P < 0.05) (Fig. 7A). Interestingly, these data are consistent with our observation that the levels of Cftr mRNA are much lower in Ae2 than Ae2a,b+/+ mouse cholangiocytes (13.7% versus 100%; P < 0.001) (Fig. 7B).
On the other hand, ATP stimulation led to significantly higher Cl− secretion in Ae2a,b−/− compared with Ae2 mouse cholangiocytes (AUC 49.68 ± 6.02 vesus 193.3 ± 5.58, respectively; P < 0.05) (Fig. 7C). It is noteworthy that this effect was correlated with higher relative values for increased intracellular Ca2+ in response to ATP in Ae2a,b−/− compared with Ae2 mouse cholangiocytes (AUC of fluorescence values: 68.05 ± 6.62 versus 32.54 ± 3.25, respectively; P < 0.001) (Fig. 7D).
Biliary HCO3− secretion is important for the alkalinization and fluidization of bile as well as for pHi homeostasis in cholangiocytes.1 In human and rat cholangiocytes, HCO secretion occurs essentially through AE2/Ae2-mediated Cl−/HCO3− anion exchange.4, 5 Previously, we hypothesized that abnormalities in the liver expression16, 17 and function18 of AE2 and concomitant alterations of HCO secretion in the biliary epithelium,32 may be involved in the pathogenesis of PBC. More recently, this hypothesis has been supported by our findings in Ae2a,b−/− mice, in which important PBC features are reproduced.20 We therefore decided to explore the transport mechanisms for HCO secretion in cultured mouse cholangiocytes from these mice. Studies involved monitoring pHi variations after two different types of perfusion maneuvers, i.e. with Cl−-free media (mainly an isethionate-based medium), and propionate-based perfusions. Although Cl−-withdrawal is designed to provoke Cl−/HCO3− exchange running in the opposite direction to the physiological one for Ae2 (with HCO entry into the cell instead of exit), this maneuver provided relevant information. It showed that, as with NRCs, normal mouse cholangiocytes display Cl−/HCO3− exchange activity that is entirely mediated by Ae2; the reverse-mode alkalinization upon Cl− withdrawal was completely absent in Ae2a,b-deficient mouse cholangiocytes (Fig. 1). Thus, in agreement with our previous data from AE2 gene silencing in normal human and rat cholangiocytes,4, 5 no additional Na+-independent Cl−/HCO exchangers (e.g., Ae1, Ae3, Slc26a3, Slc26a4, Slc26a6, or Slc26a7) or Na+-dependent exchange carriers (e.g., Slc4a8) mediate the AE activity observed in normal mouse cholangiocytes. But unlike the case with NRCs, in which alkalinization remained after simultaneous removal of Cl− and Na+, mouse cholangiocytes of both genotypes exhibited an opposite effect resulting in pHi acidification. This unexpected effect in murine cholangiocytes indicated the presence of an additional HCO3− secretory activity that is driven by Na+. Interestingly, this activity appeared more pronounced in Ae2 than Ae2a,b+/+ mouse cholangiocytes (Fig. 1). This observation was confirmed with propionate-based maneuvers (Fig. 2). These maneuvers—designed for directly measuring the acid-loading activity (e.g., HCO secretion, when cells try to recover their resting pHi from intracellular alkalinization)—also confirmed that the Na+-driven HCO3− secretory activity observed in both Ae2 and Ae2a,b−/− mouse cholangiocytes does not depend on extracellular Cl−, which is consistent with a Na+-HCO cotransport or NBC activity.25, 26 This NBC activity is electrogenic, as it is sensitive to membrane depolarization induced by high extracellular concentration of K+ during the propionate maneuvers. Such electrogenic Cl−-independent and Na+-driven HCO3− secretory activity was not detected in NRCs and therefore appears mouse-specific.
Our findings of up-regulated expression of Slc4a4 mRNA in Ae2 (versus normal mouse cholangiocytes) and absent Slc4a5 expression in both Ae2a,b−/− and Ae2 mouse cholangiocytes (Fig. 3) suggest Slc4a4 might be the responsible electrogenic NBC carrier. This concept is supported by our finding of increased expression of the SI4a4 protein in the Ae2a,b−/− cells. Slc4a4 up-regulation and subsequently increased NBC activity in Ae2 mouse cholangiocytes may account for an elevated HCO3− secreting activity in these cells, which might be viewed as an attempt to compensate Ae2 deficiency for pHi regulation. In this sense, cholangiocytes from Ae2a,b-deficient mice appear to be more capable of maintaining their resting pHi20 than Ae2a,b−/− mouse fibroblasts, in which Slc4a4 is not expressed.21
Stoichiometry of the NBC activity displayed by Slc4a4 is cell type–dependent; 1:3 stoichiometry runs for Na+and HCO efflux (acid load), whereas 1:2 works for influx of these ions into the cell (acid extrusion).26 Moreover, Slc4a4 stoichiometry may vary in the same cells, as reported for mouse kidney cells, in which protein kinase A–mediated phosphorylation in Ser982 shifts stoichiometry from 1:3 to 1:2, leading to decreased HCO3− efflux.28 This mechanism might also mediate the decrease in HCO secretory activity observed in cAMP/SM-stimulated Ae2a,b−/− mouse cholangiocytes (Fig. 4). Baseline NBC activity of these Ae2a,b-deficient cells is surprisingly high in spite of elevated cAMP production (also observed in Ae2a,b-deficient mouse fibroblasts),21 but further increased cAMP levels after cAMP/SM stimulation may trigger the decreasing effect. In the case of wild-type mouse cholangiocytes, cAMP/SM administration leads to overall increase in HCO secretion (similarly to what had been observed in NRCs with cAMP stimulation),5, 22 probably because stimulation of Ae2-mediated Cl−/HCO3− exchange overcomes the concurrent decrease in the NBC activity.
Experiments performed in isolated bile duct units with preserved bile duct lumen (thus resembling the in vivo situation) suggested that the stimulatory effect of cAMP on biliary HCO secretion is initiated by Cftr activation and Cl− efflux into bile; Cl−/HCO3− exchange would follow as a consequence of outside-to-inside transmembrane gradient of Cl− at a relatively high intracellular HCO concentration.9, 15, 33, 34 Moreover, cAMP activation of Cftr was recently associated with apical release of ATP.13 Autocrine/paracrine purinergic stimulation through the potent apically located ATP-responsive P2Y receptors and subsequent activation of Ca2+-dependent Cl− channels with additional efflux of Cl− to bile could therefore result in further increased Cl−/HCO3− exchange at the apical membrane.13, 23, 30, 31, 35 In our model of perfused cholangiocytes, however, released Cl− becomes diluted in the vast volume of perfusion media, with no chance for the alleged indirect stimulatory effects of the cAMP/SM on Cl−/HCO exchange after outside-to-inside transmembrane gradient of Cl−. Thus, cAMP appears to exert alternative stimulatory effects on the Ae2-mediated Cl−/HCO3− exchange in our Ae2 mouse cholangiocytes.
ACh was also reported to indirectly stimulate HCO3− secretion in rat isolated bile ducts.13, 15 ACh binds to M3 muscarinic receptors at the basolateral membrane of cholangiocytes, leading to increased intracellular Ca2+ levels, activation of Ca2+-dependent Cl− channels and further Cl− efflux into the bile duct lumen, and subsequent Cl−/HCO exchange for HCO3− secretion to bile.15 In our perfused Ae2 mouse cholangiocytes, we found that ACh increased HCO3− secretion similarly to cAMP. However, in contrast to cAMP, ACh showed no effect in our perfused Ae2 mouse cholangiocytes. Our data therefore suggest that ACh may stimulate HCO3− secretion through Ae2 without affecting the NBC activity. Interestingly, these effects were the opposite of those observed with ATP, which caused net inhibition of the NBC-mediated HCO secretion in our perfused Ae2a,b-deficient mouse cholangiocytes while having no overall effect on normal mouse cholangiocytes. Although different Ca2+ signaling pathways and/or different wave patterns of Ca2+ change might be involved in the differentiated effect of ATP on the NBC-mediated HCO3− transport versus that of ACh (for instance, calcium waves elicited by ATP can differ from those caused by ACh), additional Ca2+-independent mechanism may also be implicated.
We finally assessed Cl− efflux in polarized Ae2 and Ae2a,b−/− mouse cholangiocytes in reponse to cAMP and ATP stimulation (through Cftr and Ca2+-activated Cl− channels, respectively). Interestingly, cAMP stimulation elicited lower (but did not abolish) Cl− secretory activity in Ae2 than Ae2a,b+/+ mouse cholangiocytes. The diminished Cftr activity in the absence of Ae2 suggests a close functional interaction of these flux proteins in biliary cells, which is not surprising in view of their colocalization (together with aquaporin-1) in intracellular vesicles that migrate toward the apical membrane in response to cAMP.7 Thus far, no evidence for such functional interaction has been available, although previous studies in Xenopus oocytes and HEK293 cells have shown regulatory interactions of CFTR with electrogenic Cl−/HCO exchangers.36 Our data may shed some light on former findings in CFTR-deficient human and mouse cholangiocytes (isolated from patients with cystic fibrosis and Cftr−/− mice, respectively),37, 38 in which cAMP stimulation elicited no HCO3− secretion. These findings seemingly argued for CFTR/Cftr as the carrier mediating the cAMP-stimulated HCO secretion in cholangiocytes, and thus questioned AE2/Ae2 involvement. However, our present data in Ae2a,b-deficient mouse cholangiocytes, together with the aforementioned AE2-silencing results in human and rat cholangiocytes4, 5 confirm the crucial role of AE2/Ae2. In fact, our Ae2a,b-deficient cholangiocytes were completely unable to respond transporting HCO3− upon Cl− removal—with and without cAMP stimulation—while they exhibited cAMP-stimulated Cl− current, even though this was certainly diminished. Reduced cAMP-stimulated Cl− current in Ae2a,b-deficient mouse cholangiocytes might be due to both Cftr–Ae2 interaction failures and low Cftr gene expression. ATP stimulation, however, elicited higher values of Cl− currents in Ae2 mouse cholangiocytes, which suggests overactivation of Ca2+-dependent Cl− channels attempting to maintain Cl− homeostasis in these cells as reported for cystic-fibrosis cholangiocytes.37 This might be related with the fact that the ATP-stimulated increase in intracellular Ca2+ levels was more pronounced in Ae2a,b−/− than Ae2 mouse cholangiocytes.
In conclusion (see working model on Fig. 8), our data indicate that Ae2, in close interaction with Cftr, mediates biliary HCO3− secretion in mouse cholangiocytes through Cl−/HCO exchange, which is directly activated by both cAMP and ACh. Additionally, mouse cholangiocytes possess a Na+-HCO3− cotransport activity, which under circumstances of diminished Ae2 activity allows the biliary cells to maintain basal pHi and HCO secretion. Unlike Ae2, NBC activity in mice appears to be negatively regulated by cAMP and ATP. Finally, Ae2 deficiency in Ae2a,b−/− mouse cholangiocytes is associated with impaired Cftr activity and overactivation of Ca2+-dependent Cl− channels.
Ae2 mouse cholangiocytes resemble PBC cholangiocytes concerning baseline pHi homeostasis. Ae2a,b-deficient mouse cholangiocytes can maintain their resting pHi through the supplementary NBC-mediated HCO3− secreting activity, which is absent in human cholangiocytes.4, 18 However, PBC cholangiocytes may in turn maintain resting pHi because they generally display residual AE2 activity,18 due to diminished (rather than absent) AE2 expression in most PBC samples.17 Such a similar outcome might contribute to the fact that our Ae2 mice indeed reproduce several features of PBC20 and may therefore be employed as an animal model of the human disease.
We are grateful to Sarai Sarvide for her excellent technical assistance. We also thank A. Ferrer, J. T. Salas, and S. Melero for their valuable help.