Potential conflict of interest: Dr. Beuers is on the speakers' bureau of and received grants from Falk. He is also on the speakers' bureau of Zambon, Gilead, and Roche. He received grants from Intercept.
This review focuses on the hypothesis that biliary HCO secretion in humans serves to maintain an alkaline pH near the apical surface of hepatocytes and cholangiocytes to prevent the uncontrolled membrane permeation of protonated glycine-conjugated bile acids. Functional impairment of this biliary HCO umbrella or its regulation may lead to enhanced vulnerability of cholangiocytes and periportal hepatocytes toward the attack of apolar hydrophobic bile acids. An intact interplay of hepatocellular and cholangiocellular adenosine triphosphate (ATP) secretion, ATP/P2Y- and bile salt/TGR5-mediated Cl−/ HCO exchange and HCO secretion, and alkaline phosphatase–mediated ATP breakdown may guarantee a stable biliary HCO umbrella under physiological conditions. Genetic and acquired functional defects leading to destabilization of the biliary HCO umbrella may contribute to development and progression of various forms of fibrosing/sclerosing cholangitis. (HEPATOLOGY 2010)
The pathogenesis of chronic cholestatic liver diseases such as primary biliary cirrhosis (PBC), primary sclerosing cholangitis (PSC) and other fibrosing cholangiopathies remains enigmatic.1 Without adequate therapy, the prognosis is dismal and current treatment strategies may achieve stabilization but no resolution.1
Genetic factors contribute to the development of chronic cholestatic liver disease as indicated by sibling studies in PBC.2 An increased risk for first-degree relatives of PBC and PSC patients to develop the same disease also indicate a genetic background. Notably, various mutations of genes involved in bile formation present with a sclerosing/fibrosing cholangitis-like phenotype. Among these genes are ABCB4 coding for the phospholipid floppase, MDR3 (Mdr2 in mice),3 and CFTR coding for a cyclic adenosine monophosphate (cAMP)-sensitive Cl− channel. These gene mutations are associated with the clinical entities of ABCB4 deficiency and cystic fibrosis–associated liver disease, respectively.1
Most recently, anion exchanger 2 (AE2), a variant of the Cl−/HCO exchanger, has been shown to influence prognosis in patients with PBC under treatment with ursodeoxycholic acid (UDCA).4 This finding supports the view that impaired AE2 activity and thereby reduced biliary HCO secretion may play a key role in the pathogenesis of PBC.5-9
A variant of GPBAR1, the gene coding for the G-protein–coupled bile acid receptor 1, also called TGR5, appeared as a likely disease gene in the first genome-wide association analysis of primary sclerosing cholangitis.10 TGR5 is expressed on apical cholangiocyte membranes and is putatively involved in cAMP-dependent modulation of cholangiocellular HCO secretion.
Thus, functional modifications in proteins involved in apical transport of pH modifying bile contents may contribute to development and progression of chronic forms of sclerosing/fibrosing cholangitis such as PBC, PSC, cystic fibrosis–associated liver disease, and ABCB4 deficiency.
Strategies of the Cholangiocyte to Survive in an Unfriendly Environment
The cholangiocyte is exposed to millimolar concentrations of hydrophobic bile salts,11, 12 which are toxic to other cells such as hepatocytes at moderate micromolar levels.13 Resistance against these noxious compounds and their cytolytic potential is therefore essential. Which strategies help cholangiocytes survive in the unfriendly environment of bile?
Formation of Mixed Micelles of Bile Salts and Phospholipids.
One protective mechanism is the formation of mixed micelles of phospholipids and bile salts in bile.11 High millimolar amounts of bile salts are buffered by micelle formation with phospholipids. However, although this mechanism protects cells from bile salts in micelles, it has no effect on the toxicity of bile salt monomers that are always present at submicellar concentrations. Formation of mixed micelles is critically dependent on adequate biliary phospholipid secretion. Its impairment by mutations of ABCB4/MDR3 leads to progressive familial intrahepatic cholestasis (PFIC type 3) in children and in milder forms to sclerosing cholangitis, ductopenia, and occasionally biliary cirrhosis in adults.3 Thus, micelle formation in bile appears to be crucial for bile ductular integrity.
Biliary HCO Secretion: A Key Defense Mechanism?
A second protective mechanism known as dilution of bile or flushing of bile is more speculative. This mechanism involves secretion of an alkaline, HCO-rich, mainly cholangiocyte-derived fluid11, 14 that reduces the concentration of toxic compounds in bile. Flushing leads to a decrease of micellar bile salts but does not result in complete depletion. Thus, flushing should hardly affect the concentration of potentially toxic bile salt monomers below their critical micellar concentration in bile, although this remains to be proved. In addition, if the sole purpose was dilution, cholangiocytes (and periportal hepatocytes) could initiate other mechanisms of fluid secretion15 rather than secrete alkalinizing HCO by way of anion exchangers such as AE2.
Biliary HCO secretion serves a number of well-known functions: it sustains bile flow and confers the gallbladder and intestinal mucous layer its proper viscosity; it facilitates the disposal of certain endobiotics and xenobiotics; and it generates part of the alkaline tide necessary for optimal digestion of various nutrients within the intestine. Human biliary HCO secretion by far exceeds that of rodents and is responsible for 25%-40% of total bile flow versus 5%-10% or less in various rodents.16 Biliary HCO secretion in man is up-regulated after meal ingestion, thus increasing bile pH from ≈7.3 during fasting to ≈7.5 while bile salt concentrations in bile nearly double. What is the purpose of this enormous HCO secretion by biliary epithelia, particularly in humans?
Glycine conjugates of bile salts with a pKa of ≈4 are the major dihydroxy bile salts in human bile that predominate over taurine conjugates with a pKa of ≈1-2.12 Both taurine and glycine conjugates of bile salts are resistant to cleavage by pancreatic enzymes during intestinal passage in man.11 Rodents have a more hydrophilic, less toxic bile salt pool with mainly taurine conjugates11 and secrete fewer phospholipids into bile.17 On the extracellular side, mammalian membranes carry a net negative surface charge. To establish electroneutrality, protons are attracted, which would cause a more acidic pH close to the apical surface of cholangiocytes. In this relatively acidic environment, it can be expected that considerable amounts of glycine-conjugated bile salts will be protonated. These apolar, protonated, glycine-conjugated bile acids might pass cell membranes by simple diffusion.18 Indirect evidence for this assumption comes from early experimental work in gastric mucosa cells, which are continuously exposed to an acidic environment. In mouse gastric mucosa cells, glycochenodeoxycholate (pKa 4.2) induced mucosal injury only at pH 1 and 3, but not pH 5, as observed in light and electron microscopic studies.19 Taurocholic acid (pKa 1.8) at pH 1, but not taurocholate at pH 7, disrupted gastric mucosal barrier in dogs by way of simple passive bile acid uptake.20 Moreover, glycocholic acid accumulation in gastric mucosal cells of rabbits and guinea pigs was by far more pronounced at an acidic than at a neutral pH.21 In line with these observations, bile acids at pH 4.0, but not pH 7.4, have been shown to induce oxidative stress and DNA damage in human esophageal epithelial cells.22
In cholangiocytes exposed to millimolar concentrations of bile salts, acidification of bile at the apical membrane constitutes a real danger to cellular structures, including cell membranes and mitochondria. Leakage of cytochrome c out of mitochondria is a well-recognized stimulus for apoptosis. Cholangiocytes are thus under a constant threat to become damaged and eliminated by way of apoptosis, although other forms of bile acid–induced cholangiocyte death cannot be excluded.23
Cholangiocyte Survival Strategy: The Biliary HCO Umbrella
We hypothesize that cholangiocytes protect their apical surface against protonated apolar hydrophobic bile acid monomers by maintaining an alkaline pH above the apical membrane (Fig. 1). We think that a vital step in this process is the secretion of HCO at amounts high enough to form a HCO umbrella on the outer surface of the apical membrane.
Isoforms of the Cl−/HCO exchanger, AE2, are responsible for the vast majority of biliary HCO secretion. Membrane-bound carboanhydrase may propagate the HCO umbrella at the apical surface, which keeps the pH of bile high. A recent proteomics study also identified putatively soluble carboanhydrase in human bile.24 The protective HCO umbrella would markedly raise the pH of the luminal fluid near the apical surface and lead to deprotonation of apolar hydrophobic bile acids, rendering them unable to permeate membranes in an uncontrolled fashion.
This protective function of the biliary HCO umbrella might be equivalent to the protective layer of membrane-bound and secreted mucins in the stomach, the colon or the gallbladder mucosa cells.25 Human gallbladder mucosal cells express various membrane-bound (MUC3, MUC1) and secreted (MUC5B, MUC6, MUC5AC, MUC2) mucins.25 In contrast, cholangiocytes of the smallest intrahepatic bile ductules do not show relevant mucin expression, whereas large bile ducts express MUC3 and MUC5B and may occasionally express MUC1, MUC2, MUC5AC, and MUC6.25 Thus, the biliary HCO umbrella may form the key protective mechanism of human intrahepatic apical cholangiocyte membranes against apolar protonated hydrophobic bile acids. In line with this assumption, AE2 immunoreactivity in human liver has been demonstrated on apical membranes of hepatocytes as well as small and large cholangiocytes,26 whereas cholangiocytes of small bile ductules in experimental animals have been shown to contribute little to biliary HCO formation.27 We think that this protective mechanism is especially well developed in the human biliary tree as an adaptation to the human bile salt pool characterized by high levels of glycine-conjugated hydrophobic bile salts (pKa 4-5)—in contrast to, for example, the murine bile salt pool, which is dominated by taurine-conjugated hydrophobic bile salts (pKa 1-2)—although it probably also functions at a lower intensity in our evolutionary relatives. This is supported by the observation that rats can dramatically up-regulate their cholangiocyte HCO production.14
Failure to keep bile pH high enough to deprotonate bile acids supposedly has a detrimental effect on cholangiocytes. The importance of maintaining an alkaline pH to avoid cholangiocyte damage by bile acids is underlined by the unique vascularization of the biliary tree. Purely arterial blood is supplied by the biliary plexus. Focal ischemia can have severe effects on the biliary epithelium and its secretory function. The detrimental effect of ischemia on cholangiocytes and formation of ischemia-induced portal casts28 could be explained by a collapse of the biliary HCO umbrella potentiating bile acid toxicity.
Biliary HCO formation is under tight control of local factors such as bile salts or purinergic agonists as well as of visceral neurohormonal factors including secretin, cholinergic and adrenergic agents, vasoactive intestinal peptide (VIP), glucagon, glucagon-like peptide-1, and somatostatin. We restrict our discussion to local factors contributing to formation of a stable HCO umbrella. For the role of visceral neurohormonal factors which, like secretin, also may depend on local factors to induce HCO formation,29 we refer to recent reviews on the neurohormonal control of the adaptive cholangiocyte response.30
Local Control of the Biliary HCO Umbrella
The bile salt sensing receptor TGR5 (GPBAR-1) is localized on the tip of the cilia of apical mouse and human cholangiocyte membranes31, 32 reaching beyond the hypothetical HCO umbrella, thus sensing real bile composition. An obvious function of luminal bile salt sensing would be to modulate the cellular defense strategies by varying HCO secretion through direct stimulation of secretory channels or proteins and/or insertion/retrieval of key carriers/channels such as AE2 and cystic fibrosis transmembrane conductance regulator (CFTR).33 Recent in vitro studies strongly support this view: stimulation of TGR5 was shown to increase Cl− secretion of human gallbladder mucosa by way of CFTR in a cAMP-dependent way.34 The reportedly low levels of the apical bile salt transporter expression in cholangiocytes would argue in favour of a sensor function rather than a high throughput transport function of apical bile salt transporter. Rates of HCO formation as the major human biliary defense mechanism against bile acid toxicity in the intrahepatic biliary tree would then be under tight control of at least two types of bile acid sensors.
Adenosine Triphosphate and P2Y Receptors.
Extracellular adenosine triphosphate (ATP) is increasingly recognized as an important signaling molecule in the regulation of bile secretion and composition.35, 36 ATP and its metabolites adenosine diphosphate (ADP), adenosine monophosphate (AMP), and adenosine are present in human bile collected from the common bile duct.37 Cholangiocytes express purinergic receptors from the P2Y family on their membranes and cilia, and these receptors translate the signal of adenosine nucleotides in bile into intracellular cAMP levels as well as changes in cytosolic-free calcium [Ca++]I and activation of different conventional, novel, and atypical protein kinase C (PKC) isoforms.32, 38-41 Binding of ATP to apical P2Y receptors leads to enhanced Cl− excretion through yet undefined Cl− channels by way of Ca++-, PKC-, or cAMP-dependent mechanisms in rats and humans.40, 41 Extracellular Cl− then stimulates Cl−/HCO exchange over the membrane.42
Extracellular ATP has also been shown to up-regulate the inflammatory cytokine interleukin-6 in bile duct epithelium in a cAMP- and Ca++-dependent way through the activation of purinergic receptors43 and to recruit macrophages, possibly through induction of intercellular cell adhesion of molecules on the epithelium.44 P2Y receptors on the basolateral membrane are probably involved in recognition of extracellular ATP due to destruction of neighboring cells. The activation of receptors on the apical membrane leads to a net alkalinization of bile through Ca++- and/or cAMP-mediated stimulation of Cl− secretion.33 Forskolin-induced HCO secretion was shown to be dependent on luminal ATP. In isolated bile ductules, ATP hydrolyzing apyrase reduced forskolin- and thus cAMP-induced HCO secretion by ≈80%, and P2Y blockade with suramin abolished intracellular Ca++ increase and HCO secretion.33
Hepatocytes and cholangiocytes release ATP in a paracrine/autocrine fashion by yet unresolved molecular mechanisms.35, 45 Release through undefined ATP channels and CFTR-mediated ATP release35, 46 but also ATP release through exocytosis of ATP-enriched vesicles47 or even maxi-anion channels described in macula densa cells of the rabbit kidney, rat cardiomyocytes, and mouse astrocytes48 have been discussed. Whatever the molecular mechanisms of hepatobiliary ATP release are, it is attractive to speculate that cholestatic injury per se may hamper biliary ATP release and thereby ductular HCO secretion as targeting of vesicles, vesicular exocytosis, and membrane insertion of transport proteins into their target membrane are impaired in cholestatic liver.49, 50 Thus, cholestatic injury might even be a primary culprit for a defective biliary HCO umbrella. In this regard, the recent finding of bile flow–induced mechanosensitive Ca++- and PKCζ-dependent ATP release and associated ATP-dependent Cl− secretion from human biliary cells as well as rat cholangiocytes deserves particular attention.51 It is attractive to link mechano-sensitive ATP and Cl− secretion to the stabilization of the biliary HCO umbrella through stimulation of Cl−/HCO exchange when higher amounts of bile salts as a major driving force of stimulated hepatocellular bile flow reach the cilia of cholangiocytes.
Given their putative role in stabilization of the biliary HCO umbrella, but also their potential proinflammatory effects, ATP levels in bile have to be tightly controlled. Alkaline phosphatase catalyzes the dephosphorylation of ATP to ADP, AMP, and eventually adenosine in various tissues. In the intestine, alkaline phosphatase mainly located in the apical intermicrovillous glycocalyx together with P2Y1 receptors and ATP-evoked HCO secretion form a surface microclimate pH regulatory system, which protects the mucosa effectively against acid injury and bacteria.52 The P2Y1 purinergic receptor mediates ATP-evoked HCO secretion in the rat intestinal epithelium.52 Enhanced HCO secretion renders the extracellular pH more alkaline, thereby enhancing the catalytic capacity of alkaline phosphatase, which in turn increases the rate of ATP degradation to ADP and AMP, forming a negative feedback loop.52 Similar to the intestine, hepatic alkaline phosphatase is mainly located on the apical membrane of the glycocalyx-covered biliary epithelium (Fig. 2) and is anchored to the plasma membrane by way of glycosylated phosphatidylinositol. Biliary alkaline phosphatase diminishes biliary HCO secretion and total bile flow in rats, whereas levamisole, an inhibitor of alkaline phosphatase, increases the activity of the Cl−/HCO exchanger, AE2, HCO secretion, and bile flow.53 As extracellular ATP and ADP stimulate HCO secretion in a CFTR-dependent way, a protective HCO umbrella can be maintained by a balance between ATP-dependent HCO secretion and HCO-sensitive alkaline phosphatase activity, mediating ATP breakdown similar to the intestinal surface microclimate pH regulatory system.52
The role of other ATP, ADP, and AMP dephosphorylating enzymes such as CD73 (which is also known as ecto 5′ nucleotidase) in stabilizing the biliary HCO umbrella deserves further study. CD73 is involved in the generation of allergic inflammation and modulation of Cl− secretion in the airways.54
Therapeutic Interventions to Protect the Biliary HCO Umbrella
UDCA is the established first-line treatment of PBC and intrahepatic cholestasis pregnancy.1 UDCA exerts anticholestatic and antifibrotic effects in various cholestatic disorders55 and stimulates biliary HCO secretion in cholangiopathies at moderate doses.7 It also acts as a posttranscriptional secretagogue in hepatocytes55 as well as cholangiocytes.56 UDCA-induced stimulation of biliary HCO secretion may include purinergic signaling33, 35 and Ca++/cPKCα/PKA-dependent49, 55, 57 or possibly mitogen-activated protein kinase–dependent targeting58 of key transporters and channels such as AE2 and CFTR, similar to its posttranscriptional secretagogue activity on the bile salt and conjugate export pumps BSEP and MRP2 in cholestatic hepatocytes.55, 57 In rodent studies, UDCA and taurine-conjugated UDCA were shown to be inducers of hepatocyte ATP release into bile.59 UDCA-induced apical ATP release in cholangiocytes is probably CFTR-dependent and induces an intracellular rise in Ca2+ through P2Y-mediated purinergic signaling, leading to enhanced Cl− secretion and eventually stimulation of HCO secretion through Cl−/HCO exchange.35
The C23 homologue of UDCA, norursodeoxycholic acid (norUDCA), is an effective anticholestatic, anti-inflammatory, and antifibrotic agent in experimental sclerosing cholangitis as observed in Mdr2−/− mice.60, 61NorUDCA is a potent stimulus of biliary HCO secretion in humans62 and experimental animals.61 This has been explained by a cholehepatic shunt mechanism,11 a mechanism not observed for UDCA at therapeutic doses. Whatever the exact molecular mechanisms leading to potent stimulation of biliary HCO secretion by norUDCA, it is attractive to speculate that norUDCA exerts its anticholestatic, anti-inflammatory, and antifibrotic effects at the level of the biliary tree60, 61 by effectively stabilizing the biliary HCO umbrella against the toxic effects of protonated apolar bile acids.
The Biliary HCO Umbrella: Summary and Outlook
We hypothesize that biliary HCO secretion in humans serves to maintain an alkaline pH near the apical surface of hepatocytes and cholangiocytes to prevent the uncontrolled membrane permeation of protonated glycine-conjugated bile acids. Functional impairment of biliary HCO formation or its regulation may lead to enhanced vulnerability of cholangiocytes and periportal hepatocytes toward the attack of hydrophobic bile acids. An interplay of hepatocellular and cholangiocellular ATP secretion, ATP/P2Y- and bile salt/TGR5-mediated Cl−/ HCO exchange and HCO secretion, and alkaline phosphatase-mediated ATP breakdown may guarantee a stable HCO umbrella under physiological conditions.
Our hypothesis offers an attractive mechanistic link between AE2 deficiency/functional impairment in PBC patients5-8, 63 and development of fibrosing cholangitis of interlobular bile ductules in these patients. Impaired biliary HCO formation as in PBC would render small ductules most vulnerable for bile acid–induced cell damage, because they do not express mucins. Thus, immunological alterations in PBC could be the consequence rather than the cause of bile acid–induced cholangiocyte damage in PBC as proposed.2
A defective biliary HCO umbrella could furthermore contribute to explain the heretofore enigmatic pathogenesis of various other fibrosing cholangiopathies. Genome screening of patients with PSC has disclosed GPBAR-1/TGR5 as a susceptibility gene10 that, when defective, may affect the biliary HCO umbrella. TGR5 is expressed on cilia of intrahepatic and extrahepatic bile ducts, the site where bile duct alterations in PSC are observed.
Cystic fibrosis–associated liver disease due to CFTR deficiency and sclerosing cholangitis/nonanastomotic bile duct stricturing in the posttransplantation setting after vagal denervation both involve potential impairment of HCO formation. The vulnerability of the denervated biliary tree in the liver graft after transplantation may in part originate from a not yet fully developed arterial circulation around the bile ducts and the associated difficulty to maintain an alkaline pH at the apical surface of cholangiocytes.
The same mechanism of defective biliary HCO secretion may even hold for the biliary cast syndrome after ischemic or septic bile duct injury in the intensive care setting28 when acute hypoxia in the biliary plexus may lead to disruption of the biliary HCO umbrella, and subsequently to cholangiocyte damage due to the unhindered actions of protonated glycine-conjugated bile acids.
The putative impact of intact pH regulation and purinergic signaling at the level of the apical membrane of the cholangiocyte links disruption of the biliary HCO umbrella to immunological findings in autoimmune disorders of the biliary tract. In individuals with a genetic or acquired predisposition to impaired formation of a stable biliary HCO umbrella, up-regulation of purinergic signaling, ATP-dependent HCO secretion, and alkaline phosphatase expression can be expected as cholangiocytes (and hepatocytes) attempt to induce a stable biliary surface microclimate pH regulatory system. Consequently, extracellular ATP as a strong chemotactic molecule could then attract immune cells, thus leading to (auto-) immune attack against cholangiocytes as a consequence of an unstable biliary HCO umbrella.
Our hypothesis needs confirmation by experimental studies both in vitro and in vivo. Fundamental questions are: (1) Does a pH gradient exist at the apical cholangiocyte membrane, and if so, which elements contribute to it? (2) Are glycine-conjugated bile salt uptake and bile salt–induced cholangiocyte damage pH-dependent? (3) Is biliary pH lower in chronic fibrosing/sclerosing cholangiopathies such as PBC or PSC than in subjects without chronic cholangiopathies? (4) If so, does medical treatment (such as UDCA in PBC) normalize biliary pH? Confirmation of the concept of a biliary HCO umbrella would have clinical impact both in further unraveling the pathogenesis of chronic fibrosing cholangiopathies and in developing therapeutic strategies that would focus on strengthening the biliary HCO umbrella in fibrosing cholangiopathies beyond the effects observed with UDCA so far.
We gratefully acknowledge the stimulating discussions and critical reading of the manuscript by Alan F. Hofmann, Gustav Paumgartner, and Bruno Stieger.