Calcium balance is essential for a multitude of physiological processes, ranging from cell signaling to maintenance of bone health. Adequate intestinal absorption of calcium is a major factor for maintaining systemic calcium homeostasis. Recent observations indicate that a reduction of gastric acidity may impair effective calcium uptake through the intestine. This article reviews the physiology of gastric acid secretion, intestinal calcium absorption, and their respective neuroendocrine regulation and explores the physiological basis of a potential link between these individual systems.
The average adult human body contains ∼1.6% calcium, which relates to ∼1,120 g in a 70-kg individual (743). Ninety-nine percent of the calcium is stored in bone and teeth and is therefore inaccessible to most physiological processes (743). Although the amount of the immediately accessible 11 g (1%) of calcium may seem miniscule, this fraction represents a pivotal constituent of our body. It serves a broad diversity of roles, which range from intracellular signaling and maintenance of membrane integrity to muscle contraction and neuronal transmission.
To allow for these calcium-dependent processes to function, our body undertakes extensive measures to keep the intracellular and extracellular calcium concentrations and the gradient between these two compartments stable. The extracellular calcium concentration is typically clamped at ∼1.1 mM, whereas the intracellular environment is kept at a 10,000 times lower concentration. In consequence, relatively small disturbances in calcium homeostasis can lead to severe symptoms, such as cardiac arrhythmias or cognitive dysfunctions. To maintain eucalcemia, our body is therefore tightly regulating the balance between calcium absorption by the intestine and calcium excretion by the kidney. In addition, calcium is deposited in or extracted from bone, which serves as a dynamic calcium reservoir. These three organ systems, i.e., the intestine, the kidney, and bone, are precisely controlled by a complex endocrine network, which primarily consists of the calcitropic hormones: 1,25-dihydroxyvitamin D [1,25(OH)2-vitamin D], parathyroid hormone (PTH), and calcitonin.
This review mainly focuses on the question as to how calcium enters the body through the intestine and how this mechanism is regulated via the endocrine system. Furthermore, the process of gastric acid secretion as related to calcium homeostasis will be reviewed in detail. This may seem surprising, as gastric acid secretion and intestinal calcium absorption are two distinct physiological processes, which on first examination may not seem to be interdependent. However, recent clinical studies suggest that there may be a relationship between reduced gastric acid secretion and increased risk for sustaining bone fractures, which asks the question whether we need gastric acid to absorb calcium efficiently through the intestine, or whether the stomach exerts endocrine functions that impact bone health. Indeed, it has been put forward several decades ago that gastric acid solubilizes calcium that is then complexed with other dietary constituents, thereby allowing for a more efficient absorption in the intestine (18, 520, 699, 797). Furthermore, it is long known that a partial or complete resection of the stomach results in decreased bone density, also leading to fractures (58, 305, 732, 876). The stomach, the intestine, and bone are therefore functionally more intertwined than one may initially assume. This review will independently analyze the processes of gastric acid secretion, intestinal calcium absorption, and their respective neuroendocrine control and will conclude with a critical attempt at illustrating where these two seemingly independent organ systems intersect in terms of calcium homeostasis and bone health.
II. GASTRIC ACID SECRETION
The stomach is a unique organ that fulfills multiple roles. The main function of the gastric mucosa is to secrete concentrated hydrochloric acid, which provides a chemical barrier against ingested pathogens and aids in the digestion of foodstuffs. To achieve these functions, the gastric gland contains specialized cells that pump protons into the gastric lumen in an effort to acidify the contents of the stomach. These cells are known as parietal cells, or oxyntic cells. Since concentrated acid is a noxious substance, the gastric mucosa has to undertake extensive measures to protect itself from tissue injury. The protection is accomplished by secreting mucus from mucus neck cells, but also by tightly regulating the secretion of acid (see sect. IIB). A variety of specialized endocrine cells in the gastric mucosa are involved in the regulation of gastric acid secretion. A perturbation of either protective mechanism can lead to severe tissue damage, resulting in gastric ulcers. This section discusses the process of how gastric acid is secreted by reviewing the molecular mechanism underlying acid secretion in the parietal cell and its neuroendocrine regulation.
A. Apical Ion Transport in the Parietal Cell
The gastric parietal cell is responsible for acidifying the stomach by secreting concentrated acid. Gastric acid secretion depends on the apical extrusion of three ions. Protons are pumped into the gastric lumen by a proton pump, the gastric H+-K+-ATPase, to acidify the gastric content to a pH of as low as 1. Chloride is secreted via apical chloride channels to ensure formation of HCl and to provide the counter-ion conductance to protons. Lastly, potassium leaves the parietal cell apically in a recycling mechanism, thereby fueling reciprocal proton transport by the H+-K+-ATPase (FIGURE 1). It has been demonstrated in numerous investigations that disruption of one of these ion transport mechanism renders the parietal cell incapable of secreting gastric acid (705, 820, 1013, 1029).
The gastric H+-K+-ATPase belongs to the family of P2-type ATPases, which also includes the ubiquitous Na+-K+-ATPase and the sarcoplasmic reticulum Ca2+-ATPase (SERCA). As the name implies, it exchanges one intracellular hydrogen ion for one extracellular potassium ion at the expense of ATP. ATP is provided to the pump by a large network of mitochondria, which occupy up to 40% of the cell volume, making the parietal cell one of the most mitochondria-rich cells in the body (292). In the process of proton extrusion, the H+-K+-ATPase can overcome a massive acid gradient of 6 pH units, which is necessary to achieve sufficient gastric acidification. The pump itself is a heterodimer, consisting of a α subunit and a β subunit, while the individual pumps assemble as (αβ)4 tetramers on the parietal cell surface (1). The α subunit consists of 10 transmembrane domains and contains the catalytic site, which mediates ion exchange. The β subunit stabilizes the α subunit and is heavily glycosylated (41, 1105). Mutational analysis of the glycosylated asparagine residues suggests that these sites are critical for adequate membrane delivery of the entire pump (41, 1105). Furthermore, the β subunit prevents a reversal of ion transport by a “ratchet”-like mechanism, which allows H+-K+-ATPase to pump against the imposed high proton gradient (4, 294). Both subunits share a significant degree of homology to Na+-K+-ATPase (697, 1012). This close relationship to other P2-type ATPase has historically been exploited for homology modeling of H+-K+-ATPase based on the crystal structure of SERCA, which had been acquired in several conformational states (762, 815, 1092, 1093). Recently, however, direct structural information on H+-K+-ATPase has been obtained by electron crystallography, also in the presence of the acid pump antagonist SCH28080 (2–4).
In the resting parietal cell, H+-K+-ATPase is stored in tubulovesicles throughout the cell (292). Following neuronal or hormonal stimulation (see sect. IIB), these vesicles are postulated to fuse with the apical pole, which is characterized by multiple microvilli-lined membrane invaginations, the so-called secretory canaliculi (292). This distinct apical morphology of the parietal cells maximizes cell surface and thereby allows for insertion of a high number of proton pumps per cell following stimulation. The changes in membrane morphology and insertion of H+-K+-ATPase are extremely dynamic to ensure fine regulation of gastric acid secretion (973). H+-K+-ATPase containing tubulovesicle fusion relies on SNARE complex formation. In particular, the SNARE proteins syntaxin 3/7/12/13, VAMP2/8, and SNAP-25 were implicated to be candidates mediating this process (548–550, 624). The functional significance of these proteins was, for example, demonstrated in primary rabbit parietal cell cultures expressing a SNAP-25 mutation, which was shown to reduce their capacity to secrete gastric acid (548).
Apart from SNARE proteins, the small GTPases of the rab family (rab2/11a/25/27b) are involved in the regulation of H+-K+-ATPase vesicle trafficking (147, 293, 386, 387, 1049, 1070). Functional data especially substantiate the importance of rab11a and rab27b. In parallel to SNAP-25 defective cells, parietal cells transfected with a rab11a and rab27b mutant secrete acid less effectively (293, 1049).
After stimulation, in the off-phase of gastric acid secretion, H+-K+-ATPase has to be retrieved from the plasma membrane for recycling (336). It is plausible that the initial step of this process relies on the formation of clathrin-coated pits and subsequent vesicle budding. Indeed, clathrin was identified fairly early on H+-K+-ATPase containing tubulovesicles, although a functional role was not demonstrated (813). One of the multiple clathrin binding proteins is Huntingtin interacting protein 1 related (Hip1r) which aids in vesicle formation and membrane trafficking (309). It is strongly expressed in parietal cells, especially in the vicinity of secretory canaliculi (522). Functionally, Hip1r-deficient animals present with a decreased number of parietal cells, loss of tubulovesicles, and decreased acid output (522, 561).
2. Chloride secretion
Apical chloride secretion provides the second component for the formation of concentrated HCl and maintains overall electroneutrality during acid secretion. The importance of chloride efflux for the process of gastric acid secretion has been established in the 1980s. Patch-clamp measurements demonstrated the presence of chloride conductance on the apical pole of the parietal cell in Necturus, the human parietal cell line HGT-1, and rabbit parietal cells (259, 935, 940). All reports demonstrated a sensitivity of the chloride current to cAMP or histamine, which is a common second messenger promoting acid secretion or a direct acid secretagogue, respectively (259, 935, 940). Simple flux measurements in isolated parietal cell vesicles had indicated the presence of a chloride conductance pathway even earlier (232, 895, 1169). In these early experiments, inhibition of chloride flux with chloride channel blockers also abolished proton transport which underlines the necessity of intact chloride secretion for acid secretion to take place (232, 895, 1169). However, the molecular identity of the chloride pathway remained elusive. Today, at least three candidates have been put forward as potential mediators of apical chloride secretion in the parietal cell: the cystic fibrosis conductance regulator (CFTR), chloride channel protein 2 (ClC-2), and solute carrier 26 A 9 (SLC26A9) (FIGURE 1).
CFTR represents a common apical chloride conductance pathway in a broad variety of epithelia, such as the airways, intestine, and pancreas. Its mutation is responsible for the most widespread inherited disease, namely, cystic fibrosis (CF), which results in increased mortality due to secretory defects and concomitant infections. The presence of CFTR has been confirmed in gastric mucosa by in situ hybridization, albeit at low quantities (1044). Nevertheless, functional measurements in isolated gastric glands demonstrated a decreased acid secretory capacity in animals carrying the most common mutation responsible for CF (ΔF508) (1013). Furthermore, acid secretion was reduced in wild-type animals when a specific CFTR inhibitor was applied (1013). Although these observations may suggest a direct involvement of CFTR in the process of chloride secretion, it is plausible that CFTR rather has a regulatory effect on H+-K+-ATPase (1013). In other tissues, CFTR can interact with a variety of ion transport proteins, such as NHE, forming regulatory complexes, making an interaction with H+-K+-ATPase plausible (1013).
ClC-2 has been proposed as an alternative chloride secretion pathway to CFTR in other epithelia, such as the lung and intestine (207, 404, 675, 766). ClC-2 has been cloned from rabbit gastric mucosa, which led to the hypothesis that the channel may also be involved in acid secretion (706). However, follow-up investigations revealed that the role of ClC-2 is much less clear. The studies revealed controversial results regarding the channel's expression in the gastric mucosa (488, 706, 1001). While the initial observations reported mRNA and cDNA expression in rabbit gastric mucosa, no protein could be detected in human and rat gastric glands (488, 706, 1001). The importance of ClC-2 in the stomach has further been severely challenged by the creation of a ClC-2 (−/−) animal model. Although ClC-2-deficient animals present with a distinct phenotype characterized by testicular and retinal abnormalities, no defect in acid secretion was observed (118).
Lastly, evidence suggests that chloride may leave the apical pole via SLC26A9, a chloride-bicarbonate antiporter. Both SLC26A9 and an antiporter from the same anion exchanger family (SLC26A6) have been detected in the tubulovesicles of parietal cells (845, 1179, 1180). Concerning the functional involvement, the authors speculate about two potential roles SLC26A9 may play in parietal cell physiology. Being a chloride-bicarbonate exchanger, its activation would entail alkalinization of the gastric lumen by bicarbonate efflux and simultaneous chloride uptake (1180). Since this would neutralize H+-K+-ATPase-mediated proton extrusion, it has been suggested that SLC26A9 activates in the off-phase of acid secretion to neutralize tubulovesicular pH during vesicle retrieval (1180). Alternatively, SLC26A9 may function as a chloride secretion pathway that contributes to acid secretion. This hypothesis is based on the observation that SLC26A9 can also exhibit the behavior of a bona fide chloride channel, rather than an anion antiporter (88, 281). Undoubtedly, further functional investigations are needed to delineate its exact role in the parietal cell. Its genetic disruption, however, leads to a severely altered parietal cell morphology that is characterized by dilation of gastric glands, loss of tubulovesicles, and decreased acid output (1180). Although these results do not answer whether SLC26A9 serves as an apical chloride efflux pathway, they indicate that it may be necessary for normal parietal cell function.
3. Potassium recycling
Even before the identification of H+-K+-ATPase, it has been observed that potassium is necessary for acid secretion to take place (335). To prevent the luminal depletion of potassium, which would impair proton pumping by H+-K+-ATPase, potassium has to leak through potassium channels or transporters into the gland lumen to ensure adequate supply to H+-K+-ATPase (FIGURE 1). This process is referred to as potassium recycling. Early flux measurements in isolated H+-K+-ATPase containing parietal cell vesicles had already indicated the presence of a large potassium conductance during H+-K+-ATPase activity (1169). The exact molecular identity of the potassium efflux pathway is, however, under debate. The list of candidates that have been put forward to be responsible for potassium recycling during acid secretion is long and includes KCNQ1 (Kv7.1), KCNJ10 (Kir4.1), KCNJ15 (Kir4.2), KCNJ2 (Kir2.1.), and KCC4.
KCNQ1 is a typical “shaker”-like six transmembrane spanning domain voltage-gated potassium channel (1144). It was initially identified in the heart, where its mutation can be responsible for cardiac arrhythmias (1144). Yet, studies in KCNQ1 (−/−) animals revealed no electrocardiographical abnormalities (641). Rather than suffering from cardiac abnormalities, these animals surprisingly exhibited a distinct gastric phenotype with gastric hyperplasia, dilated gastric glands, vacuolated parietal cells, hypochlorhydria, and hypergastrinemia (641). This observation led to the speculation that KCNQ1 may be the channel responsible for potassium recycling. Subsequently, immunohistochemical studies confirmed a colocalization of the channel with H+-K+-ATPase, and acid secretion was shown to be inhibited by pharmacological blockade (253, 391). Direct measurement of acid secretion in KCNQ1 (−/−) mice with modified Ussing chambers (pH stat) later confirmed the initially observed hypochlorhydria (1029). Interestingly, luminal substitution of potassium could rescue the acid secretory deficit, indicating that hypochlorhydria ensued from a true lack of apical potassium secretion rather than a general morphological defect of the KCNQ1 (−/−) parietal cell (1029).
KCNQ1 is a peculiar channel in that it has a low conductance in acidic environments. In the context of the extreme acidic milieu surrounding the parietal cell, this would impede its function as a potassium recycling pathway. To circumvent this limitation, KCNQ1 attaches to a regulatory subunit (KCNE2), which modulates the channel's gating properties and current amplitude (253, 391, 1087). Coassembly with KCNE2 activates KCNQ1 at acidic pH values and thus facilitates the process of potassium secretion into the gland lumen (391, 436). The importance of KCNE2 for proper channel function is underlined by the observation that KNCE2 (−/−) animals display a phenotype similar to KCNQ1 (−/−) mice, i.e., hypochlorhydria, altered parietal cell morphology, and hypergastrinemia (917).
B) KIR CHANNELS.
Apart from KCNQ1, several members of the inward-rectifier potassium channel (Kir) family have been proposed to be involved in gastric acid secretion, albeit the amount of functional evidence supporting a role of these channels is smaller and the field is divided about the relative contribution of each channel. Kir 2.1, 4.1, 4.2, and 7.1 were all confirmed on an mRNA level in gastric mucosa (353, 431, 707). On a protein level, immunohistochemistry demonstrated colocalization of Kir 2.1, 4.1, and 4.2 with H+-K+-ATPase (353, 431, 556, 707). Cell fractionation experiments further indicated trafficking of Kir 4.1 and 4.2 to the cell surface, following parietal cell stimulation (431, 556). A most recent observation monitored acid secretion in Kir 4.1 (−/−) mice (1028). Surprisingly, loss of Kir 4.1 results in augmented rather than impaired acid secretion, accompanied by upregulated H+-K+-ATPase expression (1028). This makes a contribution of Kir 4.1 to potassium recycling highly unlikely. Instead, it has been proposed that the channel may balance excessive potassium loss through KCNQ1 and may be involved in membrane recycling (1028). In summary, more investigations will be necessary to clarify the roles of the individual Kir channels.
Apart from being secreted through channels, potassium and chloride may exit the parietal cell through transporters. This alternative hypothesis is corroborated by a recent observation of Fuji et al. (352). The group reported that the K-2Cl cotransporter KCC4 coimmunoprecipitates with H+-K+-ATPase in apical membrane fractions of parietal cells (352). Furthermore, flux measurements in H+-K+-ATPase containing vesicles showed decreased chloride and proton transport under pharmacological blockade of KCC4, suggesting a functional coupling of KCC4 to H+-K+-ATPase (352). Although the hypothesis that both potassium and chloride leave the cell via a transporter is intriguing, the observation is, as of now, solitary and needs further experimental validation.
B. Control of Acid Secretion
Gastric acid secretion is subjected to precise regulation. The complex regulatory machinery that orchestrates the secretion of gastric acid consists of hormonal (gastrin, somatostatin), paracrine (histamine, somatostatin), and neuronal components (FIGURE 2). The need for this tight regulation is highlighted by conditions that lead to a hypersecretion of gastric acid, such as Zollinger-Ellisson syndrome (ZES; gastrinoma). Gastric hypersecretion can overcome the measures our body undertakes to protect itself from the acid and thereby lead to peptic ulcers. A fine on-demand regulation of acid secretion is thus pivotal to ensure the balance between an adequately low intragastric pH and tissue protection.
According to the well-established model of acid secretion, the parietal cell is activated by neuronal input from the vagus nerve, endocrine input from gastrin-producing G cells, and paracrine input from histamine-producing enterochromaffin-like (ECL) cells (FIGURES 1 AND 2). The distinct substances released by these cells, i.e., acetylcholine, histamine, and gastrin, directly or indirectly stimulate the parietal cell by inducing insertion of H+-K+-ATPase at the apical membrane and are thus commonly referred to as acid secretagogues. The main inhibitor of parietal cell acid secretion is somatostatin, which is secreted by the D-cells of the gastric mucosa (FIGURES 1 AND 2). Because of the complexity of the network that controls the release of acid into the stomach, it has been historically challenging to dissect the relative role of each individual regulatory component. Without a doubt, knockout models have greatly aided us in the last years to gain a more profound understanding of this process, despite their limitations of chronic compensation. The subsequent chapter aims to summarize the key players in our canonical model of acid regulation.
1. Cholinergic stimulation/vagus nerve
Since the seminal experiments conducted by Pavlov on dogs, we know that the mere prospect of food ingestion or sham-feeding is sufficient to trigger the secretion of gastric acid (833). This first of three phases of acid secretion is called the cephalic phase and is mostly mediated through the vagus nerve (595, 725, 910). Hence, before the advent of pharmacological inhibitors, vagotomy has been an effective surgical procedure to control acid-related disorders (301).
The parietal cell receives neuronal input from the vagus nerve that is relayed via cholinergic postganglionic enteric fibers in the enteric nervous system (ENS) (FIGURES 1 AND 2). In addition, the vagus nerve activates G-cells to release gastrin, resulting in an indirect stimulation of the parietal cell. Direct cholinergic activation occurs mostly via muscarinic M3 receptors, which have been identified on the surface of the parietal cell (507, 541, 846). The M3 receptor is a classic seven-transmembrane domain GPCR. Predictably, knockout of M3 receptors leads to an impairment of gastric acid secretion and compensatory hypergastrinemia due to negative feedback (9). Following acetylcholine binding, M3 receptor activation mostly causes an increase in intracellular calcium concentrations (44, 1163). Calcium rises in response to PLC-mediated IP3 generation and subsequent mobilization from intracellular stores (190). The primary kinases activated by the M3 receptor are protein kinase C (PKC) and calcium/calmodulin-dependent protein kinase II (CaMKII) (136, 196, 314–316, 773, 774, 1095). While activation of CaMKII has a clear stimulatory effect on acid secretion, PKC has been reported to have dual effects, although reports of an inhibitory role predominate numerically (23, 73, 136, 196, 313, 314, 316, 597, 755, 773, 1095). It has been postulated that the expression of different PKC isoforms may account for this dichotomy (313, 314). Current evidence suggests that the PKC-α isoform has a suppressing effect by trans-inhibiting CaMKII activity, whereas PKC-ϵ increases the baseline levels of intracellular calcium, thereby sensitizing the parietal cell to subsequent stimulation (313, 314). Apart from PKC and CaMKII activation, cholinergic signaling activates parietal cell MAPKs, which is partially a downstream effect of PKC activation (771, 1039, 1062, 1063). MAPK activation seems to have a biphasic effect on acid secretion (acute inhibition and chronic augmentation) and also serves as a mediator of trophic responses in the parietal cell. For example, prolonged MAPK activation (72h) has been shown to serve as a maturation and differentiation signal leading to a transformation of parietal cell morphology in vitro (1039). The change in morphology is accompanied by a downregulation of H+-K+-ATPase gene expression (1039). As of now, it is challenging to put these findings into a physiological perspective.
In addition to M3 receptors, M1 receptors have also been implicated to play a role in the process of acid secretion. This hypothesis was derived from the observation that the M1 receptor is expressed in gastric mucosa and that its blocker pirenzepine can inhibit gastric acid secretion (29, 466). Most evidence pointed to an expression of M1 on ECL-cells, where it was speculated to regulate the release of histamine (437, 507). More recent findings somewhat surprisingly report that pirenzepine also suppresses acid secretion in M1-deficient animals. Furthermore, these animals show a normal phenotype in terms of acid output (10). These observations question both the involvement of M1 receptors in acid secretion and the specificity of pirenzepine.
Lastly, knockout studies point towards a contribution of the M5 receptor to the regulation of acid secretion, as its deletion correlates with decreased acid output (10). Yet, M5 receptor mRNA could only be detected in whole stomach homogenates, but not in gastric mucosa per se, making its localization to the submucosal enteric plexus more likely (10).
Gastrin has been discovered in 1906 by John S. Edkins, who injected gastric extracts of pig and cat stomachs into the jugular vein of cats and observed a subsequent increase in acid secretion (298). Gastrin is a peptide hormone that is produced in specialized G-cells, located in the antral section of the stomach (FIGURE 2) and endocrine cells in the duodenum, small intestine, colon, pancreas, testis, and pituitary. It is the main mediator of the so-called gastric phase of acid secretion, which initiates when the ingested food enters the stomach. The gastric phase accounts for the majority of the acid secretory response of the stomach.
The gastrin cDNA encodes a 101-amino acid pre-pro-hormone that undergoes extensive posttranslational processing (113, 519, 551, 552, 1161). In brief, the pre-pro-hormone is first cleaved NH2-terminally to create progastrin and then truncated to two main core proteins, G17 and G34, which can exist in glycine extended (G17-Gly; G34-Gly) or terminally amidated (G17-NH2, G34-NH2) forms. Furthermore, a fraction of progastrin (∼47% in humans) is sulfated at Tyr66 in the course of its passage through the Golgi apparatus, thereby giving rise to sulfated and nonsulfated isoforms of gastrin (22). Sulfation has no influence on the acid secretory response, as the affinity to the gastrin receptor remains unchanged (399, 596). G17-NH2 is the main circulating form that mediates the secretory effects of gastrin. Although the glycine-extended forms have a low affinity towards the gastrin receptor (CCK2) and thus play no role in gastric acid secretion (they are four to five orders of magnitude less potent in inducing acid secretion), it is still important to acknowledge their existence (178, 722). First, they serve as substrates for the synthesis of amidated gastrin and are cosecreted with gastrin by the G-cells (1040, 1051). Second, they potentiate the acid secretory response to amidated gastrin, although they have no intrinsic ability to induce acid secretion (178). Third, progastrin and glycine extended gastrins were shown to act as a proliferative signal, especially in the colon (20, 482, 994, 1145). This is also of pathophysiological relevance as both forms can promote cancer growth by presumably inhibiting apoptosis and inducing angiogenesis (71, 87, 900). For example, it was shown that overexpression of progastrin in mice is a predisposing factor for the development of colorectal or bronchoalveolar cancers (587, 1017).
B) REGULATION OF RELEASE.
Gastrin is released by the G-cell in response to a variety of stimuli of different origin. Direct neuronal stimulation of the G-cell occurs via ACh and gastrin releasing peptide (GRP), which are released by postganglionic neurons of the enteric nervous system. The postganglionic fibers themselves receive input from the efferent fraction of the vagus nerve (86, 485). On the other hand, food-related signals, such as calcium, amino acids, and amines, can also directly trigger gastrin secretion (FIGURE 2) (257). The secretory stimuli culminate in an increase in intracellular calcium concentrations, leading to vesicle fusion and gastrin secretion. The main inhibitory signal for gastrin secretion is somatostatin, which reaches the G-cells in a paracrine fashion from neighboring D-cells (632, 975).
With regard to the neuronal control of gastrin secretion, it is generally thought that vagal stimulation increases the release of gastrin, although some conflicting evidence exists (310, 694). Latest experiments that assessed local gastrin concentrations utilizing microdialysis, however, clearly show an increase in gastrin levels following acute electrical vagal stimulation (310). The vagus nerve then synapses on neurons of the ENS, which are thought to release either the neurotransmitter ACh or GRP on a G-cell, leading to secretion of gastrin (295, 635, 694, 960, 978). It should be noted that a recent investigation failed to observe increased gastrin levels, following exogenous GRP administration in humans (456). Yet, GRP itself serves as a clear acid secretagogue, although potentially not via gastrin (456). Whether these conflicting observations are attributable to species differences (most earlier observations utilized rodent models) remains to be elucidated. The ENS is also thought to mediate parietal and G-cell activation in response to mechanical distension of the stomach (455, 962, 977). The neurohormonal response to gastric stretch is an integral part of the gastric phase of acid secretion. Closer examination, however, reveals that the reports are very conflicting in that it is not clear whether a pure mechanical distension stimulates or inhibits gastrin release (455, 664, 803, 962, 977). A biphasic model characterized by initial inhibition of gastrin secretion under low volumes followed by stimulation under high volumes has been suggested, but awaits further confirmation (977).
Dietary components, such as amino acids and calcium, can directly promote the secretion of gastrin and can thus sustain acid secretion in the gastric and intestinal phase as digestion progresses (652, 1074). A rise in serum calcium concentrations evokes a similar effect. The correlation between calcium and gastrin is discussed in a separate section (see sect. VD2). It has been unclear for a long time as to how these dietary components activate the G-cell. An involvement of the ENS has been proposed as the most likely explanation in the past. More recent observations, however, strongly indicate that the calcium-sensing receptor (CaSR) represents the molecular link between luminal dietary constituents and G-cell activation (325). The CaSR and its role in the stomach are discussed separately and shall only be summarized at this point (FIGURE 8) (see sect. IVD). First, the same dietary components, i.e., amino acids, amines, and calcium, which have all been shown to trigger gastrin release also function as activators of CaSR (652, 1074). Second, CaSR is expressed on the apical and basolateral side of the G-cell, which allows it to act as nutrient sensor both in the gastric lumen and the circulation (142, 182, 886). Third, direct activation of the CaSR is known to stimulate acid secretion (145, 291, 373). Finally, and most importantly, CaSR (−/−) animals lack the gastrin secretory response to intraluminal instillation of peptone, calcium, and phenylalanine (325). In light of this evidence, it is highly likely that CaSR is the long elusive luminal nutrient sensor that regulates the secretion of gastrin from the G-cell.
The plasma levels of gastrin are closely tied to the intragastric pH. Low intraluminal pH is a potent inhibitor of gastrin release, which serves as a negative-feedback mechanism to impede an overproduction of acid. Conversely, a more alkali intragastric pH induces the secretion of gastrin, which accounts for the commonly observed hypergastrinemia in states of acid suppression, such as during proton pump inhibitor (PPI) therapy. The pH dependency of serum gastrin levels is mainly relayed via somatostatin, as acid directly stimulates somatostatin release (see sect. IIB4). Somatostatin, released by neighboring antral D-cells, in turn acts as the main inhibitor of gastrin secretion (FIGURE 2). The physical proximity to G-cells allows for a fine paracrine regulation of gastrin release. Although it is generally accepted that intragastric pH mostly modulates local somatostatin levels, the G-cell may also directly sense intragastric pH via CaSR. CaSR is acid sensitive, and it has been shown that isolated rat G-cells secrete less gastrin when the extracellular pH is dropped from 7.4 to 5.5 (569). However, more investigations are needed to substantiate this evidence. Furthermore, gastrin release is also inhibited by neuronal regulation by the ENS. The neurotransmitter galanin has been demonstrated to exert a direct inhibitory effect on isolated G-cells (695, 961).
C) CELLULAR EFFECTS.
Following secretion, gastrin enters the bloodstream and acts on its target cells in an endocrine fashion. Its half-life is determined by its rate of elimination from the plasma which mainly occurs by metabolism in the kidney, gut, and brain (419, 420). The importance of renal elimination is corroborated by the observation that patients with renal failure present with higher plasma gastrin levels (818, 1075).
The two primary target cells of gastrin are the histamine-secreting ECL cell and the parietal cell. Gastrin exerts its functions via binding to the cholecystokinin receptor type 2 (CCK2), a seven transmembrane domain G protein-coupled receptor, which is expressed on mature parietal and ECL cells, but also on gastric stem cells (560, 596, 608, 769, 772, 904). On the ECL cell, gastrin binding causes the release of histamine, which in turn stimulates the parietal cell in a paracrine fashion (FIGURE 2) (see sect. IIB3) (412). This activation cascade is commonly referred to as the gastrin-histamine axis. Evidence for a direct, i.e., nonhistamine-relayed, activation of H+-K+-ATPase in the parietal cell by gastrin exists, but is far less substantiated (459, 1024, 1025). Gastrin may sensitize the parietal cell to subsequent secretagogue stimulation, rather than acting as a bona fide secretagogue itself. Canonically it is widely accepted that gastrin exerts its physiological effects mostly via activation of ECL cells (24, 1131). Knock-out of gastrin leads to a severe impairment of basal and stimulated acid secretion (179, 347). Apart from stimulating acid secretion, gastrin serves as a pivotal proliferative signal for the gastric mucosa in general (60, 410, 534, 631, 816). It is commonly observed that elevated plasma gastrin levels lead to substantial mucosal proliferation (60, 410, 534, 631, 816). This phenomenon has been extensively described in various knockout animals suffering from hypochlorhydria and concomitant hypergastrinemia, but also in patients with ZES (39, 515, 584, 1066). The source of mucosal cell proliferation is progenitor cells located in the isthmus region of the gastric gland (545). Expression of the CCK2 has been confirmed on several gastric progenitor cells (560, 769). Furthermore, gastrin has been shown to stimulate cell migration from the progenitor region along the gastric gland axis (578). Mucosal hyperplasia thus ensues most likely via a direct activation of precursor cells by gastrin. Although hypergastrinemia causes a generalized mucosal hyperplasia, ECL cells seem to be particularly regulated by gastrin, as their relative fraction compared with other mucosal cells increases under prolonged gastrin exposure (60, 410, 631). Conversely, the absence of the CCK2 almost entirely eliminates mature ECL-cells from the gastric mucosa (177, 622). [Somewhat surprisingly this does not occur when gastrin itself is knocked out (179, 347).] It should be noted that the M3 receptor seems to be necessary as a cofactor mediating the trophic effects of gastrin, as its absence is associated with a normal mucosal phenotype despite elevated serum gastrin levels (see above) (9). The mechanism underlying this interdependency between gastrin and the M3 receptor is as of now elusive. The cholinergic and gastrin systems also seem to be intertwined with regard to acid secretion. In the absence of the CCK2 receptor, the parietal cell's acid secretory response to the secretagogue carbachol (ACh analog) is abolished, while the response to histamine remains intact (543). Again, one can only speculate about the molecular basis of this interaction.
In conclusion, gastrin is the most important activator of acid secretion in the stomach. The role of gastrin, and especially its glycine extended forms, has evolved beyond being a mere acid secretagogue to being an important global regulator of cell growth and differentiation. Furthermore, the regulation of gastrin by the levels of plasma calcium provokes the question as to whether gastrin itself in turn has an impact on global calcium homeostasis. A subsequent section makes an attempt at addressing this question (see sect. VD2).
3. Histamine/ECL cell
Histamine has been discovered as early as 1910 by Dale, Barger and Laidlow in extracts of ergot fungi (63, 237). In 1920, Popielski for the first time described its effect on the secretion of gastric acid (866). He observed that subcutaneous administration of histamine resulted in increased acid secretion (866). Furthermore, he concluded that this effect was independent of the vagus nerve, as secretion still took place after vagotomy and administration of atropine. This led Popielski to postulate that histamine exerts its effects directly on the level of the gastric gland (866). The hypothesis that histamine acts in a paracrine fashion on parietal cells and that its release is regulated by the levels of gastrin has been put forward for the first time by Emmelin and Kahlson in 1944 (306). At this point, the cellular source of histamine was still obscure. It was only in the late 1960s that histamine had been histochemically localized to the ECL cells of the gastric gland (413, 1084).
A) SYNTHESIS AND REGULATION OF RELEASE.
Histamine is the effector of the gastrin-histamine axis and directly stimulates the parietal cell to secrete hydrochloric acid (FIGURE 2). Histamine is derived from the amino acid histidine, which is enzymatically converted to histamine by l-histidine decarboxylase (HDC) (957). The effects of genetic HDC deletion are predictably severe: animals lacking HDC have a low basal acid output that does not respond to exogenous administration of gastrin (1066).
Histamine is stored in secretory granules of the ECL cell and is released into the surrounding milieu in response to stimulation by gastrin and neuronal signals. Stimulation by gastrin occurs via activation of its GPCR CCK2 (772, 904). Gastrin affects the ECL cell in multiple ways. First, gastrin exposure increases the levels of HDC expression by enhancing its transcription and inhibiting its degradation, to allow for increased synthesis of histamine (268, 331). The molecular mechanism underlying increased HDC transcription is fairly well understood. Following CCK2 activation, increased transcription of HDC is mediated via a PKC- and ERK-dependent pathway (470, 472). The HDC gene promoter is then activated by at least three distinct nuclear factors which bind to gastrin response elements, resulting in gene transcription (889, 890). Apart from augmenting gene transcription, gastrin regulates the degradation of HDC, which further increases intracellular enzyme levels (331, 1214). Second, gastrin enhances the transcription of the vesicular monoamine transporter type 2 (VMAT2; SLC18A2), which is responsible for accumulating histamine in the secretory vesicles (376). Similarly to HDC, this effect depends on PKC and ERK activation and binding of a nuclear factor to a gastrin response element in the VMAT2 promoter region (164, 1154). It should be mentioned that gastrin regulates the transcription of a plethora of other genes which serve a diverse array of roles, ranging from growth to metabolism (346). Amongst many others these include chromogranin A, which is essential for granule packaging and is a precursor of pancreastatin (see sect.VD3) (231). Third, gastrin induces the fusion of secretory granules and the release of histamine into the gland environment. Secretion follows a biphasic elevation of intracellular calcium concentrations after activation of CCK2 (1201). The biphasic increase has been proposed to result from initial IP3-mediated release from intracellular stores, which is followed by subsequent influx of calcium via L-type calcium channels from the extracellular space (1201). The importance of intracellular store mobilization has been contested by a different group, which proposed that solely influx trough L-type, and to a lesser extent N-type, calcium channels triggers the secretory response (673). Lastly, gastrin has a trophic effect on the ECL cell (see sect. IIB2).
Apart from gastrin, ECL cells are stimulated by pituitary adenylate cyclase activating polypeptide (PACAP), which is a neuropeptide expressed in the ENS of the gastric mucosa (737, 1054). PACAP has homology to vasoactive intestinal polypeptide (VIP) and binds to a distinct receptor (PAC-1) on the ECL cell (1207, 1208). Binding of PACAP to PAC-1 induces release of histamine (672, 798, 944, 1207). Similar results have been obtained with VIP, which is attributable to partial agonism at PAC-1 (798, 941). Historically, investigations yielded controversial results with regard to the effects of exogenously administrated PACAP on acid secretion. Both an inhibition and stimulation of acid secretion following PACAP injection are reported (760, 862, 944, 1207). This discrepancy is most likely attributable to the fact that PACAP can also act as an agonist of the VIP receptor (VPAC) on the somatostatin-secreting D-cell, leading to a concomitant suppression of acid secretion by somatostatin release (1207). Indeed, if an anti-somatostatin antibody is injected into rats simultaneously with PACAP, acid secretion is elevated threefold from baseline (compared with 1.5-fold in the absence of an anti somatostatin antibody) (1207). Evidence points to the fact that the PACAP-stimulated release of somatostatin is of particular importance in the mouse, as most studies showing a suppression of acid secretion after PACAP administration were conducted in murine models.
PACAP has very similar effects on the ECL cell as gastrin. Similarly to gastrin, PACAP causes histamine release by increasing intracellular calcium concentrations via calcium influx through L-type, but also ligand-gated calcium channels (673). In further analogy to gastrin, PACAP upregulates the expression of HDC and exerts trophic effects on the ECL cell (590, 729, 810). Contradictory results with regard to the effects of acetylcholine on histamine release exist. It has been reported that acetylcholine can either stimulate or has no effect on the secretion of histamine in in vitro experiments on isolated ECL cells (481, 672, 674, 941, 946). In vivo application of muscarinic agonists, followed by measurement of histamine concentrations using microdialysis, also yielded no evidence for cholinergic stimulation (798). Conversely, it is well accepted that adrenergic stimulation leads to an increase in histamine release; however, the physiological relevance of adrenergic activation of ECL cells is not entirely clear (636, 672, 674, 798, 871, 941).
The ECL cell is inhibited by a variety of substances, the most prominent of which is somatostatin (204, 590, 798, 941). Somatostatin is produced in D-cells of the oxyntic mucosa and reaches the ECL cell in a paracrine fashion where it binds to the somatostatin receptor (SST2 and potentially SST5) (FIGURE 2) (570, 873). Receptor binding leads to inhibition of histamine exocytosis via blockade of mostly L-type calcium channels (105). This impedes the elevation of intracellular calcium concentrations caused by ECL activators, such as gastrin (see above) (873). In addition, somatostatin also inhibits the proliferation of ECL cells (570). Somatostatin can thus be seen as the global hormonal antagonist to gastrin with regard to ECL cell function and proliferation. The neuronal inhibition of ECL cells is mainly carried out by the neuropeptide galanin (105, 672, 798, 1209). Galanin is localized to neurons of the ENS and demonstrated an inhibitory effect on histamine secretion in in vitro and in vivo models (105, 302, 672, 731, 798, 1209). Similarly to somatostatin, the molecular mechanism underlying its inhibitory effect is an interference with calcium signaling via closure of L-type calcium channels (105). Lastly, prostaglandin E and nitric oxide also act as inhibitors of histamine release (105, 554, 798, 1002). Although neuropeptide YY (PYY) and calcitonin gene-related peptide (CGRP) have also been implicated in playing a role in ECL cell regulation, a detailed discussion is omitted in light of contradictory results which range from stimulation to inhibition of secretion (672, 674, 798, 1210).
B) CELLULAR EFFECTS.
As mentioned earlier, stimulation of the ECL cell is translated into an elevation in intracellular calcium concentrations, leading to exocytosis of preformed histamine-containing secretory vesicles. The molecular mechanism of vesicle fusion with the apical membrane relies on the formation of the core SNARE complex, consisting of syntaxin, synaptobrevin, and SNAP-25. Synaptotagmin presumably acts as a calcium sensor relaying the intracellular calcium signal to the vesicle fusion protein apparatus. The expression of all SNARE complex proteins has been confirmed in the ECL cell (471, 477, 1215). In accordance with these findings, introduction of the neurotoxins tetanus toxin light chain and botulinum toxin, which cleave constituents of the SNARE complex apparatus and thereby render it nonfunctional, result in inhibition of histamine secretion (477).
Very small amounts of histamine are sufficient to induce acid secretion. Histamine acts via the H2 receptor on the parietal cell, which has been discovered by Sir J. W. Black in 1972 (106). For this seminal discovery, he was later awarded the Nobel Prize in Physiology and Medicine. The H2 receptor belongs to the family of seven-transmembrane domain GPCRs. Its activation predominantly leads to increases in the intracellular levels of cAMP, but also of calcium, which serve as stimulatory signals for H+-K+-ATPase trafficking (67, 189, 738, 840, 1026, 1143). In analogy, pharmacological agents that elevate cAMP, such as IBMX or forskolin, induce acid secretion (1026, 1191). The increase in cAMP is due to activation of adenylate cyclase via Gs. The role of calcium in the process of histamine secretion remains a controversial matter. First, the mechanism leading to histamine-induced increases in intracellular calcium has been subject of discussion. Evidence exists that histamine can, apart from adenylate cyclase, also activate PLC, leading to calcium release from intracellular stores (607, 1142, 1143). Conversely, it has been suggested that the observed increases in intracellular calcium are a byproduct of cAMP-mediated PKA activation, which in turn can regulate the opening of calcium channels (144, 189, 840). Second, it is questionable to what degree the calcium signal is an integral and necessary part of the acid secretory response to histamine (738, 840). Chelation of the transitory calcium increases with BAPTA abolishes only the secretory response of isolated gastric glands to histamine by ∼40%, while it completely eliminates the response to cholinergic stimulation (738). Also, live fluorescence imaging in isolated glands showed no spatiotemporal correlation between the histamine-induced increases in calcium and the onset of acid secretion, thereby questioning an involvement of calcium in the secretory response (840).
H2 receptor knockout animals effectively illustrate the significance of the histamine-gastrin axis in gastric physiology. Lack of the H2 receptor leads to a complete failure of gastrin or histamine to induce acid secretion (584). The secretory response to carbachol, however, remains intact (584). Hypergastrinemia develops as a feedback mechanism with the aim of reestablishing acid secretion, leading to mucosal hypertrophy (584). In light of the central role of the H2 receptor in parietal cell physiology, it has been successfully used as a pharmacological target with the aim of suppressing gastric acid output (see sect. IIC2).
Somatostatin was isolated for the first time in 1973 from ovine hypothalamus and characterized as an inhibitor of growth hormone release from the pituitary gland (124). A few years later somatostatin was identified in endocrine cells of the stomach, which we now know as D-cells (638).
A) SYNTHESIS AND REGULATION OF RELEASE.
Somatostatin is a peptide hormone that exists in two primary forms that differ in their respective peptide length. The most abundant form in the gastric mucosa is somatostatin-14 (consisting of 14 amino acids), whereas somatostatin-28 only constitutes a minute fraction of the total gastric somatostatin content (198, 1125). The two forms of somatostatin are cleavage products of a larger 116-amino acid pre-prohormone (pre-prosomatostatin), which in turn is processed to the 92-amino acid-long prosomatostatin (1000). It should be mentioned that other cleavage products, such as antrin or somatostatin-28(1–12) exist and are secreted together with somatostatin (77, 885). Their physiological significance is, however, less well understood.
Somatostatin is the global antagonist of the acid secretagogues. It is produced by intestinal and gastric D-cells, the latter of which exist in two populations in the stomach (61): an antral population locally inhibits the release of gastrin from G-cells, whereas a population localized to the acid-producing oxyntic mucosa directly regulates the parietal cell and inhibits histamine release from the ECL cell (FIGURE 2) (19). The morphology of the D-cell is characteristic in that it possesses long cytoplasmic processes, which allow it to communicate with and regulate neighboring cells in a paracrine fashion (620, 632). It is worthwhile to distinguish the two populations of gastric D-cells, as each population possesses unique physiological properties (1202).
The antral D-cell is mostly regulated by the local concentrations of gastrin, cholecystokinin, and intraluminal pH. Gastrin induces somatostatin secretion from D-cells, which causes reciprocal inhibition of gastrin release from neighboring G-cells, thereby creating a local negative-feedback loop (976, 1011, 1202). The molecular mechanism underlying this loop is, however, less clear. CCK2 receptor is, if at all, only expressed at very low levels in the antral mucosa (749, 905, 967). It has been proposed that gastrin stimulates somatostatin release in the antrum in a receptor-independent mechanism (1202). This may be accomplished via direct cell-cell contacts between the G- and the D-cell, which have been demonstrated with electron microscopy (620). Conversely, evidence for cholecystokinin and its stimulatory role for somatostatin release via CCK1 is more substantiated (749, 905, 967, 1202). Cholecystokinin is structurally closely related to gastrin (both share an identical 5-amino acid COOH terminus) and also exists in various peptide lengths (767). It is secreted by I-cells of the small intestine following protein and fat-rich chyme entering the duodenum, and thus represents a classical mediator of the intestinal phase of acid secretion (594). As its name implies, cholecystokinin has originally been described as a stimulator of gallbladder contraction; however, its inhibitory influence on gastric acid secretion is now well accepted and extensively described (593, 1203). Cholecystokinin can bind to both the CCK1 and CCK2 receptor with almost equal affinity, whereas the actions of gastrin are almost exclusively mediated by the CCK2 receptor. The dual affinity of cholecystokinin would imply a possible stimulatory effect on acid secretion via activation of CCK2 on ECL cells; however, in vivo the inhibitory effect mediated by activation of CCK1 and CCK2 on D-cells prevails (593, 966, 1202, 1203).
One of the most important stimulators of D-cell secretion is the intragastric pH. A seminal observation demonstrating a correlation between gastric acidity and the amount of secreted somatostatin was made in dogs in the 1970s. It has been shown that the amount of somatostatin directly increases in antral venous blood following gastric HCl infusion, while somatostatin levels were unaffected in venous blood from the oxyntic mucosa (982). Similar observations were later made in isolated mouse stomach, however without topographic discrimination (975). Two main hypotheses as to how somatostatin is regulated by intragastric pH exist. The first states that the D-cell can directly act as a pH sensor, and the second postulates that the pH sensing is mediated by neurons, which in turn act on D-cells. To accomplish putative direct pH sensing, several antral D-cells are equipped with a distinct morphological feature. They possess apical projections that are in contact with the glandular lumen, potentially allowing them to constantly monitor the intraluminal milieu (620). These D-cells have been termed open type. Conversely, the D-cells of the oxyntic mucosa are mostly of the closed type, meaning that they are embedded in the mucosa without luminal contact. The molecular identity of the putative apical pH sensor remains elusive. However, the presence of CaSR, which has pH sensing properties, was recently confirmed in preliminary studies on the D-cell and may represent a possible candidate for this mechanism (770). Apart from directly acting on D-cells, the effect of pH on somatostatin secretion may be mediated via afferent spinal neurons. Over 80% of the spinal afferent neurons contain the neuropeptide CGRP (397, 758, 1047, 1116). Perfusion models of antral sleeves have shown that the acid-induced rise in somatostatin is accompanied by a concomitant increase in the concentrations of the neuropeptide CGRP (708). Furthermore, application of a CGRP receptor blocker inhibited the release of somatostatin following acid exposure (708). As D-cells are known to express the CGRP receptor, an involvement of CGRP in acid sensing is plausible (558). Again, this provokes the question of how CGRP-containing neurons may sense acidity on a molecular basis. The acid-sensitive channels transient receptor potential vanilloid channel (TRPV1) and the acid-sensing ion channel 3 (ASIC3) had been proposed as molecular acid sensors; however, latest experiments have shown that the increase in CGRP still occurs in the genetic absence of the channels (47, 96, 163).
Neuropeptides of the gastric ENS that stimulate the secretion of somatostatin include PACAP and VIP, which both bind to the VPAC receptor expressed on D-cells (199, 657, 1207). The presence of VIP and PACAP containing neurons, which integrate signals from the vagus nerve, has been demonstrated in the gastric mucosa (302, 737). Furthermore, cholinergic signals can act on the antral D-cell via the M3 receptor to promote secretion of somatostatin (140). This is in sharp contrast to D-cells from the oxyntic mucosa that are inhibited by cholinergic signals (197, 200, 1182). As mentioned earlier, the D-cells in the oxyntic mucosa also differ in their morphology. D-cells in the oxyntic mucosa are of the closed type and have thus not been implicated to participate in acid sensing. They exert their acid-suppressive effects by the paracrine regulation of ECL and parietal cells. Further functional divergence between antrum and the oxyntic mucosa has been demonstrated in the regulation of the somatostatin mRNA. For example, suppression of acid secretion with omeprazole in fasted animals markedly decreased somatostatin mRNA levels in the antrum, whereas the levels in the oxyntic mucosa were affected to a much lesser extent, which further corroborates the hypothesis that the antral cells are involved in luminal chemosensation (945).
B) CELLULAR EFFECTS.
The effects of somatostatin on its target cells are mediated by the SST2 receptor. Knockout of the receptor causes a 10-fold increase in basal acid output, which exemplifies the pivotal role somatostatin plays as a global suppressant of acid secretion (715). Somatostatin acts on all main cell types that are involved in the process of acid secretion, i.e., parietal cells, ECL cells, and G-cells (FIGURE 2). The inhibition of the G- and ECL cells has been discussed in the respective sections. In the parietal cell, somatostatin has a clear direct inhibitory effect on secretagogue-induced acid secretion (827, 1177). This effect is partially attributable to activation of Gi, leading to inhibition of adenylate cyclase and a subsequent decrease in intracellular cAMP levels (827).
In conclusion, somatostatin acts as the global brake on acid secretion. By acting on G-cells, ECL cells, and parietal cells, it exerts its inhibitory action on every link in the regulatory chain leading to the secretion of gastric acid.
5. Other substances
A variety of other substances have been shown to have either direct or indirect effects on acid secretion. In the interest of conciseness, their physiological effects will only be discussed briefly at this point.
Secretin is a 27-amino acid peptide hormone that is synthesized in duodenal S-cells and secreted into the circulation in response to a low duodenal pH or passage of digestive products, such as fat (195, 955, 1153). A subpopulation of secretin-producing cells is also present in the gastric mucosa, where it may influence acid secretion in a paracrine manner (191–193). Given its secretory stimulus, it is regarded as a classic effector of the intestinal phase of acid secretion. When it was first discovered in 1902 by Bayliss and Starling (interestingly secretin was the first hormone ever to be discovered), it was noted that secretin induces pancreatic bicarbonate secretion, which leads to a buffering of the gastric acid entering the duodenum (68). In the stomach, secretin acts as an inhibitor of gastric motility and acid secretion (141, 194, 269, 374, 532, 564, 656, 677, 1107). The exact mechanism as to how secretin attenuates the secretion of acid is not exactly known, and several hypotheses have been put forward. For example, it has been shown that secretin induces the secretion of somatostatin from isolated D-cells (141). Increases in somatostatin levels were also observed in isolated perfused stomach models (205, 374). Others have proposed that secretin activates vagal primary afferent neurons, which in turn leads to neuronal modulation of acid secretion (656, 659). In opposition to this theory, it has also been demonstrated that the inhibitory effects of secretin are independent of vagotomy (677).
Oxyntomodulin is a peptide hormone produced in the mammalian intestine. It is closely related to glucagon and contains its entire amino acid sequence, extended by a COOH-terminal octapeptide (66). In isolated parietal cells, oxyntomodulin acts as an activator of acid secretion (959). The integrated response to oxyntomodulin is, however, opposite. Systemic injection decreases gastric acid secretion in rat, cat, and human test subjects (53, 157, 285, 524, 525, 965). The inhibitory effect is most likely mediated via somatostatin release (53).
It was recognized in the early 1950s that serotonin was present in the antral mucosa of dog stomachs (323). Serotonin is stored in granules of enterochromaffin cells of the antrum (1099). It is released into the circulation and the gastric lumen in response to vagal stimulation (107, 649). Intraluminal acidification serves as another stimulus for serotonin release (1196). Serotonin has an inhibitory effect on the secretion of gastric acid (107, 153, 521, 650, 720, 903). It is still poorly understood where serotonin interferes with acid secretion.
Neurotensin is a 13-amino acid neuropeptide that was originally isolated from calf hypothalamus (161). In the periphery, it is also produced and secreted postprandially by specialized endocrine cells (N-cells) of the small intestine (863). Various investigations have demonstrated that neurotensin suppresses the secretion of gastric acid and delays gastric emptying (25, 108, 486, 985). This has been shown by direct systemic injection, but also by immunoneutralization of endogenous neurotensin in a reverse approach (25, 108, 486, 985). In disagreement with these findings, other investigators could only inhibit acid secretion at unphysiologically high serum concentrations of ∼750 pmol (747). Of note, physiological postprandial neurotensin levels were measured to be ∼15 pmol by the same investigators, questioning the role of neurotensin as a physiological endocrine inhibitor of acid secretion (747). Neurotensin is also located to nerve fibers of the enteric nervous system in the stomach, indicating that it may act as a local neuronal rather than an endocrine regulator. It has been proposed that neurotensin may induce the secretion of somatostatin and thereby exert its inhibitory action on acid secretion (53, 414). Most recently, however, the low-affinity neurotenstin type 2 receptor (NTS2) has been identified on the parietal cell, suggesting a direct influence.
Ghrelin is a recently discovered 28-amino acid peptide hormone that is synthesized in P/D1 cells of the fundus (242). Since its discovery, multiple functions have been ascribed to it, ranging from being a regulator of appetite to being a modulator of bone remodeling. Its effects on bone are described in a separate section of this review (see sect. VD1). Apart from these functions, ghrelin also has been implicated to affect gastric acid secretion, although it remains a matter of discussion in which direction, as peripherally administered ghrelin has been reported to stimulate, inhibit, or not affect acid secretion (278, 355, 654, 719). The reason for these dichotomic results is largely unclear. The fact that ghrelin circulates in acylated and desacylated forms adds further complexity to the subject (491). Indeed, acylated ghrelin has been shown to stimulate acid secretion following peripheral injection, whereas desacylated ghrelin remained without effect (938). Further investigations are needed to clarify the controversy surrounding ghrelin and its influence on gastric acid secretion.
F) NITRIC OXIDE.
Nitric oxide (NO) is an important signaling molecule that plays a role in multiple physiological processes, such as vasodilation or the immune response. Indeed, NO has been shown to mediate the hyperemic response of the gastric mucosa that occurs during acid secretion (553). However, NO also directly influences the production of acid. The effect of NO on acid secretion is most likely inhibitory (81, 82, 555, 1002; opposed by Ref. 426). NO is produced by various forms of NO synthases, one of which has been localized at high concentrations in cells in the vicinity of parietal cells, allowing for a putative paracrine regulation (80). NO has been proposed to exert its inhibitory action by either directly inhibiting the parietal cell or by suppressing the release of histamine from ECL cells (81, 82, 555, 1002). Intracellular increases in cGMP concentrations have been observed in both cell types after NO exposure, suggesting that guanylate cyclase is an intracellular target for NO (82, 1002).
Interleukins (IL) are cytokines that mainly coordinate immune responses. In particular, IL-1β has been shown to impact gastric acid secretion. IL-1β is a general proinflammatory cytokine that plays an important role in the stomach in the context of Helicobacter pylori infection. H. pylori infection triggers an elevation of IL-1β levels as part of the host's immune response (65). Peripheral injection of IL-1β can profoundly suppress gastric acid secretion (912, 947, 1059, 1101, 1132). Multiple explanations for this observation have been put forward. It has been suggested the IL-1β acts in the CNS, as intrathecal injection also has an acid-suppressive effect (948, 949). Others have suggested that IL-1β promotes formation of prostaglandins or NO, which in turn inhibit acid secretion (312, 947, 1101). Yet, a direct effect on parietal cells and ECL cells is the most likely explanation, as both cell types express the IL-1 receptor and have been shown to be inhibited in their function in isolated cell models (69, 70, 872, 958).
C. The Pharmacological Suppression of Acid Secretion
Decreasing gastric acidity is indicated in many pathological contexts, including gastric reflux disease or peptic ulcer disease. This target can be achieved by two main pharmacological approaches: 1) the inhibition of gastric acid secretion or 2) the intraluminal neutralization of already secreted gastric acid (antacids). Gastric acid secretion can be attenuated by either directly blocking its final molecular effector, namely, H+-K+-ATPase (PPIs and acid pump antagonists), or by interfering with the neurohormonal signaling pathway leading to its secretion (H2 antagonists). The following section attempts to discuss the four most common substance classes employed to increase intragastric pH.
1. Direct pharmacological inhibition of H+-K+-ATPase
The inhibition of H+-K+-ATPase-mediated proton transport represents the main contemporary pharmacological strategy for reducing gastric acidity. An increase of gastric pH is the main factor ameliorating acid-related disorders and has been show to directly correlate with healing rates of, for example, GERD (74). Two main substance groups exert their acid-reducing effect via inhibiting H+-K+-ATPase function: PPIs and acid pump antagonists (APAs). Both substances achieve this aim by distinct mechanisms.
Omeprazole was the first clinically available PPI (324). The first patent on omeprazole was filed in 1979 by the Swedish company Astra AB (today AstraZeneca). The introduction as a prescription PPI followed in 1989. Today, the omeprazole enantiomer esomeprazole (S-omeprazole) generates the second highest revenue of all pharmaceuticals in the United States and is only surpassed by the statin atorvastatin (512). Furthermore, in the United States, PPIs are available as over-the-counter formulations, making them accessible for the broad public. This is partially made possible by the high safety profile of PPIs with a low incidence of unspecific adverse effects. Recently, however, concerns about the long-term effects of chronic acid suppression have emerged with regard to its impact on bone health (see sect. VA).
PPIs are delivered as pro-drugs through the bloodstream to the parietal cell. They are weak bases (pKa ∼4), which can easily pass the cell membrane and accumulate in acidic compartments, such as the secretory canaliculus of the parietal cell. The pro-drug is then converted to the pharmacologically active cyclic sulphenamide by the acidic pH in the secretory canaliculus (670, 1007, 1135). Their specific accumulation in acidic milieus and their pH-catalyzed conversion to active substances confers specificity and thus a high safety profile to PPIs. Once activated, PPIs bind covalently via disulfide bonds to H+-K+-ATPase, thereby inhibiting its capacity to pump protons (89, 90, 1005, 1006, 1008). The pattern of the cysteine residues, which are involved in PPI binding, differ among the respective members of the PPI family: cysteine-813 reacts with all PPIs. In addition, omeprazole reacts with cysteine-892, lansoprazole with cysteine-321, and pantoprazole and tenatoprazole, respectively, with cysteine-822 (89, 90, 1005, 1006, 1008). Since the binding is covalent and irreversible, the inhibitory effect of PPIs lasts long beyond their plasma half-life, which usually ranges between 0.5 and 2 h depending on the specific PPI (581, 1036). PPIs are generally metabolized by the hepatic cytochrome P-450 system, in particular CYP2C19 and CYP3A4. This is of particular clinical importance, as CYP2C19 polymorphisms are known to exist. These polymorphisms can impact the pharmacokinetics of PPIs by affecting the metabolic rate of CYP2C19, which may have consequences for the optimal therapeutic regimen (582). Esomeprazole and rabeprazole seem to be less dependent on CYP2C19 metabolism (516, 983). Apart from the pattern of cysteine reactivity, half-life and metabolism, PPIs also vary in oral bioavailability (581).
Suppression of acid secretion can never be complete, as H+-K+-ATPase is subjected to a constant turnover (half-life ∼50 h) and needs to be stimulated for the conversion of the PPI to take place (936). Nevertheless, PPIs are highly effective in reducing gastric acidity. Depending on the PPI and the regimen, overall intragastric pH can be elevated by several pH units, up to a pH of 6 (compared with 1–2 at baseline) (137, 425, 1036). For an excellent summary of PPI efficacy, please refer to Reference 1036.
APAs represent the second class of H+-K+-ATPase inhibitors. Unlike PPIs, they do not undergo irreversible binding, but rather act as potassium competitive antagonists. The duration of inhibition is thus directly dependent on the plasma concentration of the inhibitor. As predicted by homology modeling, mutational analysis, but also recent structural data, APAs bind in the luminal cavity of H+-K+-ATPase in the vicinity of the potassium entry site where they exert their inhibitory action (2, 42, 761, 1104, 1106). Although the inhibition of acid secretion has been shown to be very effective, these substances are generally not in clinical use (577). For example, clinical trials of the APA AZD08650 have shown no additional therapeutic effect compared with the PPI gold-standard, which resulted in abandonment of the drug in a clinical setting (262, 539).
2. H2 antagonists
The development of H2 blockers is inseparably intertwined with Sir Black's discovery of the H2 receptor on the gastric parietal cell at the Smith Kline and French Laboratories (now GlaxoSmithKline) (106). In his original publication, Sir Black also describes burimamide as a competitive H2 antagonist that can effectively inhibit pentagastrin-stimulated gastric acid output in human volunteers (106). Further development of the antagonist led to the synthesis of cimetidine, which was first commercially introduced in 1976 in the United Kingdom, followed by the United States in 1977. Other commonly used members of H2-antagonist family now include ranitidine, famotidine, and nizatidine.
H2 antagonists prevent histamine-mediated stimulation of the parietal cell by competitively interfering with its receptor. Although this effectively terminates the gastrin-histamine axis, the parietal cell is still susceptible to cholinergic stimulation via the M3 receptor. This partial inhibition mainly accounts for the lower clinical efficacy of H2 antagonists compared with PPIs, which directly target H+-K+-ATPase as the final target of all parietal cell stimuli (384). For example, a meta-analysis concluded that patients treated for bleeding peptic ulcers are about twice as likely to suffer from persistent or recurrent bleeding if treated with H2 antagonists compared with PPIs (384). Another meta-analysis also demonstrated a higher efficacy of PPIs in treating esophagitis (83% healing rate with PPIs compared with 52% with H2 antagonists)(567). Today, H2 antagonists are largely superseded by PPIs due to their higher clinical efficacy. Furthermore, with the exception of famotidine, H2 antagonists are extensively metabolized in the liver by the CYP-450 system, leading to substantial drug-drug interaction profile (for review, see Ref. 506).
Antacids directly neutralize gastric acid allowing immediate short-term control of heartburn. They exist in various salt formulations, the most common of which are carbonate salts, such as CaCO3, MgCO3, or NaHCO3. The use of calcium carbonate as a dietary calcium supplement is discussed in a separate section (see sect. VC). Although each formulation possesses its own spectrum of side effects, a notable condition in the context of this review is the milk-alkali syndrome, which is the result of a concomitant overingestion of calcium and alkali, such as CaCO3. Although the syndrome became less prevalent with the introduction of modern ulcer therapies, it still poses a significant risk for patients that may ingest CaCO3 as a calcium supplement for the prevention of osteoporosis or as an antacid on a regular basis. Milk-alkali syndrome presents with the triad of metabolic alkalosis (carbonate), hpercalcemia (calcium), and renal insufficiency. The above average ingestion of calcium leads to increased plasma calcium levels due to excess absorption and impaired renal secretion. PTH levels are low due to negative feedback (6). Hypercalcemia further causes renal vasoconstriction, which decreases the glomerular filtration rate, and renal fluid loss because of activation of the CaSR, which in turn has loop-diuretic-like effects (see sect. IVD3). The activation of the CaSR is further potentiated by the metabolic alkalosis, which increases its sensitivity to calcium. All abnormalities are usually reversible after withdrawal of the offending agent and adequate treatment.
III. INTESTINAL CALCIUM ABSORPTION
The intestine is responsible for the absorption of dietary calcium into our systemic circulation. Although the kidney also plays a pivotal role in calcium homeostasis by retaining and balancing systemic calcium via regulating its excretion into the urine, renal calcium handling will not be the subject of this review.
Current recommendations suggest that an average 40-yr-old adult should ingest ∼1,000 mg of calcium on a daily basis (218). In the United States, this requirement is mostly met (57). Up to 72% of the dietary calcium intake is attributable to dairy products (218). Typically, the intestine absorbs between 25 and 35% of the ingested calcium (1193). This occurs via two distinct pathways: 1) a paracellular pathway and 2) a transcellular pathway.
Calcium absorption via the paracellular route is tied to a downhill concentration gradient between the luminal and the extracellular compartment and occurs throughout the entire intestine (although solvent drag induced paracellular flux may also play a role at low luminal calcium concentrations; Refs. 266, 1071, 1098). Conversely, transcellular absorption can also take place against an uphill gradient, but requires molecular machinery in the form of distinct calcium transport proteins which are expressed on the apical and basolateral membranes of the enterocyte. This process directly requires energy in the form of hydrolyzable ATP and is alternatively termed “active” transport (versus “passive” paracellular transport). The proximal small intestine, i.e., the duodenum and the jejunum, is the main site for transcellular calcium absorption (825).
Since a concentration gradient is not a prerequisite for this process, transcellular transport allows us to absorb calcium even when the calcium concentration in the chyme is fairly low. The relative importance of each respective absorption pathway thus alternates with the amount of ingested calcium (46, 824, 1224). The rate of paracellular calcium uptake is canonically thought to remain constant, while transcellular transport can be upregulated under conditions of dietary calcium restriction (46, 824, 1224). This regulation occurs via the active metabolite of vitamin D [1,25(OH)2-vitamin D], which serves as a stimulator for transcellular calcium uptake to prevent systemic calcium depletion.
A. Transcellular Calcium Absorption
1. Calcium entry
Evidence for active transport of calcium across the intestinal epithelium was established very early. With the use of a calcium radioisotope, various groups demonstrated 1,25(OH)2-vitamin D-dependent calcium transport against an imposed concentration gradient in the small intestine of the rat (952, 954, 1150). It has also been observed that active transport can be induced by a low-calcium diet, which we now know to stimulate the production of 1,25(OH)2-vitamin D (1133, 1134). The degree of transport was highest in the duodenum and decreased in the more distal segments (952). The duodenum is still considered the primary site where the bulk of transcellular transport occurs. To conduct active transcellular calcium absorption, the enterocyte has to be equipped with an apical calcium entry pathway, a mechanism for cytosolic calcium shuttling, and a basolateral calcium exit pathway (FIGURE 3). As the cytosolic concentration of free calcium is kept at a constant low level with typical concentrations of ∼100 nM, apical calcium influx follows its electrochemical gradient into the cell. In contrast, the extrusion of calcium on the basolateral membrane against an uphill gradient either directly requires ATP or energy stored in the sodium gradient.
A) THE SEARCH FOR THE APICAL CALCIUM ENTRY PATHWAY.
The molecular identity of the apical calcium entry pathway was unclear for a long time. Early experiments in isolated duodenal brush-border vesicles revealed that calcium uptake was passive, saturable, sensitive to ruthenium red, 1,25(OH)2-vitamin D dependent, and functionally optimal at a pH of 7.5 (740). This black box characterization suggested that a “specific carrier” was responsible for calcium absorption and in retrospect already provided us with accurate key characteristics of the transient receptor potential vanilloid channel type 6 (TRPV6), which was later established as the primary apical calcium uptake channel (740). In subsequent attempts to further unravel the nature of the calcium uptake mechanism, various voltage-gated L-type calcium channel blockers were used (474, 717, 838). Although isolated duodenal cells accumulated less calcium following application of the inhibitors and 1,25(OH)2-vitamin D stimulation (717), in vivo calcium entry proved to be fairly insensitive to their effects (with the exception of verapamil, which demonstrated some degree of inhibition if applied at very high concentration in the millimolar range) (474, 838). The conflicting reports may be attributable to the different experimental models that were used, as isolated single cells do not allow discrimination between apical and basolateral transport mechanisms. Furthermore, it has been argued that L-type calcium channels may play a role in the stimulatory pathway of vitamin D, rather than in calcium uptake per se (717). In conclusion, an involvement of L-type calcium channels seemed rather inconclusive and the identity of the calcium entry channel remained elusive.
The cloning of the calcium transport protein subtype 1 (CaT1) in 1999 finally marked a turning point in the search for the elusive intestinal calcium entry channels (838). The work was pioneered by Hoenderop and colleagues who had identified the main calcium entry protein in the kidney (epithelial calcium channel type 1, ECaC) via an expression cloning strategy a few months earlier (474). CaT1 was identified by a similar approach. A rat duodenal cDNA library was functionally screened using a calcium uptake assay in a Xenopus oocyte expression system (838). This screening process yielded the 727-amino acid protein CaT1, which showed a 75% sequence homology to rabbit ECaC. Homology analysis also demonstrated a relationship to the vanilloid reptor type 1 (VR1), a nonspecific cation channel that is activated by capsaicin, the pungent ingredient in chili peppers, and mostly mediates pain signaling through afferent sensory neurons (838). The nomenclature changed over time as new channel proteins were identified, and today we consider CaT1, ECaC, and VR1 to be members of the same transient potential receptor vanilloid (TRPV) ion channel family. Literature now refers to ECaC as TRPV5, to CaT1 as TRPV6, and to VR1 as TRPV1.
The structure of TRPV6 was predicted to have six transmembrane domains and four ankyrin repeat domains, which serve as cytoskeletal linking sites (838). Channel conductance was not dependent on other ions and was inhibitable by a low extracellular pH (838). Calcium uptake was reduced by as much as ∼70% at a pH of 5.5, which confirmed the early black box data observations made in intestinal brush-border vesicles (838). This behavior seemed counterintuitive to the investigators, as TRPV6 expression was highest in the duodenum, which is exposed to an acid load from the stomach (838). Duodenal pH has been reported to be as low as ∼6.1–6.6 but is even lower acutely after gastric emptying (∼5.4), which would entail significant inhibition of TRPV6-mediated calcium uptake (284, 290, 838). Conductance was also modestly sensitive (10–15% inhibition) to L-type calcium channel blockers at high concentrations, which partially clarified the preceding ambiguous observations made by other groups with these inhibitors (318, 337, 717, 838).
Subsequently, the human analog of rat TRPV6 was cloned (it had 97% sequence homology and was rather confusingly named ECaC2), and its expression was confirmed in human duodenum (64). The generation of a TRPV6 antibody allowed for first localization studies, which confirmed an apical localization of TRPV6 and thus reaffirmed its role as the primary apical calcium entry pathway in the intestine (1220). Tissue expression of TRPV6 varies among species (461). In humans, TRPV6 was identified in the duodenum, jejunum, stomach, esophagus, kidney, placenta, mammary gland, pancreas, prostate, testis, and salivary gland, but was also found to be upregulated in a series of malignancies, including prostate, breast, colon, and ovarian cancer (461, 792, 836, 839, 1168, 1220). The role of TRPV6 in these tissues mostly remains to be elucidated.
TRPV6 displays some unique biophysical properties. As mentioned earlier, TRPV6 consists of six transmembrane segments, like many voltage-gated cation channels. Unlike these channels, TRPV6 lacks the voltage sensor domain in the fourth α-helix and is in a constitutively open state at resting membrane potential. Analysis of the quarterny structure shows that it assembles in tetramers (476). The ankyrin domains were implicated to be responsible for tetramer formation; however, more recent structural data obtained by crystallography question this hypothesis (311, 848). It is thought that heterotetramers can also be formed with subunits of closely related TRPV5 (476). In contrast to most other members of the TRP channel family, TRPV6 is a very selective channel for calcium (475, 1058, 1197). The relative permeability of TRPV6 for calcium over sodium (PCa/PNa) is >100, whereas TRPV1, for example, has a selectivity of PCa/PNa ∼10 (although this has recently been shown to be variable) (206, 475, 1197). This high selectivity is crucial for the maintenance of a constant membrane potential in the enterocyte during calcium absorption. TRPV6 is strongly inward rectifying, which has been attributed to magnesium ions plugging the channel pore for outward ion movement during states of depolarization (1126). Magnesium also exerts a voltage-independent inhibitory effect on TRPV6 current; however, the mechanism underlying this observation is unclear (475, 1126). Furthermore, TRPV6 gating is sensitive to intracellular calcium. Increases in calcium were shown to inhibit the channel, resembling a negative-feedback mechanism (475). Although global calcium concentrations in the cell largely remain constant, the channel microenvironment is exposed to fluctuations in local calcium. This feedback loop may be important for finely tuning the amount of calcium influx and preventing calcium overload of the enterocyte. The putative mechanism of this regulation will be discussed later in this section.
In 2001 it was demonstrated that ATP modulates TRPV6 activity by preventing channel rundown (475). Recent work by Al-Ansary et al. (15) suggests that ATP directly binds to the channel, thereby inhibiting inactivation by locking the channel in the open conformation. Binding of ATP may be antagonized by channel phosphorylation through PKC (15). In conclusion, TRPV6 is precisely regulated by its own microenvironment. Calcium, magnesium, and protons have an inhibitory effect on the channel, whereas ATP prevents channel inactivation.
It is important to remember that apart from being a nutrient, calcium is a crucial intracellular messaging molecule. Therefore, the enterocyte has to tightly control its intracellular concentration and adapt uptake to energy status and basolateral extrusion. To ensure this delicate intracellular homeostasis, TRPV6 is associated with and regulated by a variety of auxiliary proteins. The first protein that has been identified to interact with TRPV6 was the S100A10-annexin2 complex (1112). The S100A10-annexin2 complex has been implicated to play a role in protein trafficking and membrane anchoring. Its association with TRPV6 is essential for channel trafficking to occur, as a disruption of the interaction leads to cytosolic scattering of the channel (1112). Formation of the S100A10-annexin2 complex and its association with TRPV6 involves activation of PKA and calcineurinA (CnA) (117). Another protein that is required for TRPV6 membrane insertion is rab11a (1111). In analogy to S100A10-annexin2, perturbation of rab11a binding results in decreased surface expression of TRPV6 and channel retention in the cytosol (1111). Furthermore, the protein kinases SGK1 and WNK3 are known to promote membrane insertion of TRPV6; however, their exact site of action is still unclear (114, 1031). Once trafficked to the membrane, physical channel stability is maintained by a variety of auxiliary proteins. The COOH-terminal tail of TRPV6 contains a PDZ protein binding motif. PDZ proteins serve as membrane anchors and protein scaffolds and mediate the assembly of multiprotein complexes, thereby regulating channel activity. The PDZ protein sodium hydrogen exchanger regulating factor (NHERF4) (aka PDZK2) has recently been identified to interact with TRPV6 (574, 1113). It has been speculated that NHERF4 serves as a scaffold for TRPV6 at the apical pole, which is underlined by the observation that knockdown of NHERF4 by RNAi leads to decreased TRPV6 current in HEK293 cells (574, 1113). Apart from trafficking to the apical membrane, channel number can be regulated by internalization and degradation. One possibility of decreasing functional channels at the cell surface is protein ubiquitination. The process of ubiquitination involves enzymatic tagging of target proteins, thus directing them to the proteasome or lysosome for rapid degradation. The degradation tag is transferred to the target protein by E3 ubiquitin protein ligases, such as Nedd4–2. It is already well understood how Nedd4–2-mediated ubiquitination can decrease the number, but also directly modulate the activity, of the epithelial sodium channel (ENaC) in the kidney (825). A recent study demonstrated that a similar mechanism applies to TRPV6 (1212). Coexpression of Nedd4–2 and TRPV6 in oocytes resulted in decreased calcium flux and channel numbers (1212). Nedd4–2-mediated TRPV6 ubiquitination could thus serve as a mechanism of rapidly regulating channel retrieval from the plasma membrane (1212).
Several other proteins have been identified to associate with TRPV6 that do not affect channel numbers, but rather directly regulate channel activity. The nisnap1 gene was identified in the late 1990s; however, its function remained elusive (992). Recently, TRPV6 was shown to interact with nisnap1 (971). It is expressed in the intestine, but not the kidney, which is unusual as TRPV6 and TRPV5 are generally regulated by similar auxiliary proteins. Electrophysiological measurements showed that nisnap1 inhibits TRPV6 without affecting its surface expression (971). Very similar functional properties were attributed to RGS2, a protein that is mainly known to alter the GTPase activity of G proteins. In analogy to nisnap1, RGS2 can inhibit TRPV6 at the plasma membrane without affecting its trafficking dynamics (970).
As previously discussed, the activity of TRPV6 is finely regulated by intracellular calcium concentrations. Elevations in intracellular calcium levels exert an autoinhibitory effect on channel opening (475). Shortly after the identification of TRPV6, calmodulin (CaM) was shown to bind to the channel in a calcium-dependent manner, and it was speculated that it mediates the calcium feedback response (461, 791). Subsequent investigations demonstrated that CaM may indeed represent the molecular calcium sensor (263, 619). Elegant FRET studies reported that CaM dynamically associates with TRPV6 in the presence of calcium and that this association is terminated when intracellular calcium is depleted (263). The association between TRPV6 and CaM also correlates with decreased current flux through the channel (263). Although the ankyrin repeat domains of the channel were thought to possibly serve as binding sites for CaM, more recent structural insights provide rebutting evidence for this hypothesis (848). Instead, the COOH-terminal tail of TRPV6 has been suggested as the site where CaM binding occurs (263, 619, 791). Hence, CaM can tune ion flux through TRPV6 and most likely mediates the channel's sensitivity to intracellular calcium.
Phosphorylation by kinases and dephosphorylation by phosphatases represent a common cellular strategy for rapidly modulating channel gating properties. The non-receptor tyrosine kinase Src has emerged as a candidate for the direct phosphorylation of TRPV6, thereby increasing channel conductance (1042). Src was previously shown to modulate TRPV4 activity (1178). The effects of Src are antagonized by the phosphatase PTP1B. Both enzymes act on the tyrosine residues Y161/162, which are located in the NH2-terminal tail of the channel (1041).
An interesting interaction without direct relevance for the intestine has been reported between renal TRPV5 and klotho (170). Klotho is a β-glucuronidase with a transmembrane anchor that can be cleaved, resulting in shedding of the enzymatically active domain of the protein into the urine (170, 612). Klotho can then increase TRPV5-mediated calcium uptake by hydrolyzing N-linked oligosaccharides on extracellular channel domains, thereby preventing channel retrieval from the plasma membrane (170). Although a recent report indicates that klotho can also affect TRPV6 activity in vitro, it is not expressed in the intestine and thus may only interact with renal TRPV6 (612, 683). Although this finding has no implications for intestinal calcium uptake, various bacteria and neutrophils produce β-glucuronidase, which may affect TRPV6 activity in the intestine (359, 1004).
Recently, Stumpf et al. (1046) described an association between cyclophilin B (CyB) and TRPV6 in the placenta. Coexpression of both proteins in oocytes increased calcium uptake (1046). CyB was also detected by Western blot analysis in the small intestine and colon; however, colocalization and functional studies in intestinal tissue remain to be performed (1046).
C) THE TRPV6 KNOCKOUT CONUNDRUM.
As will be discussed in more detail later, 1,25(OH)2-vitamin D is one of the key hormonal regulators of systemic calcium homeostasis (see sect. IVA). Increased levels of 1,25(OH)2-vitamin D lead to enhanced intestinal calcium absorption. Shortly after cloning of TRPV6, it was recognized that the channel is positively regulated at the mRNA level by 1,25(OH)2-vitamin D (614, 1030, 1109, 1110). The TRPV6 promoter has multiple binding sites for the 1,25(OH)2-vitamin D receptor (VDR) and is thus directly sensitive to increases in 1,25(OH)2-vitamin D levels (736). Consequently, VDR-deficient animals display a marked decrease in TRPV6 mRNA levels, whereas administration of 1,25(OH)2-vitamin D in wild-type animals results in an increase in mRNA transcription (1030, 1109, 1110). In conjunction with the rapidly expanding characterization of the channel itself, these observations supported the dogma that TRPV6 is the essential player in transcellular calcium uptake and that this pathway is strongly regulated by 1,25(OH)2-vitamin D.
With the introduction of novel genetic techniques, a TRPV6 (−/−) mouse was created in 2007 by Bianco et al. (94). The animals presented with decreased intestinal calcium uptake, as measured by serum concentrations of a calcium radioisotope following gavage, decreased femoral bone density (9.3%) and increased 1,25(OH)2-vitamin D levels, as a result of feedback regulation (94). These findings were in accordance with the postulated role of TRPV6 as the primary 1,25(OH)2-vitamin D-sensitive calcium uptake mechanism in the intestine. However, the generation of another TRPV6 (−/−) animal line by Benn et al. (76) a year later spawned a controversy in the field. In these animals, baseline calcium uptake was identical between wild-type and (−/−) groups. Surprisingly, calcium uptake could be increased in the (−/−) group by a low calcium diet and, more importantly, by 1,25(OH)2-vitamin D (an observation that was also confirmed by another group; Ref. 615) (76). These findings profoundly questioned the role of TRPV6 in calcium absorption and suggested that a different molecular target of 1,25(OH)2-vitamin D mediates the increase in calcium uptake after exposure. All these observations were in sharp contrast to the initial report by Bianco et al. (94). However, Benn et al. (76) also reported that compared with wild-type animals calcium uptake was reduced in TRPV6-deficient animals that were fed a low-calcium diet. This suggests that the animals may not be able to adequately increase absorption, when confronted with a low dietary availability of calcium, which in turn suggests a role of TRPV6 during states of dietary calcium insufficiency.
These controversial findings allow several conclusions and speculations. 1) As with any model of targeted gene disruption, a compensatory mechanism may be in place that masks and distorts the physiological importance of TRPV6. The animals may upregulate other, yet unidentified, calcium transport mechanisms to compensate for the loss of TRPV6 function. 2) Rather than being a constitutively active pathway, transcellular calcium absorption through TRPV6 only occurs in states of low dietary calcium intake. This hypothesis has been put forward very early and has been reaffirmed by a recent study that observed an increase in bone turnover in TRPV6 (−/−) animals on a low-calcium diet (46, 666, 824, 1224). Mobilization of bone calcium may be necessary in these animals to maintain systemic calcium concentrations, which are double challenged by the lack of TRPV6 and the insufficient calcium intake (666). Furthermore, it has been extensively described that a low-calcium diet induces gene expression of TRPV6 [presumably through an increase in 1,25(OH)2-vitamin D levels], suggesting a role of the channel during states of insufficient dietary calcium supply (76, 132, 639). 3) 1,25(OH)2-vitamin D may have many more targets in the intestinal mucosa than previously anticipated and may also regulate calcium uptake through the paracellular pathway (76, 615). In conclusion, the generation of TRPV6 (−/−) animals has sustainably challenged our model of transcellular calcium absorption by questioning the relative importance of TRPV6 and by introducing new potential targets of 1,25(OH)2-vitamin D regulation.
D) GENETIC POLYMORPHISMS OF TRPV6.
Single nucleotide polymorphisms (SNPs) are variations in the genomic sequence that occur in one single base. If the frequency of a SNP or a specific set of SNPs (haplotype) increases in a population over time compared with other SNPs, it can be concluded that this set of SNPs is associated with an evolutionary advantage, meaning that this gene locus was under selection. Recent reports from various groups indicate that TRPV6 was subjected to strong selection (12, 502, 1023, 1050). Traces for selection can be found in any non-African population (12, 502, 1023). The selective event in Europeans has presumably occurred ∼7,000 years ago and coincides with the development of agriculture (502). This correlation also holds true for other populations (502). It has been speculated that a change in diet or resistance to emerging disease led to selection and fixation of the now conserved haplotype (502). Interestingly, selection took place in parallel in many populations, pointing towards a strong selective stress that developed independently in each region (502).
The electrophysiological properties of ancestral and derived TRPV6 were investigated by two groups (502, 1050). Hughes et al. (502) observed no statistically significant divergence in channel behavior. The derived channel only displayed a tendency towards decreased sensitivity to the autoinhibition by calcium, albeit not significant (P = −0.094) (502). The authors speculated that differences in protein-protein interactions may have led to selection (502). Conversely, Sudo et al. (1050) reported increased calcium influx through the derived channel, which may constitute an evolutionary advantage by facilitating dietary calcium uptake. Regardless of the controversial functional data, these insights indirectly underline the importance of TRPV6 and provide a counterweight to the conclusions drawn from the TRPV6 (−/−) animal model. It is unquestionable that TRPV6 was under parallel independent selection in many regions across the planet. This provokes the question of how strong selection can occur for a derived channel whose physiological difference to the ancestral form seems to be very subtle, yet complete disruption of the channel in the mouse model only causes a very mild phenotype. This inconsistency adequately exemplifies the caveat that underlies the conclusions drawn from (−/−) animals and reminds us of the fact that we are far from understanding the exact physiology of transcellular calcium uptake.
TRPV6 may not be the only channel mediating apical calcium absorption in the intestine. A recent investigation provides another explanation for the initially observed partial sensitivity of transcellular calcium uptake to L-type calcium channel blockers. The voltage-gated calcium channel Cav1.3, a member of the L-type calcium channel family, was recently identified in the apical membranes of the distal jejunum and proximal ileum (751). Previous observations were emulated, as the investigators again demonstrated a decrease in calcium absorption following application of L-type calcium channel inhibitors in the corresponding segments (751). The authors argued that uptake through Cav1.3 may have previously been misinterpreted as paracellular calcium uptake (751). However, it should be noted that the calcium uptake assay used in this report did not discriminate between transcellular and paracellular calcium movement. Subsequently, it has been observed that L-type inhibitor-sensitive calcium flux is linked to stimulation of glucose uptake through the glucose transporter type 2 (GLUT2) (692, 750). Cav1.3 may serve as an alternative calcium entry pathway that is active in states of luminal calcium abundance and that coregulates glucose absorption (692, 750). Further investigations will be needed to determine the contribution of Cav1.3 to dietary calcium absorption.
2. Calbindin-D 9k
The following section will address the question of what happens to dietary calcium after it enters the enterocyte. It should be considered that transcellular calcium uptake and cytosolic calcium homeostasis are two contradicting requirements for the enterocyte. A prominent transcellular flux of free calcium will inevitably interfere with housekeeping functions of the cell, such as intracellular signaling. Furthermore, it has been calculated from in vivo data that if calcium were to diffuse freely (simple diffusion) through the cytosol, uptake rates would only be 1/70th of the actually measured values (129). Simple diffusion would constitute a bottleneck in the process of transcellular calcium absorption if cytosolic calcium concentrations should be kept low. A partial solution to this problem was found as early as 1966, when Wasserman et al. (1151) identified a 1,25(OH)2-vitamin D-inducible calcium binding protein in the chick intestine. The authors observed that calcium radioisotopes traveled faster across a cellophane membrane if suspended in intestinal homogenates from rachitic, i.e., 1,25(OH)2-vitamin D deficient, chicks than if suspended in the homogenates from rachitic animals treated with 1,25(OH)2-vitamin D (1151). This indicated that calcium was bound to a protein in the enterocyte and that expression of this protein was controlled by 1,25(OH)2-vitamin D (1151, 1152). Subsequently, the calcium binding protein was further characterized (1073, 1149, 1152). Expression levels were shown to be highest in the duodenum and to gradually decrease in more distal segments, which correlated with the degree of calcium uptake that has been attributed to each intestinal segment respectively in prior functional investigations (1073). The mammalian isoform, which had a lower molecular mass of ∼9 kDa (hence the name calbindin-D 9k), compared with the 28 kDa of the avian isoform, was later identified (138, 280, 283, 358). Concerning the functional role of calbindin-D 9k, it had already been speculated very early after its discovery that it may serve as a calcium shuttling protein (951). Indeed, initial calculations and later very basic experimental data confirmed that calbindin-D 9k may mediate facilitated diffusion of calcium between the two poles of the enterocyte, in analogy to the transport of oxygen by myoglobin in the muscle (321, 602). In a very fundamental investigation, calcium flux was measured between chambers that were separated by dialysis membranes. A 51% increase in trans-chamber calcium flux occurred in the presence of calbindin-D 9k (321). However, it is hard to relate this number to physiological values, given the simplification underlying the experimental model. It has subsequently been calculated that calbindin-D 9k may facilitate diffusion of calcium by a factor of up to ∼60 (129). Rather than envisioning calbindin-D 9k as a protein that individually shuttles calcium ions across the cell, one should imagine calbindin-D 9k as a calcium gradient amplifier (129, 602). As the intracellular concentration of free calcium is in the nanomolar range, the intracellular gradient of free calcium between the apical and basolateral pole can also only be in the nanomolar range. Diffusional flux, however, directly correlates with the concentration gradient and can in consequence only be very small. Calbindin-D 9k is expressed in the enterocyte in the micromolar range and dynamically binds and releases calcium (714). Since the association between calbindin-D 9k and calcium is dynamic, the local concentration of calbindin-D 9k-bound calcium will directly correlate with the concentration of free calcium. It is thus the concentration gradient of calbindin-D 9k-bound calcium throughout the cell that determines the flux rate of calcium. As this gradient can be in the micromolar range, calbindin-D 9k serves as a calcium gradient amplifier (129, 602). It should be noted that the real experimental data on calbindin-D 9k are fairly scarce and that the bulk of scientific effort has gone into mathematical modeling of facilitated diffusion and calcium binding kinetics (129, 321, 322, 602, 1021).
Functionally, several correlations between calbindin-D 9k and calcium uptake were identified. The calbindin-D 9k content of each intestinal segment correlates linearly with its ability for calcium uptake (129, 824, 1021). Although this relation holds true on the experimental and the mathematical modeling level, it does not prove causality (129, 824, 1021). Furthermore, 1,25(OH)2-vitamin D and a low-calcium diet induce calbindin-D 9k on both the mRNA and protein level, suggesting that it is involved in the process of regulated calcium absorption (289, 604). In accordance with these findings, VDR-deficient animals have a decreased calbindin-D 9k mRNA content that cannot be rescued by exogenous 1,25(OH)2-vitamin D administration (663, 1194). Although a 1,25(OH)2-vitamin D responsive element (VDRE) had been identified in the 5′-flanking region of the calbindin-D 9k gene, later mutational analysis demonstrated that this site was not essential for transcriptional regulation of calbindin-D 9k by 1,25(OH)2-vitamin D (217, 240). Hence, it is not clear how 1,25(OH)2-vitamin D exactly regulates calbindin-D 9k on a molecular level. Interestingly, intestinal calbindin-D 9k mRNA levels can be rescued in VDR (−/−) animals by a high-calcium/phosphorus/lactose diet, while extraintestinal calbindin-D 9k mRNA levels are unaffected (663). This suggests that, apart from endocrine regulation through 1,25(OH)2-vitamin D, intestinal calbindin-D 9k is also regulated by local factors (663). A direct short-term stimulatory effect of oral calcium intake on calbindin-D 9k levels had been observed before in a wild-type background (647). The mechanisms underlying the effect of diet on calbindin-D 9k levels are, however, still obscure.
Despite the correlation between calbindin-D 9k levels and 1,25(OH)2-vitamin D and the mathematical and experimental models, which support its role as a calcium shuttling protein, no clear evidence for the involvement of calbindin-D 9k in transcellular calcium absorption existed. This had sparked controversy among investigators. In fact, a very early study concluded that although 1,25(OH)2-vitamin D induced both calbindin-D 9k and calcium absorption, a discrepancy existed in the time course of both effects (424). Later, synthetic 1,25(OH)2-vitamin D derivatives, designed for a maximized effect on cell differentiation and suppressed calciotropic activity, failed to increase serum calcium levels while causing a sevenfold increase in calbindin-D 9k mRNA [native 1,25(OH)2-vitamin D increases both serum calcium and calbindin-D 9k mRNA](604). In 2006, a calbindin-D 9k (−/−) mouse was created by Kutuzova et al. (613). The animals presented with no apparent phenotype abnormalities and had normal serum calcium concentrations (613). Subsequent analysis of these animals showed that they responded equally well to 1,25(OH)2-vitamin D with regard to calcium absorption as wild-type animals (13). Of note, calcium absorption was assessed by measuring serum calcium following gavage of a calcium radioisotope and did not discriminate between transcellular and paracellular absorption (13). A different group replicated the findings made in the calbindin-D 9k (−/−) mouse and also did not find any appreciable variation in phenotype or serum calcium concentrations (639). However, the investigators observed that during preweaning, when both calcium demand and absorption are at their peak, gene expression of TRPV6 and the basolateral calcium extruder, plasma membrane calcium ATPase isoform 1b (PMCA1b), were highly induced in the calbindin-D 9k (−/−) animals (128, 639). This has been interpreted as a compensatory upregulation of calcium transport proteins to ensure sufficient uptake in a state of high metabolic demand for calcium (639). Subsequently, a TRPV6 and calbindin-D 9k double (−/−) animal was created. The mature animals show no disturbances in calcium homeostasis, although, similarly to TRPV 6 (−/−) animals, they cannot increase calcium uptake in response to a dietary calcium challenge to the same extent as wild-type animals (76).
In light of the insights gained through knockout animals, the role of calbindin-D 9k remains somewhat unclear. Again, it is difficult to dissect the physiological function of calbindin-D 9k in these animal models, given the assumption that the organism will pursue compensation for the loss of a gene product. In conclusion, it is undisputed that calbindin-D 9k is regulated by 1,25(OH)2-vitamin D, that it can bind calcium and that theoretical modeling and very fundamental experimental models verify that it can facilitate diffusion of calcium across the enterocyte. The (−/−) models suggest that animals can compensate for the loss of calbindin-D 9k and maintain normal calcium homeostasis (13, 76, 613, 639). Furthermore, TRPV6, calbindin-D 9k and double (−/−) animals can still increase their calcium uptake in response to 1,25(OH)2-vitamin D or a low-calcium diet, respectively, albeit not to the same extent as wild-type animals (13, 76, 615). One may conclude that neither of these proteins is necessary for transcellular uptake; however, it seems more likely that 1,25(OH)2-vitamin D has more targets than previously postulated and that compensatory mechanisms are in place.
3. Basolateral calcium extrusion
Once calcium reaches the basolateral membrane, it is extruded into the extracellular space, which completes the process of intestinal absorption. As extracellular calcium concentrations are higher than cytosolic concentrations, this process requires energy. Two proteins are responsible for this task. 1) The PMCA extrudes calcium at the direct expense of ATP, whereas 2) the sodium-calcium exchanger (NCX) utilizes the energy stored in the sodium gradient to transport calcium out of the cell (FIGURE 3). It is this basolateral exit process that requires energy during transcellular calcium uptake and that allows us to absorb calcium against an uphill gradient when dietary concentrations are low.
A) THE PLASMA MEMBRANE CALCIUM ATPASE.
The plasma membrane calcium ATPase (PMCA) is a virtually ubiquitous protein that is responsible for intracellular calcium homeostasis by pumping calcium into the extracellular milieu, thereby keeping intracellular concentrations low. The pump belongs to the family of P-type primary ion transport ATPases, which among others also includes gastric H+-K+-ATPase (see sect. IIA1). Four isoforms of the protein exist; however, variety is increased by splicing (562, 1043). The PMCA1b splice variant is most predominant in the small intestine and the duodenum in particular, which suggest an involvement of this isoform in the process of transcellular calcium absorption (341, 492). Given the ubiquitous expression of PMCA, a detailed review of its structure and function will be omitted. For a current and detailed review, please refer to Reference 1043.
PMCA was first characterized in the 1960s in the membrane of erythrocytes (956). Very early experiments in basolateral membranes isolated from enterocytes identified a calcium-dependent enzyme with phosphatase activity, which served as first evidence for PMCA in the intestine (100, 742). Subsequent investigations in basolateral vesicles from rat intestine concluded that two distinct calcium transport mechanisms existed in their membranes: an ATP-dependent (PMCA) and a sodium-dependent (NCX) mechanism (382, 457, 573). Furthermore, it was shown that inhibition of calmodulin halved the amount of ATP-dependent calcium transport (573). Today, we know that PMCA is highly regulated by CaM, which can increase both the affinity of the pump to calcium and its turnover speed by a factor of up to 10 (526, 627, 628, 756).
Interestingly, it has also been suggested that calbindin-D 9k can directly stimulate calcium transport via PMCA (1136, 1137). In addition to intracellular proteins, PMCA is also quantitatively regulated by endocrine factors. 1,25(OH)2-vitamin D can increase both the PMCA mRNA and protein content in enterocytes (35, 36, 62, 146, 382, 489, 614, 823, 1109; contested by Ref. 1138). The data on the effects of a low-calcium diet on PMCA transcription are more controversial, as one study suggests induction whereas two other studies observed reduced mRNA levels (146, 583, 639). Of note, these observations were made in different species (chick vs. mouse). The sensitivity of the tissue to 1,25(OH)2-vitamin D is also reported to decrease with age, which may represent another confounding factor underlying these observations (35, 36). In conclusion, the transcriptional regulation of PMCA1b by 1,25(OH)2-vitamin D and its decreasing expression along the length of the small intestine correlate well with the canonical model of transcellular calcium absorption. It is challenging to investigate the functional contribution of PMCA1b in the process of calcium absorption, given its role as a housekeeping protein. PMCA1 is crucial during development, which results in embryonic death if knocked out (814, 1198). Conversely, heterozygote (+/−) animals present with no apparent phenotype, albeit parameters linked to calcium homeostasis were not assessed (814).
B) THE NCX.
Long before the molecular identities of PMCA and NCX were known, it had been observed that calcium uptake in the intestine was dependent on extracellular sodium (713). The authors concluded that “a sodium, calcium-exchange diffusion carrier may exist at the basal membrane of the cell,” which proved to be a remarkably accurate prediction (713). As mentioned before, later investigations delineated between an ATP and a sodium-dependent transport of calcium on the basolateral enterocyte membrane (382, 457, 573).
For a recent review on the structure and function of NCX, please refer to Reference 687. In brief, NCX exists in three isoforms (NCX1–3), which are expressed in a broad variety of cell types (687). The function of NCX has mainly been investigated in excitatory tissues, given its role as a high-capacity calcium extrusion mechanism following excitation. To transport calcium against its strong electrochemical gradient, NCX has to utilize three sodium ions to accomplish extrusion of one calcium ion (85, 460). NCX1 is the predominant isoform in the small intestine, where it has been detected on the mRNA and the protein levels (275, 688). However, NCX had been postulated to play a more important role during calcium absorption in the kidney than in the enterocyte, hence not much data on intestinal NCX are available to us (473). Furthermore, genetic disruption of NCX1 is embryonically lethal, which imposes some limitations on our experimental methodology (600). A more recent study provides some functional evidence for NCX in the intestine. By measuring intracellular calcium concentrations with a fluorescent indicator dye, Dong et al. (275) demonstrated that calcium uptake in a sodium-free environment (to run NCX in its reversed configuration) was significantly decreased when NCX was pharmacologically inhibited. Despite our knowledge that NCX exists in enterocytes and that it can extrude calcium under experimental conditions, it is difficult to assess its relative contribution to transcellular calcium absorption.
In addition, NCX is not subjected to regulation by 1,25(OH)2-vitamin D, which further contributes to the ambiguity concerning its importance in the process of calcium absorption. This has been observed very early, when exogenous administration of 1,25(OH)2-vitamin D to vitamin D-deficient animals did not increase sodium-dependent calcium transport, while doubling transport through PMCA (382). Furthermore, it was recently shown that a calcium-depleted diet decreases duodenal NCX1 mRNA levels (583).
Of note, two members of the potassium-dependent sodium calcium exchanger (NCKX) family, namely, NCKX 3 and 4, were also identified in the small intestine (658, 687, 688). However, further functional investigations will be needed to clarify their role.
B. Paracellular Calcium Absorption
If we plot the amount of duodenal calcium absorption as a function of the luminal calcium concentration, we can observe that the absorption curve is comprised of two distinct kinetic components: a saturable/exponential component and a nonsaturable/linear component (824, 825, 1134). The saturable component represents transcellular calcium uptake through the enterocyte, as both the number of calcium transport proteins and their turnover rate is limited. The nonsaturable component reflects calcium uptake through the paracellular pathway. The saturable component is less pronounced in the jejunum and disappears completely in the ileum, indicating that transcellular calcium absorption is restricted to the proximal segments of the intestine, as discussed previously (825) [this has been contested more recently, when transcellular flux was also noted in the ileum (46)]. Compared with the transcellular pathway, the paracellular route has not received much scientific attention. It has been put forward that the rate of paracellular calcium absorption is constant across the length of the intestine and is neither sensitive to 1,25(OH)2-vitamin D nor a low-calcium diet (824, 825, 1224). However, provided that enough dietary calcium is available to saturate the transcellular pathway, observations indicate that net calcium absorption is highest in the ileum, which has been attributed to the sojourn time, rather than alterations in paracellular permeability (290, 710). In the rat, the chyme spends some 74% of its transit time in the ileum, which allows for a long exchange period between lumen and plasma (290). The diffusion rate itself is fairly low and only amounts to 2% of the rate if calcium were to diffuse freely between intestinal lumen and plasma (290). This effect is a consequence of the diffusion barrier that tight junctions, which act as functional gating molecules of the paracellular pathway, impose on calcium flux.
Epithelial tight junctions are adhesion points between two neighboring cells that seal their intercellular space against a lumen, thereby restricting ion and water movement between the two compartments. As water cannot be directly transported, its movement is tied to ion fluxes, which in turn are determined by their respective concentration gradients across the epithelium. Each epithelial cell expresses an annulus of tight junction proteins at the apical end of the lateral membrane. Tight junctions are protein complexes that consist of a variety of intra- and transcellular proteins. A detailed review of their structure would exceed the scope of this article (for a recent review, please refer to Ref. 999). Briefly, the composition of the involved proteins determines the pore size, ion selectivity, and thereby the relative leakiness of each tight junction and the whole epithelium in general. Furthermore, tight junctions create a lateral diffusion barrier for membrane proteins and help to maintain the functional polarity of the epithelial cell. It is important to recognize that tight junctions are not static complexes, but can be rather dynamically regulated (999). This allows the epithelium to change its ion and water permeability in response to various stimuli (999).
As discussed previously, the rate of the nonsaturable component of calcium absorption is documented to be fairly constant and insensitive to 1,25(OH)2-vitamin D, which indicated that 1,25(OH)2-vitamin D may have no effect on paracellular transport (824, 825, 1224). However, in the late 1990s, Chirayath et al. (201) postulated that 1,25(OH)2-vitamin D can increase paracellular calcium flux in confluent Caco-2 cell cultures, which form an epithelia-like monolayer with tight junctions. This conclusion is based on the observations that 1,25(OH)2-vitamin D decreased the transepithelial electric resistance, which is often used as a measure of tight junction permeability, and induced bidirectional, i.e., also “serosal” to “mucosal,” calcium flux in the cultured monolayers (201). A more recent investigation further substantiated this hypothesis. It has been shown that 1,25(OH)2-vitamin D regulates the mRNA levels of some tight junction proteins in rat duodenum, which may alter their gating characteristics (354, 614). For example, the mRNA levels of claudin-3, a protein that directly determines tight junction permeability, were decreased 2.2-fold following 1,25(OH)2-vitamin D administration (614). Conversely, claudin-2 and -12 mRNA and protein levels were increased (354). Functional observations in Caco-2 monolayers further demonstrate that overexpression of claudin-2 and -12 increases electrical conductivity and facilitates paracellular calcium flux (354). All of these events are tied to a 1,25(OH)2-vitamin D-mediated promotion of transcriptional events. However, it is known that 1,25(OH)2-vitamin D can also exert short-term nongenomic effects (1122) (see sect. IVA6b). A recent investigation presents evidence for augmented solvent-drag paracellular calcium flux after acute administration of 1,25(OH)2-vitamin D (1098). Solvent-drag mediated flux may occur at low calcium concentrations in the absence of a calcium gradient. In this model, calcium is dragged through the paracellular space by water flux that is fueled by the hyperosmolar milieu in the paracellular space (266, 626). 1,25(OH)2-vitamin D was shown to induce this calcium flux in the presence of initially equimolar calcium concentrations between the mucosal and serosal compartments (1098).
Over the last years, evidence for the regulation of the paracellular pathway by 1,25(OH)2-vitamin D has slowly accumulated. It is apparent that the canonical dogma which postulates that 1,25(OH)2-vitamin D only targets the transcellular pathway has to be revisited. Still, more experimental investigations will be needed to ultimately clarify the effects of 1,25(OH)2-vitamin D on paracellular calcium transport. Should tight junctions indeed be subjected to regulation by 1,25(OH)2-vitamin D, this mechanism may partially explain the sensitivity of calcium uptake in TRPV6 and calbindin-D 9k (−/−) animals to 1,25(OH)2-vitamin D.
C. Alternative Models of Transcellular Calcium Absorption
Two alternative models of transcellular calcium absorption have been put forward. One suggests that calcium is transported through the enterocyte by vesicles (vesicular transport model) rather than facilitated diffusion, and the other postulates that 1,25(OH)2-vitamin D can induce intestinal calcium absorption in a rapid, nongenomic fashion via a putative 1,25(OH)2-vitamin D surface receptor (transcaltachia).
The assumption that calcium is transported through the cell in vesicles is corroborated by a handful of observations. It has been demonstrated that 1,25(OH)2-vitamin D treatment increased the number of supranuclear lysosomes in rachitic chicks, compared with nontreated animals (247). It was later shown that the calcium content of lysosomes can more than double following treatment with 1,25(OH)2-vitamin D, suggesting that they may play a role in the process of transcellular calcium movement (780). Compared with the cumulative evidence that is in accordance with our canonical model of calcium absorption, the role of vesicular transport of calcium is still fairly obscure.
The model of transcaltachia mainly relies on the observation that 1,25(OH)2-vitamin D can exert effects that are too acute to be attributable to transcriptional events. For example, intestinal calcium absorption was increased in chicks within 14 min of 1,25(OH)2-vitamin D exposure, an onset which is too rapid as to be of a genomic nature (785). It should be noted that rapid effects of 1,25(OH)2-vitamin D have also been suggested to influence the paracellular pathway (see above). Undoubtedly a careful evaluation of the route of calcium flux, i.e., transcellular versus paracellular, is necessary. At least, isolated intestinal cells respond with increased uptake of radiolabeled calcium to acute 1,25(OH)2-vitamin D exposure (568). It has been postulated that apart from the VDR, a membrane-bound 1,25(OH)2-vitamin D exposure receptor, the 1,25(OH)2-vitamin D-MARRS (membrane-associated, rapid response, steroid binding) protein, mediates the acute effects of 1,25(OH)2-vitamin D (568, 779, 785).
IV. REGULATION OF CALCIUM HOMEOSTASIS
Eucalcemia is maintained by the concerted effort of vitamin D, PTH, and, to a lesser extent, calcitonin. All three hormones can influence serum calcium concentrations by acting on the intestine, the kidney, or bone. 1,25(OH)2-vitamin D, the active vitamin D metabolite, primarily modulates the intestinal absorption of calcium and will therefore be discussed in most detail (FIGURE 4). Apart from hormonal regulators that influence the absorption, excretion, and deposition of calcium, our body needs a mechanism that allows it to sense the current levels of plasma calcium. This task is fulfilled by the CaSR. It oversees the precise regulation of the calcitropic hormones.
A. Vitamin D
As discussed previously, vitamin D is one of the key regulators of calcium homeostasis. Our body has two sources for vitamin D, namely, a dietary source (vitamin D2+3) and an endogenous source that relies on ultraviolet (UV) light catalyzed synthesis in the skin (vitamin D3) (FIGURE 5). Nomenclature in this case is fairly misleading, as vitamins are per definition substances that cannot be generated by our body and have to be ingested from an external source. The fact that we can synthesize vitamin D3 in our skin classifies the substance as a prohormone rather than a vitamin. As we will see, our current nomenclature is a byproduct of the historic events leading to the discovery of vitamin D.
1. Historical perspective
From a historical perspective, the identification of vitamin D is closely intertwined with attempts to understand the pathophysiology of rickets. Rickets is characterized by childhood skeletal deformities resulting from inadequate osteoid mineralization and calcification of cartilage due to decreased serum calcium levels during development. The adult form equivalent of rickets is termed osteomalacia. With the onset of industrialization, rickets became a prevalent problem in the 18th, 19th, and the beginning of the 20th century. In fact, the disease was so widespread at the beginning of the 20th century that an investigation conducted by the German pathologist Schmorl on 386 children, who had died before the age of 4 years, concluded that 90% of them had had rickets (968). Even in present times nutritional rickets still remains a major public health concern in developing countries (303, 757, 809). The seminal observations that led to the identification of vitamin D were provided by Mellanby in 1919 (733). He observed that dog pups who were fed a severely restricted diet consisting of porridge or bread were consistently developing rickets (733). Development of rickets could be averted if their diet was supplemented with cod liver oil, which we now know to contain a high concentration of vitamin D (733). Mellanby (733) concluded that “rickets is a deficiency disease which develops in consequence of the absence of some accessory food factor or factors.” Vitamin A had been discovered shortly before, and it was subsequently speculated that it may represent the factor promoting bone formation (733). This hypothesis was later rebutted by American biochemist McCollum who concluded that “a substance which is distinct from fat-soluble [vitamin] A” must be responsible for preventing rickets (727). Furthermore, he stated that his experiments “demonstrate the existence of a fourth Vitamin [vitamin D] whose specific property […] is to regulate the metabolism of bones” (727). In parallel to the unraveling of the dietary component of rickets, scientists were independently discovering the importance of sunlight for disease prevention. The Polish pediatrician Raczynski was most likely the first to demonstrate evidence for this hypothesis experimentally (881). He kept one dog pup in the shade while a littermate was kept in the sunlight. Both dogs were breastfed by their mother. After 6 wk, the bones of the dog that was kept in the shade contained 36% less calcium (881). These observations were followed up by the German pediatrician Huldschinsky, who healed rachitic children after exposing them intermittently for 2 months to the UV rays generated by a mercury vapor quartz lamp (504). Hess and Unger (447) replicated these findings by using sunlight treatment. Hess (446) later summed up the impact of these revolutionary observations, which now seem intuitive to us: “We have known that a growing plant cannot thrive in the dark, but have failed to realize that the same laws apply to growing animals.”
It was later recognized that it was sufficient to irradiate the food administered to animals rather than the whole animal to prevent development of or heal rickets. Thus initially “inert” dietary substances with no antirachitic properties could be activated by UV light (385, 448–450, 505, 1037). It was concluded that irradiation caused the conversion of a biological precursor to an active form and that the same mechanism was physiologically taking place in the skin. Initially, it was speculated that cholesterol may serve as this pro-vitamin. It was Windaus and Hess (in collaboration with Rosenheim) who were the first to uncover its exact molecular identity. They stated: “We conclude from our experiments with complete certainty that ergosterol […] represents the anti-rachitic provitamin (1166). In 1928, Windaus received the Nobel Prize in Chemistry for “the services rendered through his research into the constitution of the sterols and their connection with the vitamins.” The irradiation product of ergosterol was later purified and named vitamin D2 (ergocalciferol) (43, 896, 1164, 1167). Although these findings solved the question as to how UV irradiation generates Vitamin D2 from ergosterol, which has antirachitic properties if ingested, the molecular mechanisms underlying the antirachitic effects of cutaneous sunlight exposure still remained obscure. Since ergosterol is an exclusive component of yeast and fungal membranes, a different precursor substance had to exist in animal skin. Again, it was Windaus and colleagues who identified 7-dehydrocholesterol as the provitamin in porcine skin, which is converted to vitamin D3 (cholecalciferol) under irradiation (1165). After these discoveries, industrially produced vitamin D has rapidly been used in medical applications and as a food fortification. Today, the main portion of dietary vitamin D ingestion in the United States stems from fortified dairy products (150).
2. Intestinal vitamin D absorption
To exert its antirachitic effects, dietary vitamin D has to be absorbed into our circulation. Early everted gut sack experiments demonstrated that this process has linear, nonsaturable, and passive kinetics, suggesting that no specific carrier mechanism for vitamin D is in place (484). These observations were later replicated in in vivo models (483). Absorption is highest in the proximal and mid small intestine (484). Since vitamin D is fat soluble, its absorption mechanism is similar to that of dietary lipids. In an aqueous media, vitamin D aggregates in micelle-like structures (735). Its absorption is aided by the secretion of bile acids, which is underscored by the observation that patients suffering from cholestasis can present with vitamin D deficiency and develop bone disease, such as osteomalacia or osteoporosis (245, 445, 563, 721, 894, 953, 1018, 1080). Apart from bile salts, formation of mixed micelles containing monoglycerides and free fatty acids represents another factor that aids in vitamin D absorption (884, 1082; contested by Ref. 483). These substances increase micelle size, which promotes the solubilization of vitamin D, thereby increasing uptake (884). Clinically, pancreatic insufficiency, causing an impairment of triglyceride breakdown through insufficient lipase secretion, leads to decreased vitamin D absorption (1080, 1127). This is a particular problem in cystic fibrosis patients, who often develop pancreatic and concomitant vitamin D insufficiencies (33, 230, 625, 678, 924).
Following uptake into the enterocyte, vitamin D is packed into chylomicrons and secreted into the lymph (288, 953). It has been demonstrated that after intestinal administration of radioactive labeled vitamin D3, up to 90% of the recovered radioactive tracer was associated with the chylomicron fraction of the collected intestinal lymph (288). Furthermore, patients suffering from the autosomal recessive chylomicron retention disease (Anderson disease; OMIM 246700), which causes impairment of chylomicron processing and secretion in the enterocyte, can present with insufficient levels of fat-soluble vitamins, such as vitamins D and E (535). The chylomicron remnants, which include vitamin D, are then scavenged by the liver after the lymph has entered the circulation through the thoracic duct (287, 407). However, it has been demonstrated in vitro and in vivo in hepatectomized and normal rats that vitamin D can directly transfer from chylomicrons to a vitamin D binding protein (DBP; see sect. IVA5) in the blood plasma (286, 288). It has therefore been suggested that at least a fraction of hepatic vitamin D uptake is mediated through DBP rather than chylomicrons (286; contested by Ref. 407). In the liver, vitamin D is then hydroxylated at its 25 position to 25(OH)-vitamin D. The metabolism of vitamin D will be the subject of a later section.
The intestinal absorption of 25(OH)-vitamin D has been investigated by many groups, with the aim of optimizing vitamin D administration in a therapeutic context by bypassing the first metabolic step in the liver (219, 245, 287, 445, 645, 721, 1018, 1019, 1127). In general, enteric uptake of 25(OH)-vitamin D is more effective than that of vitamin D, which is partially attributable to a comparably lower dependency on bile acid secretion (219, 245, 287, 645, 1018, 1019). The observation that patients with cholestasis still absorb 25(OH)-vitamin D effectively, whereas vitamin D absorption is impaired, corroborates this hypothesis (1018). There is some controversy with regard to the transport of 25(OH)-vitamin D following its absorption (287, 701, 1019). It has been argued that it may be transported predominantly in the protein fraction of the lymph (287), i.e., not in chylomicrons, or that it is directly absorbed into the portal blood (701, 1019).
3. Cutaneous vitamin D synthesis
Cutaneous synthesis is our second source of vitamin D. Cutaneous production depends on exposure to UVB (290–315 nm) light (FIGURE 5). The UVB photons convert 7-dehydrocholesterol, which is located in the plasma membrane of keratinocytes, to previtamin D3 (1117–1120). This dependency on sunlight causes a seasonal variation in vitamin D3 production, with synthesis being low during the winter months when the radiation angle of the sun flattens (187, 478, 538, 1035, 1156). In consequence, changes in latitude result in similar variability in production. Further factors that decrease production include pigmentation of the skin and application of sunscreen (209, 723), whereas an increase in altitude promotes production (478). Following conversion from 7-dehydrocholesterol, previtamin D3 isomerizes to vitamin D3 (479). The isomerization process is temperature dependent and fairly slow (479, 1117). It has been calculated that the half-life for the formation of vitamin D3 is ∼2.5 h (1085). Interestingly, in vitro experiments conducted in isotropic medium demonstrated that the isomerization rate was 10 times slower than in in vivo experiments (1085). This was later attributed to the fact that amphiphatic interactions with phospholipids of the cell membrane stabilize the previtamin D3 conformer which then isomerizes to vitamin D3 (1086). The cellular microenvironment of the reaction thus greatly optimizes the isomerization to vitamin D3. After its synthesis, vitamin D3 is bound to DBP and carried though the bloodstream to its target organs. Observations made in patients indicate that the vitamin D3 plasma levels peak ∼2 days after sunlight exposure, which is due to the slow isomerization rate in the skin (7).
4. Vitamin D metabolism and its regulation
Irrespective of the source (endogenous or exogenous), vitamin D is metabolized in the liver to 25(OH)-vitamin D (FIGURE 5). Evidence for the existence of biologically active vitamin D metabolites emerged in the 1960s (686, 799). 25(OH)-vitamin D was identified by means of injecting rats with radiolabeled vitamin D and subsequent silicic acid column chromatography of lipid extracts from serum and various tissues (686). Chromatography revealed the presence of a vitamin D metabolite in serum, liver, bone, and feces in rats and in human serum (256, 686). Furthermore, it was shown that the vitamin D metabolite was biologically active by reverting rickets in diseased animals and by increasing intestinal calcium transport and bone metabolism (686, 752). The metabolite was characterized as 25(OH)-vitamin D and subsequently successfully synthesized (111, 112). It was soon recognized that hepatectomized rats were not effectively converting vitamin D to 25(OH)-vitamin D, suggesting that the liver was the main organ responsible for 25-hydroxylation (864, 865). The ability of the liver to metabolize vitamin D was also confirmed in perfused livers and tissue homogenates (490). Historically, a long controversy existed with regard to the subcellular localization of the enzyme that hydroxylates vitamin D (25-hydroxylase) in the liver, as enzyme activity was observed in both mitochondrial and microsomal fractions of liver homogenates (93, 102, 103, 258, 696, 1053). Extensive investigations indicated that both the microsomal and mitochondrial 25-hydroxylase are members of the cytochrome P-450 family (102–104, 696, 835). Both fractions demonstrated distinct enzymatic kinetics. The mitochondrial enzyme was characterized as a low-affinity high-capacity enzyme, whereas the microsomal fraction displayed high-affinity low-capacity characteristics (103, 104, 356, 696).
The mitochondrial 25-hydroxylase (CYP27A1) was first purified to homogeneity from rabbit liver mitochondria (235). It was demonstrated that this cytochrome P-450 was not specific to vitamin D and could also hydroxylate other substrates, most notably cholesterol (27-hydroxylation), which represents an important step in the formation of bile acids (235). In retrospect, CYP27A1 had been purified 4 years earlier; however, only the 27-hydroxylation of cholesterol had been investigated and its effects on vitamin D had remained obscure (1162). Subsequently, the enzyme's cDNA was cloned from rabbit, rat, and human, and its dual role in vitamin D and steroid conversion was confirmed (26, 149, 1048, 1103). The full gene structure was identified a few years later (646). Although no crystal structure of the enzyme is available to us, a homology model based on other CYP family members has been proposed (874). In the liver, CYP27A1 is expressed on the mRNA level in hepatocytes, endothelial, stellate, and Kupffer cells (1078). Furthermore, CYP27A1 expression was confirmed in a variety of other tissues, including duodenum, adrenal glands, kidney, lung, vascular endothelium, brain, retina, skin, muscle, and osteoblasts; however, their potential contribution to vitamin D metabolism remains unclear (26, 51, 135, 184, 370, 383, 508, 640, 898, 979). Interestingly, an extrahepatic conversion of vitamin D had been suggested previously to occur in the kidney and the intestine (1097). Marked differences in CYP27A1 activity can be observed between males and females. For example, CYP27A1 enzyme activity and mRNA expression were demonstrated to be increased in female rats (933, 1078). Higher expression in females was also confirmed in biopsy samples from human subjects (370). A regulation via sex hormones may underlie this phenomenon, as injection of estradiol was shown to induce CYP27A1 activity (933). Interestingly, seasonal variations in expression were also observed, which may represent a confounding factor for decreased 25(OH)-vitamin D levels during the winter months (370).
It should be noted that CYP27A1 can also hydroxylate vitamin D3 at other positions (402, 950). These include positions 27 and 26; however, the ratio for 25-:27-:26-hydroxylation has been estimated to be only 100:15:3, which demonstrates that 25-hydroxylation of vitamin D3 is the most essential reaction catalyzed by the enzyme (950). More importantly, CYP27A1 can also use its own product 25(OH)-vitamin D as a substrate to further act as a 1α-hydroxylase and produce the hormonally active form of vitamin D, namely, 1,25(OH)2-vitamin D (50, 51, 950). As will be discussed later, this reaction is normally catalyzed in the kidney by another CYP family member (CYP27B1). At the moment, it is not clear what the physiological significance of 1α-hydroxylation by CYP27A1 is.
Mutations in CYP27A1 cause the autosomal recessive disorder cerebrotendinous xanthomatosis (CTX; OMIM 213700). The disease was first described in 1937 and is characterized by cholestanol deposits that are most prominent in tendons, especially the Achilles tendon, the brain, and the lung. Patients present with progressive neurologic defects, atherosclerosis, and cataracts and commonly suffer from diarrhea. The inadequate bile acid synthesis was first noted in 1974 by Setoguchi et al. (993). Shortly after the cDNA of CYP27A1 was cloned, it was demonstrated that mutations of this enzyme were responsible for CTX (148). In agreement with the dual role of CYP27A1, CTX patients also suffer from osteoporosis, low 25(OH)-vitamin D levels, and impaired intestinal calcium absorption (83, 320). Three CYP27A1 mutations that are known to cause CTX and still lead to protein expression were recreated in vitro, and enzymatic activity was assayed. Depending on the expression system, these mutants showed lower or higher 25-hydroxylation activity than the wt enzyme, which led the authors to questions the enzyme's role in vitamin D metabolism (403). It should be considered that 1) many more (∼38) mutations underlying CTX exist, 2) by far not all patients exhibit disturbances in bone or vitamin D homeostasis, and 3) the investigated mutant enzymes may only cause disturbances in cholesterol, rather than vitamin D metabolism (83, 320, 403). In the light of these limitations, it should be questioned whether CTX represents an apt model system to evaluate CYP27A1 in the context of vitamin D metabolism.
With the introduction of novel genetic tools, a cyp27a1 (−/−) mouse was created in 1998 (918). However, the phenotype of CTX could not be reproduced, albeit fecal bile acid content was markedly decreased (918). Rather surprisingly, serum 25(OH)-vitamin D levels were increased in (−/−) animals, which either pointed to compensatory upregulation of other, maybe microsomal, enzymes or a noninvolvement of cyp27a1 in vitamin D metabolism in the mouse (918). The (−/−) mouse model was later characterized in more detail, but no new conclusions with regard to vitamin D were drawn (902).
Another caveat that needs to be considered when assessing the physiological importance of CYP27A1 for vitamin D metabolism is that the enzyme cannot 25-hydroxylate dietary vitamin D2 (402). This observation underlines the necessity for another 25-hydroxylase which physiologically metabolizes vitamin D2. The microsomal 25-hydroxylase was a clear candidate for this process; however, it has only recently received more extensive scientific attention.
The molecular identity of the microsomal 25-hydroxylase was unclear, until it was unmasked by Cheng et al. (184) in 2003 as CYP2R1. CYP2R1 is expressed on the mRNA level in a plethora of tissues, but most prominently in the liver and testes (184). As illustrated by a recent investigation, the testes may play a role in calcium homeostasis. Patients with testiculopathy (and concomitant lower CYP2R1 expression) were shown to have decreased 25(OH)-vitamin D levels and osteoporosis (333). The protein's crystal structure has recently been resolved in complex with vitamin D3 (1045). Unlike CYP27A1, CYP2R1 has been shown to 25-hydroxylate both vitamin D2 and vitamin D3, which may be a solution to the enigma of vitamin D2 metabolism (184). The physiological relevance of CYP2R1 was underscored by the characterization of a patient who presented with low 25(OH)-vitamin D levels and was shown to have a transition mutation in the CYP2R1 gene (183). Furthermore, a remarkable large-scale study has recently tried to establish a correlation between 25(OH)-vitamin D status and the genotype of 33,996 individuals. It was found that lower 25(OH)-vitamin D levels correlated with variants in the CYP2R1 gene (1146) (a smaller study came to similar conclusions, Ref. 139). This represents an outstanding finding, as the influence of CYP27A1 mutations on 25(OH)-vitamin D production is far less conclusive. Given the disproportionate quantity of scientific data, it is at this moment difficult to evaluate the relative contribution of each 25-hydroxylase to vitamin D metabolism, but more and more evidence for the importance of CYP2R1 is accumulating.
The regulation of 25-hydroxylation has been the subject of controversy in the past, which is partially due to the fact that multiple enzymes may metabolize vitamin D, and that it was challenging to experimentally discriminate between these enzyme entities. Thus evidence which supports (91, 92, 741, 1078) and questions (258, 1097) the existence of 25-hydroxylase regulation can be found. A detailed analysis of these observations is beyond the scope of this review, but the most recent investigation should be considered in more detail, given the advances in our experimental repertoire: it has been demonstrated in the rat that 1,25(OH)2-vitamin D can downregulate hepatic CYP27A1 transcription with a concomitant reduction in enzyme activity (1078). These observations strongly corroborate the hypothesis that the 25-hydroxylation step is subjected to negative-feedback regulation. The exact mechanism of this regulatory mechanism remains elusive, especially since the CYP27A1 gene is not under control of a VDRE (369, 988, 1078).
Following its synthesis, 25(OH)-vitamin D binds to DBP and is transported to the kidney, where it undergoes further conversion to 1,25(OH)2-vitamin D, the hormonally active form of vitamin D (FIGURE 5). Historically, 1,25(OH)2-vitamin D was first identified in the nuclei of intestinal cells as an uncharacterized vitamin D metabolite (429, 634). The biological activity of the metabolite was determined to be much higher than that of vitamin D, and it was eventually isolated and identified as 1,25(OH)2-vitamin D (480, 586, 633, 800). 1,25(OH)2-vitamin D is up to 10 times more potent than vitamin D. The importance of the kidney for the synthesis of 1,25(OH)2-vitamin D was soon discovered, as nephrectomized rats were neither able to convert 25(OH)-vitamin D, nor absorb calcium effectively (339, 458). Briefly thereafter, patients with chronic renal insufficiency were shown to lack the capability to metabolize 25(OH)-vitamin D, which further established the role of the kidney as the major conversion site (724). The impairment of 1,25(OH)2-vitamin D synthesis in the course of chronic renal deficiency is a cofactor in the development of renal osteodystrophy, a bone mineralization deficiency due to deranged mineral balance. Of note, the kidney is not the exclusive site of CYP27B1 expression. 1α-Hydroxylase has been detected in a variety of other tissues, including the placenta, decidua, skin, brain, vascular endothelium, pancreas, colon, but also in monocytes and dendritic cells (451–453, 603, 1204–1206). The role of extrarenal 1,25(OH)2-vitamin D synthesis is not entirely clear. Beyond modulating calcium homeostasis, vitamin D has been shown to have immunomodulatory and antiproliferative effects (see sect. IVA6). Given these observations, it has been speculated that extrarenal local 1,25(OH)2-vitamin D production and paracrine secretion may represent important factors for the maintenance of the “barrier function” in these tissues (451, 453).
In the kidney, 1α-hydroxylation of 25(OH)-vitamin D takes place in the inner mitochondrial membrane of the epithelial cells of the proximal tubule (393).
The cellular uptake mechanism of 25(OH)-vitamin D by the tubule cells will be subject of a later section. The 1α-hydroxylase responsible for the conversion is also a member of the cytochrome P-450 family (CYP27B1) and was first cloned in 1997 by St-Arnaud et al. from rat cDNA (1034). The human homolog was cloned shortly thereafter (349, 748). Interestingly, mapping of the CYP27B1 gene revealed that the locus was identical to the gene locus of the autosomal recessive disorder pseudo vitamin D deficiency rickets, type 1A (PDDR1A, OMIM 264700) (1034). PDDR is characterized by low serum calcium, secondary hyperparathyroidism, and low 1,25(OH)2-vitamin D levels (869) (of note, the original article wrongly suggests an autosomal dominant inheritance pattern). Patients can exhibit rickets or osteomalacia due to increased mobilization of calcium. The gene locus of PDDR1A had been mapped by linkage analysis; however, before the cloning of CYP27B1, it was not clear whether a defect in the enzyme itself or disturbances in its regulation were responsible for the disease (617, 618, 1223). The clear-cut phenotype of PPDR1A effectively illustrates the pivotal role of CYP27B1 and the lack of redundancy at this essential step of vitamin D metabolism. The characteristic features of PDDR1A can essentially be emulated if cyp27b1 is knocked out in a mouse model (238, 822). The serum calcium levels and the secondary hyperparathyroidism can be normalized if these animals are fed a high-calcium, phosphorus, lactose diet, albeit bone growth remains impaired (239).
The activity of CYP27B1 is subjected to tight hormonal regulation. The key regulator of 1,25(OH)2-vitamin D synthesis is PTH, which is secreted by the parathyroid glands in response to low serum calcium concentrations in an effort to increase calcium uptake and release bone calcium into the circulation. The regulation of calcium homeostasis by the parathyroid glands and the CaSR will be covered in subsequent sections (see sect. IV, B and D). Apart from this stimulatory input, CYP27B1 is under negative control by its own product. 1,25(OH)2-vitamin D represses CYP27B1 on a transcriptional level (116, 125, 443, 591, 764). Although studies in (−/−) animals suggest that the VDR is essential for autoinhibition to take place, the promoter region of CYP27B1 does not include a canonical VDRE (125, 591, 764). It is thus most likely that the transcriptional regulation through 1,25(OH)2-vitamin D is indirect (125, 591). Alternatively, so-called E-box-type elements were recently proposed to act as negative VDREs (575). Moreover, CYP27B1 can be directly regulated by the local calcium concentrations. High extracellular calcium inhibits 1,25(OH)2-vitamin D synthesis, whereas low calcium concentrations induce its production (109). It has been proposed that changes in calcium modulate VDR expression and thereby the sensitivity of cell to the local negative feedback by 1,25(OH)2-vitamin D (702). Some evidence suggests that the CaSR mediates the regulatory effects of calcium on CYP27B1 activity (FIGURE 8) (702). Other factors that regulate CYP27B1 activity include fibroblast growth factor 23 (FGF23), calcitonin, prolactin, sex steroids (at least in avian species), and phosphate (716, 913, 1067, 1068, 1217).
Apart from regulating the synthesis of the vitamin D metabolites, our body also tightly controls their degradation. The first step of vitamin D catabolism is 24-hydroxylation, which is carried out by the mitochondrial enzyme CYP24A1. The primary site for vitamin D catabolism is the kidney, but CYP24A1 is also strongly expressed in other extrarenal tissues, such as the intestine, osteoblasts, keratinocytes, prostate, placenta, brain, and heart (11, 181, 796). CYP24A1 can hydroxylate both 25(OH)-vitamin D and 1,25(OH)2-vitamin D, thereby creating 24,25(OH)2-vitamin D and 1,24,25(OH)3-vitamin D, respectively (14). 24-Hydroxylation is followed by a series of oxidation/reduction reactions which finally yield the excretable product calcitroic acid (703, 893). At least in the proximal tubule of the kidney, the regulation of CYP24A1 is reciprocal to that of CYP27B1. 1,25(OH)2-vitamin D upregulates CYP24A1, thereby stimulating its own breakdown, whereas PTH inhibits CYP24A1 (37, 180, 812, 1069, 1222). While 1,25(OH)2-vitamin D increases the transcription of CYP24A1 (the gene has two upstream VDREs), PTH exerts its inhibitory effects by decreasing CYP24A1 mRNA stability (1222). In osteoblastic and distal convoluted tubule cell lines, PTH has synergistic effects with 1,25(OH)2-vitamin D in inducing CYP24A1 (37, 1189). In analogy to CYP27B1, CYP24A1 is further regulated by FGF23, calcitonin, and phosphate (361, 513, 1175).
5. Vitamin D transport and cellular uptake
The hypothesis that vitamin D may be bound to a carrier substance in serum was first expressed in the late 1950s, when it was demonstrated that the alpha fraction of human serum had high anti-rachitic properties (1079). It was later shown that DBP is identical to the group specific component (Gc) protein, which had been characterized independently by Hirschfeld and colleagues around the same time (236, 464, 465). DBP (∼58 kDa; 458 amino acids) is synthesized in the liver and is closely related to albumin and α-fetoprotein, which are all derived from the same ancestral gene (221, 875, 1158, 1187, 1188). The crystal structure of DBP has been resolved at a resolution of 2.3 Å in complex with 25(OH)-vitamin D (1121). DBP can bind vitamin D and all of its metabolites (408). There is, however, a difference in the relative affinity for the vitamin D steroids, with the affinity for 25(OH)-vitamin D being highest (Kd ∼10−8 M), followed by 1,25(OH)2-vitamin D and vitamin D (Kd ∼10−7 M) (408). In humans, vitamin D2 and D3 metabolites are bound with equal affinity to DBP (406). Given its long plasma half-life, 25(OH)-vitamin D is also measured as the primary clinical parameter to assess the vitamin D status of patients. Although the plasma concentration of DBP is ∼4–8 μM, only <5% of the binding sites are occupied by vitamin D sterols (408, 409). DBP has a very fast turnover rate. It has been estimated that up to 28% of DBP are replaced every day (557). This turnover entails a high demand for synthesis output by the liver. In consequence, patients with liver disease demonstrate lower DBP and total vitamin D levels than healthy subjects (97, 409). This relationship is reversed during pregnancy, when the levels of DBP and vitamin D sterols are increased (97, 409, 609, 685, 892). Of note, DBP is not the exclusive carrier substance for vitamin D sterols. Albumin and lipoproteins were shown to transport a fraction (∼15%) of vitamin D (1015).
Given its multitude of functions, DBP has been regarded as an essential protein. Indeed, an analysis of over 80,000 human serum samples showed that DBP was present in all of them, which led to the hypothesis that deleterious mutations of DBP were lethal (213). Rather surprisingly, DBP (−/−) mice thrive well with growth curves identical to their littermates, although their 25(OH)-vitamin D and 1,25(OH)2-vitamin D levels (total) are severely decreased (937). If challenged with a low vitamin D diet, the DBP (−/−) animals develop secondary hyperparathyroidism, leading to defects in bone mineralization (937). It is unclear, however, to what extent albumin and lipoproteins compensate for the loss of the primary vitamin D carrier substance.
Not much is known about the cellular uptake of vitamin D. As all other steroids, free vitamin D can passively diffuse through the plasma membrane because of its lipohilic nature. However, due to the high concentration of DBP and its affinity towards vitamin D ligands, only a fraction of vitamin D circulates in the free form. For example, 0.003% of 25(OH)-vitamin D is transported as an unbound sterol in serum, which raises the question whether passive diffusion represents a sufficient uptake pathway (98). At least 25(OH)-vitamin D has been proposed to be delivered to the proximal kidney tubule for further conversion to 1,25(OH)2-vitamin D by a different and remarkable mechanism. Rather than diffusing passively, it has been proposed that the 25(OH)-vitamin D-DBP complex passes the glomerular filter and is endocytosed by the epithelial cell of the proximal tubule (FIGURE 5) (805). The endocytotic process is mediated by megalin (aka gp330), which is a multifunctional clearance receptor on the luminal membrane. Mice lacking functional megalin were shown to lose DBP and vitamin D in the urine and develop vitamin D deficiency (805). A kidney specific megalin (−/−) animal was recently created, and the observations made in the global (−/−) animal, which had a very low perinatal survival rate of 2%, could essentially be replicated (643). Furthermore, defects in cubulin, a membrane protein that colocalizes with megalin, cause a similar phenotype (806). The experimental data are further supported by clinical observations made in patients suffering from Fanconi syndrome. Fanconi syndrome is a global reabsorption deficiency of the proximal tubule, which can develop as a result of heavy metal poisoning and/or drugs or may have inherited causes. These patients were shown to lose DBP in their urine, which may reflect the inability of the tubule cell to endocytose DBP (805, 1076, 1100). A similar endocytotic uptake mechanism has been proposed for mammary cells, which can also convert 25(OH)-vitamin D to 1.25(OH)2-vitamin D(926).
6. Cellular effects of vitamin D
The cellular effects of 1,25(OH)2-vitamin D can be categorized into two major pathways, which are defined by their respective speed of onset: 1) slow genomic responses and 2) rapid nongenomic responses. Both pathways require binding of 1,25(OH)2-vitamin D to its intracellular receptor, the VDR (503, 787). The VDR (NR1I1, nuclear receptor subfamily 1, group I, member 1) belongs to the superfamily of nuclear receptors, which amongst others also includes the estrogen, testosterone, or glucocorticoid receptors. VDR was first identified in the late 1960s in the chromatin fraction of chick intestinal mucosa, where 1,25(OH)2-vitamin D increases the rate of intestinal calcium uptake (430). The 427-amino acid protein (molecular mass 48.3 kDa) was cloned in 1988 by Baker et al. (59). The crystal structure of the VDR is available to us with 1,25(OH)2-vitamin D bound to the receptor's ligand binding domain (914).
A) GENOMIC EFFECTS.
Following ligand binding, nuclear receptors typically act as transcription factors and induce or repress the transcription of certain target genes. In the case of VDR, 1,25(OH)2-vitamin D binds to the receptor, which subsequently heterodimerizes with the retinoid X receptor (RXR) (FIGURE 3) (1027, 1083). The VDR-RXR complex then interacts with a VDRE in the 5′ promoter region of the regulated gene resulting in transactivation. Alternatively, it has been proposed that VDR can bind to the VDRE before ligand binding occurs (920).
I) Intestine. The intestine is one of the primary target sites of 1,25(OH)2-vitamin D. 1,25(OH)2-vitamin D upregulates the expression of intestinal TRPV6, calbindin-D 9k, and PMCA, which are canonically regarded to mediate the process of transcellular calcium absorption (FIGURE 3) (see sect. IIIA). Furthermore, 1,25(OH)2-vitamin D may modulate calcium uptake through the paracellular route (see sect. IIIB). By increasing the amount of absorbed calcium, 1,25(OH)2-vitamin D directly elevates serum calcium levels. This represents one of the final links in the regulatory chain of calcium homeostasis which starts with sensing of low calcium levels by the parathyroid gland and ends with increased synthesis of 1,25(OH)2-vitamin D by the kidney. The significance of 1,25(OH)2-vitamin D for intestinal calcium absorption is illustrated by 1,25(OH)2-vitamin D-deficient patients, which absorb up to 80% less calcium from their meal compared with healthy individuals (996). Although similar results were obtained in animal models, probably one of the more illustrative observations has been made in VDR-deficient animals (825, 1110). Deletion of VDR correlates with a massively impaired capacity to absorb intestinal calcium resulting in low plasma calcium levels and hyperparathyroidism (1110). This phenotype is reversible by intestine-specific expression of VDR on a VDR (−/−) background (712, 1181). Thus animals that lack the VDR, except in the intestine, present with normal serum calcium and PTH levels (1181). In conclusion, it is unquestionable that 1,25(OH)2-vitamin D and its receptor are the key regulators of intestinal calcium absorption.
II) Bone. Some evidence exists that 1,25(OH)2-vitamin D may have direct influences on the formation of bone (FIGURE 4). The VDR is expressed in osteoblasts, osteoclasts, and chondrocytes (115, 533, 623, 730). Most of the direct effects of 1,25(OH)2-vitamin D are thought to be mediated by osteoblasts. 1,25(OH)2-vitamin D has been shown to promote osteoblast differentiation from mesenchymal stem cells and to regulate the synthesis of various osteoblast proteins, such as osteocalcin, alkaline phosphatase, collagen I, osteopontin, and RANKL (45, 78, 79, 665, 709, 804, 870, 925). In general, however, the effects of 1,25(OH)2-vitamin D on osteoblast maturation and protein synthesis are very pleiotropic and depended on the duration of the exposure and the differentiation stage at which the osteoblast is exposed (817). Effects of 1,25(OH)2-vitamin D on the mineralization of bone have also been reported. For example, an increase in the mineralization of extracellular matrix was observed following concomitant 1,25(OH)2-vitamin D and vitamin K exposure (599, 745). Yet, the overall direct influence of 1,25(OH)2-vitamin D on bone metabolism is somewhat obscure. This is illustrated by VDR-deficient animals. Naturally, these animals develop rickets and osteomalacia due to impaired intestinal calcium absorption. If the animals are, however, maintained normocalcemic, the skeletal phenotype is completely rescued (21, 661). Although, these observations question the importance of 1,25(OH)2-vitamin D as a direct regulator of bone metabolism, a subsequent investigation by Panda et al. (821) came to a different conclusion. The authors reported that osteoblast numbers, mineral apposition rates, and overall bone volume were reduced in normocalcemic (rescue diet) cyp27b1/VDR (double −/−) animals, suggesting that the 1,25(OH)2-vitamin D system is necessary for intact bone formation (821). The reason for these discrepant findings is generally unclear; however, differences in the length of exposure to the rescue diet have been put forward as a possible cause (821). For a more detailed review of the effects of vitamin D on bone, please refer to References 1033, 1114.
III) Kidney. The kidney is not only the major site of 1,25(OH)2-vitamin D synthesis (see above), but also represents a vitamin D target organ. The kidney acts as key regulator of calcium homeostasis by changing the amount of calcium that is reabsorbed from the primary urine. The majority of the calcium that is filtered through the glomerulus is reabsorbed in the proximal tubule through the paracellular space, with the amount of absorbed calcium gradually decreasing along the nephron. Fine regulation of calcium absorption occurs in the distal tubule and collecting duct. The mechanism by which the distal tubule conducts transcellular calcium absorption is highly analogous to the proximal intestine. The epithelial cells of the distal tubule express TRPV5 (the “sister” channel of TRPV6) as the apical calcium entry channel, calbindin-D 28k and NCX1 and PMCA1b as basolateral calcium extruders (203, 474, 610, 837, 922). In analogy to the intestine, 1,25(OH)2-vitamin D upregulates the majority of these proteins in an effort to increase renal calcium reabsorption (203, 474, 610, 837, 922).
B) NONGENOMIC EFFECTS.
In contrast to the genomic effects of 1,25(OH)2-vitamin D, which have been known for decades, the rapid cellular responses have only recently received more scientific attention. While the transcriptional events of 1,25(OH)2-vitamin D take place on a time scale of a few hours to days, the rapid nongenomic responses occur within minutes of exposure. One of the first evidences, which in retrospect can be attributed to a nongenomic response, was the observation made in 1941 by Selye that intraperitoneal injection of steroids had an anesthetic effect (990). Interestingly, the rapid responses to 1,25(OH)2-vitamin D also require the presence of the VDR, as these responses cannot be elicited in VDR deficient animals (503, 787). It should be noted that other proteins such as 1,25(OH)2-vitamin-D-MARRS have also been suggested as candidates for a membrane associated 1,25(OH)2-vitamin D receptor (782). Attempts at identifying the subcellular localization of the VDR in the rapid response context have yielded that VDR is also present in plasma membrane invaginations, the so-called caveolae (503, 802). These VDR-containing membrane microdomains have been identified in multiple tissues, including the intestine, kidney, and lung, and are identified by coexpression of caveolin-1, which is used as a marker protein for caveolae (503, 802). Functionally, not many rapid response effects of 1,25(OH)2-vitamin D have been characterized. For example, it has been demonstrated that 1,25(OH)2-vitamin D can influence ion channel gating in osteoblasts, modulate the contraction of cardiomyocytes, lead to insulin secretion in pancreatic β-cells via elevating intracellular calcium, and cause photoprotection in keratinocytes (273, 540, 1088, 1200, 1216). In the intestine, the phenomenon of transcaltachia (see sect. IIIC) has been attributed to rapid actions of 1,25(OH)2-vitamin D (801).
In addition to 1,25(OH)2-vitamin D, the VDR also binds the secondary bile acid lithocholic acid (704). Secondary bile acids are bile acids that have been metabolized by the intestinal gut flora. Lithocholic acid is toxic and has been implicated to play a role in intestinal carcinogenesis (601). It has been suggested that the VDR may serve as a secondary bile acid sensor and induce lithocholic acid breakdown through CYP3A activation (537). The noncanonical VDR stimulation by lithocholic acid may thus serve as an autoprotective mechanism (704). Apart from inducing CYP3A, lithocholic acid has been demonstrated to increase expression of TRPV6 in intestinal cell lines, which corroborates its role as a physiological VDR agonist (517).
We are far from understanding the full spectrum of the effects of 1,25(OH)2-vitamin D, and unfortunately, the scope of this review does not allow a detailed analysis of these processes. In general, our knowledge concerning the physiological role of 1,25(OH)2-vitamin D is massively expanding beyond the horizon of calcium homeostasis. Evidence is accumulating that 1,25(OH)2-vitamin D can influence the renin-angiotensin-aldosterone system and may thereby act as a regulator of blood pressure (660, 662, 1219). Furthermore, low vitamin D status is associated with an increased incidence of colorectal, ovarian, and breast cancer(363, 364, 642, 984). Vitamin D also acts as a potent modulator of the immune response and cell proliferation.
Experiments dating back to the beginning of the 1900s have demonstrated that surgical removal of the parathyroid gland results in tetany. It was recognized very early that administration of calcium could ameliorate or prevent the manifestation of tetany, and it was subsequently concluded that the parathyroid glands play an important role in calcium metabolism (690, 826, 939). In 1925, extracts from the parathyroids were for the first time shown to control tetany in dogs (216). This finding marked the discovery of PTH, which is one of the three prime hormones regulating calcium homeostasis.
1. Production and secretion
PTH is a 84-amino acid peptide hormone produced in the parathyroid glands. Its amino acid sequence was first established in 1970 in bovine (126, 788). Cloning of the human cDNA followed a decade later (440). The PTH gene encodes a 115-amino acid precursor hormone (preproPTH), which is enzymatically cleaved in a two step process to its mature 84-amino acid secreted form (565). The NH2-terminal prepro-signaling sequence is necessary for correct hormone processing and trafficking (340, 546). This knowledge is mainly founded on truncation studies and observations made in patients with mutations in the PTH gene, which can result in familial isolated hypothyroidism, a disorder characterized by hypocalcemia and low PTH levels (38, 340, 828, 1055). For example, a well-characterized T-to-C point mutation in the prepro signaling sequence that causes FIH leads to accumulation of the precursor hormone in the ER (243, 546). The impaired processing triggers ER stress, ultimately apoptosis and PTH insufficiency (243).
Following its cleavage to mature PTH, the hormone is stored in secretory vesicles and released into the circulation in response to low plasma calcium. The regulation of PTH secretion is extremely tight, given the body's need to maintain the calcium concentration within a narrow window (1.1–1.3 mM). Small alterations in calcium homeostasis can have deleterious effects, for example, on the excitability of neurons and muscles. The low plasma half-life of PTH of <5 min allows for a precise regulation of this balance (95). To achieve controlled and rapid on-demand secretion of PTH, the parathyroid is equipped with an ultrasensitive extracellular CaSR, which constantly monitors the plasma calcium levels and triggers intracellular signaling events and PTH release upon imminent drops in calcium levels (FIGURE 6) (see sect. IVD). PTH is further regulated on a transcriptional level by 1,25(OH)2-vitamin D, creating a negative-feedback loop (154, 930, 931).
PTH exerts its physiological effects via activation of a membrane-bound GPCR, the parathyroid hormone receptor type 1 (PTH1R) (PTH2R is mostly expressed in the CNS and tissues that are not involved in calcium handling and will thus not be reviewed). Of note, PTH1R is not exclusively located at the plasma membrane but can also localize to the cell nucleus. The physiological significance of nuclear PTH1R is currently unclear but may represent a novel signaling paradigm for the actions of PTH (830, 856, 857, 1155).
Full-length PTH is not required to activate PTH1R. The NH2-terminal domain of PTH mediates most of the physiological effects of the hormone and is responsible for binding in the α-β-βα binding fold of PTH1R (861). This is why clinically PTH(1–34) is used as a PTH analog with identical biological activity (753, 867, 919, 1094). Conversely, NH2-terminal truncation of PTH(1–34) to PTH(2–34) changes the characteristics to a partial receptor agonist, whereas further truncation to PTH(3–34) results in loss of biological activity (1094).
PTH1R was first cloned from opossum in 1991, which was followed by identification of the highly homologous human cDNA shortly thereafter (536, 964). In the nonactivated state the receptor is expressed as a homodimer at the cell surface, which dissociates upon PTH binding (860). PTH1R is a member of the class B (class 2, secretin family) GPCRs. As many other GPCRs, it undergoes N-linked glycosylation at four asparagine residues (1218). Mutational analysis revealed that site-specific mutation of all four sites decreases cell surface expression, whereas impairment of fewer glycosylation sites does not seem to have significant effects on trafficking or ligand binding (1218). Furthermore, the extracellular domain of the receptor includes a characteristic disulfide bond pattern involving six cysteine residues (392). These residues are thought to be essential for stabilizing the hydrophobic α-β-βα binding pocket for PTH, which is conserved among all members of class B GPCRs and has recently been crystallized in the presence of PTH (861).
Binding of PTH causes activation of at least two distinct G proteins. Gαq/11 mediates intracellular calcium release via phospholipase C (PLC) activation and increases in inositol trisphosphate (IP3), whereas Gαs activates adenylyl cyclase leading to rises in cAMP (5, 188, 808, 859).
PTH1R can be regulated on a variety of levels, ranging from trafficking and internalization to direct protein interactions at the cell surface. Desensitization of PTH1R is mediated by GRK2 binding/phosphorylation and β-arrestin binding, which uncouples the receptor from its associated G proteins and triggers its internalization (267, 326, 330, 1123). For example, knockout of β-arrestin causes increased and sustained levels of the second messenger cAMP in primary osteoblast cultures upon PTH stimulation (327). PTH1R also associates with the scaffolding protein NHERF1, which stabilizes the receptor at the cell membrane and prevents its endocytosis and desensitization (1022, 1139). This effect of NHERF1 is partially attributable to a prevention of an interaction between β-arrestin and PTH1R (1140). Colocalization of NHERF1, β-arrestin, and PTH1R has been demonstrated and suggests that NHERF1 is constitutively bound, whereas β-arrestin association is more dependent on receptor activation (580). Interestingly, it has recently become apparent that β-arrestin not only plays a role in receptor desensitization, but also mediates activation of downstream signaling cascades, such as MAPKs, in a G protein-independent manner (380). Current scientific effort thus focuses on the development of so-called “biased” PTH1R agonists, which can selectively trigger G protein- or β-arrestin-dependent signaling events, allowing for a more selective therapeutic repertoire (381, 1160). Furthermore, NHERF may also modulate the cell's response to PTH binding. In the presence of NHERF2, the cell's calcium response via PLC is augmented, whereas the cAMP response is dampened, presumably via recruitment of Gαi (698). Both NHERF and β-arrestin provide effective examples of how receptor-associated proteins can modulate canonical signaling events or even, as is the case with β-arrestin, initiate signaling events in their own right.
PTH1R can also be regulated by its external environment. The extracellular receptor domain can be cleaved by metalloproteases, resulting in receptor degradation (579). PTH binding prevents this proteolytic cleavage, thus stabilizing PTH1R at the cell surface. The physiological importance of this mechanism is not yet fully understood. However, MMPs are involved in bone remodeling and may in consequence locally regulate the sensitivity of osteoblasts to PTH (579).
3. Cellular effects
PTH exerts its effects in two primary target tissues: bone and the kidney.
In the kidney, PTH causes phostphaturia, increases calcium absorption, and induces the synthesis of 1,25(OH)2-vitamin D. In-detail analysis of renal phosphate handling is beyond the scope of this review and has been summarized previously (334, 765). In brief, phosphaturia mainly results from downregulation of the Na-Pi transporter type IIa (NaPi-IIa) at the apical membrane of the proximal tubule, thereby reducing the amount of reabsorbed phosphate from the primary urine. Rather than directly modulating the transporter's activity, PTH exposure mainly affects the number of active cotransporters on the plasma membrane. Activation of basolateral PTH1R causes retrieval of NaPi-IIa and targets it for lysosomal degradation, resulting in diminished Pi reuptake (566, 682, 847). Apart from NaPi-IIa, at least two other apical phosphate transporters are present in the proximal tubule: NaPi-IIc and PiT-2 (986, 1124). Currently, there is some evidence that PTH can also regulate NaPi-IIc and PiT-2, but additional studies remain to be conducted (855, 987). Although a contribution of these transporters to renal phosphate reabsorption is highly likely, knockout studies suggest that NaPi-IIa is responsible for ∼80% of total phosphate transport, thus representing the major uptake mechanism (72, 468).
It has been recognized for over 30 years that PTH can stimulate the synthesis of the active vitamin D metabolite 1,25(OH)2-vitamin D in the kidney (116, 338) (see sect. IVA4). 1,25(OH)2-vitamin D in consequence enhances the intestinal and renal uptake of calcium in an effort to counteract hypocalcemia, which initially led to secretion of PTH. The PTH-stimulated increase in 1,25(OH)2-vitamin D levels is achieved on a transcriptional level. PTH upregulates the transcription of CYP27B1, the mitochondrial enzyme which is responsible for the conversion from 25(OH)-vitamin D to 1,25(OH)2-vitamin D. Transcriptional upregulation occurs via PTH binding to PTH1R, leading to increases in the second messenger cAMP and activation of PKA (125, 442, 763, 764, 921).
Apart from inducing the synthesis of 1,25(OH)2-vitamin D, PTH can directly upregulate the renal reabsorption of calcium. The regulation of calcium absorption by PTH occurs in distal segments of the nephron, mostly in the distal convoluted tubule and the connecting tubule (99, 379, 1003, 1172). In close analogy to the duodenum, these segments express calcium transport proteins, which are responsible for mediating the process of active transcellular calcium absorption in the kidney, namely, TRPV5, calbindin-D 28k, and NCX1. PTH can regulate all of these protein levels on a transcriptional level (1108). Interestingly, this seems to be accomplished independently of 1,25(OH)2-vitamin D, which also positively regulates most of these transporters (see sect. IVA6) (1108). It is therefore difficult to dissect the relative contribution of each of the two hormones in the physiological regulation of the calcium transport proteins. In addition to transcriptional activation, PTH was shown to cause direct phosphorylation of TRPV5, thereby increasing its opening probability (249). The channel is phosphorylated at threonine-709 in a PKA-dependent fashion (249). Elevation of intracellular cAMP levels and concomitant PKA activation are classical downstream effects of PTH1R activation in the kidney (248, 1172).
In bone, PTH exerts a dichotomous effect depending on the pattern of exposure. It is well documented that pulsatile PTH exposure has anabolic effects on bone mass, whereas continuous release increases plasma calcium by bone catabolism (FIGURE 7) (27, 348, 401, 469, 1065). The observation that intermittent PTH administration increases bone mass has led to the use of PTH as a treatment strategy for osteoporosis (621, 671, 897). The enhanced bone formation mainly results from an increase in osteoblast numbers. This phenomenon has been partially attributed to a PTH-mediated induction of osteoblast differentiation and an inhibition of their apoptosis (75, 274, 518, 530, 531, 684, 969, 1032). Multiple mechanisms underlying the anti-apoptotic effects of intermittent PTH on osteoblasts have been suggested. Among others, these include runt-related transcription factor 2 (Runx2)-mediated transcription of survival genes and increased DNA repair (75, 969). It has further been shown that fibroblast growth factor 2 (FGF2) is partially needed as an endogenous cofactor for the anabolic effects of PTH to take place (934).
Continuous PTH exposure, on the other hand, mainly affects osteoclast numbers and activation, thereby increasing bone turnover. Since osteoclasts are canonically thought to not express PTH1R (although this view has recently been challenged, Ref. 260), the catabolic effects of PTH are relayed through osteoblast signaling. The PTH-induced crosstalk between osteoblasts and osteoclast is mainly mediated by receptor activator of nuclear factor κB (RANK), osteoprotegrin (OPG), and RANK ligand (RANKL). Both RANKL and OPG are expressed by osteoblasts and exert opposing actions on osteoclasts. RANKL promotes osteoclastogenesis by binding to RANK on osteoclasts. Conversely, OPG serves as a soluble decoy receptor for RANKL and thus inhibits its interaction with RANK, thereby suppressing osteoclastogenesis. In accordance with this model, RANKL-deficient animals develop osteopetrosis because of insufficient osteoclasts activation (592). Sustained PTH exposure affects RANK-RANKL signaling by downregulating antiresorptive OPG, while simultaneously stimulating production of RANKL by osteoblasts (FIGURE 7) (350, 497, 679, 689). The enhanced RANK-RANKL signaling induces formation of osteoclasts, which in turn leads to enhanced bone resorption and elevates serum calcium levels.
In the intestine, several observations suggest that PTH may have a direct, i.e., non-1,25(OH)2-vitamin-D mediated, effect on intestinal calcium absorption. Both isolated enterocytes and intestinal loops demonstrated an increase in calcium uptake following acute PTH exposure (781, 783, 784). However, more investigations are needed to clearly establish a direct regulatory role of PTH in the context of intestinal calcium uptake.
4. PTH fragments
It should be noted that full-length PTH(1–84) is not the only circulating form of the hormone in the body. Various PTH fragments can be found in the circulation, which partially originate directly from the parathyroid gland and partially represent products of peripheral cleavage. The parathyroid itself releases COOH- and NH2-terminal hormone fragments, which are generated by cysteine proteases (cathepsin B and H) in distinct secretory vesicles of the gland (418, 427, 693). Interestingly, the fraction of secreted hormone fragments changes with extracellular calcium conditions. It has been reported that more fragments are released under conditions of hypercalcemia, when secretion of full-length PTH is suppressed (417, 418, 605, 726). Peripheral proteolysis represents the second source of PTH fragments. This process occurs predominantly in liver and the kidney (127, 234, 989). The group of fragments that have received the most amount of scientific attention is the large NH2-terminally truncated non-PTH(1–84) fragments. PTH(7–84) is the quantitatively major member of this group, which is secreted by the parathyroids (233). The group of non-PTH(1–84) fragments can represent up to 20% of circulating PTH, but can increase in patients with renal failure dramatically to up to 50% because of impaired renal clearance (130, 131, 444, 648). This is of particular interest, as it has become recently apparent that the non-PTH(1–84) fragments exert biological activity. In general, non-PTH(1–84) fragments antagonize the effects of PTH in its primary target tissues, bone and the kidney but also the parathyroid gland directly. It has been shown that PTH(7–84) can inhibit PTH release from the parathyroid, presumably in an autocrine fashion, despite low serum calcium concentrations (493). In bone, PTH(7–84) blocks the effects of PTH on calcium mobilization in thyroparathyroidectomized rats (786). More detailed investigations revealed that PTH(7–84) reduces calcium release from bone in vivo and inhibits the formation of osteoclast-like cells in murine primary marrow cultures (272). In the kidney, PTH(7–84) can inhibit the formation of 1,25(OH)2-vitamin D, presumably via a posttranscriptional mechanism (768, 1102).
Since activation of PTH1R requires an intact NH2-terminal domain of PTH, it has been speculated that non-PTH(1–84) fragments exert their function through a pathway distinct of PTH1R. The existence of a COOH-terminal PTH receptor has thus been postulated (270–272, 786). The molecular identity of this receptor is, however, not yet resolved. Another mechanism through which non-PTH(1–84) fragments may exert their biological activity is downregulation of PTH1R. It has been demonstrated that PTH(7–84) causes internalization of PTH1R, thus offsetting the effects of PTH by decreasing the number of available receptors at the cell surface (1022).
In conclusion, non-PTH(1–84) fragments act as PTH antagonists and are secreted by the parathyroid in response to hypercalcemia.
Calcitonin is a peptide hormone that has been discovered by Copp et al. in 1962 as a factor that reduces serum calcium concentrations (160, 227). Calcitonin production was initially falsely ascribed to the parathyroid glands, and it was only later that the thyroid gland had been established as the source of calcitonin (463). The primary sites of calcitonin production are the parafollicular cells (C-cells) of the thyroid gland. Calcitonin exerts its hypocalcemic effects primarily by inhibition of osteoclast activity. It should be noted that the importance of calcitonin in day-to-day calcium homeostasis in humans is rather negligible (see sect. IVC4). For this reason, it will only be reviewed concisely.
1. Production and secretion
Calcitonin is encoded by the CALCA gene and is initially synthesized as a 141-amino acid precursor (preprocalcitonin), which is later processed to the mature 32-amino acid hormone (637). The same gene also encodes the neuropeptide CGRP. Production of either peptide is dependent on tissue-specific RNA splicing. Although encoded by a different gene, amylin also belongs to the calcitonin peptide family. All three peptides, i.e., calcitonin, CGRP, and amylin, share some overlapping functions with regard to osteoclast suppression (16, 229, 1199). Calcitonin is released from the C-cells in response to rising concentrations of plasma calcium. The CaSR is responsible for the molecular process of calcium sensing on the parafollicular cells (351, 367, 728).
2. The calcitonin receptor
Calcitonin exerts its physiological functions via activation of the calcitonin receptor. The calcitonin receptor is a seven-transmembrane domain GPCR. In particular, it is a member of the family B subfamily of GPCRs (669). It shares significant homology with other receptors of the family, which include the PTH, GHRH, PACAP, VIP, secretin, glucagon, and glucagon-like peptide receptors.
The calcitonin receptor is expressed in the two most extensively described calcitonin target tissues, i.e., osteoclasts and the kidney, but also in other adult tissues, such as the prostate, CNS, skeletal muscle, and placenta (17, 329, 428, 790, 878, 1174). Multiple isoforms of the calcitonin receptor occur in the body, which results from different RNA splicing directed by tissue-specific promoters (17, 30, 389, 606). Activation of the calcitonin receptor mostly translates into a rise of intracellular cAMP levels via Gs-dependent activation of adenylate cyclase (167, 332, 435, 669). Although the cAMP/PKA pathway appears to be dominant, activation of both PLC and PLD have also been reported (167, 332, 776).
3. Cellular effects
Shortly after the discovery of calcitonin and its hypocalcemic effects, investigators set out to identify its physiological site of action. First evidence for an effect of calcitonin on bone metabolism came from experiments on rat embryonic bone in tissue culture. It was observed that calcitonin caused a decreased basal release of calcium from these preparations (343). These observations served as the first evidence of how calcitonin lowers serum calcium levels. Furthermore, calcitonin blocked the resorptive actions of PTH on bone, albeit only temporarily (342, 343). After 4–6 days of combined treatment (calcitonin + PTH), calcium release rose again (342). This desensitization to the effects of calcitonin has been coined as the calcitonin “escape” phenomenon (342). Today we know that the transient effect of calcitonin on osteoclasts is attributable to a downregulation of surface calcitonin receptors and their synthesis (882, 1061, 1129, 1130).
As alluded to before, calcitonin directly inhibits the action of osteoclasts, causing the balance between bone absorption and formation to shift towards anabolism. Calcitonin exerts its inhibitory effects on osteoclasts via activation of its receptor, which is expressed in abundance on their surface (428, 790, 878). Exposure to calcitonin triggers distinct morphological changes in the osteoclast. Osteoclasts are highly motile cells that resorb bone via formation of so-called resorptive pits, which are membrane invaginations that are luminally acidified by active proton secretion. Calcitonin has been shown to inhibit the formation of these resorptive bays in vitro (487, 1057). Furthermore, osteoclast motility is markedly decreased, causing the cell to enter a state of functional quiescence (169). Apart from directly altering osteoclast function, calcitonin also affects osteoclast differentiation. Calcitonin inhibits the formation of multinucleated mature osteoclasts by arresting their differentiation in more immature stages (1060).
Renal calcium handling is the second organ function that is influenced by calcitonin. However, it is not entirely clear whether calcitonin causes calciuria or enhances calcium reabsorption from the urine. The conflicting results may be to some extent attributable to species differences. In humans, calcitonin likely increases the excretion calcium through the urine and thereby acts in concert with its inhibitory action on osteoclasts to lower serum calcium levels (31, 32, 143, 215, 819, 1016). Conversely, most studies demonstrating an increase in the renal reabsorption of calcium and magnesium were conducted in rats, rabbits, or mice (84, 158, 159, 265, 304, 877, 883, 1225). A single investigation established a calcium-conserving effect of calcitonin in human (160). The primary site of action for calcitonin in the rat is the thick ascending limb (TAL) of the loop of Henle, where calcitonin has been demonstrated to bind its receptor, increase local adenylate cyclase activity, and promote calcium reabsorption (166, 168). Apart from enhancing the reabsorption of calcium, calcitonin also promotes the vectorial transport of NaCl in the rat TAL, thereby amplifying the corticomedullary concentration gradient, which is a prerequisite for the subsequent concentration of urine in the collecting duct (265, 304). In the rabbit, the calcium-conserving effects of calcitonin seem to be mediated by the distal tubule (1225).
It has also been speculated that calcitonin may act directly on the collecting duct in a similar fashion to antidiuretic hormone (ADH or vasopressin), i.e., to concentrate the urine by increasing the reabsorption of water from the primary urine (251). Indeed, calcitonin was shown to increase the apical expression of aquaporin 2 (AQP2) in principal cells of the collecting duct (119). Apical insertion of AQP2 and subsequent transepithelial water movement is the primary mechanism by which ADH causes concentration of the urine to lower plasma osmolarity.
In conclusion, the direct impact of calcitonin on renal calcium handling is quite vague and may be of minor importance. However, calcitonin also has another, indirect effect on calcium homeostasis. Calcitonin was shown to be an important regulator of the expression of CYP27B1, the renal enzyme responsible for the conversion of 25(OH)-vitamin D to 1,25(OH)2-vitamin D (1009, 1217). In normocalcemic rats, CYP27B1 mRNA levels were inducible by calcitonin administration, leading to an increase in the production of 1,25(OH)2-vitamin D (1009, 1217). Furthermore, CYP24A1 was induced in HEK-293 cells following calcitonin exposure, suggesting a regulatory role of calcitonin in vitamin D catabolism (361). In addition to its putative direct effect on calcium reabsorption, calcitonin may thus affect normal calcium metabolism indirectly by modulating the levels of circulating 1,25(OH)2-vitamin D.
4. The relevance of calcitonin for calcium homeostasis
The relevance of calcitonin for day-to-day calcium balance is highly debatable. This is corroborated by fundamental observations during conditions of decreased or increased calcitonin levels, neither of which result in an appreciable phenotype in terms of calcium balance. For example, patients with medullary carcinomas of the thyroid (MTC), a tumor of the thyroid C-cells resulting in the hypersecretion of calcitonin, were shown to have normal bone mineral densities (1176). Animal studies further substantiate this conundrum. A reduction in serum calcitonin levels by thyroidectomy in rats did not impact serum calcium levels (223, 1064). In light of this evidence, the question arises as to what the physiological role of calcitonin is.
It has been suggested that calcitonin is an evolutionary remnant (462). This is substantiated by the fact that calcitonin from other species is more potent than human calcitonin. Teleost calcitonin has the highest biological activity in humans, which may be a result of their higher dependence on the hormone. For example, salmon calcitonin has an approximately sixfold higher affinity to calcitonin receptor than human calcitonin (28, 328). Furthermore, it is less effectively eliminated by the kidney, resulting in a longer plasma half-life (405). The differences in the biological activity between calcitonin forms led to the introduction of salmon calcitonin as a treatment for skeletal disorders, such as osteoporosis or Paget's disease (1077).
Given its ambiguous role in regular calcium homeostasis, it has been postulated that calcitonin may be of importance during states of high calcium demand, such as during lactation (1170). It has been demonstrated that calcitonin and CGRP (−/−) mice show greater loss of skeletal mass during lactation than wt animals (1170). Since calcitonin and CGRP are encoded by the same gene, animals were controlled by CGRP substitution, which was without effect (1170). Another mechanism by which calcitonin may be osteoprotective during lactation or pregnancy is by inducing the renal synthesis of 1,25(OH)2-vitamin D (1217).
D. The CaSR
The CaSR is a G protein-coupled membrane-bound receptor that is the primary sensor for calcium and is the first link in the regulatory chain of calcium homeostasis. By regulating the release of PTH from the parathyroid to modulate current serum calcium levels, it presides over the subsequent hierarchical cascade of vitamin D synthesis and calcium handling in the vitamin D target organs, such as the intestine, kidney, and bone. More recent investigations have demonstrated that the CaSR is not exclusively expressed in the parathyroid gland, but is also present locally in the vitamin D target organs. This diverse expression suggests that CaSR can modulate organ function locally and outside of the strict PTH-vitamin D-organ axis. Thus a short and local feedback loop is created, which allows the organs to respond rapidly to the local calcium environment. A detailed description of the receptor's role in each tissue is beyond the scope of this article, but an attempt will be made to identify the key features of CaSR in these local sites in a subsequent section (FIGURE 8). For excellent in-depth reviews, please refer to Refs. 372, 711, 906.
The importance of CaSR as a regulator of calcium balance is underlined by clinical pathologies that are caused by its mutation. Loss-of-function mutations cause familial hypocalciuric hypercalcaemia (FHH; OMIM 145980) and neonatal severe hyperparathyroidism (NSHPT; OMIM 239200), whereas activating mutations cause autosomal dominant hypocalcemia (OMIM 601198). These mutations mostly change the threshold of receptor activation in either direction. Currently, ∼300 different mutations are reported, two-thirds of which represent inactivating mutations (241). Given its broad expression pattern, the deranged sensing of blood calcium levels in these disorders not only affects the secretion of PTH from the parathyroid glands, but also causes local dysfunction in other organs, such as the kidney where calcium absorption is perturbed.
1. Structure and signaling
The CaSR was first cloned by Brown et al. (134) in 1993 from bovine parathyroid using a Xenopus oocyte expression cloning system. Cloning of the human CaSR followed 2 years later (366). The CaSR is a 1,028-amino acid protein that belongs to the superfamily of classic 7-transmembrane domain G protein-coupled receptors (134, 366). It is mainly expressed as a homodimer on the cell surface (55). The dimerization process has been shown to take place in the ER and is mediated by the formation of disulfide bonds between cysteine residues (C129, C131) and noncovalent interactions of leucine residues (L112, L156) in the extracellular domain of the receptor (54, 529, 858, 888, 1213). Following assembly in the ER, the CaSR undergoes N-linked glycosylation in the Golgi apparatus, some of which is pivotal for cell surface expression (887). The trafficking between ER and Golgi apparatus is regulated by the small GTP-binding protein Rab1 (1221). Knockdown and mutations of Rab1 in HEK cells results in decreased numbers of CaSR at the cell surface (1221). Conversely, the internalization of CaSR is thought to be mediated by ubiquitination by the E3 ligase dorfin (498).
On the cell surface, the CaSR resides in caveolin-1-rich plasma membrane domains, which also contain associated signaling proteins (571). These signaling complexes are formed with the help of scaffolding proteins. The COOH-terminal tail of CaSR binds to filamin A, an actin binding protein (48, 467). Silencing of filamin A with siRNAs results in the attenuation of MAPK signaling by the receptor (496) (see below).
In lack of a crystal structure of CaSR, the exact binding sites of calcium on the extracellular domain remain subject of speculation. So far, applicable structural data has only been obtained from the metabotropic glutamate receptor type I (mGluR1), which belongs to the same family of type C GPCRs. In this model, glutamate binds to key residues which are located in a cavity, embedded in between two lobular domains (LB1 and LB2) of the extracellular tail (611). This structural hallmark has been aptly coined the receptor's Venus fly trap module. This motif is conserved among other GPCRs of the same family. Multiple attempts to identify the calcium binding sites have been undertaken using computational homology modeling (499, 500, 1014). These models have postulated between one and five calcium binding sites (499, 500, 1014). With the employment of mutational analysis, it has been possible to validate functionality of some of these putative binding sites (500). Monitoring of the intracellular calcium response to increasing extracellular calcium levels in CaSR transfected HEK cells had indicated previously that the Hill coefficient for this response was ∼3.1, suggesting that multiple calcium binding sites may exist (834).
It should be noted that the CaSR can also be stimulated by other polyvalent cations (Mg2+, Pb2+, Cd2+, Fe2+, Ba2+, Ni2+, Co2+, or Gd3+) and larger polycationic molecules, such as spermine, spermidine, putricine, protamine, and neomycin (134, 416, 880). Furthermore, several substances can allosterically modify the receptor and potentiate its sensitivity to its direct agonists. These include pharmacological small molecule substances (calcimimetics) that are in clinical use for the treatment of conditions, such as secondary hyperparathyroidism, and L-type amino acids, which enable the CaSR to act as a nutrient sensor (220). Truncation studies have demonstrated that the Venus fly trap motif is necessary for allosteric modification by L-type amino acids (759). The affinity of calcium to the CaSR can also be modulated by changes in extracellular pH (279, 879). An acidic extracellular milieu has been shown to decrease the sensitivity of CaSR to its agonists, whereas an increased extracellular pH has converse effects (879).
The intracellular domain of CaSR contains five PKC phosphorylation sites (366). Mutational analysis demonstrated that PKC-mediated phosphorylation of the CaSR at Thr-888 blunts its response to extracellular calcium, as evidenced by inhibited calcium release from intracellular stores (56, 246). The phosphorylation occurs in response to receptor activation and thus represents an autoinhibitory feedback mechanism (246). β-Arrestins are most likely involved in the process of PKC-associated desensitization (681, 853). Conversely, dephosphorylation of the Thr-888 residue is carried out by a calyculin-sensitive phosphatase, thereby restoring the receptor's initial sensitivity (246). Another mechanism of desensitization is mediated by a G protein receptor kinase (GRK), most likely by interfering with Gαq regulated pathways (see below) (681, 853).
Once stimulated, the CaSR activates a variety of intracellular signaling cascades. Being a GPCR, most of these processes are mediated by G proteins. Specifically, Gαq/11, Gαi, and Gα12/13 have been shown to be coupled to the CaSR (40, 175, 495, 849). The expression of all subunits was confirmed in bovine parathyroid (1115). Gαi mediates the suppression of cAMP levels by inhibiting adenylyl cyclase and activates the ERK/MAPK pathway (175, 250, 375, 572). Activation of Gαq/11 results in increased intracellular calcium concentrations via activation of PLC and IP3 triggered calcium release (133, 1010). As demonstrated in HEK cells, this cascade can also activate further downstream phospholipase A2 leading to production of arachidonic acid metabolites (415). Gα12/13 is thought to regulate phospholipase D and phosphatidylinositol 4-kinase (PI 4-K); however, this interaction has only been demonstrated in heterologous cell culture system (494, 495).
2. CaSR in the parathyroid
CaSR regulates parathyroid function at three levels: 1) the release of PTH from secretory granules, 2) de novo synthesis of PTH, and 3) parathyroid cell growth.
Activation of CaSR by increasing plasma calcium results in an inhibition of PTH release, thereby lowering calcium levels. It is thought that this response is mediated by the generation of arachidonic acid metabolites via Gαq and PLA2 activation (FIGURE 6) (121, 152). Cultured porcine parathyroid cells demonstrated an increase in arachidonic acid production after CaSR stimulation while PTH release was inhibited (121). Furthermore, exogenous administration of arachidonic acid suppressed PTH release from the parathyroid cells (121). Similar effects were demonstrated for the arachidonic acid metabolites 12- and 15-hydroxyeicosatetranoic acid, suggesting that they represent the downstream effectors of arachidonic acid production (120).
Apart from directly controlling PTH release, CaSR also modulates PTH synthesis. PTH gene transcription is mainly regulated by 1,25(OH)2-vitamin D. Binding of 1,25(OH)2-vitamin D to the VDR causes a decrease in pre-pro-PTH mRNA levels creating a negative-feedback loop (154, 930, 931). However, it was recognized before the identification of the CaSR that serum calcium can modulate the actions of 1,25(OH)2-vitamin D on PTH gene transcription (930). It was shown that increases in calcium can potentiate the inhibitory effects of 1,25(OH)2-vitamin D (930). This effect is most likely mediated by CaSR, whose activation can decrease PTH transcription by augmenting the inhibitory effects of 1,25(OH)2-vitamin D. Molecularly this is achieved by upregulating the expression of the VDR (151, 162, 362, 653, 916). The current working model states that activation of CaSR causes an increase of arachidonic acid metabolites and activation of the MAPK pathway, which in turn results in increased VDR mRNA levels (FIGURE 6) (151). This allows the parathyroid to adjust its 1,25(OH)2-vitamin D sensitivity to the current plasma calcium levels.
The molecular mechanisms underlying the trophic effects of CaSR activation are less clear. Earlier observations had already suggested that hypocalcemia is associated with parathyroid cell proliferation (778). Currently, the CaSR specific calcimimetics provide the most useful insight into the regulation of parathyroid growth by CaSR. Calcimimetics administered in the context of both animal models and clinical studies of hyperparathyroidism demonstrate that activation of CaSR leads to a reduction in gland size (510, 589, 746, 1128). Conversely, inactivating mutations of CaSR result in parathyroid enlargement. Parathyroid-selective genetic disruption of Gαq was furthermore shown to cause moderate hyperparathyroidism with increased plasma PTH levels and gland hyperplasia, suggesting a role of Gαq in the regulation of parathyroid cell growth (849). Similar findings were reported in Gαq/11 double KO animals (1159).
3. CaSR in the kidney
The CaSR acts as an important regulator of ion and water homeostasis in the kidney. It should be noted that it can exert its effects on calcium transport independently of other hormonal regulators, such as PTH and 1,25(OH)2-vitamin D. The CaSR is expressed along most of the nephron, albeit in varying subcellular localizations (FIGURE 8) (907, 908). In the proximal tubule, CaSR is localized apically at the base of the brush border, where it has been implicated to play a role in phosphate transport (52, 907, 909). The primary regulator of phosphate transport in the proximal tubule of the kidney is PTH. In brief, increased PTH levels inhibit phosphate reabsorption from the lumen. Activation of the apical CaSR can partially reverse the effects of PTH and restore phosphate absorption (52). Conversely, PTH and high phosphate levels reduce CaSR expression (909). Furthermore, it is likely that the CaSR mediates the inhibitory effects of calcium on 1,25(OH)2-vitamin D synthesis in the proximal tubule (109, 702).
In the thick ascending limb of the loop of Henle, CaSR is located on the basolateral membrane (907). In this nephron segment, the receptor acts as a major modulator of monovalent and polyvalent ion absorption. Activation of CaSR leads to an inhibition of the apical renal outer medullary potassium (ROMK; Kir1.1) channel, mainly through arachidonic acid metabolites created by PLA2 (1147, 1148). Apical ROMK releases potassium ions into the lumen, which in turn are needed to fuel apical ion uptake through the Na+-K+-2Cl− (NKCC2) cotransporter. By decreasing apical potassium efflux, CaSR inhibits sodium and chloride uptake through NKCC2 (906). This correlation is reflected in much earlier observations, which report that calcium infusions can decrease tubular sodium clearance (300, 718, 1052). In addition, impairment of NKCC2 has also implications for calcium absorption. Reduced NKCC2 activity decreases the lumen-positive potential and negatively affects countercurrent multiplication, and in consequence the nephron's ability to concentrate urine (434). Both mechanisms will lead to impaired calcium absorption (434). Calcium absorption in the medullary portion of the thick ascending limb is thought to occur predominantly as passive uptake through the paracellular route (995). Similar observations have been made when blocking NKCC2 pharmacologically with the loop diuretic furosemide (299). It has thus been proposed that activation of basolateral CaSR has “loop diuretic-like” effects, reducing NaCl but also calcium absorption in the kidney (434).
In contrast to the medullary section, the cortical portion of the thick ascending limb has been proposed to have predominantly active calcium uptake properties, which are under the hormonal regulation of PTH and calcitonin (344, 509). PTH increases calcium absorption in this segment, and it has been shown that, similarly to phosphate absorption in the proximal tubule, activation of CaSR can suppress the effects of PTH (264, 754). The absorption of NaCl does not seem to be affected by CaSR (264, 754).
The distal convoluted tubule and the connecting tubule are responsible for the fine-tuning of calcium reabsorption in the kidney. To achieve this goal, they are equipped with molecular machinery, similar to that in the duodenum (TRPV5, calbindin-D 28k, NCX1, and PMCA1b) to absorb calcium against its electrochemical gradient through the transcellular pathway (680). In analogy to the proximal small intestine, these transporters are predominantly regulated through 1,25(OH)2-vitamin D, but also PTH. CaSR colocalizes with TRPV5 in this segment (1090). Its activation causes increase calcium influx through TRPV5 and may thereby locally and rapidly adapt active absorption to the urine calcium concentration (FIGURE 8) (1090).
Apart from regulating calcium and phosphate absorption in the kidney, CaSR modulates proton and water movement in the collecting duct. In the intercalated cells of the collecting duct, apical V-ATPase acidifies the urine in an effort to maintain systemic acid-base homeostasis. It has been shown that luminal calcium and neomycin can induce V-ATPase activity via activation of CaSR, thereby causing proton secretion into the urine (FIGURE 8) (901). Since the formation of calcium kidney stones is dependent on luminal pH, it has been speculated that this may represent an autoprotective mechanism that prevents nephrolithiasis (901). Furthermore, stimulation of apical CaSR in the principal cells of the collecting duct leads to decreased ADH (vasopressin)-stimulated water reabsorption through AQP2 (FIGURE 8) (942, 943). Taken together, activation of CaSR has diuretic effects via inhibiting NKCC2 in the thick ascending limb of the loop of Henle and by inhibiting AQP2-mediated water reabsorption the collecting duct.
4. CaSR in the gastrointestinal tract
The CaSR is distributed along most of the gastrointestinal tract, ranging from the stomach to the large intestine (186, 360, 744, 932, 998). We are now only slowly beginning to unravel its function in this diverse array of tissues. In the stomach, CaSR localizes to the basolateral membrane of the acid-secreting parietal cells and to all membranes of the gastrin-secreting G-cells (FIGURE 8) (142, 182, 886). Primary cultures of G-cells were shown to release gastrin after stimulation of the CaSR with calcium (142, 886). The release is mediated via calcium influx into the cytosol through nonselective cation channels opening after CaSR stimulation (142). These findings provide the molecular basis for the observation that rises in serum calcium can increase serum gastrin levels (see sect. IIB2). The apical expression of CaSR in G-cells theoretically enables it to act as a luminal nutrient sensor modulating gastric acid secretion and other parameters. In recent studies there is direct evidence showing that gastrin levels increase in mice after calcium and L-type amino acid ingestion (325). This effect was abolished in CaSR (−/−) animals (325). In healthy human test subjects, pharmacological stimulation of CaSR leads to a concomitant increase in gastrin levels and gastric acid output (165). CaSR on G-cells was thus postulated to play an important role in the gastric phase of acid secretion by maintaining acid output by maintaining gastrin secretion (325).
Apart from being expressed on G-cells, CaSR is also localized on the basolateral membrane of the acid-secreting parietal cell, where it exerts effects that are independent of gastrin and other secretagogues. Activation of parietal cell CaSR has been reported to increase H+-K+-ATPase-mediated proton secretion, thereby acidifying the gastric lumen (FIGURE 8) (145, 291, 373). This stimulatory effect was demonstrated for direct activators, such as calcium or Gd3+, but also allosteric modifiers, such as L-type amino acids (145, 291, 373). In parallel to other tissues, the intracellular activation signal for H+-K+-ATPase is mediated by rises in intracellular calcium, PLC, MAPK, and PKC (899). In conclusion, both rises in luminal and plasma calcium concentrations can induce gastric acid secretion either indirectly through gastrin release or directly through parietal cell activation. The physiological significance of this observation remains the subject of speculation, but may be linked to facilitating calcium uptake by increasing acid output. As described in a subsequent section, it has been speculated that gastric acid increases the bioavailability of ingested calcium (see sect. V). Alternatively, CaSR may primarily function as a nutrient sensor in the stomach (amino acid sensing), which maintains constant acid output in the gastric (apical G-cell sensing) and postprandial (basolateral parietal cell) phase of digestion, when circulating levels of amino acids are high (325).
In the intestine, functional investigations on CaSR have mainly been conducted in colonic epithelia, where CaSR localizes to both the basolateral and apical membranes of the colonic crypt (173, 186, 360). Expression patterns vary slightly in the small intestine, with general basolateral expression and additional weak apical expression in the villus (173, 360). Furthermore, CaSR is expressed in both the Meissner's and Auberach's plexuses. Early experiments on single perfused colonic crypts demonstrated that intracellular calcium concentrations could be increased when exposing the crypts to classic CaSR agonists and that forskolin-stimulated fluid secretion could be inhibited (186). This and subsequent investigations indicate that CaSR plays an important role as a modulator of colonic fluid secretion (185, 186). Subsequently, attempts have been made to take advantage of the “constipatory” effects of CaSR activation in pathophysiological settings. Activations of CaSR in the course of diarrheagenic enterotoxin exposure was shown to decrease fluid secretion via increased breakdown of cyclic nucleotides (371). Although the potential clinical applications of ameliorating the symptoms of secretory diarrhea are promising, more efforts will have to be made to fully unravel the physiological role of CaSR in intestinal ion and fluid transport. So far it is not clear whether intestinal CaSR can modulate calcium absorption, as is the case in the kidney.
5. CaSR in bone
It is well established that CaSR is expressed in osteoblasts, osteoclasts, and their respective precursors (172, 174, 542, 1183–1185, 1192). The functional role of CaSR in these cells is, however, less clear. Undoubtedly, both cell lines are exposed to local fluctuations in calcium concentrations making an adaptive response to the calcium environment plausible. Indeed, changes in extracellular calcium concentration have been shown to regulate various cell functions, mostly in in vitro models. Extracellular calcium can stimulate the proliferation, migration, and differentiation of osteoblasts (174, 297, 1183, 1184, 1186). Similarly, calcium was proposed as a differentiation signal for osteoclasts (542, 544, 734). Significant doubt about the in vivo importance of CaSR in bone has emerged with the generation of the CaSR (−/−) mice. Although CaSR knockout results in rickets, these animals suffer from severe hyperparathyroidism, which did not allow a discrimination between the effects of high PTH and CaSR on bone turnover (365). Concomitant genetic ablation of the parathyroid gland or PTH secretion, however, revealed that the skeletal phenotype of CaSR single mutation (−/−) could mostly be rescued, suggesting that the skeletal abnormalities were due to high circulating PTH levels rather than CaSR inactivation (598, 1096). Furthermore, CaSR does not seem to be the exclusive calcium-sensing mechanism in osteoblasts, as changes in extracellular calcium can still elicit functional responses in CaSR (−/−) osteoblasts (852). This observation has been attributed to another GPCR with calcium-sensing capabilities, namely, GPRC6A (850, 851). Although GPRC6A has a higher activation threshold for calcium, it also responds to the CaSR allosteric activator R568 (851). GPRC6A activation may thus represent a confounding factor in most in vitro studies on osteoblasts and their modulation by CaSR. Also, GPRC6A knockout leads to osteopenia, further underlining the possibility of an alternate calcium-sensing pathway in bone (850). Osteoblasts extracted from these GPRC6A-deficient animals show decreased sensitivity to extracellular calcium and in vitro mineralization defects (854).
Although these observations have profoundly questioned the physiological significance of CaSR in bone, closer examination still favors a role of CaSR in bone turnover. With the recent advances in genetic methods, an osteoblast-specific CaSR (−/−) model has been created (171). These animals have severely stunted growth and skeletal development, clearly suggesting an involvement of CaSR in normal osteoblast function (171). The previous conflicting evidence gained from global CaSR (−/−) models with survival rescue by elimination of PTH synthesis have been attributed to the possible expression of alternate CaSR splice variants, which may compensate for the deletion of full-length CaSR in these animals (171, 915). In an attempt to further elucidate the function of CaSR in osteoblasts, the reverse approach has been executed by specifically upregulating CaSR in osteoblasts with use of a constitutively active receptor mutant (296). Upregulation of CaSR results in bone loss, as evidenced by a decrease in bone volume and density, specifically of trabecular bone (296). These findings are accompanied by an increased number of osteoclasts, whereas osteoblast parameters were essentially unchanged (296). Activation of CaSR has been speculated to promote RANKL production by osteoblasts, which serves as an osteoclastogenic signal (296). Osteoblasts may thus recruit osteoclasts and induce their maturation via CaSR signaling and increased RANKL expression, which would explain the observed increase in bone turnover and osteoclast numbers in the setting of constitutive CaSR activation (241).
V. THE STOMACH AND CALCIUM
Preceding parts of this review have independently summarized the physiology of acid secretion, intestinal calcium absorption, and their respective regulation. The following section will attempt to illustrate the functional intersections between these seemingly unrelated fields. In particular, the question of whether acid is needed to absorb calcium effectively from the gut or whether the stomach contributes to the regulation of calcium homeostasis by secretion of an endocrine substance will be investigated.
A. Proton Pump Inhibitors and the Risk of Fracture
In May 2010, the Food and Drug Administration (FDA) released the following safety announcement: “Healthcare professionals and users of proton pump inhibitors should be aware of the possible increased risk of fractures of the hip, wrist, and spine with the use of proton pump inhibitors, and weigh the known benefits against the potential risks when deciding to use them” (319).
Proton pump inhibitors (PPIs, see sect. IIC1) are in widespread use for the treatment of acid-related disorders, such as gastroesophageal reflux disease (GERD) or gastric ulcer disease. They exert their curative effects by inhibiting the acid output of the stomach. Over the recent years, mostly epidemiological evidence has accumulated which links the intake of PPIs to an increased risk of sustaining fractures, especially in the elderly population. Yang et al. (1190) published one of the earliest and largest studies investigating this potential correlation in 2006. Examining a population of over 13,000 hip fracture cases and over 135,000 controls over the age of 50, the authors concluded that long-term (over 1 year) PPI use was associated with an increase in hip fractures (AOR = 1.44) (1190). Although the likelihood of sustaining a fracture following PPI intake may seem fairly low, the implications for public health are substantial. This has multiple reasons: PPIs represent the third most commonly prescribed medication in the United States and are also available as over-the-counter formulations. Furthermore, there is an ongoing debate whether PPIs are overprescribed, putting certain populations at unnecessary risk of side effects. In combination with the high incidence of osteoporotic fractures, the mean incidence of hip fractures alone between 1986 and 2005 was 957 per 100,000 women over the age of 65 per year, a small increase in risk suddenly has implications for a very large population (123).
The roots of this controversy may potentially be traced back to the 1940s and 1950s. Before the advent of PPIs, total and partial gastrectomies or vagotomies were performed to control acid-related disorders. It was soon apparent that patients who underwent these radical surgical procedures developed osteoporosis/-malacia (58, 305, 732, 876). A study that assessed the prevalence of osteomalacia in gastrectomized patients concluded that up to 12% of patients (19% of females) had histologically overt osteomalacia, although general disturbances in calcium metabolism were estimated to occur in up to 28% of patients (208, 368). Other investigations came to lower prevalence results of ∼5–10% (1091). The osteomalacia was also shown to translate into an increased incidence of fractures in these patients (795). Naturally, gastrectomy represents a radical intervention, and the reasons for this correlation may be multifactorial, but reduced acid output may be of significance.
Back in the field of PPIs, the seminal epidemiologic investigation by Yang et al. was subsequently followed up by a number of studies, which also focused on other types of fractures, other populations, and other drugs reducing gastric acid output, such as H2 receptor antagonists (202, 228, 252, 394, 400, 559, 868, 923). Although their conclusions were somewhat controversial, a recent meta-analysis supports the initial hypothesis that a correlation between PPI intake and fracture risk (hip, spine, and any-site fractures) exists (1195). The meta-analysis considered 11 studies and identified an overall odds ratio of 1.30 for all fracture types combined (1195). There was no association between H2 blocker intake and an increase in fracture risk, although some single studies supported a link (228, 1195). Another meta-analysis came to a comparable conclusion with regard to an increased fracture risk under PPI exposure (616).
B. Gastric Acid and Intestinal Calcium Uptake
A variety of reasons could theoretically account for the observation that PPIs increase the likelihood of fractures. The most prominent hypothesis assumes that the reduced acidity in the stomach impairs the intestinal absorption of dietary calcium. This assumption is based on both patient observations and experimental animal data. Alas, the number of animal studies, which in contrast to investigations in humans per default allow more radical experimental designs and genetic manipulation, is very small.
1. Effects of gastrectomy, vagotomy, and PPIs on mineral metabolism in humans
Before discussing the reports that try to correlate PPI use with calcium uptake, it is worthwhile to examine older literature on patients that had undergone partial or total gastrectomy. As discussed previously, these procedures are known to be linked to bone disease. In contrast to PPIs, which eliminate the singular factor of acid secretion, gastrectomies also influence gastric emptying, the emulsification of food stuffs, and food habits. It is thus more difficult to draw conclusions on the influence of acid secretion on calcium absorption from gastrectomized patients than from individuals on PPI therapy. A further caveat lies in the type of gastrectomy, as different surgical procedures are and were in use. Some surgeries bypass the duodenum (Billroth II, Roux-en-Y, total gastrectomy), whereas some leave the duodenal passage intact (Billroth I).