The emergence of phosphate as a specific signaling molecule in bone and other cell types in mammals

Abstract

Although considerable advances in our understanding of the mechanisms of phosphate homeostasis and skeleton mineralization have recently been made, little is known about the initial events involving the detection of changes in the phosphate serum concentrations and the subsequent downstream regulation cascade. Recent data has strengthened a long-established hypothesis that a phosphate-sensing mechanism may be present in various organs. Such a phosphate sensor would detect changes in serum or local phosphate concentration and would inform the body, the local environment, or the individual cell. This suggests that phosphate in itself could represent a signal regulating multiple factors necessary for diverse biological processes such as bone or vascular calcification. This review summarizes findings supporting the possibility that phosphate represents a signaling molecule, particularly in bone and cartilage, but also in other tissues. The involvement of various signaling pathways (ERK1/2), transcription factors (Fra-1, Runx2) and phosphate transporters (PiT1, PiT2) is discussed.

Introduction

Phosphorus is the second most abundant mineral in the human body and plays a critical role in vital processes. In body fluids, phosphorus is present mainly as inorganic phosphate (Pi) in the form of H₂PO₄ ⁻/HPO₄ ²⁻ with a 1:4 ratio at physiological pH values. In cells and tissues, it is found in a variety of organic substances, including sugar phosphates, phosphoproteins, phospholipids and nucleic acids. Approximately 85% of phosphorus is in bone, and is primarily complexed with calcium (Ca²⁺) in hydroxyapatite crystals deposited onto the collagen matrix [1]. The remainder is in soft tissue with only about 1% in extracellular fluids [2]. Hence, phosphorus has structural roles in phospholipids of cell membranes, nucleoproteins and nucleic acids [3, 4], and is essential for bone mineralization [1]. Moreover, Pi is central to the storage and liberation of metabolic energy, cellular signaling, enzyme activity, lipid metabolism, muscle and neurological functions and the delivery of oxygen to the peripheral tissues [2, 5–10]. Prolonged deficiency of Pi results in hypophosphatemia along with serious biological consequences, including impaired mineralization of bone resulting in osteomalacia or rickets, central nervous system dysfunction, increased erythrocyte membrane rigidity, abnormal leukocyte and platelet function, rhabdomyolysis and muscle weakness, and cardiac dysfunction and respiratory failure [11–15]. On the other hand, hyperphosphatemia is now known to decrease life expectancy and lead to seizures, cardiac dysrhythmias, muscle weakness and tetany, decreased visual acuity, soft tissue calcification, renal failure and eventually death [16–21]. Therefore, controlling the serum Pi concentration is critical for the well being of the organism.

The normal range of serum Pi changes from the neonatal period through adolescence to reach adult values towards the end of puberty (Table 1). Pi concentrations are highest in the neonatal period (1.88–2.4 mM), but fall rapidly in the first months of life and then more slowly to reach childhood levels (1.45–1.80 mM). The concentrations then remain fairly constant until adolescence, when they fall to adult values (0.85–1.44 mM) [22] (Table 1). A normal Pi serum value is the result of a properly controlled absorption, reabsorption and bodily Pi consumption. Absorption of Pi occurs primarily in the ileum, whereas Pi reabsorption occurs in the kidney, principally along the proximal convoluted tubules [23]. In states of neutral Pi balance, the amount of Pi absorbed in the intestine (approximately 0.8–1.5 g per 24 h) is equivalent to the amount excreted in the urine, reflecting that serum Pi homeostasis is maintained primarily through the control of renal Pi reabsorption.

Table 1.

Normal serum values of Pi, Ca, magnesium and potassium in children and adults

	Pi (mM)	Total Ca (mM)	Ionized Ca (mM)	Mg (mM)	K (mM)
Neonatal period	1.88–2.4	1.8–1.9	1.0–1.3	0.5–0.75	3.9–5.9
Childhood	1.45–1.8	2.2–2.6	1.1–1.4	0.6–0.9	3.5–5
Adulthood	0.85–1.44	2.2–2.6	1.1–1.4	0.7–0.9	3.5–5

Pi inorganic phosphate, Ca calcium, Mg magnesium, K potassium

The regulation of Pi homeostasis is classically described as a complex process involving the interplay between parathyroid hormone (PTH) and the vitamin D endocrine system. Whereas 1α,25-dihydroxyvitamin D (1α,25(OH)₂D₃) increases the efficiency of Pi absorption in the intestine [24–26], PTH decreases the efficiency of renal Pi reabsorption [27–29]. Over the past 15 years, considerable advances has been made in the understanding of the mechanisms that specifically control Pi balance [22, 30–32]. Several factors, such as fibroblast growth factor 23 (FGF-23), klotho, secreted frizzled-related protein 4 (sFRP-4), stanniocalcin-1 and -2, dentin matrix protein 1 and matrix extracellular glycoprotein have emerged as major regulators of Pi homeostasis and suggest the existence of an elaborate network of humoral interactions and feedback loops involving intestine, kidney, parathyroid gland, and bone (Fig. 1). As this topic is beyond the scope of this short review, the reader can refer to numerous recent reviews concerning these various factors and their role in controlling Pi homeostasis [14, 30, 33–36].

Fig. 1.

Simplified overview of the interrelationships between organs, hormones and transporters involved in the maintenance of phosphate homeostasis. Major regulators that inhibit (red lines) or stimulate (green lines) phosphate absorption or reabsorption, and bone mineralization are shown. The reader can refer to the text for more details, and to references [14, 30, 33–35]

Despite the considerable progress made in our understanding of Pi homeostasis, there are still many gaps in our knowledge. Particularly, a necessary first step in the regulation of Pi homeostasis is the ability of a cell or an organism to detect changes in Pi concentrations in its environment. This would imply the existence of a Pi-sensing mechanism able to detect changes in the concentration of Pi. While the existence of such a mechanism still lacks evidence in mammals, a number of recent observations suggest that Pi may act as a signaling molecule able to modulate diverse cellular functions. This review focuses on the role of Pi as a signaling molecule at both the cellular and whole-body level. Its role in bone and cartilage will be highlighted and questions regarding its action mechanisms will be discussed.

Mammalian phosphate transporters

Because a cell’s interior is electronegative relative to the exterior, the movement of Pi into the cell does not occur by simple diffusion. In mammals, Na⁺-coupled Pi cotransporters mediate the transport of Pi across cell membranes [23, 37]. Currently, identified Na⁺-Pi cotransporters in mammals have been grouped into three dissimilar families, based on their amino acid sequence and structural similarity [38, 39]. These families differ not only by their affinities for Pi, but also by their tissue distribution, physiological regulations and roles. Whilst they are often referred to in the literature as “type I, II and III” Na⁺-Pi cotransporters, a numbering system based on the chronological identification of their Pi transport activity, it must be noted that all these transporters are multi-spanning membrane proteins and are therefore classified as type III integral membrane proteins according to the classification of membrane proteins proposed by Singer in 1990 [40–42]. The human gene nomenclature database (http://www.genenames.org/) has assigned these transporters to the solute carrier series SLC17, SLC34, and SLC20. Hence, the first identified (i.e. “type I”) membrane protein bearing a Na⁺-Pi cotransport activity belongs to the SLC17 series and is known as NPT1; the second identified (i.e. “type II”) Na⁺-Pi cotransporter belongs to the SLC34 series and is known as NPT2; and the last identified (i.e. “type III”) mammalian Na⁺-Pi cotransporter belongs to the SLC20 series and is known as PiT (Fig. 2).

Fig. 2.

The three families of mammalian Na⁺-phosphate cotransporters. a Assignment to solute carrier series and schematic representation of membrane topology according to current accepted models, as indicated. b Substrate specificity and stoechiometry of Pi transport. c Affinity for phosphate (Michaelis–Menten K _m value). d Main site of expression in adults

Although the SLC17 transporters display Pi uptake capacity [43], they also transport other anions and molecules [44–48] and their physiological roles remain to be fully established [49, 50]. The SLC34 transporters include NPT2a, NPT2b and NPT2c. Npt2a (SLC34A1) is primarily located at the brush border of proximal tubular epithelial cells, is recognized as the main transporter responsible for the renal reabsorption of Pi, and is a target for most humoral factors regulating Pi homeostasis [39, 51, 52]. Npt2b (SLC34A2) is expressed broadly and has recently been shown to be the main transporter for intestinal Pi absorption [53, 54]. Npt2c (SLC34A3) is localized in the same cells as Npt2a, and is not expressed in bone [55]. Its role is still unclear, and it has been suggested it plays a role during body growth [56]. Recently, the Miyamoto group showed that Npt2c has important physiological roles in Pi and Ca homeostasis by modulating the vitamin D/fibroblast growth factor 23 axis, but also in bone mineralization [55, 57]. The SLC20 family of transporters includes PiT1 (SLC20A1) and PiT2 (SLC20A2) [58]. These widely expressed proteins were originally described as retroviral receptors, hence were named Glvr-1 and Ram-1 [59, 60], and were later shown to be Na⁺-Pi cotransporters [61]. According to their physiological function, they were renamed after their prokaryotic paralogs: PiT1 for Glvr-1 and PiT2 for Ram-1. The possible physiological roles of PiT1 and PiT2 have been recently reviewed [58, 62]. Of importance for the present article, it should be noted that PiT1 may have important roles in vascular and bone physiology [63–65], whereas PiT2 may play a role in Pi reabsorption [66].

Phosphate and skeletal tissues

During skeletal growth and bone remodeling, calcium and Pi are required for the formation of biological apatites. The rate at which mineralization occurs is dependent, in part, on the local availability of Pi and calcium [1, 67]. In the absence of these ions, or in case of abnormal absorption or reabsorption, mineralization is impaired, resulting in the formation of poorly mineralized bone that is characteristic of osteomalacia or rickets [68]. However, increasing evidence indicates that Pi is also required in events other than mineralization, proposing Pi itself as a signaling molecule capable of modulating multiple cellular functions in bone [69].

Pi modulates osteoblastic proliferation

Several studies have recently implicated Pi as an important molecule capable of modulating osteoblastic proliferation. Naviglio et al. [70] reported that Pi treatment of human osteosarcoma U2OS cells resulted in a decrease in both intracellular cAMP levels and adenylate cyclase activity, and in cell growth inhibition. The use of pharmacological agents that maintain a high level of cAMP, such as forskolin, prevented the cell growth inhibition in response to Pi. A large-scale quantitative proteomic investigation conducted in MC3T3-E1 cells, a model of preosteoblastic cells often used for studying osteoblast differentiation and activity [71], showed that a 24-h Pi treatment modulates the expression levels of a large set of proteins [69]. A pathway/function analysis of these proteins revealed wide-ranging effects from transcriptional regulators to signal transduction molecules. Several upregulated proteins in response to Pi were found to be involved in the regulation of cell cycle, proliferation and DNA synthesis [69]. Flow cytometric analysis of MC3T3-E1 cells showed that the number of cells in the S-phase increased in response to Pi treatment, suggesting an increase in the proliferation in response to Pi [69, 72, 73]. Moreover, Kanatani et al. [74] have recently shown that an increase in extracellular Pi from 2 to 4 mM, dose-dependently stimulated DNA synthesis in MC3T3-E1 cells. By using an Insulin growth factor-1 (IGF-1) antiserum or a neutralizing IGF-1 antibody, these investigators inhibited partially but significantly the Pi-induced proliferation. This suggested that Pi can modulate osteoblastic proliferation at least in part via IGF-1, which is known as one of the main growth factors involved in osteoblastic cell proliferation [75]. However, the interaction with the IGF-1 signaling pathway may not be restricted to Pi since calcium was also shown to stimulate osteoblastic proliferation in relation to IGF-1 status [76], although IGF-1-independent effects of calcium on DNA synthesis were also described [77].

Pi modulates osteoblastic differentiation and activity

Inorganic phosphate has also been shown to play a role in osteoblastic cell differentiation and activity. These are complex processes that involve the temporally-regulated expression of multiple markers. During the early osteoblastic differentiation, the expression level of alkaline phosphatase (ALP) increases. In the presence of organic phosphates, such as β-glycerophosphate, ALP contributes to the local production or liberation of phosphate, thus leading to an increased Pi concentration in the extracellular environment [67]. The potential role of Pi as a signaling molecule during the course of osteoblast differentiation was first suggested by Beck et al. [78]. They showed that a decrease in extracellular Pi formation due to the inhibition of ALP activity correlated with a decrease in osteopontin expression, a molecule involved in the regulation of mineralization [78]. Later on, they showed that osteopontin expression was strongly induced in direct response to increased Pi levels in the culture medium [79]. The authors speculated that the physiological significance of the regulation of osteopontin expression by Pi generated via ALP was to prevent the induction of an excessive mineralization before osteoblasts cease proliferation [79]. An elegant work from Yoshiko and colleagues illustrated the role of Pi in bone mineralization in vitro and in vivo through the identification of an osteoblast autonomous Pi regulatory system that regulates bone mineralization. They showed that Pi stimulated the expression of stanniocalcin 1, which itself increased PiT1 accumulation in both osteoblast cultures and in vivo, resulting in increased Pi uptake and increased mineralization [65]. Once mineralized, osteoblasts may become osteocytes or bording cells or die by apoptosis. The work of Adams et al. [80] has linked extracellular Pi to a decrease in the mitochondrial membrane permeability leading to the apoptosis of osteoblasts. It should be noted, however, that osteoblasts are much less sensitive to Pi-induced apoptosis than chondrocytes, and that calcium is required during this process [80, 81]. In a recent study conducted in the murine odontoblast-like cell line MO6-G3, we showed an up-regulation of PiT1 and PiT2 expressions by high extracellular Pi concentrations, which correlated with ERK1/2 phosphorylation and calcium/phosphate crystal formation [82]. Although more detailed studies are necessary, our observations support the idea that crystal formation more than Pi itself may be required for the increase in ERK phosphorylation and PiT1 and PiT2 expression [82]. In line with this work, we recently obtained similar data in mouse primary osteoblasts and MC3T3-E1 cells where the Pi stimulation of MGP expression via the ERK1/2 pathway [83] correlated with the presence of calcium in the medium and could be abolished by the use of crystal formation inhibitors (Khoshniat et al., unpublished observation).

Pi modulates bone resorption

Bone resorption is due to the resorbing activity of specialized cells called osteoclasts [84]. Osteoclast precursors come from hematopoietic cells and need to couple with osteoblasts in order to differentiate into mature resorbing osteoclasts. Thus, the receptor activator of nuclear factor kappa B (NF-κB) (RANK) ligand (RANKL) at the surface of osteoblasts interacts with RANK at the surface of osteoclasts and allows the differentiation into mature osteoclasts [85–88], while osteoprotegerin is a molecule that inhibits osteoclast differentiation by binding to RANKL and preventing the RANK–RANKL interaction [86, 89].

Inorganic phosphate depletion is well known to stimulate osteoclastic bone resorption [90–92] while increased Pi inhibits both osteoclast differentiation and the bone resorption process [93]. The effect of an increase in Pi concentration has recently been studied in a culture of bone marrow cells. These data show that high extracellular Pi concentrations inhibit both the osteoclast differentiation and the osteoclast bone resorbing activity [94]. Furthermore, it has been shown that Pi inhibits the differentiation of osteoclastic precursors by stimulating osteoprotegerin expression in osteoblastic cells, and that extracellular Pi inhibits osteoclastic activity at least in part by the direct induction of osteoclasts apoptosis [94]. These findings were confirmed and expanded by Mozar et al. in a recent study where they showed that increased extracellular Pi concentration inhibited both osteoclastic differentiation and bone resorption activity induced by RANKL. Pi was found to inhibit the RANKL-induced JNK and Akt activation, while RANKL-induced p38 and ERK 1/2 phosphorylation were not affected [95].

Pi modulates chondrocyte differentiation

Early studies suggested a possible role of Pi in modulating chondrocyte differentiation. Using an organ culture technique as a model, Bingham and Raisz [96] showed in 1974 that bone shaft growth and mineralization from rat fetuses increase with a Pi treatment from 1.5 to 4.5 mM. Such concentrations of Pi lead to crystal formation by precipitation. However, using a low-calcium medium, they showed that growth and mineralization were still impaired [96]. Along the same line of evidence, a study from Kakuta et al. supports the view that metabolic regulation of the Pi pool size may be a rate-limiting factor in the mineralization of cartilage. These investigators analyzed sections of epiphyseal growth cartilage which revealed that very low levels of Pi are present in pre-mineralized cartilage. At the mineralization front, however, a large increase in Pi is correlated with mineral formation together with a decrease in the concentration of low molecular weight compounds that could provide the initial source of Pi for the development of mineral [97]. In a more recent study, Wang et al. [98] examined the potential role of Pi and calcium in the regulatory pathways that mediate chondrogenesis and cartilage maturation, using the CFK2 cells as a chondrocytic model. They showed that the response to external Pi induced a rapid up-regulation of PiT1 protein expression at the plasma membrane, associated with an increase in Pi transport. However, this effect was temporally restricted to an early stage of CFK2 differentiation, suggesting that extracellular Pi plays a role in the commitment of chondrogenic cells to differentiation [98]. The effect of Pi on the maturation of chondrocytes was also shown by the suppression of type II collagen and PTH receptor expression, and acceleration of type X collagen expression in the ATDC5 cell line, a model of growth plate chondrocytes [99, 100]. Additional data obtained in osteoarthritic cartilage showed that IL-8 promotes hypertrophic terminal chondrocyte differentiation by modulating chondrocyte PiT1 expression and sodium-dependent Pi uptake [101]. These results demonstrate that Pi can play a role in the early and late differentiation of chondrocytes, by stimulating the expression of various chondrocytic markers. The exact molecular mechanisms involved in these effects have not yet been deciphered and need further investigation.

The terminal differentiation of growth plate chondrocytes involves a hypertrophic conversion associated with matrix mineralization and ultimately on their apoptosis [102]. During the calcification process, chondrocytes experience high levels of extracellular Pi [103]. Several studies have shown that exposing chondrocytes to a high level of Pi leads to their terminal maturation and subsequent matrix mineralization [100, 101, 104–107]. The role of Pi in these processes is highlighted by the mineralization defects that could be observed in the case of inherited hypophosphatemias [108]. Finally, the work of Mansfield et al. [106] has linked the elevation in environmental Pi concentration and the concomitant rise in intracellular Pi levels to rapid chondrocytic death through the apoptotic pathway. This effect of Pi is increased in the presence of calcium [109].

Phosphate and extraskeletal tissues

The ability of Pi to alter gene expression or cell function is not restricted to bone-forming cells. Regulation of osteopontin expression by Pi has also been demonstrated in NIH3T3 fibroblasts [79], cells that do not retain an ability to differentiate into osteoblast-like cells, showing that the regulation of osteopontin by phosphate is not restricted to cells in the osteoblast lineage.

Inorganic phosphate has been linked to the regulation of Runx2 and osteocalcin in human vascular smooth muscle cells [110, 111]. When cultured in media containing normal physiological levels of Pi (1.4 mM), human aortic smooth muscle cell (HSMC) grew in monolayers and did not mineralize. In contrast, HSMC cultured in media containing Pi levels comparable to those seen in hyperphosphatemic individuals (>1.4 mM) showed dose-dependent increases in cell culture calcium deposition. High Pi also enhanced the expression of osteocalcin and Runx2. This data suggest that high Pi may directly stimulate HSMC to undergo phenotypic changes that predispose to calcification and may help explain the phenomenon of human metastatic calcification under hyperphosphatemic conditions. The effects of high Pi were suggested to be mediated by PiT1. When HSMC were stably transduced with shRNA directed against PiT1, Pi uptake as well as Pi-induced mineralization and osteogenic differentiation markers Runx2 and osteopontin were reduced [112].

Low-external phosphate concentration has been shown to stimulate the SLC34 Na⁺-Pi cotransporter in the murine kidney [113, 114]. Kido et al. showed that feeding mice with a low-Pi diet for 4 days led to a twofold increase in Npt2a mRNA abundance. They identified and characterized a DNA sequence responsible for the Pi response, designated as a phosphate response element in the Npt2a gene. This response element shares a region with 9 of 10 bp identity to yeast phosphate-responsive transcription factor Pho4 binding element. At the center of this region is a CACGTG motif sufficient to confer the transactivation by dietary Pi deprivation. The authors have also identified the mouse transcription factor mE3 (TFE3), with structural features very similar to those of Pho4, that is up-regulated by low Pi diet. These results emphasize the role of Pi as a direct regulator of gene transcription.

The group of Cho has published a series of papers during the past 5 years describing the effect of high or low dietary Pi on the growth of diverse organs, including brain [115, 116], liver [117] and lungs [118–121]. Their studies clearly show that Pi alters the growth of these organs, most probably through the Akt/mTOR pathway, as well as the Raf/MEK/ERK pathway. However, the precise mechanisms are still to be clarified since both high and low Pi diets act similarly on both of these pathways. Moreover, the Pi transporters that may be involved in this process are not clearly identified since, for example, either Npt2a [118] or Npt2b [119–121] have been implicated in the same organ, and SLC20 transporters were not investigated.

Finally, changes in extracellular Pi (and intracellular Pi) were thought to be responsible for the increased survival of cultured fetal rat cortical neurons after an excitotoxic or oxidative insult [122]. Pi was also implicated as a triggering signal for PTH induction and secretion in parathyroid tissue in vivo and in vitro [123–125].

Phosphate as a signaling molecule: mechanistic issues

As the biological role of Pi as a signaling molecule is emerging in mammals, questions regarding its mechanism of action have arisen. One of the first issues to address is whether any known signaling pathways are activated and if there is the need for a nuclear response. Another critical aspect is whether or not Pi needs to be transported into the cell to fulfill its signaling role. A close examination of the available data reveals that answering this latter point is not straightforward. Moreover, in the event that Pi acts as an extracellular signal, the requirement of membrane proteins capable of mediating the effects of Pi within the cell has to be conceptualized.

What are the signaling pathways involved?

The specific intracellular signaling pathways activated by Pi are poorly elucidated. Beck et al. have found that Pi selectively activates the extracellular signal regulated kinase (ERK1/2) signaling pathway to mediate its effect on osteopontin expression in MC3T3 cells [83, 126]. Increased extracellular Pi concentration caused a rapid and biphasic phosphorylation of ERK1/2 but did not activate the other mitogen-activated protein kinase (MAPK) signaling proteins, p38 or the c-jun N-terminal kinase (JNK). Other pathways including PI3-kinase, protein kinase A, and protein kinase G were not involved. These observations are in agreement with a study demonstrating an increase in ERK1/2 phosphorylation, but not p38 or JNK, in response to bisphosphonates that is further induced by the addition of 3 mM Pi [127]. In a recent study, we obtained similar results and confirmed that Pi effect induces the specific activation of the ERK1/2 signaling pathway in osteoblasts [83]. We have shown that this pathway is required for the effect of Pi on the expression of matrix gla protein (MGP), a known inhibitor of calcification. We have also obtained similar results in ATDC5 and primary chondrocytes. Most importantly, we have shown that, as in osteoblasts, Pi stimulates the ERK1/2 pathway in chondrocytes to mediate its effects on MGP expression, suggesting a negative feedback loop for the control of mineralization induced by Pi in cartilage [128]. Although further investigation is needed to fully understand the mechanisms by which an increased intracellular Pi might regulate gene expression and ultimately cell function, this data collectively demonstrates that several specific signaling pathways are involved in the effects of Pi in the regulation of gene expression.

Are transcription factors involved?

Very few data are available on the transcription factors that might be involved in the gene regulation by extracellular Pi, and most of the studies have been conducted in bone-forming cells. In osteoblastic MC3T3-E1, chondrocytic ATDC5 and osteocytic MLO-Y4 cell lines, Runx2 was shown to be exported from the nucleus to the cytoplasm in response to extracellular Pi, but its expression was not regulated at the mRNA level by Pi treatment [129]. Accordingly, osteocalcin gene expression was repressed by the lack of Runx2 in the nucleus after Pi treatment [129], which appears contradictory to the physiological expression of osteocalcin occurring during MC3T3-E1 differentiation [73]. Runx2 was also suggested to mediate the effects of Pi in an in vitro model of vascular calcification [110]. In this study, treatment with high level Pi enhanced the expression of the osteoblastic differentiation markers osteocalcin and Runx2 by vascular smooth muscle cells (VSMCs). The ability of Pi to regulate gene transcription and cellular function represents a potentially novel extracellular signaling mechanism. Accordingly, Beck et al. [73] have identified a discrete set of genes increased by Pi in MC3T3-E1 cells including the transcription factor Nrf2 which was shown to be regulated at the transcriptional level. These authors further compared transcriptome and proteome datasets with the goal of identifying a number of candidate proteins that are post-transcriptionally regulated by elevated Pi [69]. From their data, Fra-1, a member of the AP-1 family of transcription factors, seems to be such a candidate as it is not regulated at the mRNA level but its protein level increases in response to Pi. Moreover, the use of inhibitors of transcription and translation did not affect the effects of Pi on Fra-1 expression, thus suggesting a post translational regulation of Fra-1 in response to Pi [69]. Accordingly, previous studies have shown that phosphorylation of Fra-1 by the MAPK ERK1/2 resulted in Fra-1 protection from proteosomal degradation [130] and subsequent post-translational stabilization [131]. Interestingly, in a recent report, we have observed that the regulation of MGP expression by Pi through the ERK1/2 signaling pathway involves the stimulation of the DNA binding activity of Fra-1 to specific sequences in the MGP promoter [83].

Does Pi have to enter the cell?

It is commonly believed that Pi must enter the cell to affect cell functions, hence involving Pi transporters at the plasma membrane. To investigate this issue, most studies have used phosphonoformic acid (PFA), a well-documented competitive inhibitor of Na⁺-dependent Pi transport mediated by SLC34 (Npt) Pi transporters [132, 133]. Limited data are available on the effect of PFA on SLC20 (PiT) proteins. Bai et al. [134] expressed mouse PiT2 in Xenopus oocytes and documented a 40% decrease in Pi uptake by using a 50-fold excess of PFA (5 mM). However, when used at concentrations which inhibit renal SLC34 transporters, Ravera et al. [135] found no effect of PFA on Pi uptake in any of the PiT isoforms studied, confirming previous observations of Villa-Bellosta et al. [136]. Using Pi uptake and electrophysiology on PiT mRNA-injected Xenopus oocytes, Ravera et al. [135] unequivocally showed that PFA, in contrast to its action as a competitive inhibitor of SLC34 Pi transporters, does not inhibit Pi transport mediated by PiT. Confirming these data, a recent study from Villa-Bellosta et al. [137] showed, by using Pi uptake assays, that PFA is a very poor inhibitor of Pi uptake mediated by rat PiT1 and PiT2 both in oocytes and in native rat VSMCs. The authors conclude that, in contrast to its action as a competitive inhibitor of SLC34 Na⁺-Pi cotransporters, PFA does not inhibit Pi transport mediated by PiT. PiT transporters, which prefers H₂PO₄ ⁻ as a substrate, and SLC34, which prefers HPO₄ ²⁻, transport most efficiently in the acidic and alkaline ranges, respectively [62]. This difference in preferred Pi species may explain why PFA inhibits Pi transport by SLC34, but not by SLC20.

It has been reported that PFA blocks Pi-induced calcification in smooth muscle cells [110, 112] as well as matrix calcification in osteoblast-like cells [138] and osteoblast subcultures [65]. As PFA does not block PiT1-mediated Pi transport, the effect of PFA in inhibiting calcification is probably more related to the ability of phosphonates and bisphosphonates (with which PFA shares a similar structure [139]) to inhibit the formation of calcium phosphate crystals, than to inhibition of Pi transport through PiT1 [137]. The studies aimed at illustrating the role of extracellular Pi in signaling processes were conducted in fibroblasts, osteoblasts, chondrocytes and VSMCs in which only SLC20 Pi transporters (and not SLC34) were expressed. Most of these studies used PFA as an inhibitor of Pi transport and drew conclusions based on this property that now need to be re-considered in view of the lack of PFA effect on PiT-mediated Pi transport. As a consequence, and based on the actual available literature, it is not possible to get a clear answer as to whether Pi needs to enter the cells to mediate its signaling effect, and whether PiT proteins are implicated in this process. We and others have recently generated mice carrying conditional and null alleles of the Pi transporter PiT1 [140, 141]. Since there is no specific PiT1 inhibitor, such animal models will be invaluable tools to investigate the role of SLC20 Pi transporters in mediating the effects of Pi on cell functions.

Phosphate sensing in mammals

Although long-term Pi homeostasis is controlled by a variety of hormones, peptides and small molecules that regulate the efficiency of Pi absorption in the intestine and Pi excretion in the renal tubule, recent data suggest the existence of a Pi-sensing mechanism on the surface of cells that is able to detect changes in Pi concentration [142].

In vitro, renal and intestinal cells are able to respond to changes in Pi concentration in the extracellular medium. A low-Pi medium increased Pi uptake, whereas a high-Pi medium decreased Pi uptake in renal [143] and intestinal cells [144, 145]. Osteoblast and bone-marrow stromal cells respond to changes in external Pi concentrations by altering the expression, localization or secretion of growth factors, transcription factors or enzymes [129, 144, 146, 147].

In vivo, a chronic elevation of Pi concentration in the serum, which can result from genetic syndromes or renal insufficiency, is associated with severe defects in humans, such as vascular calcification and secondary hyperparathyroidism [63, 148]. Genetic disorders due to mutations in GALNT3 [149], FGF23 [150] or Klotho [151] genes cause familial tumoral calcinosis, a rare autosomal recessive disorder manifesting with severe hyperphosphatemia and widespread ectopic calcifications [152]. Likewise, targeted deletion of either FGF23 or Klotho in mice leads to hyperphosphatemia and vascular calcification [153–155]. Hyperphosphatemia and vascular calcification are also prevalent in patients with end-stage renal disease (ESRD) [148]. Elevated serum Pi concentration (greater than 2.1 mM) was shown to be positively correlated with mortality in ESRD patients [156], and even small elevations in Pi concentration in the high-normal range (1.1–1.45 mM) have been correlated with increased risk of cardiovascular mortality in ESRD patients and in the general population [157, 158]. A recent review of the literature shows that there is a graded independent association between serum Pi levels and mortality, mainly cardiovascular events, and the progression of renal disease [20]. Lowering serum Pi levels with a non-calcium-containing Pi binder slows the progression of vascular calcification in ESRD patients [159, 160].

The hyperphosphatemia seen in ESRD patients will also eventually lead to secondary hyperparathyroidism with attendant increased PTH synthesis, secretion, and parathyroid cell proliferation. Secondary hyperparathyroidism can be controlled by a Pi-restricted diet, as was shown in dogs [161, 162]. Importantly, these studies showed that dietary Pi restriction corrected the secondary hyperparathyroidism independently from changes in serum calcium and 1α,25(OH)₂D₃ levels. In addition, in vitro studies have shown that the effect of serum Pi on the parathyroid was direct. Although high Pi concentration stimulated PTH secretion, this direct effect required the maintenance of tissue architecture [123, 124, 163]. The work of Silver and colleagues suggests that the effect of Pi on PTH gene expression is post-transcriptional and involves the regulated interaction of parathyroid cytosolic proteins to a defined cis-acting sequence in the PTH mRNA [164]. Two of the PTH mRNA-binding proteins, acting as stabilizing or degrading factors, were identified as AU-rich binding factor (AUF1) and upstream of N-ras (Unr). Their regulated binding to the PTH cis-acting element located on its 3′UTR determine the PTH mRNA half-life [164].

The existence of a Pi-sensing mechanism in the parathyroid gland was suggested in the experiments of Martin et al. [165]. These investigators showed that, when uremic rats were administered a low-Pi diet by force-feeding, this induced a decrease in PTH and serum Pi concentration within 15 min. The infusion of Pi into the duodenum or the circulation led to a rapid increase in PTH, without change in Pi or calcium concentration, suggesting that the parathyroid gland is capable of sensing Pi in the absence of changes of serum calcium. In addition, Berndt et al. [166] recently demonstrated that Pi administration into the duodenum of intact or parathyroidectomized rats was associated with very rapid change in the fractional excretion of phosphorus in the kidney, with no change in PTH, FGF-23 or s-FRP-4 concentration. Thus, the intestine is able to sense an increase in luminal Pi concentration and signal to the kidney [166]. Enteric-renal signaling mechanisms also exist in the case of other ions, including sodium [167, 168], potassium [169] and calcium [170].

The molecular nature of the Pi sensor in mammals still remains to be identified. Bacteria and yeasts both have well-developed Pi sensing mechanisms [171–173]. In Escherichia coli, the phosphate transporter PstS and other periplasmic proteins (PstC, Pst and PstB/PhoU) detect low external Pi concentration and increase the efficiency of Pi retention in the bacteria [172]. In Saccharomyces cerevisiae, a low external Pi concentration inactivates the Pho80–Pho85 complex by the cyclin-dependent kinase inhibitor Pho81, resulting in the activation of the transcription factor Pho4 and subsequent induction of the Pi transporter Pho84, along with other PHO genes [173]. Recent data now identify Pho84 as the essential component of the Pi-sensing system, mediating the rapid activation of PKA following Pi restriction [174]. Pho84 was also identified as a Pi transceptor, by mediating PKA activation following Pi triggering, without transporting Pi [175]. Although the response to Pi variations in the environment has been well characterized in unicellular organisms, no equivalent Pi-sensing system has been found to date in mammals. In mammals, extracellular Pi regulates the activity of specific Na⁺-dependent Pi transporters and modulates the cell cycle, as in unicellular organisms [176–178]. In parathyroid glands, PTH secretion is regulated by extracellular Pi [123, 124]. This effect of Pi can be mediated by specific molecules in the parathyroid cell membrane or by metabolic signals associated with increased cellular Pi concentrations. One such mediator may be a Na⁺-Pi cotransporter. The work of Miyamoto and colleagues identified PiT1 as being the only Na⁺-Pi cotransporter present in the parathyroids and being regulated by changes in extracellular Pi concentrations. As such, they proposed that PiT1 could play a role in sensing Pi in parathyroid cells, although formal demonstration is still lacking [179–181]. On the other hand, Salaün et al. proposed that PiT2 could serve as a Pi sensor. They showed that PiT2 transport-deficient mutants were still able to “sense” Pi by retaining their oligomerization properties in response to high extracellular Pi concentrations [182]. It is not known whether a Pi-sensing mechanism in mammals would involve a local cell-to-cell regulation or a central or hormonal control. Since PiT1 and PiT2 are expressed ubiquitously, their putative role in sensing extracellular Pi may relate to a cell-to-cell-based regulation, enabling a fine tuning depending on individual Pi cellular needs. Of course, the involvement of PiT transporters and their mode of action remains entirely a hypothesis, and physiological evidence is lacking.

Conclusion

Considerable progress has been made over the past decade in our understanding of the mechanisms of Pi homeostasis and skeleton mineralization. Still, very little is known about the initial events involving the detection of a change in the Pi serum concentration and the subsequent appropriate regulation cascade. Future studies should be directed at uncovering the nature and identity of the Pi-sensing mechanism. This knowledge is essential to the proper understanding of the global regulation of Pi homeostasis and would most certainly change the current concepts of this field. It is now clear that Pi represents a signal molecule capable of modulating the expression or regulation of multiple factors necessary for diverse biological processes. Increased knowledge of Pi biology and mode of action is essential for the development of therapeutic strategies that may prevent the consequences of hyperphosphataemia in patients with chronic renal failure.

Abbreviations

1α,25(OH)₂D₃

1α,25-dihydroxyvitamin D

ALP

Alkaline phosphatase

Pi

Inorganic phosphate

IGF-1

Insulin growth factor-1

ESRD

End-stage renal disease

FGF-23

Fibroblast growth factor 23

HSMC

Human aortic smooth muscle cell

MGP

Matrix gla protein

PTH

Parathyroid hormone

PFA

Phosphonoformic acid

RANKL

Receptor activator of nuclear factor kappa B (NF-κB) (RANK) ligand

VSMC

Vascular smooth muscle cell