Biology of Fibroblast Growth Factor 23: From Physiology to Pathology

Abstract

Fibroblast growth factor (FGF)23 is a phosphaturic hormone produced by osteocytes and osteoblasts that binds to FGF receptors in the presence of the transmembrane protein αKlotho. FGF23 mainly targets the renal proximal tubule to inhibit calcitriol production and the expression of the sodium/phosphate cotransporters NaPi2a and NaPi2c, thus inhibiting renal phosphate reabsorption. FGF23 also acts on the parathyroid glands to inhibit parathyroid hormone synthesis and secretion. FGF23 regulation involves many systemic and local factors, among them calcitriol, phosphate, and parathyroid hormone. Increased FGF23 is primarily observed in rare acquired or genetic disorders, but chronic kidney disease is associated with a reactional increase in FGF23 to combat hyperphosphatemia. However, high FGF23 levels induce left ventricular hypertrophy (LVH) and are associated with an increased risk of mortality. In this review, we describe FGF23 physiology and the pathological consequences of high or low FGF23 levels.

The study of genetic and acquired hypophosphatemic diseases led to the discovery of the phosphaturic hormone fibroblast growth factor (FGF)23 (Shimada et al. 2001). FGF23 is produced in bone and acts on renal proximal tubules to inhibit phosphate reabsorption and calcitriol production. FGF23 binds to FGF receptors (FGFRs) whose specificity depends on the presence of the transmembrane protein αKlotho (Urakawa et al. 2006). FGF23 primarily acts up on organs involved in phosphocalcic metabolism, but does also affect other organs (Faul et al. 2011). In this review, we describe the physiology of FGF23 and the pathological conditions linked to abnormal FGF23 levels.

FGF23 BIOLOGY IN PHYSIOLOGY

Protein Structure

Human FGF23, a 32-kDa glycoprotein, is encoded by a gene consisting of 9386 nucleotides on chromosome 12p13 with three exons and two introns (Goetz et al. 2007; Bhattacharyya et al. 2012). FGF23 is mainly produced by osteocytes and osteoblasts. However, low levels of FGF23 messenger RNA (mRNA) have been detected in the brain, thymus, spleen, small intestine, heart, and testis. The physiological role of FGF23 produced in these organs is undetermined (Yoshiko et al. 2007). FGF23 production has also been noted in response to trauma in liver (Prie et al. 2013), kidney (Zanchi et al. 2013), and heart (Andrukhova et al. 2015).

Intact FGF23 (iFGF23) is cleaved by an unknown protease at 176RXXR179S180 (“R”: arginine; “S”: serine), (Shimada et al. 2002) into inactive 18-kDa amino-terminal and 12-kDa carboxy-terminal fragments (Fig. 1). During maturation, the protein is O-glycosylated by GALNT3 (UDP-N-acetyl-α-d-galactosamine: polypeptide N-acetylgalactosaminyl transferase 3) at several sites, especially Thr178 (Kato et al. 2006). Glycosylation protects FGF23 from proteolysis and facilitates its secretion. The posttranslational regulation of FGF23 also involves a phosphorylation process. The kinase family with sequence similarity 20-member C (FAM20C) phosphorylates FGF23 at Ser180. This inhibits FGF23 glycosylation by GALNT3, thus promoting its proteolytic cleavage (Tagliabracci et al. 2014).

Figure 1.

The fibroblast growth factor (FGF)23 protein. FGF23 is a 32-kDa glycoprotein containing 251 amino acid (aa) residues including a signal sequence (24 aa), an amino-terminal FGF homology domain (155 aa), and a specific carboxy-terminal sequence (72 aa). After cleavage of the signal sequence, the intact FGF23 (iFGF23) molecule can be secreted or cleaved by a yet-unknown protease between aa 179 and 180 into inactive amino-terminal and carboxy-terminal fragments. The cleavage site (red) is located as follows: 176 RXXR 179/S180 (“R”: arginine; “S”: serine). Two types of enzyme-linked immunosorbent assays (ELISAs) are commercially available to detect FGF23. iFGF23-specific assays use antibodies recognizing two epitopes flanking the proteolytic cleavage site (green). The carboxy-terminal FGF23 assays detect both iFGF23 and its carboxy-terminal fragments by recognizing two epitopes in the carboxyl terminus (orange). The units differ between the two types: pg/ml for iFGF23 and RU/mL for carboxy-terminal FGF23 (cFGF23).

The plasma half-life of iFGF23 is probably between 20 and 60 min as suggested by the rapid normalization of levels after excising a tumor in tumor-induced osteomalacia (TIO) (Takeuchi et al. 2004; Khosravi et al. 2007).

Two types of enzyme-linked immunosorbent assay (ELISA) kits are available to measure FGF23 (Fig. 1) (Wolf and White 2014). Neutralizing antibodies showed that the FGF23 carboxyl terminus binds αKlotho and the amino terminus binds to the FGFR (Yamazaki et al. 2008).

Actions of FGF23

FGF23 Receptor

The FGF23 receptor is a homodimer of tyrosine kinase FGFR (types 1c, 3c, or 4) and the αKlotho coreceptor. FGFRs are ubiquitously expressed so coexpression with αKlotho provides cell specificity for FGF23 actions (Urakawa et al. 2006). Figure 2 illustrates FGF23 signaling (Beenken and Mohammadi 2012).

Figure 2.

Fibroblast growth factor (FGF)23/FGFR1/αKlotho signaling. The FGF23 receptor consists of a homodimer of tyrosine kinase FGF receptors (FGFRs) (types 1c, 3c, or 4) and the αKlotho coreceptor. αKlotho contains two extracellular domains: a transmembrane domain and an intracytoplasmic domain. Homodimerization and FGF23 binding results in transphosphorylation of the tyrosine kinase domains and initiates the cascade of intracellular signaling. FGFs initiate signaling through many pathways, including phospholipase C (PLC)-γ–protein kinase C (PKC), CRKL, Ras-mitogen-activated protein kinase (MAPK), and PI3-Akt. However, the most common pathway for FGF23 is the MAPK pathway type ERK1/2 (extracellular signal-regulated kinases 1 and 2), which is activated by phosphorylation of FGFR substrate-2α (FRS2α), constitutively associated with the juxtamembrane domain of the FGFR.

Kidney, parathyroid glands, osteoblasts, and osteocytes express FGFR and αKlotho and are targeted to regulate mineral ion homeostasis and FGF23 itself. The specificity of the FGFR/αKlotho homodimer explains why αKlotho^−/− mice show high circulating FGF23, likely caused by resistance, and yet are phenotypically similar to Fgf23^−/− mice with severe growth retardation, accelerated aging, vascular and ectopic calcifications, bone mineralization defects, increased serum phosphate, and increased calcitriol levels (Kuro-o et al. 1997). Deletion of Fgf23 from αKlotho^−/− mice does not worsen the phenotype, confirming that FGF23 requires αKlotho (Nakatani et al. 2009).

FGF23 Actions on the Kidney

FGF23 actions on the proximal tubule

FGF23 primarily acts on the renal proximal tubule to inhibit renal phosphate reabsorption and calcitriol production.

Approximately 85% of phosphate (Pi) filtered at the glomerulus is reabsorbed in the proximal tubule through a transcellular pathway involving sodium/phosphate cotransporters NaPi2a, NaPi2c, and PiT2. Fgf23^−/− mice display hyperphosphatemia caused by increased renal reabsorption in the presence of elevated NaPi2a expression in the proximal tubule, and increased serum calcitriol levels (Shimada et al. 2004b). Conversely, mice overexpressing human FGF23 displayed hypophosphatemia owing to reduced expression of NaPi2a in renal proximal tubules and increased urinary phosphate excretion, low serum calcitriol levels, and rachitic bone (Bai et al. 2004; Larsson et al. 2004; Shimada et al. 2004c). Thus, FGF23 is an important regulator of renal Pi handling.

FGF23 also inhibits expression of NaPi2c and PiT2 in proximal tubular cells (Tomoe et al. 2010) and decreases NaPi2a and NaPi2c mRNA levels (Segawa et al. 2003). However, at least in mice, the phosphaturic action of FGF23 is mainly determined by down-regulation of apical membrane NaPi2a.

The Na⁺/H⁺ exchange regulatory factor 1 (NHERF1) is a scaffolding protein with two tandem PDZ domains. PDZ1 interacts with NaPi2a to stabilize it on the apical membrane. FGF23 signaling phosphorylates NHERF1 via ERK1/2 and serum glucocorticoid-regulated kinase 1 (SGK1) (Gattineni et al. 2009; Weinman et al. 2011; Andrukhova et al. 2012). This disrupts NHERF1 binding to NaPi2a, leading to its endocytosis and lysosomal degradation, ultimately decreasing renal Pi reabsorption. How FGF23 inhibits NaPi2c and PiT2 expression is undetermined.

FGF23 fragments are thought to be biologically inactive, but the carboxy-terminal could compete with iFGF23 for the Klotho/FGFR complex, hence antagonizing the phosphaturic activity of FGF23 in vivo as observed in healthy rats and in a mouse model of phosphate wasting disorders (Goetz et al. 2010).

FGF23 decreases the conversion of 25-hydroxy vitamin D into its active form 1,25-dihydroxy vitamin D in proximal tubular cells by reducing 25-hydroxyvitamin D-1α-hydroxylase mRNA, and increases the catabolism of 1,25-dihydroxy vitamin D into 1,24,25-trihydroxy vitamin D by increasing 25-hydroxyvitamin D-24-hydroxylase mRNA (Shimada et al. 2004a). These two effects lead to a decrease in circulating calcitriol levels and intestinal absorption of phosphate and calcium. Noteworthy, ablating the vitamin D activation pathway reverses the skeletal and biological anomalies of Fgf23^−/− mice (Sitara et al. 2006), showing that the phenotype is a vitamin D–mediated process (Razzaque et al. 2006).

This dual action of FGF23 on the proximal tubule—inhibition of Pi reabsorption and calcitriol production—decreases serum phosphate (Fig. 3).

Figure 3.

Main actions and regulatory pathways of fibroblast growth factor (FGF)23. FGF23 acts on the proximal tubule to inhibit sodium/phosphate cotransporters NaPi2a and NaPi2c and, consequently, renal phosphate reabsorption. FGF23 also inhibits renal 1 α-hydroxylase and stimulates renal 24-hydroxylase leading to a decrease in serum calcitriol with a subsequent decrease in intestinal phosphate (and calcium) absorption resulting in a decrease in serum phosphate. In turn, FGF23 is stimulated by serum calcitriol, an increase in serum phosphate, and an increase in phosphate intake. FGF23 also inhibits parathyroid hormone (PTH) synthesis and secretion, whereas PTH stimulates FGF23 production.

αKlotho is highly expressed in the distal tubule (Farrow et al. 2009), but not in the main target of FGF23, the proximal tubule (Andrukhova et al. 2012). This puzzling circumstance may occur as a paracrine action instigated by FGF23 and αKlotho in the distal tubule. This hypothesis is supported by findings that deleting αKlotho in mouse distal tubules leads to high serum phosphate levels, elevated FGF23 levels and increased expression of NaPi2a in proximal tubular cells (Olauson et al. 2012). A second hypothesis is that the limited αKlotho expression in proximal tubular cells may be sufficient to initiate intracellular signaling. These two mechanisms are not mutually exclusive. Mice with αKlotho ablated from the proximal tubules have impaired urinary phosphate excretion and increased NaPi2a in the brush border membrane but only mild hyperphosphatemia, indicating the existence of compensatory mechanisms (Ide et al. 2016).

FGF23 actions on the distal tubule

Experiments with mice indicate that FGF23 stimulates the transient receptor potential cation channel subfamily V member 5 (TRPV5) expressed on the apical membrane of proximal tubule cells via with-no-lysine kinase (WNK)4 activation to increase calcium reabsorption (Andrukhova et al. 2014b). In young healthy girls, iFGF23 positively correlated with serum calcium and negatively correlated with urinary calcium/creatinine ratio, suggesting that this also happens in humans (Mitchell et al. 2017). FGF23 is therefore also a calcium-conserving hormone.

Moreover, Fgf23^−/− and αKlotho^−/− mice express less membrane sodium chloride cotransporter Na-Cl cotransporter (NCC) in renal distal tubules, leading to renal sodium wasting, reduced plasma volume, and lower blood pressure despite elevated aldosterone secretion. FGF23 injection into normal mice results in up-regulation of renal NCC expression via WNK4 signaling, leading to increased renal sodium reabsorption, plasma expansion, hypertension, and heart hypertrophy (Andrukhova et al. 2014a). These data suggest that FGF23 acts also as a sodium-preserving hormone. Noteworthy, human data (volume expansion in patients with arterial hypertension) do not support a direct feedback loop between volume status and FGF23 (Humalda et al. 2016).

FGF23 Actions on the Parathyroid Glands

Animal and in vitro data have shown that FGF23 directly acts on the parathyroid glands to decrease parathyroid hormone (PTH) synthesis and secretion (Ben-Dov et al. 2007; Krajisnik et al. 2007). FGF23 decreased PTH production in rat glands and increased expression of the calcium-sensing and vitamin D receptors, pathways that inhibit PTH, and reduced parathyroid cell proliferation (Canalejo et al. 2010). However, there are conflicting results regarding proliferation and FGF23, as another study reported that FGF23 induced parathyroid cell proliferation (Kawakami et al. 2017). In cultured bovine parathyroid cells, FGF23 decreased Pth mRNA levels but increased 1α-hydroxylase expression, the opposite of proximal tubule calcitriol effects (Krajisnik et al. 2007). This increase in the local calcitriol production could partially explain the inhibition of PTH synthesis. Moreover, a novel, Klotho-independent, calcineurin-mediated FGF23 signaling pathway has been identified in parathyroid glands that suppresses PTH (Olauson et al. 2013).

However, the FGF23 inhibitory action is less clear in clinical conditions in which increases in FGF23 occur without a concomitant decrease in PTH, as in TIO or X-linked hypophosphatemic rickets (Blau and Collins 2015). Importantly, failure of very high-circulating FGF23 levels in chronic kidney disease (CKD) to decrease PTH has been attributed to a decrease in FGFR1 and Klotho in the parathyroid glands (Komaba et al. 2010).

Potential Direct Actions of FGF23 on Bone

Whether FGF23 acts directly on bone is a difficult question to address because in vivo modifications of FGF23 levels also disrupt serum phosphate and calcitriol levels and any action of FGF23 on bone mineralization remains to be clarified.

Fgf23^−/− mice have high serum phosphate, calcium, calcitriol levels, and ectopic calcifications, but show a paradoxically severe defect in skeletal mineralization. Fgf23/NaPi2a double knockout mice had completely normal serum phosphate levels compared with hyperphosphatemic Fgf23^−/− mice. However, these mice showed the same defects in bone mineralization, suggesting a direct action of FGF23 on bone independent of serum phosphate levels (Sitara et al. 2008). Importantly, it is suggested that increased osteopontin expression in bone is a pathogenic factor mediating the mineralization defect and alterations in bone metabolism observed in Fgf23^−/− mice (Yuan et al. 2014).

This accords with a clinical study associating high levels of circulating FGF23 and improved indices of skeletal mineralization in pediatric patients undergoing peritoneal dialysis (Wesseling-Perry et al. 2009). Conversely, in vitro experiments show that FGF23 inhibits matrix mineralization independent of its systemic effect on phosphate metabolism (Wang et al. 2008) and osteoblast differentiation while having a biphasic effect on osteoclasts (inhibition of osteoclast formation and stimulation of osteoclast activity) (Allard et al. 2015).

Regulation of FGF23

Systemic Regulation of FGF23

The major systemic regulators of FGF23 levels are calcitriol, PTH, and dietary phosphate (Fig. 3), although how phosphate and PTH affect FGF23 production remains poorly understood.

Calcitriol

Calcitriol directly stimulates Fgf23 transcriptional expression in osteoblasts and osteocytes by binding to vitamin D response elements (VDREs) in the Fgf23 gene (Kolek et al. 2005; Liu et al. 2006). Early unsuccessful searches for VDREs in the human Fgf23 promoter were limited by the assumption that transactivation elements would be within 3–5 kb of the promoter. However, nine potential VDREs have been identified in the genomic interval containing Fgf23. These VDREs are located quite distant to the transcription start site, but six are capable of mediating direct transcriptional activation of a reporter gene when bound by a calcitriol-liganded VDR complex (Saini et al. 2013).

Calcitriol may stimulate FGF23 by indirect mechanisms. In vitro studies with osteocyte-like cells show that calcitriol reduced Dmp1 mRNA, an inhibitor of FGF23 expression, through a VDR-dependent pathway (Nociti et al. 2014).

Phosphate

In animal (Perwad et al. 2005; Saito et al. 2005) and human studies (Ferrari et al. 2005; Antoniucci et al. 2006; Burnett et al. 2006), high-phosphate diets stimulated FGF23 secretion. Increased phosphate supplementation correlated with increases in serum phosphate and FGF23 levels. Similarly, decreased dietary phosphate results in decreased circulating FGF23 levels in mice and healthy men and women (Yu et al. 2005; Burnett et al. 2006).

However, intravenous phosphate injection did not affect FGF23 levels in humans (Ito et al. 2007), questioning whether phosphate directly regulates FGF23. Acute parenteral or enteral phosphate loads in humans with normal renal and parathyroid functions increased FGF23 10 h after the first significant increase in phosphate levels and distinctly later than the first significant increase in PTH or phosphaturia (Scanni et al. 2014). This suggests that serum phosphate acts directly and less rapidly than PTH to stimulate FGF23 secretion or indirectly via PTH.

In vitro data yield conflicting results for direct phosphate regulation of FGF23. Phosphate did not modify Fgf23 mRNA levels in rat osteosarcoma (ROS)17/2.8 and UMR-106 osteoblast cell lines (Liu et al. 2006). However, experiments with UMR-106 cells indicate that phosphate directly stimulates Fgf23 transcription by stimulating nicotinamide adenine dinucleotide phosphate (NADPH)-induced reactive oxygen species production and the MEK-ERK pathway (Hori et al. 2016). Note that a phosphate-sensing mechanism in FGF23-secreting cells has not been identified.

PTH

PTH treatment reportedly increases serum FGF23 in animals (Lavi-Moshayoff et al. 2010; Lopez et al. 2011; Rhee et al. 2011; Fan et al. 2016), but some reports indicate the opposite (Liu et al. 2006; Saji et al. 2009; Samadfam et al. 2009).

PTH infusion in humans increases FGF23 over 18 h (Burnett-Bowie et al. 2009). Other studies show an acute decrease within 6 h (Gutierrez et al. 2012). The conflicting results from animal models and humans regarding regulation of FGF23 by PTH could be caused by confounding factors such as serum phosphate or calcitriol levels. Stimuli from the bone-formation rate are also possible. This hypothesis is supported by intermittent administration of PTH (1–34) in mice resulting in an increase in bone formation and reduced circulating FGF23 levels. Continuous administration of PTH results in a net increase in bone resorption and increased FGF23 levels (Samadfam et al. 2009).

In vitro models are essential to understand the effects of PTH. Administering PTH to rat osteoblast-like UMR106 cells led to increased FGF23 expression (Lavi-Moshayoff et al. 2010). PTH, PTH-related protein, or a cyclic adenosine monophosphate (cAMP)-stable analog increase Fgf23 transcripts in a time- and dose-dependent manner in calvarial osteocyte cell cultures (Rhee et al. 2011). PTH increases Nurr1 (nuclear orphan receptor, nuclear receptor-associated protein1) mRNA levels by activating the protein kinase A (PKA) pathway before elevation of FGF23 mRNA levels in UMR-106 cells and in vivo (Meir et al. 2014; Fan et al. 2016). Nurr1 binds to potential Nurr1-binding sites in the FGF23 promoter to induce transcription. Administering calcimimetic R568 to rats with experimental CKD decreased PTH expression, calvaria Nurr1 mRNA and protein levels, and Fgf23 mRNA.

Calcium

Pth^−/− and Pth/CaSr double knockout mice experiments show that rising serum calcium increased serum FGF23 by a PTH- and CaSR-independant mechanism. Calcium-mediated increases in circulating FGF23 required a threshold level of serum phosphorus of ∼5 mg/dL, whereas phosphorus-mediated increases were markedly inhibited if serum calcium was below 8 mg/dL (Quinn et al. 2013). Increasing dietary calcium in parathyroidectomized rats increased serum calcium and FGF23, decreased calcitriol, but did not affect phosphorus (Rodriguez-Ortiz et al. 2012). A high-calcium diet also increased serum FGF23 in Cyp27b1^−/− mice in the absence of calcitriol and in Gcm2^−/− mice with low PTH levels (David et al. 2013).

Conversely, rats on a low-calcium and vitamin D diet showed hypocalcemia associated with low FGF23 despite high PTH and calcitriol levels (Rodriguez-Ortiz et al. 2012). Another genetically modified mouse study suggests that hypocalcemia leads to a decrease in FGF23 production (David et al. 2013).

These data indicate a positive correlation between serum calcium levels and FGF23 production. Treatment of MC3T3-E1 cells with calcium stimulated Fgf23 expression (David et al. 2013). Moreover, in vitro data confirm these findings, as Fgf23 promoter activity in cultured osteoblasts was inhibited by the L-calcium-channel inhibitor nifedipine and stimulated by calcium ionophores (David et al. 2013).

Circulating αKlotho

αKlotho has a circulating endoproteolytic cleavage product of the membranous form. Circulating αKlotho (cKl), produced in the kidney (Olauson et al. 2017), inhibits renal phosphate reabsorption, likely with a β-glucuronidase-mediated proteolytic degradation, and reduced surface expression of NaPi2a in the proximal tubule (Hu et al. 2010), while increasing TRPV5 activity and calcium reabsorption in the distal tubule (Chang et al. 2005). The phosphaturic effect of cKl occurs in Fgf23^−/− mice, suggesting direct action by cKl distinct from αKlotho. However, cKl also modulates renal phosphate reabsorption by inducing FGF23 synthesis in an FGFR-dependent manner (Smith et al. 2012).

Hypoxia, iron deficiency, and inflammatory cytokines

The role of iron in FGF23 regulation was discovered with the observation that iron deficiency triggers the onset of autosomal dominant hypophosphatemic rickets (ADHR), a genetic disorder caused by a mutation in the proteolytic cleavage site of FGF23. This was confirmed in mice fed control or iron-deficient diets. Wild-type and mice with an ADHR phenotype on low-iron diets had significantly elevated bone Fgf23 mRNA. However, ADHR mice, in which FGF23 is resistant to cleavage, had increased iFGF23 and carboxy-terminal FGF23 (cFGF23), leading to hypophosphatemic osteomalacia. Wild-type mice had elevated cFGF23, but not iFGF23, and thus no FGF23-induced hypophosphatemia (Farrow et al. 2011). This direct action of iron was confirmed in in vitro experiments in which iron chelation with UMR-106 cells significantly increased Fgf23 mRNA. The increase was the result of the stabilization of hypoxia-inducible factor 1α (HIF-1α) protein, a key metabolic sensor that regulates genes in response to iron deficiency. This work was the first to identify FGF23 production and cleavage as two distinct levels of FGF23 regulation in which increased FGF23 production is balanced by an increase in FGF23 cleavage to ensure normal phosphate homeostasis. Accordingly, cFGF23, but not iFGF23, plasma concentrations increased in women with iron-deficiency anemia (Wolf et al. 2013). Intravenous administration of saccharated ferric oxide and iron polymaltose induced a decrease in cFGF23 levels by restoring normal iron stores while transiently increasing iFGF23 levels associated with urinary phosphate wasting and hypophosphatemia (Wolf et al. 2013; Wolf and White 2014). However, intravenous iron dextran or oral iron preparations did not cause these effects. Therefore, specific iron formulations may reduce FGF23 cleavage more than production, causing transient hypophosphatemia (Wolf and White 2014).

Hypoxia also stimulates FGF23 production independently of iron deficiency. Fgf23 expression increased in UMR-106 cells cultured under hypoxia. Rats housed under hypobaric atmosphere experienced increased cFGF23 and stable iFGF23 plasma concentrations (Clinkenbeard et al. 2014). Moreover, a consensus sequence for HIF-1α, a stimulator of Fgf23 expression, was identified within the proximal promoter of Fgf23 (Zhang et al. 2016). Increasing cFGF23 secondary to hypoxia accords with the finding that patients with sickle cell disease anemia, which is a state of tissue hypoxia, have normal iFGF23 and increased cFGF23 levels (Courbebaisse et al. 2017).

Chronic inflammation in mice mimics chronic iron deficiency with high circulating cFGF23 but normal iFGF23 (David et al. 2016). Inflammation is associated with functional iron deficiency, so in vitro experiments investigated the role of inflammation on FGF23 production. In various bone cell lines, proinflammatory stimuli (tumor necrosis factor [TNF], interleukin [IL]-1β, TWEAK, and bacterial lipopolysaccharides [LPS]) dose-dependently up-regulated FGF23 (Ito et al. 2015; David et al. 2016), probably through HIF-1α signaling (David et al. 2016).

Adipokines

Several lines of evidence suggest that circulating adipokines from adipose tissue may modulate FGF23 production (Wagner et al. 2017).

Lean adipose tissue secretes adiponectin in correlation to total fat mass, whereas expanded adipose tissue generates leptin and inflammatory cytokines such as tumor necrosis factor α and IL-6. Adiponectin causes a significant reduction in FGF23 secretion in IDG-SW3 osteocytes (Rutkowski et al. 2017). Inversely, leptin increases production in primary cultured rat osteoblasts (Tsuji et al. 2010), and TNF, as seen above, increased FGF23 in a dose-dependent manner in IDG-SW3 osteocytes (Ito et al. 2015).

Adipokines also regulate FGF23 via calcitriol. Treating UMR-106 cells with leptin potentiated calcitriol up-regulation of FGF23, whereas IL-6 treatment blunted it (Saini et al. 2013).

Acidosis

Metabolic acidosis directly stimulates Fgf23 expression in mouse primary calvarial osteoblasts (Krieger et al. 2012). This requires the intracellular calcium signaling pathway and proper prostaglandin synthesis (Krieger and Bushinsky 2017), the same initial signaling pathway used in acidosis-induced bone resorption.

Local Regulation of FGF23

Systemic factors may also act locally to modulate FGF23 production by osteocytes and osteoblasts. An example is phosphate released from hydroxyapatite crystals during bone resorption. Calcitriol produced locally by osteoblasts may also stimulate FGF23 independently of circulating calcitriol (Nguyen-Yamamoto et al. 2017).

Studies using FGFR inhibitors (Wohrle et al. 2011, 2013) and a mouse model with Fgfr1 deleted from osteocytes (Xiao et al. 2014) report a decrease in FGF23, suggesting that FGFR1 signaling stimulates FGF23. In vitro data from osteoblasts showed that FGF2 activates Fgf23 promoter activity via the FGFR1 pathway (Han et al. 2015). αKlotho expression in osteocytes (Rhee et al. 2011) suggests that FGF23, through FGFR1, creates a positive feedback loop. In fact, αKlotho expression in long bone is required to increase Fgf23 expression in osteocytes in a uremic mouse model (Kaludjerovic et al. 2017). However, in a mouse model with αKlotho deleted in osteocytes, circulating FGF23 levels and Fgf23 transcripts in bone were unchanged (Komaba et al. 2017). Whether FGFR1 induces FGF23 by a systemic or autocrine/paracrine process remains to be determined.

GALNT3 glycosylation protects FGF23 from proteolytic cleavage and FAM20C phosphorylates FGF23 to inhibit glycosylation. The roles of other local factors involved in FGF23 regulation are not fully clarified. Three main factors are the metalloendopeptidase phosphate-regulating gene with homologies to endopeptidase on the X chromosome (PHEX also known as HYP), dentin matrix acidic phosphoprotein 1 (DMP1), and matrix extracellular phosphoglycoprotein (MEPE).

PHEX and DMP1 are expressed in osteoblasts and osteocytes and inhibit FGF23 synthesis (Quarles and Drezner 2001). Inactivating mutations of PHEX and DMP1 lead to hypophosphatemia in humans. Comparing compound- and single-mutant Dmp1 and Phex mice indicates that FGF23 is regulated through a PHEX–DMP1 common pathway involving FGFR signaling (Martin et al. 2011). Deleting FAM20C down-regulates DMP1 to significantly up-regulate FGF23 in human and mouse osteogenic cell lines, suggesting that FAM20C endogenous activity may suppress FGF23 by enhancing DMP1 expression (Wang et al. 2012). Moreover, in vitro experiments with preosteocytes show that sclerostin inhibits the expression of DMP1 and PHEX, although the effect on FGF23 expression is unknown (Atkins et al. 2011).

Human MEPE was cloned from a tumor resected from a patient with TIO (Rowe et al. 2000). A micropuncture study using rat renal proximal tubules showed that MEPE inhibits phosphate transport (Shirley et al. 2010). Mice with Mepe null mutations had increased bone mass and hyperphosphatemia and lowered FGF23, whereas overexpressing Mepe carboxy-terminal ASARMs (acidic serine aspartate-rich MEPE-associated motifs) induced localized hypomineralization and hypophosphatemia and increased FGF23 (Zelenchuk et al. 2015). This suggests that MEPE induces hypophosphatemia via its ASARM. Furthermore, PHEX may modulate MEPE and DMP1 by binding to ASARM peptides in these SIBLING proteins (Quarles 2012).

Finally, inorganic pyrophosphate (PPi) metabolism coupled with mineralization may affect FGF23 expression. PPi, an inhibitor of calcification, is generated by the ectonucleotide pyrophosphatase/phosphodiesterase 1 (ENPP1). Ankyrin 1 (ANK1), a multipass transmembrane channel, controls transport of PPi into the extracellular matrix for conversion to Pi for matrix mineralization by the alkaline phosphatase enzyme, tissue-nonspecific alkaline phosphatase (TNAP) (Ho et al. 2000). Inactivating mutations of ENPP1 can cause type 2 autosomal recessive hypophosphatemic rickets with hypophosphatemia and high FGF23 (Levy-Litan et al. 2010). The mechanism linking ENPP1 to FGF23 expression is unknown. It is hypothesized that inactivation of ENPP1 causes ectopic calcifications by reducing PPi levels in soft tissues. This results in phosphate depletion in bone leading to osteomalacia and increased FGF23 expression (Quarles 2012). Inactivation of Ank1 in mice causes impaired mineralization of extracellular matrix and a large increase in FGF23 expression (Chen et al. 2011).

FGF23 BIOLOGY AND DISEASES

FGF23 and Diseases Related to Mineral Metabolism

Normal Renal Function

Increases in circulating FGF23 in patients with normal renal function can result from tumors secreting FGF23 or monogenic diseases. High circulating FGF23 levels induce renal phosphate leak, hypophosphatemia, and inappropriately low serum levels of calcitriol, resulting in rickets in children and osteomalacia in adults. Untreated patients do not present a concomitant increase in urinary calcium excretion because the lower serum calcitriol levels reduce calcium intestinal absorption and hyperabsorptive hypercalciuria. However, excessive treatment with oral phosphate and active vitamin D supplements may lead to iatrogenic hypercalciuria and a severe increase in urinary phosphate excretion leading to nephrolithiasis and nephrocalcinosis.

A decrease in circulating FGF23 levels or resistance to FGF23 defined as the inability of FGF23 to properly act through its receptor are responsible for decreased urinary phosphate excretion, hyperphosphatemia, and inappropriately high serum calcitriol levels, resulting in soft tissue calcifications. This phenotype has been described as familial tumoral calcinosis (FTC) and is inherited only.

High intact FGF23 levels

• Acquired increase in FGF23 secretion: TIO.

  TIO is a rare acquired disease. Fewer than 200 cases have been reported and this rarity typically delays diagnosis by months or years. This pathology is defined by a tumor (typically benign and mesenchymal) secreting a phosphatonin, usually FGF23 (Jonsson et al. 2003). TIO is responsible for multiple pathological fractures and major muscle weakness directly related to hypophosphatemia. It is very difficult to locate the tumor, which is small, slow growing, and located in varying sites. Total body magnetic resonance imaging, computed tomography, or scintigraphy using radiolabeled somatostatin analog are often required. Resection stops secretion of the phosphatonin and corrects the biological and clinical effects.

• Inherited increase in FGF23 secretion. Three pathologies, collectively hypophosphatemic rickets, are primarily caused by high-FGF23 circulating concentrations:

  1. X-linked hypophosphatemia (XLH) is the most common form of hereditary hypophosphatemic rickets. XLH is caused by inactivating mutations of the PHEX gene whose precise mechanism of action remains to be defined (Francis et al. 1995; Jonsson et al. 2003). Osteopontin was recently identified as the first protein substrate for PHEX, showing in the murine model of XLH (Hyp mice) an increase in osteopontin that contributes to the osteomalacia (Neves et al. 2016).
  2. ADHR is rarer and linked to gain-of-function missense mutations in FGF23, rendering it resistant to inactivating cleavage (White et al. 2000).
  3. Autosomal recessive hypophosphatemic rickets (ARHR) type 1 is caused by inactivating mutations in the FGF23 transcription inhibitor DMP1 (Feng et al. 2006).
  Two other genes could also be responsible for hypophosphatemic rickets and high FGF23 levels:

  1. Three patients from a Bedouin family developed ARHR secondary to a loss-of-function mutation in the ENPP1 gene. This new form of ARHR is type 2 ARHR (Levy-Litan et al. 2010). Inactivating mutations in ENPP1 cause generalized arterial calcification of infancy (GACI) (Rutsch et al. 2003), with some patients presenting hypophosphatemic rickets (Lorenz-Depiereux et al. 2010). However, these three patients had no ectopic calcifications, suggesting that type 2 ARHR is a direct consequence of ENPP1 mutations.
  2. A de novo translocation with a break-point adjacent to αKLOTHO has been reported in one patient presenting clinical and biochemical features of hypophosphatemic rickets, resulting from inappropriate renal phosphate losses with high plasma αKLOTHO and FGF23. This patient subsequently developed primary hyperparathyroidism (Brownstein et al. 2008). The elevated FGF23 in this patient is difficult to explain but may be caused by increased circulating αKLOTHO or to increased PTH levels. Conversely, the αKLOTHO deficiency associated with increased FGF23, likely because of resistance, suggests a complex interaction (Ichikawa et al. 2007).
  Three phenotypes more complex than hypophosphatemic rickets may be associated with hypophosphatemia caused by renal phosphate loss with increased circulating levels of FGF23:

  1. McCune–Albright syndrome (fibrous dysplasia) is caused by somatic activating mutations of the Gαs subunit (subunit of the stimulatory G protein) encoded by the GNAS1 gene and is characterized by bone abnormalities, skin pigmentation, and endocrine abnormalities, including hyperthyroidism, acromegaly, and Cushing’s syndrome. Renal phosphate loss occurs in approximately 50% of patients as a result of an increase in FGF23 secondary to FGF23 production in fibrous bone lesions (Riminucci et al. 2003).
  2. Raine syndrome (osteosclerotic bone dysplasia) is caused by mutations in FAM20C. Clinical manifestations include bone mineral hyperdensity, early ossification, cerebral calcifications, and pulmonary hypoplasia (Rafaelsen et al. 2013). This pathology is typically lethal 3 weeks postpartum, but has been described in patients at puberty. Patients develop hypophosphatemia as a result of elevated FGF23.
  3. Osteoglophonic dysplasia is caused by activating mutations in FGFR1. Patients present with craniosynostosis, prominent supraorbital ridges and depressed nasal bridges, rhizomelic dwarfism, and nonossifying bone lesions. Three patients had osteoglophonic dysplasia and hypophosphatemia as a result of a massive renal phosphate leak because the constitutive activation of FGFR1 probably excessively mediates FGF23 down-regulation of NaPi2a and NaPi2c. One patient had high FGF23, whereas low circulating FGF23 levels secondary to FGFR1 constitutive activation would be expected (White et al. 2005). However, FGFR1 signaling has been shown to induce FGF23 secretion in osteocytes (Xiao et al. 2014). Consequently, FGFR1 actions on FGF23 secretion are difficult to predict.

Low intact FGF23 levels: Familial tumoral calcinosis (FTC)

FTC is a rare, autosomal recessive disease characterized by hyperphosphatemia associated with higher concentrations of calcitriol and phosphocalcic deposits within periarticular, subcutaneous (Fig. 4), and vascular tissues caused by loss-of-function mutations in FGF23 or GALNT3 subjecting FGF23 to rapid degradation (Benet-Pages et al. 2005; Ichikawa et al. 2009). Thus, levels of iFGF23 are low and cFGF23 levels are elevated. One case report indicates that a loss-of-function mutation in αKLOTHO may cause FTC and FGF23 resistance with high intact and cFGF23 levels (Ichikawa et al. 2007). Of note, this patient also had primary hyperparathyroidism that could be explained by the loss of FGF23 inhibitory effect on PTH through the FGFR1/Klotho pathway (Komaba et al. 2010).

Figure 4.

Consequences of genetic fibroblastic growth factor (FGF)23 deficiency (familial tumoral calcinosis [FTC]). This image represents extraosseous calcifications of the right buttock of a 42-year-old patient diagnosed with FTC. FTC was confirmed by genetic analysis revealing homozygous mutations in the gene encoding FGF23 (c.211A>G or p.Ser71Gly). The patient had decreased circulating intact FGF23 (iFGF23) but increased carboxy-terminal FGF23 (cFGF23) levels caused by enhanced FGF23 degradation resulting in increased serum phosphate and inappropriately high calcitriol levels. At the time of diagnosis, the extraosseous calcifications comprised a plain massive heterogeneous lesion in the right buttock (167 × 204 mm) so that sitting was impossible. An attempt to surgically remove the buttock’s lesion several years ago led to an increase of the tumoral calcinosis (TC) and a pseudocystic development within the lesion.

Altered Renal Function

The most common condition with elevated FGF23 is CKD. Uremic patients may have an increase in FGF23 concentrations up to 1000-fold. During CKD, FGF23 expression increases in osteocytes (Pereira et al. 2009; Shimada et al. 2010), which requires Klotho expression in bone (Kaludjerovic et al. 2017). Data in rats have shown that the diseased kidney produces Fgf23 (Zanchi et al. 2013; Mace et al. 2017) although this does not seem to contribute to increased circulating FGF23 after experimental kidney injury (Mace et al. 2017). During CKD progression, cFGF23 decreases in favor of iFGF23, suggesting decreased cleavage (Smith et al. 2012). The main cause of elevated FGF23 in CKD patients is phosphate retention because of decreased urinary phosphate excretion. Several conditions associated with CKD, such as acidosis, tissue hypoxia secondary to anemia, and iron deficiency, may also contribute to the increase in FGF23 levels.

FGF23 secretion increases very early in CKD when the estimated glomerular filtration rate is lower than 90 mL/min/1.73 m². FGF23 increases usually antedate hyperparathyroidism and the increase in serum phosphate (Larsson et al. 2003; Ix et al. 2010; Isakova et al. 2011). This suggests that FGF23 plays a key role in the development of secondary hyperparathyroidism (Fig. 5).

Figure 5.

Putative mechanism of secondary hyperparathyroidism during the course of chronic kidney disease (CKD). Elevated fibroblastic growth factor (FGF)23 levels correlate with increased urinary phosphate excretion in patients with CKD and likely serve to maintain serum phosphate levels in the normal range by inhibiting the reabsorption of phosphate in the proximal tubules of the remaining functional nephrons. Concomitantly, FGF23 inhibits 1α-hydroxylase and stimulates 24-hydroxylase in the proximal tubule to decrease calcitriol levels. This decrease in serum calcitriol levels leads to a decrease in intestinal phosphate absorption, thereby facilitating the normalization of serum phosphate levels. However, this is accompanied by a decrease in intestinal calcium absorption, resulting in lower serum calcium levels in late CKD stages. The decrease in serum calcitriol and calcium levels weakens the suppression of parathyroid hormone (PTH), driving the development of secondary hyperparathyroidism. In CKD patients, very high FGF23 levels are observed concomitantly to high PTH levels, suggesting a loss of PTH suppression by FGF23 (orange arrow). This resistance to FGF23 in the parathyroid glands is probably caused by the decreased expression of both fibroblastic growth factor receptor (FGFR)1 and Klotho in hyperplastic glands from uremic patients.

This hypothesis is supported by a study in CKD rat models treated with an anti-FGF23-inactivating antibody (Hasegawa et al. 2010). Treated rats had reduced fractional excretion of phosphate, increased serum phosphate, and normalized serum calcitriol (due to an increased 1α hydroxylase and decreased 24-hydroxylase renal expression), leading to decreased serum PTH. However, the neutralizing anti-FGF23-Ab also increased mortality and aortic calcifications in CKD rats owing to a dose-dependent increase in serum phosphate (Shalhoub et al. 2012). This finding limits the use of anti-FGF23-Ab in CKD patients for whom the primary role of FGF23 is to delay hyperphosphatemia.

FGF23 inhibits PTH synthesis and secretion but very high FGF23 and PTH levels are observed in CKD, suggesting FGF23 resistance caused by the known decreased expression of both FGFR1 and Klotho in hyperplastic glands from uremic patients (Komaba et al. 2010).

FGF23 levels also increase secondary to acute kidney injury (AKI). There is a rapid increase in iFGF23 levels in patients with AKI (Christov 2014). Animal studies confirm these data and further show that this increase secondary to AKI is independent of PTH and vitamin D (Christov et al. 2013). Phosphate regulation of FGF23 in AKI was, however, difficult to determine because of the inevitable increase in serum phosphate.

FGF23 and Diseases Not Related to Mineral Metabolism

Higher FGF23 levels are independently associated with mortality risk among patients on hemodialysis treatment (Gutierrez et al. 2008). High FGF23 levels contribute to left ventricular hypertrophy (LVH), anemia, inflammation, and a decreased host defense in response to infection. Some of these have been reported in non-CKD populations (Souma et al. 2016).

FGF23 and LVH

FGF23 circulating levels are associated with LVH (Farrow et al. 2011; Seeherunvong et al. 2012) and an increased risk of cardiovascular mortality (Gutierrez et al. 2008). The correlation between FGF23 and LVH has also been described in non-CKD patients (Mirza et al. 2009). Data from animal models confirm the causative relationship between high FGF23 levels and LVH. Intramyocardial or intravenous injection of FGF23 in mice resulted in LVH. Interestingly, in a rat model of CKD, FGFR antagonist treatment attenuated LVH independent of blood pressure (Faul et al. 2011). However, FGF23 also induces renal sodium reabsorption in mice thus increasing plasma volume and potentially cardiovascular risk (Andrukhova et al. 2014a). Treating neonatal rat ventricular cardiomyocytes with iFGF23 increased cell-surface area in a dose-dependent manner (Faul et al. 2011). The hypertrophic action of FGF23 on cardiomyocytes is independent of Klotho and mediated by FGFR4 (Grabner et al. 2015).

Patients with sickle cell disease have elevated carboxy-terminal and normal intact FGF23 levels. Carboxy-terminal levels were positively and independently associated with the left ventricular mass index. Moreover, cFGF23 induces hypertrophy in adult rat cardiomyocytes in vitro via an FGFR-dependent but Klotho-independent pathway (Courbebaisse et al. 2017), challenging the fact that only iFGF23 has biological actions.

FGF23 and Erythropoiesis

Ablating Fgf23 in mice leads to a significant increase in erythropoiesis (Coe et al. 2014). FGF23 loss stimulates a hypoxic bone marrow environment that activates erythropoietin-induced erythropoiesis in mutant mice. In vivo and in vitro FGF23 treatment inhibits erythropoiesis (Coe et al. 2014). Hypoxia from anemia stimulates FGF23 production via HIF-1α signaling, suggesting that FGF23 not only decreases erythropoiesis but is stimulated by anemia (Clinkenbeard et al. 2014; Zhang et al. 2016). These findings are consistent with clinical studies in CKD patients (Tsai et al. 2016) and sickle cell disease patients (Courbebaisse et al. 2017), which show an inverse association between cFGF23 and hemoglobin levels.

FGF23 and Immunity

Ectopic Fgf23 expression occurs in proinflammatory macrophages (Bacchetta et al. 2013; Masuda et al. 2015; Han et al. 2016). FGF23 enhances TNF-α in macrophages and may act as a proinflammatory cytokine. IL-1β and other proinflammatory cytokines increase Fgf23 transcription by activating HIF-α signaling in vitro (Ito et al. 2015; David et al. 2016), suggesting a connection between FGF23 and the inflammatory process. FGF23 inhibits 1-α hydroxylase expression in monocytes to decrease local calcitriol synthesis, which has microbicidal properties (Bacchetta et al. 2013; Masuda et al. 2015). In vivo and in vitro data show that FGF23 directly inhibits neutrophil arrest at the site of infection through an FGFR2-dependent pathway to diminish host defense (Rossaint et al. 2016). These findings agree with clinical studies in CKD patients showing that higher FGF23 levels are independently associated with higher levels of inflammatory markers (Munoz Mendoza et al. 2012) and a higher risk of infectious events (Chonchol et al. 2016).

CONCLUSION

Several facets of the complex biology and regulation of FGF23 remain to be clarified. FGF23 has well-known hormonal actions to decrease circulating phosphate and calcitriol levels. It has a role in many other physiological and pathological processes involving mineral metabolism. It acts on the heart and the immune system, suggesting new research regarding FGF23 actions beyond mineral metabolism. FGF23 is associated with an increased risk of mortality, likely because of its contribution to LVH, anemia, inflammation, and decreased host defense response to infection. It may cause secondary hyperparthyroidism during CKD. However, targeting FGF23 to avoid such complications is difficult because FGF23 neutralization also impairs phosphate homeostasis with a subsequent increase in serum phosphate that leads to an increased risk of vascular calcifications.