FGF23 Signalling and Physiology

Abstract

Fibroblast growth factor 23 (FGF23) is a phosphotropic hormone that belongs to a subfamily of endocrine FGFs with evolutionarily conserved functions in C. elegans and fruit flies. FAM20C phosphorylates FGF23 post-translationally, targeting to proteolysis through subtilisin-like proprotein convertase FURIN, resulting in secretion of FGF23 fragments. O-glycosylation of FGF23 through GALNT3 appears to prevent proteolysis, resulting in secretion of biologically active intact FGF23. In the circulation, FGF23 may undergo further processing by plasminogen activators. Crystal structures show the ectodomain of the cognate FGF23 receptor FGFR1c binds with the ectodomain of the co-receptor alpha-KLOTHO. The KLOTHO-FGFR1c double heterodimer creates a high-affinity binding site for the FGF23 C-terminus. The topology of FGF23 deviates from that of paracrine FGFs, resulting in poor affinity for heparan sulfate, which may explain why FGF23 diffuses freely in the bone matrix to enter the bloodstream following its secretion by cells of osteoblastic lineage. Intact FGF23 signalling by this canonical pathway activates FRS2/RAS/RAF/MEK/ERK1/2. It reduces serum phosphate by inhibiting 1,25-Dihydroxyvitamin D synthesis, suppressing intestinal phosphate absorption, and by downregulating the transporters NPT2a and NPT2c, suppressing phosphate reabsorption in the proximal tubules. The physiological role of FGF23 fragments, which may be inhibitory, remains unclear. Pharmacological and genetic activation of canonical FGF23 signalling causes hypophosphatemic disorders, while its inhibition results in hyperphosphatemic disorders. Non-canonical FGF23 signalling through binding and activation of FGFR3/FGFR4/calcineurin/NFAT in an alpha-KLOTHO-independent fashion mainly occurs at extremely elevated circulating FGF23 levels and may contribute to mortality due to cardiovascular disease and left ventricular hypertrophy in chronic kidney disease.

Introduction

Fibroblast growth factor 23 (FGF23) is a phosphotropic hormone that belongs, along with FGF19 and 21, to a subfamily of endocrine FGFs. (Château et al., 2010) FGF23 is secreted by osteoblasts and osteocytes into the systemic circulation and acts in the kidney, parathyroid, heart, bone and possibly other organs. (Shimada et al., 2004, Krajisnik et al., 2007, Faul et al., 2011, Murali et al., 2016b) This property sets FGF23 apart from paracrine FGFs, which mediate cell and organ differentiation locally. FGF23 also stands apart from the intracrine FGFs, which act intracellularly independently of an FGF receptor (FGFR) by interacting with voltage-gated sodium channels and neuronal scaffolding proteins to potentiate neuronal excitability. (Goldfarb et al., 2007, Itoh et al., 2016)

FGF23 was first discovered in the murine brain (Yamashita et al., 2000) and soon thereafter loss-of-function (LOF) mutations were identified by linkage analysis to be the cause of autosomal dominant hereditary rickets (ADHR) (ADHR-Consortium, 2000) and tumour-induced osteomalacia (TIO) (Shimada et al., 2001). ADHR and TIO are rare renal phosphate wasting disorders characterized by, through different mechanisms, hypophosphatemic rickets and osteomalacia, which leads to bowing of the long bones, short stature, bone pain, muscle weakness and fractures. (Carpenter et al., 2020)

FGF23 has since been shown to be a critical phosphaturic hormone which, along with parathyroid hormone (PTH), regulates phosphate recycling and synthesis of calcitriol (1,25-Dihydroxyvitamin D or 1,25(OH)₂D) in the kidneys. (Shimada et al., 2004) Canonical FGF23 signalling requires the obligatory co-receptor alpha KLOTHO (KL), a transmembrane protein with extracellular glucuronidase activity, (Tohyama et al., 2004) for binding to the FGF receptor 1c (FGFR1c). (Urakawa et al., 2006) Still, some FGF23 signalling occurs independently of KL and is often referred to as non-canonical FGF23 signalling.

Here we review the current understanding of the actions of FGF23 and the signalling pathways involved. FGF23, in turn, is regulated by phosphate, inflammation and energy metabolism, which we will not discuss here, but has been addressed in several recent reviews. (Faul, 2012, Clinkenbeard and White, 2017, Chande and Bergwitz, 2018, Michigami et al., 2018, Takashi and Fukumoto, 2020)

Evolutionary conservation of KL/FGF23 function

Orthologs of the FGF signalling pathway are found in C. elegans and D. melanogaster. (Château et al., 2010) Even KL and FGF23 functions are evolutionary conserved. C. elegans has two FGF orthologs, egg-laying-defective 17 (Egl-17) (Burdine et al., 1997) and lethal-756 (Let-756); (Roubin et al., 1999) one FGFR ortholog, Egl-15; (DeVore et al., 1995) and two KL orthologs, Klo-1 and Klo-2 (Table 1) (Château et al., 2010). Egl-15 activity is negatively regulated by N-glycosylation, which inhibits ligand and heparan sulfate binding to this receptor, (Polanska et al., 2009) in a similar manner to human FGFR activity. (Duchesne et al., 2006) Egl-15 has also been shown to associate with Klo-1 by co-immunoprecipitation using monoclonal Klotho or Egl-15 antibodies, (Polanska et al., 2011) though further work is necessary to determine if Egl-15 binds to Klo-1 in a fashion similar to KL and FGFR and if Klo-1 and Klo-2 are necessary for Egl-15 activity. Activation of Egl-15 by Egl-17 or Let-756 results in muscle lysis. (Szewczyk and Jacobson, 2003) Loss of function mutation in Egl-17 leads to defective migration and differentiation of sex myoblasts, (Burdine et al., 1997) and loss of function mutation in Let-756 leads to larval developmental arrest. (Roubin et al., 1999) These effects resemble the roles of FGFs in morphogenesis in higher species and suggest these orthologs may share properties similar to those of multiple FGFs in higher species. Interestingly, like ablation of FGF23 and KL in higher species, LOF mutations in Klo-1 and Klo-2 also reduced stress tolerance and longevity in C. elegans. (Polanska et al., 2009, Château et al., 2010) Conversely, the gain of function in Egl-15 leads to fluid accumulation in the animal’s pseudocoelom, hinting at the roles of FGF23 and KL in mineral excretion in higher species. (Huang and Stern, 2004)

Table 1:

Orthologs to human FGF, FGFR, and KL in C. elegans and D. melanogaster, including their general function and consequences of alterations of activity, are listed as mentioned in the text.

Species	Ortholog	General Function	Gain/Loss of Function Consequences
C. elegans	FGF: Egl-17 Let-756 (Szewczyk and Jacobson)	Egl-17: Longevity, oxidative stress resistance Let-756: Fluid homeostasis Both: Muscle lysis	LOF Egl-17: defective migration and differentiation of sex myoblasts LOF Let-756: larval developmental arrest LOF both Egl-17 and Let-756: suppression of muscle lysis
	FGFR: Egl-15 (Szewczyk and Jacobson)	Receptor for Egl-17 and Let-756	GOF: Muscle lysis, abnormal fluid filling in pseudocoelom
	KL: Klo-1 Klo-2 (Château et al.)	Improve longevity and resistance to oxidative stress	Unknown
D. melanogaster	FGF: bnl, pyr, ths	bnl: promote Pi excretion	LOF during development: absent trachea resulting in death Injection of human FGF23 in adults: increased MFS2 expression, decreases blood Pi
	FGFR: btl, htl	btl: bnl receptor htl: pyr and ths receptor	LOF during development: absent trachea resulting in death LOF in adults: blocks hypophosphatemic response to injection of human FGF23
	KL: CG7901 (Rose et al.)	Unknown	Unknown

LOF = Loss-of-function mutation

GOF = Gain-of-function mutation

FGF = Fibroblast growth factor

FGFR = Fibroblast growth factor receptor

KL = Klotho

Egl-15/17 = Egg-laying-defective 15/17

Let-756 = Lethal-756

Klo = Klotho-ortholog

bnl = Branchless

pyr = Pyramus

ths = Thisbe

btl = Breathless

htl = Heartless

MFS2 = Major facilitator superfamily 2

Likewise, D. melanogaster has three FGF orthologs, branchless (bnl), pyramus (pyr), and thisbe (ths); two FGFR orthologs, breathless (btl) and heartless (htl); and one KL ortholog (CG9701) which were identified through the DRSC integrative ortholog prediction tool (DIOPT) (Table 1). (Rose et al., 2019) As in humans, these FGFR orthologs signal via the MAPK pathway. (Sen et al., 2011) We recently showed that flies microinjected with human FGF23 become hypophosphatemic. This hypophosphatemia appears to be due to upregulation of the type I sodium-dependent Pi co-transporter major facilitator superfamily 2 (MFS2) in their Malpighian tubules. (Rose et al., 2019) Ablation of MFS2 prevents this FGF23 effect and also blocks the formation of fly calcium Pi kidney stones. Based on expression in the tubule and stimulation of MFS2 gene expression by dietary Pi and FGF23, MFS2 may function as an excretory Pi-transporter in the Malpighian tubules. This function is different from the situation in mammalian kidneys, where the type II sodium-dependent Pi co-transporters NPT2a and NPT2c help reclaim Pi from the urine and may reflect the unique anatomy of the fly renal system, which lacks a connection to a glomerulus as seen in higher species, (Rose et al., 2019) and absence of skeletal mineral stores for Pi. The roles of the FGF, FGFR, and KL orthologs in D. melanogaster in Pi homeostasis require further study, although in particular bnl may promote Pi excretion as transgenic overexpression of bnl increases MFS2 expression. (Rose et al., 2019)

Structure and biochemistry

FGF23 is a 251-amino acid, 32 kDa protein, with high N-terminal homology to other FGFs. (Yamashita et al., 2000) Phosphorylation of FGF23 through the extracellular kinase family member 20C (FAM20C) at amino acid S180 prevents post-translational O-glycosylation by polypeptide N-acetylgalactosaminyltransferase 3 (GALNT3)(Fig. 1). Lack of O-glycosylation makes FGF23 susceptible to proteolysis by the subtilisin-like proprotein convertase FURIN. Proteolysis of FGF23 results in the secretion of inactive FGF23 fragments. (Tagliabracci et al., 2014) FGF23 phosphorylation can be detected in bone sections and FAM20C-transfected cells (Lindberg et al., 2015) underlining its biological importance. Conversely, phosphorylation and proteolysis by FURIN are prevented by O-glycosylation at position T178 through GALNT3, which results in the secretion of the biologically active intact FGF23. (Tagliabracci et al., 2014) The regulation of transcription, production, and post-translational modification of FGF23 has been recently reviewed in greater detail. (Edmonston and Wolf, 2020)

Figure 1. Post-Translational Modification of FGF23.

FGF23 is a 251-amino acid 32 kDa protein. Phosphorylation by FAM20C at amino acid S180 is cleaved by FURIN into biologically-inactive FGF23 fragments. Unprocessed transcripts can also be O-glycosylated by GALNT3 into biologically-active FGF23. Cleavage by FURIN is suppressed by O-glycosylation of FGF23 and may be regulated by the elusive endocrine Pi sensor. The C-terminal fragment may serve as the endogenous inhibitor of KL-FGFR complex formation.

FGF23 = Fibroblast growth factor 23

FAM20C = Extracellular kinase family member 20C

FURIN = Subtilisin-like proprotein convertase

GALNT3 = N-acetylgalactosaminyltransferase 3

KL-FGFR = Klotho – Fibroblast growth factor receptor complex

Regulation of FGF23 bioactivity by GALNT3 and FAM20C is still poorly understood but may be controlled by the elusive endocrine Pi-sensor. (Takashi and Fukumoto, 2020) Impaired processing and inactivation of FGF23 play an essential pathophysiological role in the development of FGF23 excess in chronic kidney disease (CKD). Excess circulating FGF23 causes left ventricular hypertrophy (LVH) in CKD patients (Faul et al., 2011) and an autosomal dominant hypophosphatemic rickets (ADHR) as recently reviewed. (Wolf and White, 2014)

Furthermore, C-terminal fragments of FGF23 may serve a physiological role as an endogenous inhibitor to the formation of the KL-FGFR complex (Goetz et al., 2010) and subsequently full-length FGF23 signalling, which may thus induce hypophosphaturia (Goetz et al., 2010) and raise serum iron levels (Agoro et al., 2020), but this and the signalling pathways involved remain poorly understood.

Likewise, whether biologically meaningful cleavage and inactivation of intact FGF23 in the circulation occurs is poorly understood. FGF23 cleavage and inactivation may involve the actions of both tissue-type and urokinase plasminogen activators. Genetic ablation of Pai-1, an inhibitor of plasminogen activator, for example, can rescue the phenotype of the Kl-null mouse. (Eren et al., 2017) As a result, these authors speculated that inhibition of PAI-1 could be used pharmacologically to prevent FGF23 excess and the development of LVH in patients with CKD.

FGF23 signalling

FGFs signal through the four FGF tyrosine kinase receptors (FGFR1–4) and thereby activate the RAS-MAPK and the PI3K-AKT pathways. (Ornitz and Itoh, 2015) While FGF23 can activate FGFR3 and FGFR4 in a KL-dependent (Kurosu et al., 2006) or -independent fashion (Grabner et al., 2015, Murali et al., 2016b), FGFR1c appears to be the primary endocrine receptor for FGF23. Recent crystal structure data suggest that this receptor isoform, together with KL, but independent of its beta-glucuronidase enzymatic activity, forms a binding pocket for the C-terminal region of c-FGF23. (Chen et al., 2018) Heparan sulfate, which unlike other paracrine FGFs has a weak affinity to FGF23 itself, facilitates the dimerization of two 1:1:1 FGF23–FGFR1c–KL complexes into a symmetric 2:2:2:2 FGF23–FGFR1c–KL–HS signal transduction unit. (Chen et al., 2018) The osteocyte extracellular matrix is a heparan sulfate-rich environment, which limits systemic diffusion of paracrine FGFs, and these authors hypothesize that reduced affinity to heparan sulfate promotes systemic diffusion of FGF23. (Beenken and Mohammadi, 2012)

Commonly, FGF23 signalling through KL-dependent binding to FGFR1c is referred to as ‘canonical’. Conversely, FGF23 signalling through KL-independent binding to FGFR3 and FGFR4 is referred to as ‘non-canonical’. See Fig. 2 for details of canonical and non-canonical signalling and Table 2 for the various animal models that helped shape the current understanding of FGF23 signalling and physiology.

Figure 2. Canonical and Non-Canonical FGF23 Signaling.

FGF23 is a peptide hormone secreted by bone that affects multiple organ systems. A. In the proximal tubules, FGF23 caused internalization of sodium type 2 co-transporter NPT2a in a FGFR1c/KL-mediated manner by activation of ERK1/2 and SGK1 (red, canonical). FGF23 also decreased activation of vitamin D in a KL-independent fashion (blue, non-canonical). B. In the distal tubules, FGF23 causes KL-dependent stimulation of calcium and sodium reabsorption. C. In the parathyroid glands, FGF23 causes suppresses of PTH secretion, likely through KL-dependent and -independent pathways. D. In the heart, FGF23 causes LVH in a KL-independent manner in a FGFR4 mediated fashion mediated by PLCgamma/NFAT/calcineurin. E. In osteoblasts, FGF23 inhibits bone mineralization by decreased expression of the phosphatase TNAP in a KL-independent fashion.

FGF23 = Fibroblast growth factor 23

FGFR = Fibroblast growth factor receptor

NPT2a = Sodium-phosphate cotransporter 2a

ERK1/2 = Extracellular signal-regulated kinases 1/2

SGK1 = Serum/glucocorticoid regulated kinase 1

WNK4 = Serine/threonine-protein kinase WNK4

NFAT = Nuclear factor of activated T-cells

PLCgamma = Phosphoinositide phospholipase C gamma

TNAP = tissue nonspecific alkaline phosphatases

TRPV5 = transient receptor potential vanilloid 5 (calcium channel protein)

NCC = Sodium-chloride cotransporter

Table 2:

Overview of the mouse models that have contributed to our understanding of phosphate homeostasis. Each strain is compared to its appropriate wild-type control.

Model	Serum					Urine
	Calcium	Phosphate	FGF23	PTH	1,25(OH)2D	Calcium	Phosphate
Kl−/− (Yoshida et al., 2002, Segawa et al., 2007)	↑	↑	↑↑	↓	↑	↑	-
Kap- Kl−/− (Ide et al., 2016)	-	-	-	-	-	-	-
Pepck-Kl−/− (Ide et al., 2016)	-	-	↑↑	-	-	-	↓
Slc34a1-Kl−/− (Ide et al., 2016)	↑	↑	-	-	-	-	↓↓
Ksp-Kl−/− (Olauson et al., 2012)	-	↑	↑	↓	-	↑	-
Fgf23−/− (Shimada et al., 2004, Hesse et al., 2007)	↑	↑	↓↓	↓	↑	-	-
Hyp (Murali et al., 2016a, Eicher et al., 1976)	↓	↓	↑	-	-	N.d.	↑
Fgfr1−/− (Gattineni et al., 2009)	N.d.	-	↑	-	-	N.d.	N.d.
Fgfr3−/− (Gattineni et al., 2009)	N.d.	-	-	-	↑	N.d.	N.d.
Fgfr4−/− (Gattineni et al., 2009)	N.d.	-	↑	-	-/↑	N.d.	N.d.
Ksp-Fgfr1−/− (Han et al., 2016)	-	↓	-	↑	-	↑	↑
gGt-Fgfr1−/− (Han et al., 2016)	-	↑	-	-	-	-	-
Ndrg1-CreERT2; Fgfr1–4−/− (Takeshita et al.)	-	↑	↑	↓	↑	N.d.	N.d.

FGF23 = Fibroblast growth factor 23

PTH = Parathyroid hormone

1,25-(OH)₂D = Calcitriol, 1,25-Dihydroxyvitamin D.

Kl^−/− = Klotho-null mouse model

Kap-Kl^−/− = Kl-ablation in the proximal tubules mediated by Kidney androgen-regulated protein (Kap)-Cre-recombinase

Pepck-Kl^−/− = Kl-ablation in the proximal tubules mediated by Phosphoenolpyruvate carboxykinase (Pepck)-Cre-recombinase

Slc34a1-Kl^−/− = Kl-ablation in the proximal tubules mediated by Solute carrier family 34 member 1 (Slc34a1)-Cre-recombinase

Ksp-Kl^−/− = Kl-ablation in the distal tubules mediated by Kidney-specific cadherin (Ksp)-Cre-recombinase

Hyp = Phosphate-regulating gene with homologies to endopeptidases on the X chromosome (PHEX)-deficient mouse model of X-linked hypophosphatemia (XLH)

Fgfr^−/− = Fibroblast growth factor receptor-null mouse model

Ksp-Fgfr1^−/− = Fgfr1-ablation in the distal tubule by Kidney-specific cadherin (Ksp)-Cre-recombinase

gGt-Fgfr1^−/− = Fgfr1-ablation in the proximal tubule by Gamma-glutamyltransferase (gGt)-Cre-recombinase

Ndrg1-CreERT2;Fgfr1–4¹⁻⁴ = Tamoxifen-induced Fgfr1–4-ablation in the proximal tubule by N-myc downstream-regulated gene-1 (Ndrg1)-Cre-recombinase

N.d. = No data.

Canonical FGF23 signalling

The phosphaturic effects of FGF23 in the kidney tubules are KL-dependent. In the proximal renal tubules, membrane levels of the sodium-phosphate co-transporters, NPT2a and NPT2c, are reduced by FGF23. The internalization and degradation of these phosphate transporters increases phosphate excretion and was shown to depend on the activation of the extracellular signal-regulated kinases 1/2 (ERK1/2) and serum/glucocorticoid-regulated kinase-1 (SGK1) pathways, (Andrukhova et al., 2012) resulting in phosphorylation of the Na+/H+ exchange regulatory factor (NHERF1)(Fig. 2A). (Deliot et al., 2005) An initial question was the site of action of FGF23 in the kidneys since KL is expressed highly in the distal renal tubules. (Kuro-o et al., 1997) KL was furthermore shown to co-localize with phosphorylated-ERK1/2 in the distal tubules following FGF23 injection of mice. (Farrow et al., 2009) This was corroborated by co-localization of high KL expression with the calcium channel protein transient receptor potential vanilloid 5 (TRPV5) in the distal tubules. (Chang et al., 2005) Furthermore, targeted deletion of KL in the distal tubule with Ksp-Cre elevated serum Pi, which is consistent with resistance to FGF23 in the distal tubules, while interestingly 1,25(OH)₂D remained normal (Table 2). (Olauson et al., 2012)

However, low levels of KL expression in the proximal tubule was ultimately demonstrated along with FGFR1, 3, and 4 expression. (Andrukhova et al., 2012) Multiple groups have now shown that FGF23 stimulates both fibroblast growth factor receptor substrate 2 (FRS2) and ERK1/2 in the proximal tubule. (Urakawa et al., 2006, Andrukhova et al., 2012) Targeted ablation of KL using three proximal tubule-specific promoters driving Cre, Kap, Pepck, and SLC34a1 (Table 2), increased serum Pi and reduced urine Pi, suggesting that FGF23 regulates re-absorption of Pi in a KL-dependent manner in the proximal tubules. (Ide et al., 2016) Likewise, conditional ablation of Fgfr1 in the distal tubule (Ksp-Cre) caused hypophosphatemia and hypercalciuria whereas ablation in the proximal tubule (gGt-Cre) resulted in hyperphosphatemia (Fig. 2B). At the same time, 1,25(OH)₂D levels were unchanged. (Han et al., 2016) Tamoxifen-induced conditional ablation of Fgfr1–4 in the proximal tubule (Ndrg1-CreERT2) likewise caused hyperphosphatemia, although along with an increase in 1,25(OH)₂D levels, which will be discussed further. (Takeshita et al., 2018)

Conversely, global ablation of Fgfr3 and 4 decreased 1,25(OH)₂D levels in mice (Table 2). (Gattineni et al., 2009) These results suggest that 1,25(OH)₂D and 1-alpha-hydroxylase regulation may be mediated through FGFR3 and FGFR4 in a KL-independent manner, which would formally be considered non-canonical (see below). However, KL-deficient mice have highly increased serum 1,25(OH)₂D and 1-alpha-hydroxylase levels, which resemble findings in FGF23-deficient mice. This notion argues that FGF23 suppresses 1-alpha-hydroxylase at least in part in a KL-dependent fashion. (Shimada et al., 2004)

FGF23 affects calcium homeostasis in the distal tubules, which express the calcium channel protein transient receptor potential vanilloid 5 (TRPV5) in the apical membrane, a transporter that reabsorbs calcium from the urine. (Hoenderop et al., 1999) KL deglycosylates and thereby retains TRPV5 in the apical distal tubule membrane. (Chang et al., 2005) This process appears to be stimulated by FGF23: recombinant FGF23 injections increased membrane TRPV5 expression and decreased urinary calcium excretion in wild-type but not Klotho-deficient mice, which appears to be mediated by FGFR, ERK1/2 and serum/glucocorticoid-regulated kinase-1 (SGK1). (Andrukhova et al., 2012) Furthermore, KL-deficient and FGF23-deficient mice showed profound hypercalciuria. FGF23 stimulates formation of TRPV5 – serine/threonine-protein kinase 4 (WNK4) complexes. (Fig. 2B).

Sodium is reabsorbed in the distal tubules by a thiazide-sensitive sodium-chloride cotransporter (NCC). (Gamba et al., 1993) This transporter appears to be regulated by FGF23 in a KL-dependent FGF23 fashion. Expression of NCC and thus sodium reabsorption is decreased in an FGF23-deficient mouse model, exhibit decreased, and conversely injection of FGF23 into wild-type mice results in increased sodium reabsorption and volume expansion, an effect not observed with inhibition of ERK1/2 or SGK1 or FGF23 injection into KL-null mice. (Andrukhova et al., 2014) FGF23 has also been shown to reduce expression of angiotensin-converting enzyme 2 (ACE2), a component of the renin-angiotensin system, and thus has been proposed to participate systemically in blood pressure regulation and to prevent angiotensin-related cardiac remodeling as recently reviewed. (Kovesdy and Quarles, 2016) The contribution of FGF23 excess to volume expansion seen in CKD may warrant further study.

Parathyroids in Kl-null mice have a reduced set-point for calcium (Mace et al., 2018) which may be due to decreased expression of Na/K-ATPase and is independent of FGF23 (Imura et al., 2007). In addition, FGF23 reduces PTH secretion in the parathyroid gland, which may have KL-dependent and -independent mechanisms (Fig. 2C). A murine chronic kidney disease (CKD) model exhibited decreased Kl and Fgfr1 expression. (Ben-Dov et al., 2007) Presence of secondary hyperparathyroidism in these mice was interpreted by the authors as indirect evidence for Klotho-dependence of PTH suppression by FGF23. However, FGF23 was able to suppress PTH in a parathyroid-specific Kl-null mouse model. (Olauson et al., 2013) Suppression of PTH secretion by FGF23 was abolished by addition of the calcineurin-NFAT pathway inhibitor cyclosporine A, arguing that this action of FGF23 is, at least partially, KL-independent (Olauson et al., 2013) Furthermore, broad FGFR inhibition by PD173074 in Wistar rats (Mace et al., 2018) and inhibition of ERK1/2 by U0126 (Ben-Dov et al., 2007) or by recombinant dual-specific phosphatases (Dusp) (Román-García et al., 2012) in ex vivo parathyroid cultures prevented suppression of PTH secretion in response to recombinant FGF23.

There is some evidence that canonical FGF23 signalling may finally play a role in the pathophysiology of tumour-induced osteomalacia (TIO). TIO is caused by mesenchymal tumours often described as hemangioperitomas, which secrete FGF23 and cause a paraneoplastic renal phosphate wasting syndrome with osteomalacia and muscle weakness. (Carpenter et al., 2020) RNA-seq analysis of FGF23-producing tumours revealed fibronectin (FN)- FGF1 or FN-FGFR1 gene rearrangements, which may activate FGFR1 in an autocrine fashion that stimulates FGF23 production by these tumours. (Lee et al., 2016) Increased KL expression may constitute a second positive feedback loop to increase FGF23 production in TIO tumours, suggesting that the autocrine stimulation of FGF23 and the manifestations of TIO may be KL-dependent, however this requires further study. (Kinoshita et al., 2018)

Non-Canonical FGF23 signalling

KL-independent activation of the MAPK, calcineurin/NFAT, and PLC pathways by FGF23 is essential for the actions of FGF23 other than regulation of renal phosphate excretion and is referred to as ‘non-canonical’ FGF23 signalling. The first evidence for KL-independent effects of FGF23 was the discovery that this hormone can cause LVH in Kl-null mice, which have extremely elevated circulating FGF23 levels. (Faul et al., 2011, Grabner et al., 2015) Changes consistent with LVH in isolated rat cardiomyocytes can be prevented by treatment with inhibitors of PLCy and calcineurin/NFAT. (Faul et al., 2011) It was later shown that this effect of FGF23 is mediated by FGFR4, and in turn, genetic ablation or pharmacological inhibition of FGFR4 reduced the development of LVH in isolated murine cardiomyocytes and rats with CKD, respectively (Fig. 2D). (Grabner et al., 2015) Myocardiocyte-specific conditional ablation of FGFR4 in a mouse model also prevents FGF23-induced LVH. (Han et al., 2020) Interestingly, LVH is not observed in the Hyp mouse model of X-linked hypophosphatemia (XLH), (Liu et al., 2018) which may be because FGF23 in Hyp and XLH is only mildly elevated and because hyperphosphatemia, a possible co-factor for the development of LVH in CKD, is missing in this disorder.

Skeletal muscle like cardiac muscle expresses FGFR1–4, KL, and even FGF23; however, no effect of FGF23 on fibre diameter and ex-vivo excitation-contraction testing was shown. (Avin et al., 2018)

KL-independent signalling was also described in the skeleton. FGF23 inhibits expression of tissue nonspecific alkaline phosphatases (TNAP) (Murali et al., 2016b) in femurs obtained from newborn WT and Kl-null mice (Fig. 2E). Further work using specific FGFR inhibitors showed that TNAP suppression depends on FGFR3 and activation of ERK1/2. (Murali et al., 2016b) Suppression of TNAP is also thought to be a possible mechanism for the mineralization defects seen in XLH, and inhibiting FGF23 or FGF23-FGFR3 binding was able to partially restore TNAP expression and the mineralization defect in Hyp mice. (Murali et al., 2016a) Furthermore the osteoclast markers matrix metalloproteinase 13 (MMP13) and cathepsin K (CTSK), are elevated in Hyp mice, and thought to be one mechanism why treatment with anti-FGF23 antibodies decreased bone resorption. (Tokarz et al., 2018) Further work is needed to show which signalling pathway is involved and if this is a Klotho-independent effect.

FGF23 may also play a role as a negative regulator of erythropoiesis. This role was initially discovered by the observation of increased red blood counts in FGF23-null mice, which may be mediated by increased circulating erythropoietin (EPO) levels. (Coe et al., 2014) Conversely, injection of FGF23 into wild-type mice decreased both EPO and functional erythroid cell counts. (Coe et al., 2014). Elevated levels of FGF23 in patients with CKD who suffer from anaemia may contribute to failure of the kidneys to produce erythropoietin (EPO). (Babitt and Lin, 2012) Thus EPO closes a feedback loop whereby iron deficiency stimulates EPO and FGF23, which in turn inhibits EPO and erythropoiesis. KL and FGFR1, 3, and 4 are highly expressed in erythroid cells. (Coe et al., 2014) Kl-null mice have increased EPO and erythrocyte counts, although hematopoiesis is also impaired in KL deficiency. (Vadakke Madathil et al., 2014) However, whether regulation of EPO by FGF23 is KL-dependent and mediated by one of the FGFR remains to be seen.

FGF23 finally impairs neutrophil activity in partially nephrectomized mice, a model for CKD, that was restored by neutralizing FGF23 or by inhibition of FGFR2. Binding of FGF23 to FGFR2 was shown to be KL-independent in neutrophils, and leads to activation of protein kinase A and repression of the small GTPase Rap1, resulting in suppression of chemokine- and selectin-mediated integrin activation in neutrophils. (Rossaint et al., 2016) Impaired neutrophil activity may be part of the impaired inflammatory responses commonly seen in CKD. (Recio-Mayoral et al., 2011)

Summary and future questions

FGF23 is an endocrine FGF secreted by bone, which stimulates phosphate excretion in the kidneys like PTH, while different from PTH, also inhibits synthesis of 1,25(OH)₂D. Both FGF23 actions lower Pi levels in the circulation and require canonical signalling through FGFR1c and the co-receptor KL. Conversely, FGF23 targets in the heart, parathyroids, bones, and erythrocytes are regulated mainly through FGFR3 and 4 in a KL-independent fashion, which is also referred to as non-canonical FGF23 signalling. FGF23 plays a significant role in the pathophysiology of many phosphate concentration disorders, including CKD, XLH, and TIO. We still need to understand better the functions of FGF23 fragments, and the signal transduction pathways involved particularly for non-canonical FGF23 signalling to treat these disease processes better. It is also unclear why iron and phosphate homeostasis are co-regulated through FGF23 and EPO, and what role inflammation plays for the regulation of FGF23 bioactivity.