BoneKEy-Osteovision | Commentary
Disordered secretion of FGF23 links three disorders of phosphate transport
Gordon J Strewler
DOI:10.1138/20060225
Commentary on:
Kato K, Jeanneau C, Tarp MA, Benet-Pages A, Lorenz-Depiereux B, Bennett EP, Mandel U, Strom TM, Clausen H. Polypeptide GalNAc-transferase T3 and familial tumoral calcinosis. Secretion of fibroblast growth factor 23 requires O-glycosylation. J Biol Chem. 2006 Jul 7;281(27):18370-7. Liu S, Zhou J, Tang W, Jiang X, Rowe DW, Quarles LD. Pathogenic role of Fgf23 in Hyp mice. Am J Physiol Endocrinol Metab. 2006 Jul;291(1):E38-49.
Disordered processing of FGF23 leads to two mirror-image disorders. Mutations that impair O-glycosylation (), as well as several missense mutations in the FGF23 sequence (), lead to defective secretion of FGF-23, giving rise to familial tumoral calcinosis (OMIM 211900), a disorder characterized by hyperphosphatemia and massive deposition of calcium phosphate in soft tissues. The converse, excessive levels of FGF23 with phosphate wasting and rickets, occurs in autosomal dominant hypophosphatemic rickets because of defective cleavage of FGF23 at a prohormone convertase site (). A recent paper () ties the two disorders directly together by showing that O-glycosylation of FGF23 occurs at the proprotein convertase cleavage site and prevents cleavage. The results indicate that this cleavage site is a control point for regulation of phosphate and vitamin D metabolism by FGF23. In another recent paper (), the Hyp mouse model of X-linked hypophosphatemic rickets is also linked to disordered control of FGF23 secretion by the osteocyte.
To develop a model of glycosylation of FGF23, Kato et al. expressed FGF23 in the mutant CHO cell line IdlD, which is deficient in an epimerase which catalyzes the reaction from UDP-glucose to UDP-galactose. The defect can be complemented by addition of GalNac plus Gal. Secretion of FGF23 from CHO IdlD cells required O-glycosylation, as demonstrated by the requirement for addition of GalNAc and Gal to the mutant cells. Secretion of FGF23 was dramatically enhanced by coexpression of GalNAc-T3, the glycotransferase that was implicated in familial tumoral calcinosis (). Thus, GHO IdlD cells do not produce an endogenous GalNAc-T3 with specificity for FGF23.
Next, the acceptor substrate site for O-glycosylation was identified by MALDI-TOF of model peptides from FGF23 that encompass the prohormone convertase site. Several recombinant GalNAc transferases targeted Thr171, but only GalNAc-T3 incorporated GalNAc into Thr178. The results agree with previous studies by Shimada et al. in which Thr178 and Ser180 were shown as the most likely sites of O-glycosylation of a similar peptide ().
Thr178 is within the prohormone convertase site R176HTR179, raising the possibility that O-glycosylation of Thr178 affects the susceptibility of FGF23 to cleavage. In experiments to test this possibility, Kato et al. showed that several glycopeptides modified at Thr178 could be efficiently cleaved by furin in vitro, but mono- and disialylated peptides were completely resistant to furin cleavage, suggesting that the major inhibitory effect of O-glycosylation is exerted by α2,6-sialylation of GalNAc (NeuAcα2-6GalNAcα1-O-Thr178). Interestingly, however, α2,6-sialylation of O-glycans is rarely complete.
The results identify a novel mechanism for regulation of secretion of FGF23, in which competition between proteolytic cleavage and O-glycosylation determines the rate of secretion of FGF23. In the absence of glycosylation, secretion of FGF23 is virtually abolished because of efficient proteolysis of the full length molecule. Were glycosylation completely efficient, on the other hand, it is likely that mutations in R176 or R179 would not produce a phenotype. The fact that mutations in these arginines lead to FGF23 excess and a syndrome of phosphate wasting indicates that competition between proteolysis and glycosylation is indeed active in humans, and also suggests that the derangement in synthetic control that results from arginine mutations outflanks whatever homeostatic mechanisms regulate secretion of FGF23. That is, in general an increase in the efficiency of secretion of a hormone might be expected not to create a syndrome of hormone excess because negative feedback regulation should compensate, but negative feedback does not prevent phosphate wasting in autosomal dominant hypophosphatemic rickets.
The role of the C-terminal domain of FGF23 (aa180-251), which is unique among the FGFs, may be in part to stabilize FGF23 during biosynthesis and secretion. The C-terminal domain could have additional functions, e.g. to stabilize FGF23 in the circulation, or to bind klotho, a step that appears to be essential for the biological activity of FGF23 (). It appears that other FGFs that lack this domain also interact with klotho, however. Another possibility is that the C-terminal domain, which accumulates in the blood of individuals with familial tumoral calcinosis, is itself a feedback signal to FGF23 synthesis.
A recent paper by Liu et al. () examines FGF23 secretion from bone cells. To understand better the regulation of Fgf23 gene transcription at the cellular level, Liu et al. created a mouse reporter strain in which green fluorescent protein (eGFP) was knocked in to replace exon 1 of the FGF23 coding sequence. In these reporter mice, cells that express Fgf23 gene transcripts are green, but the Fgf23 gene is inactivated. The resulting Fgf23(+/-) and Fgf23(-/-) mice had phenotypes very similar to previous reports of removal of the Fgf23 gene ().
The Fgf23(-/+) reporter mice were crossed with Hyp mice. Hyp is a mouse model of X-linked hypophosphatemia (XLH) caused by loss-of-function mutations in a cell surface protease, Phex () and characterized by high levels of FGF23 (). In keeping with earlier reports (), the phenotype of the double knockout mice resembled that of Fgf23(-/-), with hyperphosphatemia and high levels of 1,25(OH)2D. Thus, FGF23 is required in order for the Hyp mutation to affect phosphate and vitamin D metabolism, placing Fgf23 genetically downstream of Hyp. Rickets was also abolished in double knockout mice, consistent with a role of hypophosphatemia and/or low levels of 1,25(OH)2D in the pathogenesis of rickets. The double knockout mice did not have excessive osteoid, as do Hyp mice, but mineralization was virtually absent. Therefore, FGF23 lack and the absence of Phex independently impair mineralization.
Reporter mice expressed Ffg23 at two bone sites, osteocytes and CD31-positive endothelial cells in venous sinusoids of bone, but not in surface osteoblasts. A previous in situ hybridization study () localized FGF23 in osteocytes and venous sinusoids; active osteoblasts at fracture healing sites also expressed the FGF23 gene. The reporter construct was also expressed in the thymus and ventrolateral hypothalamus. Introduction of the Hyp mutation into reporter mice led to a striking increase in Fgf23 expression in osteocytes, but had no effect on expression in venous sinusoids. The increase in osteocyte FGF23 expression was greatest in Fgf23(-/-)/Hyp, indicating that expression of FGF23 is not required in order for Fgf23 gene transcription to be upregulated by the absence of Phex. This result puts to rest the attractive idea that FGF23 levels are high in Hyp and XLH because processing of FGF23 by Phex is impaired. Although intact FGF23 is not a substrate for Phex (), internal Phex cleavage sites have been demonstrated in the FGF23 sequence, at least one of which would be exposed by prohormone convertase cleavage at R176HTR179 (), but the results of Liu et al. essentially rule out models of the pathogenesis of Hyp (or XLH) in which FGF23 peptides feed back to upregulate gene transcription.
Instead, the results of Liu et al. favor models in which Phex inhibits Fgf23 gene transcription in osteocytes by cleaving a bone matrix substrate in their vicinity. Based on abstracts submitted to the 2006 Annual Meeting of the American Society for Bone and Mineral Research, dentine matrix protein 1 is a good candidate for the phex substrate, as its absence also leads to renal phosphate wasting.
The observation that Fgf23(-/-) mice have markedly impaired mineralization (), despite dramatic hyperphosphatemia, raises the possibility that FGF23 has a local role in the control of mineralization, conceivably by regulating calcium or phosphate flux across mineralizing osteoblasts. Although Liu et al. prepared mineralizing bone nodules from Fgf23(-/-) mice (), they did not comment on the ability of FGF23-null cells to mineralize their matrix in vitro. Alternatively, abnormalities of the extracellular milieu, such as the high levels of 1,25(OH)2D, could be responsible for the mineralization disorder in Fgf23(-/-) mice, as suggested by data in a recent review by Razzaque et al. ().
Together, the work of Kato et al. and Liu et al. implicate disordered synthesis and secretion of FGF23 in three independent disorders of phosphate and vitamin D homeostasis. Osteocytes, now featured players in bone remodeling as well as mechanotransduction, are evidently featured players in phosphate and vitamin D homeostasis as well. Who would have thought?
This work is licensed under a
Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.