INSERM U426 and Institut Fédératif de Recherche Cellules Epithéliales, Faculté de Médecine Xavier Bichat, Paris, France
Keywords: FGF23; phosphatonin; renal phosphate reabsorption
Genetic defects in familial hypophosphataemias associated with low renal phosphate reabsorption
The kidney is the principal organ responsible for phosphate homeostasis. It regulates serum phosphate by modulating phosphate reabsorption. This process takes place in the renal proximal tubule and is regulated by dietary phosphate and PTH. A large amount of evidence designates the sodium-phosphate cotransporter NPT2a activity as the rate limiting and regulated step in phosphate reabsorption [1,2]. Thus, NPT2a has been considered an attractive candidate gene in familial hypophosphataemia associated with low renal phosphate reabsorption. However, no familial hypophosphataemia due to a primary decrease in renal phosphate reabsorption has yet been associated with a defect in the NPT2a gene. X-linked hypophosphataemic rickets (XLH) is associated with mutations in the PHEX gene [3] (Table 1). Mutations in the coding region of the NPT2a gene have been excluded in hereditary hypophosphataemic rickets with hypercalciuria (HHRH) [4,5] (and the gene defect is not yet identified). Mutations in the NPT2a gene have been identified only in unrelated adult patients presenting with hypophosphataemia associated with osteoporosis or renal stone disease (D. Prié, unpublished observation). These observations imply that other genes are involved in the regulation of renal phosphate excretion. These genes could encode for as yet unidentified sodium-phosphate cotransporters. Alternatively, products of these genes could directly or indirectly regulate NPT2a activity. Supporting this possibility, experimental and clinical data derived from XLH and one clinically distinct disorder, tumour-induced osteomalacia (TIO), provided evidence for the presence of a novel hormone activity, termed phosphatonin, that regulates phosphate homeostasis through a PTH-independent mechanism [3,6,7]. Is FGF23a novel factor independently identified by three groups in the last year [810]the long sought after humoral factor phosphatonin?
|
Fibroblast growth factor 23 (FGF23)
In the search for mouse fibroblast growth factor 15 (FGF15) cDNA sequence using the protein sequence as a bait, T. Yamashita et al. [8] isolated a new FGF, which was termed FGF23 (as it was the 23rd documented protein belonging to the FGF family). FGF23 is a 251 amino acid protein with a putative molecular mass of 28 kDa, presenting with a hydrophobic 24 amino acid terminal domain typical of a signal sequence. In agreement with the presence of a signal sequence, it was found to be a secreted protein. Human FGF23 cDNA was subsequently isolated (72% amino acid identity to the mouse sequence), and localized to chromosome 12p13. FGF23 mRNA was detected mainly in the brain and thymus from adult mouse by real-time quantitative PCR. In the brain, it was found to be expressed preferentially in the ventrolateral thalamic nucleus, possibly making FGF23 unique, playing a role in the function of this organ. Approximately at the same time, FGF23 was identified, using a positional cloning approach, as the gene responsible for autosomal dominant hypophosphataemic rickets (ADHR) in humans [9]. In addition, using molecular biological screening, a third team identified FGF23 as a humoral factor secreted by tumours inducing hypophosphataemia and osteomalacia (TIO) [10]. It therefore became apparent that FGF23 could play an important role in the regulation of phosphate homeostasis.
FGF23 and ADHR
ADHR is an autosomal dominant disorder characterized by low serum phosphorus concentration, inappropriately normal 1,25(OH)2D concentrations, rickets, osteomalacia, lower extremity deformities, short stature, bone pain and dental abscesses [11,12]. The clinical manifestations of ADHR are usually severe, and similar but more variable than those observed in XLH. Incomplete penetrance, delayed onset, and loss of phenotype have been reported. Missense mutations that cosegregated with rickets were found in the FGF23 gene in four unrelated families with ADHR [9]. Two families shared the same mutation. These mutations affected one of two closely spaced arginine residues (R176 and R179) present within a potential subtilisin-like proprotein convertase minimum consensus cleavage site (RXXR motif). Hybridization of multiple tissue northern blots was negative for FGF23, but RTPCR analysis demonstrated transcription of FGF23 at low levels in specific tissues, although not in the kidney and bone. These results led to the speculations that FGF23 may be a circulating factor with an important role in phosphate homeostasis and that the ADHR phenotype was caused by a gain-of-function mutation in the FGF23 gene.
FGF23, TIO and XLH
These speculations were corroborated by FGF23 identification as a causative factor of TIO. TIO is a rare paraneoplastic syndrome, first described by McCance in 1947 [6]. The clinical manifestations include bone pain, fractures, fatigue and proximal muscle weakness, with hypophosphataemia caused by low phosphate reabsorption and inappropriately normal 1,25(OH)2D concentrations [7]. The underlying tumours are usually benign; most are of mesenchymal origin (hemangiopericytomas, fibromas, angiosarcomas). When found, resection of the tumour invariably cures the hypophosphataemia, indicating the secretion by the tumour of a humoral phosphaturic factor, historically termed phosphatonin. (Note: octreotide imaging was recently reported to be a valuable diagnostic tool in patients with phosphate wasting but with no family history and with no clinically apparent tumour. In the same report, the authors proposed somatostatin analogue therapy for tumour-induced hypophosphataemia in patients in whom surgery cannot be performed [23]). In support of this hypothesis, studies have shown that tumour extracts or conditioned media from tumoural cells inhibit phosphate transport in vitro, and produce phosphaturia and hypophosphataemia in vivo, although conflicting results were reported. Interest in the pathogenesis of TIO was intensified further when the identification of the gene responsible for XLH reinforced the hypothesis of a circulating phosphaturic factor. By that time, it was clear that the NPT2a gene, located on chromosome 5, was not directly involved in XLH. XLH, a dominant genetic disorder, is the most common form of hereditary renal phosphate wasting. Its hallmark is isolated renal phosphate wasting with inappropriately normal calcitriol concentration, and its clinical and biochemical abnormalities largely overlap with that of both TIO and ADHR. The gene mutated in XLH, named PHEX (phosphate-regulating gene with homology to endopeptidases located on the X-chromosome), encodes a membrane-associated metalloprotease of the M13 family [3], which includes neutral endopeptidase 24.11, endothelin converting enzymes 1 and 2, and the Kell blood group of antigen. Since PHEX is mainly expressed in bone but is not expressed in the kidney, a humoral mechanism has been invoked to explain how mutations in this gene caused hypophosphataemia, low phosphate reabsorption and a decrease in NPT2a activity. The prevailing hypothesis suggested that PHEX normally inactivated the putative humoral phosphaturic factor phosphatonin, and that loss-of-function mutations in PHEX in XLH led to increased circulating levels of phosphatonin.
The phenotypic similarities between ADHR, XLH and TIO were compatible with the possibility that disregulation of the same phosphate regulating pathway was involved in the pathogenesis of these three diseases. Indeed, TIO tumours were found by northern blot analysis to express high levels of FGF23 mRNA, and western blot analysis confirmed the presence of a 32 kDa FGF23 protein [13]. Definite identification of FGF23 as a factor that is secreted by tumours to cause hypophosphataemia was provided by several lines of evidence [10]: (i) FGF23 cDNA was cloned from a haemangiopericytoma that caused TIO; (ii) administration of full-length recombinant FGF23 specifically decreased serum phosphate concentrations in mice within 12 h; and (iii) continuous production of FGF23 in nude mice following implantation of FGF23-producing Chinese hamster ovary (CHO) cells reproduced clinical, biochemical and histological features of TIO. In contrast, continuous in vivo production in mice of dentin matrix protein-1 (DMP1) and matrix extracellular phosphoglycoprotein (MEPE), two factors also produced by TIO tumours, did not cause hypophosphataemia. Analysis of FGF23-derived molecules produced by CHO cells (by detecting the FGF23 C-terminus fragment using western blotting) revealed that two recombinant products were secreted, with molecular masses of approximately 30 and 10 kDa. Following amino-acid sequencing, it was determined that the larger product was a mature FGF23 protein lacking the 24 amino acid signal sequence, while the smaller product had a 180Ser residue at its N-terminus. The preceding amino-acid sequence was the consensus proteolytic cleavage sequence RXXR, Arg176His177Thr178Arg179, supporting the possibility of proteolytic processing. Subsequently it was demonstrated that ADHR mutations in Arg176 or Arg179 stabilize FGF23, as illustrated by the secretion of ADHR-mutated FGF23, which is less sensitive to protease cleavage than wild-type FGF23 [14].
FGF23, a phosphaturic hormone
At this point an attractive plot emerges that, as well as providing new insights into the pathways regulating phosphate homeostasis, may also unravel some mysteries regarding the pathogenesis of TIO, XLH and ADHR. It may be proposed that FGF23 is a phosphate-regulating hormone secreted by one or several tissues, with a native phosphaturic action. The level of FGF23 in the blood would be determined at least in part by the rate of its cleavage by PHEX protease at the Arg179/Ser180 site, inactivating FGF23. Overproduction of FGF23 by tumours (as in TIO), mutations that prevent its cleavage by PHEX (as in ADHR) or mutations that inactivate PHEX (as in XLH) would all increase the level of active FGF23, thereby leading to hyperphosphaturia and hypophosphataemia, and rickets/osteomalacia.
Although attractive, how solid is the theory?
Questions
First, as noted above, a series of evidence points to NPT2a being the sodium-phosphate cotransporter responsible for the bulk of phosphate reabsorption by the renal proximal tubule, and it also points to the fact that NPT2a activity is regulated by dietary phosphate and PTH. However, although FGF23 is expected to inhibit renal sodium-phosphate reabsorption directly, studies addressing this issue in OK cells, a cell line that expresses NPT2a and is sensitive to both PTH and changes in medium phosphate content, led to conflicting and inconclusive results [10,15]. Thus, a direct action of FGF23 on NPT2a still needs further clarification.
Secondly, it remains to be proven for certain that FGF23 is a substrate for PHEX. Similar to the effect of FGF23 on NPT2a activity, studies addressing this issue provided different results [15,16]. In that context, whether phosphataemia regulates the levels of biologically active FGF23 remains to be studied.
Thirdly, the role of FGF23 in vitamin D metabolism, and in the pathogenesis of the inappropriately normal levels of 1,25(OH)2D in TIO, ADHR and XLH, remains to be explored. In Hyp and Gy mice, which both bear mutations in the Phex gene and serve as models for human XLH, inappropriately normal serum 1,25(OH)2D levels are also observed and have been attributed to a defective stimulation of renal 1OHase by hypophosphataemia [17,18]. In addition, decreased mRNA expression of 1
OHase was observed in kidneys isolated from CHO-FGF23 mice [10]. These results support a role (direct or indirect) for FGF23 or PHEX substrate in vitamin D metabolism. It is interesting to note, however, that renal phosphate wasting can be dissociated from abnormal calcitriol synthesis. Indeed, calcitriol synthesis is appropriate to the hypophosphataemic challenge in mice invalidated for the Npt2a gene, which produces levels of phosphate wasting comparable to the inactivation of the Phex gene in Hyp mice [19]. Furthermore, high normal 1,25(OH)2D levels and hypercalciuria are observed in patients with hypophosphataemia associated with heterozygote missense mutations in the NPT2a gene (D. Prié, unpublished results). This phenotype resembles that of heterozygous Npt2a-deficient mice, which exhibit increased urinary phosphate and calcium excretion, elevated plasma 1,25(OH)2D levels, and urolithiasis [20]. These observations suggest that hypophosphataemia due to NPT2a mutations in mice and humans leads to an appropriate regulation of 1
OHase by phosphataemia. Comparison of these phenotypes suggests that FGF23/phosphatonins/PHEX substrates have at least two independent renal effects, inhibition of phosphate reabsorption and impairment of the regulation of calcitriol.
Fourthly, FGF23 may not be the only phosphatonin. In this regard, further investigation is required to define the role of other factors such as MEPE and DMP1, which have also been isolated from tumours associated with hypophosphataemia [10,21]. Alternatively, FGF23 may not be the final phosphatonin, and may stimulate the secretion of a final phosphate-regulating factor. Candidate molecules include stanniocalcin-1 and -2 [22].
The fifth point is that no information is available concerning the FGF23 receptor, so it may be a different receptor than one of the four known ones shared by other FGFs.
The sixth point is that ADHR, XLH and TIO share a similar, but not identical, phenotype. If indeed disregulation of the same phosphate regulating pathway is involved in the pathogenesis of these three diseases, it will be a challenge to elucidate the sources of their phenotypic differences.
Finally, although its action on phosphate homeostasis and renal phosphate excretion represents the first function attributed to FGF23, it does not necessarily follow that it is a phosphate-regulating hormone or that its actions are confined to the kidney. Ultimately, the elucidation of the normal physiological role of FGF23 will help define its importance in biology and medicine.
Notes
Correspondence and offprint requests to: Caroline Silve, Unité de Recherche INSERM U426, 16 rue Henri Huchard, BP416, F-75870 Paris Cedex 18, France. Email: silve{at}bichat.inserm.fr
References