Skip to main content
NCBI home page
The Journal of Clinical Endocrinology and Metabolism logoLink to The Journal of Clinical Endocrinology and Metabolism
. 2009 Oct 16;94(11):4267–4274. doi: 10.1210/jc.2009-0961

Defective O-Glycosylation due to a Novel Homozygous S129P Mutation Is Associated with Lack of Fibroblast Growth Factor 23 Secretion and Tumoral Calcinosis

Clemens Bergwitz 1, Santanu Banerjee 1, Hilal Abu-Zahra 1, Hiroshi Kaji 1, Akimitsu Miyauchi 1, Toshitsugu Sugimoto 1, Harald Jüppner 1
PMCID: PMC2775647  PMID: 19837926

Abstract

Background: Homozygous mutations in fibroblast growth factor (FGF23) have recently been described as the genetic cause of one form of hyperphosphatemic tumoral calcinosis (HFTC). However, it remained unclear to date how these mutations lead to loss of biologically active FGF23 in the circulation.

Methods: We here report a novel homozygous mutation, c.385T>C in FGF23 exon 2, which changes codon 129 from serine to proline (S129P) in a previously described individual affected by HFTC. The S129P mutation as well as two known FGF23 mutations, S71G and S129F, were introduced into an expression vector encoding wild-type (wt) human (h) FGF23 to yield [P129]hFGF23, [F129]hFGF23, and [G71]hFGF23; whole lysates, glycoprotein fractions, and conditioned media from HEK293 and COS-7 cells expressing these constructs were subjected to Western blot analysis using affinity-purified goat anti-hFGF23(51-69) and anti-hFGF23(206-222) antibodies.

Results: We detected 25- and 32-kDa protein species in total lysates of HEK293 cells expressing wt-hFGF23. The 32-kDa band, representing O-glycosylated hFGF23, was not detectable in the glycoprotein fraction of lysates from HEK293 cells expressing [P129]hFGF23, and in comparison with wt-FGF23 only small amounts of [P129]hFGF23 were secreted into the medium. Similar results were obtained for cells expressing [G71]hFGF23 and [F129]hFGF23.

Conclusion: Our data for the first time directly show that FGF23 mutations associated with HFTC impair O-glycosylation in vitro resulting in poor secretion of the mutant hormone thereby explaining the characteristic hyperphosphatemic phenotype of homozygous carriers in vivo.

Tumoral calcinosis is caused by missense mutations in FGF23 which impair O-glycosylation of the mutant hormone, leading to poor secretion and loss-of-function.

The term hyperphosphatemic familial tumoral calcinosis (HFTC) (Online Mendelian Inheritance in Man 211900), refers to several related disorders first described by Giard (1898) (1) and Duret (1899) (2). Most affected patients show similar clinical and biochemical abnormalities, namely inappropriately enhanced renal tubular absorption of phosphate leading to hyperphosphatemia. Increased activity of the renal 1-α-hydroxylase furthermore results in an elevated serum 1,25 dihydroxyvitamin D3 level, which leads to increased intestinal absorption of calcium and phosphate. The resulting increase in serum calcium-phosphate product is thought to cause the characteristic calcium-phosphate deposits in different tissues. Osteogenic cells and osteoid formation are absent, which distinguishes HFTC from disorders of heterotopic ossification (3).

Considerable progress has been made in determining the molecular causes of the different forms of HFTC, which for the purpose of this article are referred to as types 1, 2, and 3. The first form to be defined at the molecular level was HFTC type 1 (HFTC1), which is caused by homozygous loss-of-function mutations in the gene encoding uridine diphosphate-N-acetyl-α-d-galactosamine-polypeptide N-acetylgalactosaminyl-transferase 3 (GALNT3) (4). GALNT3 is responsible for O-glycosylation and proper secretion of intact fibroblast growth factor 23 (FGF23), as recently shown using CHO cells deficient in an enzyme required for O-glycosylation (5). The second form, HFTC2, is caused by homozygous missense mutations in FGF23 that also lead, through yet-unknown mechanisms, to impaired secretion of the intact hormone. HFTC1 and HFTC2 are both characterized by low or undetectable circulating levels of intact FGF23, whereas the levels of C-terminal FGF23 fragments are generally elevated (6,7,8).

A third form of HFTC, HFTC3, was recently reported in a 13-yr-old girl with a disorder resembling HFTC1 and HFTC2, who had extremely high circulating levels of intact and C-terminal FGF23. A homozygous, presumably inactivating mutation in α-KLOTHO, the co-receptor for FGF23, was identified in this individual (9).

FGF23 is a glycoprotein secreted by osteocytes and osteoblast in the skeleton that acts on the renal proximal tubules to decrease expression of the sodium-phosphate co-transporters NPT2a and NPT2c (10,11,12), thus reducing reabsorption of phosphate from the glomerular filtrate (13,14,15). Recent work using neutralizing anti-FGF23 antibodies indicates that the N-terminal portion of FGF23 interacts with FGF receptor 1c, whereas the C-terminus binds to the co-receptor Klotho, and both interactions appear to be important for FGF23 bioactivity in vitro and in vivo (16). FGF23 is O-glycosylated in the 162–228 region (17), and O-glycosylation appears to protect FGF23 from cleavage by subtilisin-like proprotein convertases when using recombinant peptides in vitro (5) and after expression in CHO cells (18). Although residues 162–228 are important for the biological activity of FGF23, their deletion does not affect secretion (19). Furthermore, all HFTC2 mutations identified to date reside in the N-terminal portion of FGF23; these include H41Q (20), Q54K (19), S71G (6,21), M96T (22), and S129F (23). The S71G mutation is associated with accumulation of mutant FGF23 in the Golgi apparatus (6), and secretion of Flag-tagged versions of [G71]hFGF23 and [F129]hFGF23 by HEK293 cells can be rescued by lowering the culture temperature to 25 C or when compounded with R176Q (8), a mutation that appears to stabilize human FGF23 (hFGF) by protecting a subtilisin-furin cleavage site (5,18).

We here describe a novel homozygous missense mutation in FGF23, S129P, that was identified in a previously reported HFTC patient (24). We further show that the S129P modification impairs O-glycosylation of the mutant FGF23 protein when expressed in HEK293 or COS-7 cells, thus leading to severely impaired secretion of [P129]hFGF23. Similar results were obtained for [G71]hFGF23 and [F129]hFGF23, suggesting a common mechanism for three mutations within the N-terminal region of FGF23 that lead to HFTC2. These findings establish that this variant of HFTC is a disorder of abnormal O-glycosylation (25).

Patients and Methods

Patients

The Japanese index case was previously reported by Yamaguchi et al. (24), and institutional review board approval and informed consent had been obtained before conducting genetic studies. He presented at age 10 months with periarticular eruptions displaying chalky drainage (44% calcium phosphate, 12% calcium carbonate, 44% protein). His laboratory studies at presentation showed hyperphosphatemia (7.0 mg/dl; normal 2.5–4.5), normal serum calcium (9.7 mg/dl; normal 8.5–10.5), and mildly impaired renal function (creatinine, 0.8 mg/dl; blood urea nitrogen, 8 mg/dl, creatinine clearance 88 ml/min.), normal immunoreactive PTH (0.5 ng/ml; normal <0.546), normal alkaline phosphatase (70 U/liter; normal 36–100), normal 25 hydroxyvitamin vitamin D (20 ng/ml; normal 14–42), and normal 1,25 dihydroxyvitamin D3 (44 pg/ml; normal 20–60). The urinary 24-h excretion of calcium was normal (70.8 mg/d; normal 50–400), with a urinary calcium to creatinine ratio of 0.079 mg/mg (normal <0.2); the tubular threshold of phosphate absorption was 8.2 mg/dl (normal 2.5–4.5). Treatment consisted in aluminum hydroxide and acetazolamide to reduce intestinal and renal phosphate absorption, respectively, which led to slow remission of his periarticular lesions. His intact FGF23 level, which was measured in 2004 while on medication, was 10–38 pg/ml (normal range 26 ± 8.4; Kainos Laboratories, Tokyo, Japan). His C-terminal FGF23 level was 15638 ± 4149 RU/ml (normal range 50 ± 51; Immutopics, San Clemente, CA). Despite continuation on aluminum hydroxide, dietary phosphate restriction to 500 mg/d, and acetazolamide treatment until the present time, he suffered from painful sc calcifications at the hip as well as renal calcifications with mild impairment of his renal function (serum creatinine is now 1.3 mg/dl).

Mutational analysis

The coding exons of FGF23, including the intron/exon borders, was amplified by PCR (PCR kit; QIAGEN, Valencia, CA) in four overlapping fragments using genomic DNA isolated from peripheral lymphocytes of the index case. The obtained PCR products were purified using spin columns (QIAGEN), and nucleotide sequence analysis was performed using 20 ng per 100 bp DNA and nested forward or reverse primers (100 ng each) at the Massachusetts General Hospital DNA Sequencing Core Facility. Numbering of the sequence abnormalities found in FGF23 is according to cDNA accession no. NM_020638.2 starting at the initiator ATG.

In vitro expression of mutant and wild-type hFGF23

Mutations were introduced into the pcDNA3.1 mammalian expression vector containing the human cDNA encoding FGF23 using the Quickchange method (Stratagene, La Jolla, CA). Plasmid DNA was transfected into HEK293 cells using Effectene (QIAGEN). Medium was changed to serum-free medium after 24 h, and cells and medium were harvested after 48 h for Western blot analysis. Intact and C-terminal FGF23 levels were determined in cell culture supernatants by ELISAs (Immutopics).

Western blot analysis

Cells were rinsed once with PBS and lysed with Triton X-100 buffer containing protease inhibitors and stored at −70 C until further analysis. For some experiments medium was concentrated 50 times by Macrosept-10K-omega (Pall, Port Washington, NY) and subjected to dialysis against 1× PBS and lyophilization. Forty-two microliters of cell lysates or medium were combined with 8 μl 6× SDS-PAGE loading buffer, heated to 99 C × 5 min, loaded on 15% SDS-PAGE, electroblotted onto polyvinyl difluoride (PVDF; Millipore, Bedford, MA) membrane, blocked with 5% nonfat milk per 0.5% Tween 20 per 1× PBS at 4 C for 4 h or overnight, hybridized with a goat anti-human FGF23 (51-69) or a goat anti-human FGF23 (206-222) anti-body (kindly provided by Immutopics), and developed using the Vector ABC anti-goat detection system (Vector Labs, Burlingame, CA)/Western Lightening chemiluminescence reagent (PerkinElmer, Norwalk, CT).

Enzymatic deglycosylation experiments

Ni-Agarose chromatography was used to purify recombinant [Q176]hFGF23(25-251)V5His in pcDNA3.1V5His from medium of Free-Style 293-F cells (Invitrogen, Carlsbad, CA), and recombinant (presumably nonglycosylated) [M24, Q176] hFGF23(24-251)V5His in pCRT7/CT-Topo from the BL21 TrxB bacterial strain. Approximately 1 μg of each recombinant peptide was then subjected to fractionated digestion to sequentially remove N-linked and O-linked carbohydrates under denaturing conditions and using the E-DEGLY kit (Sigma, St. Louis, MO). The expected fragment lengths are: pcDNA3.1-hFGF23(25-251) expressed in HEK293 cells, 227 amino acids approximately 25 kDa; recombinant [Q176]hFGF23(25-251)V5His (45 additional amino acids) purified from Free-Style 293-F cells, 272 amino acids approximately 30 kDa; and recombinant [M24, Q176]hFGF23(24-251) V5His (30 additional amino acids) purified from the BL21 TrxB bacterial strain, 258 amino acids approximately 28 kDa.

Wheat germ lectin purification of HEK293 glycoprotein fraction

Three days after transfection with pcDNA3.1 containing mutant and wild-type hFGF23(25-251), lysates were subjected to purification of the glycoprotein fraction using wheat germ-lectin agarose according to the manufacturer’s instructions. Forty-two microliters total lysates and glycoprotein fractions were combined with 6× SDS-Page loading buffer, heated at 99 C × 5 min, and run on 15% SDS-PAGE, and electroblotted onto PVDF membranes (Millipore). The PVDF membranes were blocked with 5% nonfat milk/0.5%Tween 20/1× PBS, hybridized with a biotinylated goat anti-hFGF23 (186-206) antibody at 4 C overnight, and developed using the Vector ABC antigoat detection system (Vector Labs)/Western Lightening chemiluminescence reagent (PerkinElmer).

Results

Mutational analysis identifies a novel homozygous missense mutation S129P in FGF23

Genomic DNA isolated from peripheral blood leukocytes was subjected to PCR using primers positioned to amplify all three exons and the exon-flanking intronic sequences as shown in Fig. 1A. Direct nucleotide sequence analysis of the obtained PCR products revealed a homozygous transition C>T at nucleotide 385, leading to a serine to proline change at codon 129 (Fig. 1B), which is highly conserved in FGF23 among the available species (Fig. 1C). S129P was not found in 174 control chromosomes. The unaffected mother of the index case is heterozygous for this mutation (the father was not available for analysis).

Figure 1.

Figure 1

Open in a new tab

A, Primers flanking exons 1–3 of FGF23 were used to PCR amplify the entire coding sequence and exon-flanking intronic sequences using blood genomic DNA from the index case. B, Nucleotide sequence analysis revealed a homozygous transition C>T at nucleotide 385, leading to a serine to proline change at codon 129, which is highly conserved among the available species (C).

Expression analysis of wild-type (wt) and mutant forms of FGF23 in HEK293 and COS-7 cells

The S129P mutation and the previously reported tumoral calcinosis mutations S71G (6,21) and S129F (23) were introduced individually into the expression vector pcDNA3.1V5His or along with the autosomal dominant hypophosphatemic rickets (ADHR) mutation T176Q. This vector uses the native signal sequence of FGF23 but introduces a C-terminal V5His-tag to permit purification for deglycosylation experiments (see below). Because it is possible that the V5His-tag modifies expression and secretion of FGF23, we reintroduced the native termination codon (designated Stop) into those plasmids used for expression analysis in HEK293 or COS-7 cells when purification of the peptides was not required. Western blot analysis of COS-7 cell lysates transiently transfected with wt-hFGF23(1-251)-Stop-pcDNA3.1V5His showed 25 and 32 kDa protein species, which were detected by antibodies raised against amino acids 51-69 and 206-222, respectively, of hFGF23. The 32-kDa protein species was secreted into the medium along with N- and C-terminal fragments (Fig. 2, A and B). However, these low-molecular-weight bands tended to be faint, which may be related to their degradation during cell culture. Alternatively, folding of these FGF23 fragments may have affected recognition by the antibodies before or during SDS-PAGE or during electroblotting or immobilization. The 25- and 32-kDa protein species were both secreted by COS-7 cells expressing [Q176]hFGF23(1-251)Stop, but significant amounts of peptide were lacking in medium conditioned by COS-7 cells expressing [G71]hFGF23(1-251)Stop, [P129]hFGF23(1-251)Stop, or [F129]hFGF23(1-251)Stop. Similar results were obtained in HEK293 cells (not shown).

Figure 2.

Figure 2

Open in a new tab

COS-7 cells were transiently transfected with [Q176] hFGF23(1-251)Stop, [G71]hFGF23(1-251)Stop, [P129]hFGF23(1-251)Stop, or [F129]hFGF23(1-251)Stop in pcDNA3.1V5His. After 2 d, cultures were switched to serum-free medium, and medium was harvested and cells were lysed on d 3. Lysates and medium were subjected to 15% SDS-PAGE, transferred to PVDF membranes, and probed with goat anti-hFGF23(51-69) (A) or goat anti-hFGF23 (206-222) antibodies (B). COS-7 cell lysates and media lacked the 32-kDa modified protein species (star) when expressing the FGF23 variants carrying the tumoral calcinosis mutations S71G, S129P, or S129F; in contrast, the 32-kDa species was observed when analyzing cells expressing the wt-FGF23. The 25-kDa protein species (arrow) was present in lysates for mutant and wt-hFGF23 but absent in the respective media.

To determine, whether one of the ADHR mutations, Q176, can rescue the mutant FGF23 identified in tumoral calcinosis, we constructed expression plasmids encoding for [G71, Q176]hFGF23(1-251)Stop, [P129, Q176]hFGF23(1-251)Stop, or [F129, Q176]hFGF23(1-251)Stop in pcDNA3.1V5His. Lysates of COS-7 cells and conditioned medium were subjected to Western blot analysis, and PVDF membranes were again probed with goat anti-hFGF23(51-69) (Fig. 3A) or goat anti-hFGF23(206-222) antibodies (Fig. 3B). Q176 failed to rescue secretion of all three tumoral calcinosis mutations, yet two novel intracellular N- and C-terminal FGF23 species of approximately 14 kDa were observed for the double mutants. The ELISA-assays detecting intact FGF23 alone, or intact and C-terminal FGF23, respectively, likewise failed to detect double-mutant FGF23 in conditioned media (Fig. 3C). Different units for the standard used in these assays make a direct molar comparison of secreted intact and C-terminal amounts difficult. However, relative to wild-type FGF23, secretion of the mutant forms of FGF23 appeared to be similarly reduced when assessed by either assay.

Figure 3.

Figure 3

Open in a new tab

COS-7 cells were transiently transfected with pcDNA3.1V5His plasmids encoding hFGF23(1-251)Stop comprising one of the following mutations: [Q176], [G71], [P129], [F129], [G71, Q176], [P129, Q176], [F129, or Q176]. After 2 d, cultures were switched to serum-free medium, and medium was harvested and cells lysed on d 3. Lysates and medium were subjected to 15% SDS-PAGE, transferred to PVDF membranes, and probed with goat anti-hFGF23(51-69) (A) or goat anti-hFGF23(206-222) antibodies (B). COS-7 cell lysates and media lacked the 32-kDa modified protein species (star) when expressing the FGF23 variants carrying the tumoral calcinosis mutations S71G, S129P, S129F, or double mutations; the 25-kDa protein species (arrow) is present in lysates expressing mutant and wt-hFGF23 but undetectable in the respective media. Note that a nonspecific protein band (arrowhead) was detected in medium but not in cell-lysates. Two novel 14-kDa N- and C-terminal protein species are observed for double-mutant FGF23. C, Intact and C-terminal FGF23 ELISA (Immutopics) of conditioned media from HEK293 cells expressing the above wild-type and mutant forms of hFGF23. C-terminal and intact FGF23 are similarly reduced when compared with wild-type. Q176 is unable to rescue secretion of hFGF23.

To further elucidate the mechanism by which the tumoral calcinosis mutations impair secretion of FGF23, we examined their impact on glycosylation of the peptide. Enzymatic deglycosylation of recombinant wt-[Q176]hFGF23(25-251)V5His purified from medium conditioned by Free-Style 293-F cells with 1,4-β-galactosidase, N-acetyl-glycosidase, sialidase, or O-glycosidase, but not N-glycosidase F (PNGaseF), resulted in a shift in molecular weight, whereas recombinant [M24, Q176]hFGF23(24-251)V5His in pCRT7/CT-Topo purified from the BL21 TrxB bacterial strain showed no change in size (Fig. 4A). This finding is consistent with O-glycosylation of wt-hFGF23(25-251) and suggests that the longer 32-kDa protein species observed in lysates of HEK293 cells is O-glycosylated. To directly show that O-glycosylation of FGF23 is impaired by the tumoral calcinosis mutations, we isolated the glycoprotein fractions from HEK293 cells using wheat germ lectin agarose. wt-FGF23(1-251) was detected in the glycoprotein fraction and whole lysates of transfected HEK293 cells. Conversely, the tumoral calcinosis mutants, although equally well detected in whole lysates, were not found in the glycoprotein fraction (Fig. 4B). Small amounts of the presumably O-glycosylated 32-kDa mutant protein species could be detected in the medium after it had been concentrated 50-fold, but the 25-kDa protein species was not observed in the medium despite equal expression in whole lysates (Fig. 4C). The shorter N- and C-terminal fragments of FGF23, which had been observed earlier in cells transfected with the same plasmids (Figs. 2 and 3), were not found, which may be related the 10-kDa cutoff of the spin columns that were used to concentrate conditioned media. These results suggest that the tumoral calcinosis mutations prevent secretion of FGF23 by blocking O-glycosylation.

Figure 4.

Figure 4

Open in a new tab

A, Enzymatic deglycosylation of recombinant FGF23: Ni-agarose chromatography was used to purify: recombinant (presumably glycosylated) [Q176]hFGF23(25-251)V5His in pcDNA3.1V5His from medium of Free-Style 293-F cells (Invitrogen) and recombinant (presumably nonglycosylated) [M24, Q176]hFGF23(24-251)V5His in pCRT7/CT-Topo from the BL21 TrxB bacterial strain. N-linked and O-linked carbohydrates were removed under denaturing conditions using approximately 1 μg of peptide and the E-DEGLY kit (Sigma). Expected glycan-free fragment lengths are: recombinant [Q176]hFGF23(25-251)V5His (45 additional amino acids) purified from Free-Style 293-F cells, 272 amino acids approximately 30 kDa; and recombinant [M24, Q176]hFGF23(24-251)V5His (30 additional amino acids) purified from the BL21 TrxB bacterial strain, 258 amino acids approximately 28 kDa. Immunoblot analysis was performed as described in Patients and Methods and in the previous figures. B, Glycoprotein purification of total lysates from HEK293 cells expressing [G71]hFGF23(1-251)Stop, [P129]hFGF23(1-251)Stop, [F129]hFGF23(1-251)Stop, or wt-hFGF23(1-251)Stop using WGL-agarose, followed by 10% SDS-PAGE analysis using a biotinylated goat anti-hFGF23(186-206) antibody. Whereas whole lysates of HEK293 cells expressing mutant and wt-hFGF23 showed equal expression of the 25-kDa protein species, only the wt-FGF23 32 kDa protein species was detected in the glycoprotein fraction after purification with wheat germ lectin agarose (star). C, A 32-kDa, presumably glycosylated, but not the more abundant intracellular 25-kDa protein species was detected after 50 times concentration of media conditions with HEK293 cells transfected with the tumoral calcinosis mutations of hFGF23 (star).

Discussion

We here describe a novel homozygous missense mutation in FGF23 in a previously reported individual affected by tumoral calcinosis (24). His healthy mother is a heterozygous carrier, consistent with an autosomal recessive mode of inheritance, and the mutation was not found in 174 control chromosomes. Furthermore, a mutation affecting the same codon, S129F, was previously reported in an unrelated individual with tumoral calcinosis (23). Thus, S129P likely causes hyperphosphatemic tumoral calcinosis in our patient.

The mechanism by which S129P and other HPTC2 mutations lead to loss of function of FGF23 is unclear. Consistent with the patient’s low circulating intact and elevated C-terminal FGF23 levels, HEK293 or COS-7 cells showed poor secretion of intact [P129]hFGF23 (Figs. 2, A and B and 3, A and B). Interestingly, total lysates of transfected HEK293 or COS-7 cells showed normal intracellular levels of 25-kDa mutant FGF23, whereas the longer 32-kDa protein species observed after transfection with wild-type FGF23 (Fig. 2) was not detectable. Similar results were obtained with two other tumoral calcinosis mutations, [G71]hFGF23 and [F129]hFGF23, suggesting that a similar mechanism underlies the poor secretion of mutant FGF23.

The 32-kDa protein species detected with antibodies raised against N-terminal and C-terminal portions of FGF23 is larger than predicted from the amino acid sequence of hFGF23(25-251) (Figs. 2, A and B, and 3, A and B), suggesting that wild-type FGF23 is modified and that this modification is lost in patients carrying one of the FGF23 mutations leading to tumoral calcinosis. Intact FGF23 produced in mammalian HEK293 cells was deglycosylated by treatment with O-glycosidase but not with PNGaseF (Fig. 4A). In contrast, FGF23 produced in bacteria, which is presumably nonglycosylated, was smaller in size and enzymatic treatment did not lead to a detectable change in molecular weight. These data indicate that recombinant wild-type FGF23 purified from HEK293 cells is O-glycosylated. The shifts in apparent molecular weight that were induced by enzymatic deglycosylation are further consistent with the observed 7-kDa difference between the 25- and 32-kDa protein species.

We next sought to directly show that the novel and known tumoral calcinosis mutations impair O-glycosylation of FGF23. For this purpose, we isolated glycoprotein fractions from HEK293 lysates expressing wild-type FGF23 using wheat germ lectin agarose. Only the 32-kDa protein species was present in the glycoprotein fraction, whereas both the 25-kDa (presumably unmodified) and 32-kDa protein species were both present in total lysates. These findings indicate that the 32-kDa protein species is indeed modified by a glycan moiety. The 32-kDa protein species was not detectable by Western blot analysis in the glycoprotein fractions of HEK293 cells expressing the mutant FGF23s, indicating that O-glycosylation is impaired by the tumoral calcinosis mutations.

Our experiments did not address the mechanism underlying impaired O-glycosylation of the HFTC2 mutants. However, previous reports indicated that O-glycosylation occurs within the C-terminal 162–228 portion of FGF23 (17), whereas all HFTC2 mutations identified to date reside in the N-terminal portion of FGF23; in fact, some of these mutations, such as H41Q (20) and Q54K (19), do not even affect potential sites for O-glycosylation. Thus, HFTC2 mutations likely cause misfolding of FGF23, which delay or impair O-glycosylation, as suggested by Larsson et al. (8).

It is possible that the lack of O-glycosylation is a consequence or the cause of poor secretion of the mutant FGF23 by HEK293 cells. The former scenario would likely result in secretion of large quantities of unmodified 25-kDa FGF23 into the medium of cells transfected with mutant forms of FGF23s. However, even after 50 times concentration, the 25-kDa protein species was absent in supernatants collected from HEK293 cells expressing mutant FGF23 (Fig. 4C). Instead, small amounts of the 32-kDa protein species were observed in concentrated medium, indicating that O-glycosylation occurs, albeit inefficiently for the mutant FGF23s, and that mutant FGF23 can be secreted once glycosylated.

Interestingly, secretion of a partially glycosylated 25-kDa FGF23 protein species can be observed after introduction of the ADHR mutation Q176 into wt-hFGF23, which presumably protects the partially glycosylated protein against cleavage by furin/subtilisin endopeptidases (18). Yet HEK293 cells failed to secrete hFGF23 carrying both mutations, suggesting that tumoral calcinosis mutations either completely abolish glycosylation or that they impair secretion through other mechanisms, possibly involving degradation. Consistent with the latter hypothesis, we observed the intracellular accumulation of two novel N- and C-terminal 14-kDa protein species (Fig. 3, A and B), which may result from cleavage at alternative sites within FGF23. Furthermore, these fragments may be unstable unless the RXXR motif is protected by the introduction of Q176. Larsson et al. (8) were able to rescue secretion of [G71]hFGF23 and [F129]hFGF23 by introducing Q176 and lowering the temperature at which cells are cultured. When compared with our expression plasmids, their vectors introduced an artificial signaling sequence containing an N-terminal Flag tag, which may have favored secretion under these conditions, further indicating that secretion and degradation of FGF23 are intricately linked. However, our data suggest that degradation is the consequence rather than the cause of impaired secretion because intact FGF23 is the dominant protein species in COS-7 or HEK293 cell lysates and wild-type and mutant FGF23 are observed in similar quantities.

Based on the above findings, we would like to suggest a working model to illustrate the role of O-glycosylation by GALNT3 for proper secretion of hFGF23 (Fig. 5A). When O-glycosylation is impaired by loss-of-function mutations in GALNT3 or HFTC2 mutations in FGF23, secretion of intact hFGF23 may be impaired. Nonglycosylated FGF23 may be targeted for degradation by subtilisin/furine-like endopeptidases at the RXXR motif and possibly novel midprotein cleavage sites and inactive fragments of the mutant hormone may spill over or may be actively secreted into the circulation (Fig. 5B). The ADHR mutation Q176, on the other hand, may protect glycosylated and partially glycosylated intact FGF23, leading to intracellular accumulation and excess secretion of intact FGF23. By protecting the subtilisin/furine-like cleavage site, the ADHR mutation may also lead to the intracellular accumulation and secretion of an otherwise unstable 25-kDa protein (Fig. 5C). This 25-kDa form of FGF23 may be biologically active, as shown with bacterial recombinant (nonglycosylated) FGF23 (26). However, it appears to escape a reduction in synthesis and secretion, as would be expected as a normal counterregulatory measure in the setting of hypophosphatemia and low 1,25 dihydroxyvitamin D3 levels, which are commonly observed in patients with ADHR.

Figure 5.

Figure 5

Open in a new tab

Working model to illustrate the role of O-glycosylation by GALNT3 for proper secretion of hFGF23 (A). When O-glycosylation is impaired by loss-of-function mutations in GALNT3 as in HFTC1 or mutations in FGF23 as in HFTC2, secretion of intact hFGF23 may be impaired. Nonglycosylated FGF23 may be targeted for degradation by subtilisin/furine-like endopeptidases. hFGF23 fragments may spill over or be actively secreted into the circulation, particularly if gene expression of hFGF23 is stimulated by feedback up-regulation (B). The ADHR mutation Q176 may protect glycosylated, partially, and nonglycosylated intact FGF23, leading to accumulation and excess secretion of intact FGF23. By protecting the subtilisin/furine-like cleavage site, the ADHR mutation may also lead to intracellular accumulation of otherwise unstable 25- and 14-kDa protein species and secretion of partially glycosylated 25-kDa protein species, which may have biological activity but escape negative feedback suppression of gene transcription (C).

In conclusion, we here describe a novel homozygous missense mutation, S129P, in a previously reported Japanese individual with HFTC2 and provide first direct evidence for the conclusion that this mutation impairs O-glycosylation of the mutant FGF23 protein when expressed in mammalian cell lines, thus explaining its poor secretion. Similar results were obtained for [G71]hFGF23 and [F129]hFGF23, suggesting a common mechanism for all three mutations that lead to development of tumoral calcinosis and indicating that HFTC2 is a disorder of O-glycosylation (25).

Footnotes

This work was supported by the National Institutes of Health Grants PO1 DK11794, Project IV (to H.J.), and 1K08 DK078361; Young Investigator Awards by the National Kidney Foundation and the American Association for Clinical Investigation (to C.B.).

Disclosure Summary: The authors have nothing to declare.

First Published Online October 16, 2009

Abbreviations: ADHR, Autosomal dominant hypophosphatemic rickets; FGF23, fibroblast growth factor 23; GALNT3, UDP-N-acetyl-α-d-galactosamine-polypeptide N-acetylgalactosaminyl-transferase 3; hFG23, human FGF23; HFTC, hyperphosphatemic familiar tumoral calcinosis; PVDF, polyvinyl difluoride; wt, wild type.

References

  1. Giard JM 1998 Sur la calcifcation hibernale. Compes Rend Seanes Soc Biol So 34:1013–1015 [Google Scholar]
  2. Duret MH 1899 Tumeurs multiples et singulieres des bourses sereuses. Bull Mem Soc Ant Paris 74:725–731 [Google Scholar]
  3. Eddy MC, Jan De Beur SM, Yandow SM, McAlister WH, Shore EM, Kaplan FS, Whyte MP, Levine MA 2000 Deficiency of the α-subunit of the stimulatory G protein and severe extraskeletal ossification. J Bone Miner Res 15:2074–2083 [DOI] [PubMed] [Google Scholar]
  4. Topaz O, Shurman DL, Bergman R, Indelman M, Ratajczak P, Mizrachi M, Khamaysi Z, Behar D, Petronius D, Friedman V, Zelikovic I, Raimer S, Metzker A, Richard G, Sprecher E 2004 Mutations in GALNT3, encoding a protein involved in O-linked glycosylation, cause familial tumoral calcinosis. Nat Genet 36:579–581 [DOI] [PubMed] [Google Scholar]
  5. Kato K, Jeanneau C, Tarp MA, Benet-Pagès A, Lorenz-Depiereux B, Bennett EP, Mandel U, Strom TM, Clausen H 2006 Polypeptide GalNAc-transferase T3 and familial tumoral calcinosis. Secretion of fibroblast growth factor 23 requires O-glycosylation. J Biol Chem 281:18370–18377 [DOI] [PubMed] [Google Scholar]
  6. Benet-Pagès A, Orlik P, Strom TM, Lorenz-Depiereux B 2005 An FGF23 missense mutation causes familial tumoral calcinosis with hyperphosphatemia. Hum Mol Genet 14:385–390 [DOI] [PubMed] [Google Scholar]
  7. Ichikawa S, Lyles KW, Econs MJ 2005 A novel GALNT3 mutation in a pseudoautosomal dominant form of tumoral calcinosis: evidence that the disorder is autosomal recessive. J Clin Endocrinol Metab 90:2420–2423 [DOI] [PubMed] [Google Scholar]
  8. Larsson T, Davis SI, Garringer HJ, Mooney SD, Draman MS, Cullen MJ, White KE 2005 Fibroblast growth factor-23 mutants causing familial tumoral calcinosis are differentially processed. Endocrinology 146:3883–3891 [DOI] [PubMed] [Google Scholar]
  9. Ichikawa S, Imel EA, Kreiter ML, Yu X, Mackenzie DS, Sorenson AH, Goetz R, Mohammadi M, White KE, Econs MJ 2007 A homozygous missense mutation in human KLOTHO causes severe tumoral calcinosis. J Clin Invest 117:2684–2691 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Segawa H, Yamanaka S, Ohno Y, Onitsuka A, Shiozawa K, Aranami F, Furutani J, Tomoe Y, Ito M, Kuwahata M, Imura A, Nabeshima Y, Miyamoto K 2007 Correlation between hyperphosphatemia and type II Na-Pi cotransporter activity in klotho mice. Am J Physiol Renal Physiol 292:F769–F779 [DOI] [PubMed] [Google Scholar]
  11. Baum M, Schiavi S, Dwarakanath V, Quigley R 2005 Effect of fibroblast growth factor-23 on phosphate transport in proximal tubules. Kidney Int 68:1148–1153 [DOI] [PubMed] [Google Scholar]
  12. Segawa H, Onitsuka A, Furutani J, Kaneko I, Aranami F, Matsumoto N, Tomoe Y, Kuwahata M, Ito M, Matsumoto M, Li M, Amizuka N, Miyamoto KNpt2a and Npt2c in mice play distinct and synergistic roles in inorganic phosphate metabolism and skeletal development. Am J Physiol Renal Physiol 297:F671–F678 [DOI] [PubMed] [Google Scholar]
  13. Strom TM, Jüppner H 2008 PHEX, FGF23, DMP1 and beyond. Curr Opin Nephrol Hypertens 17:357–362 [DOI] [PubMed] [Google Scholar]
  14. Liu S, Gupta A, Quarles LD 2007 Emerging role of fibroblast growth factor 23 in a bone-kidney axis regulating systemic phosphate homeostasis and extracellular matrix mineralization. Curr Opin Nephrol Hypertens 16:329–335 [DOI] [PubMed] [Google Scholar]
  15. Fukumoto S, Yamashita T 2007 FGF23 is a hormone-regulating phosphate metabolism—unique biological characteristics of FGF23. Bone 40:1190–1195 [DOI] [PubMed] [Google Scholar]
  16. Yamazaki Y, Tamada T, Kasai N, Urakawa I, Aono Y, Hasegawa H, Fujita T, Kuroki R, Yamashita T, Fukumoto S, Shimada T 2008 Anti-FGF23 neutralizing antibodies show the physiological role and structural features of FGF23. J Bone Miner Res 23:1509–1518 [DOI] [PubMed] [Google Scholar]
  17. Shimada T, Muto T, Urakawa I, Yoneya T, Yamazaki Y, Okawa K, Takeuchi Y, Fujita T, Fukumoto S, Yamashita T 2002 Mutant FGF-23 responsible for autosomal dominant hypophosphatemic rickets is resistant to proteolytic cleavage and causes hypophosphatemia in vivo. Endocrinology 143:3179–3182 [DOI] [PubMed] [Google Scholar]
  18. Frishberg Y, Ito N, Rinat C, Yamazaki Y, Feinstein S, Urakawa I, Navon-Elkan P, Becker-Cohen R, Yamashita T, Araya K, Igarashi T, Fujita T, Fukumoto S 2007 Hyperostosis-hyperphosphatemia syndrome: a congenital disorder of O-glycosylation associated with augmented processing of fibroblast growth factor 23. J Bone Miner Res 22:235–242 [DOI] [PubMed] [Google Scholar]
  19. Garringer HJ, Malekpour M, Esteghamat F, Mortazavi SM, Davis SI, Farrow EG, Yu X, Arking DE, Dietz HC, White KE 2008 Molecular genetic and biochemical analyses of FGF23 mutations in familial tumoral calcinosis. Am J Physiol Endocrinol Metab 295:E929–E937 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Masi L, Gozzini A, Franchi A, Campanacci D, Amedei A, Falchetti A, Franceschelli F, Marcucci G, Tanini A, Capanna R, Brandi ML2005 A novel recessive mutation in fibroblast growth factor-23 (FGF23) causes a tumoral calcinosis. J Bone Surg Am 91:1190–1198 [DOI] [PubMed] [Google Scholar]
  21. Larsson T, Yu X, Davis SI, Draman MS, Mooney SD, Cullen MJ, White KE 2005 A novel recessive mutation in fibroblast growth factor-23 causes familial tumoral calcinosis. J Clin Endocrinol Metab 90:2424–2427 [DOI] [PubMed] [Google Scholar]
  22. Chefetz I, Heller R, Galli-Tsinopoulou A, Richard G, Wollnik B, Indelman M, Koerber F, Topaz O, Bergman R, Sprecher E, Schoenau E 2005 A novel homozygous missense mutation in FGF23 causes familial tumoral calcinosis associated with disseminated visceral calcification. Hum Genet 118:261–266 [DOI] [PubMed] [Google Scholar]
  23. Araya K, Fukumoto S, Backenroth R, Takeuchi Y, Nakayama K, Ito N, Yoshii N, Yamazaki Y, Yamashita T, Silver J, Igarashi T, Fujita T 2005 A novel mutation in fibroblast growth factor 23 gene as a cause of tumoral calcinosis. J Clin Endocrinol Metab 90:5523–5527 [DOI] [PubMed] [Google Scholar]
  24. Yamaguchi T, Sugimoto T, Imai Y, Fukase M, Fujita T, Chihara K 1995 Successful treatment of hyperphosphatemic tumoral calcinosis with long-term acetazolamide. Bone 16(Suppl 4):247S–250S [DOI] [PubMed] [Google Scholar]
  25. Jaeken J 2003 Komrower Lecture. Congenital disorders of glycosylation (CDG): it’s all in it! J Inherit Metab Dis 26:99–118 [DOI] [PubMed] [Google Scholar]
  26. Goetz R, Beenken A, Ibrahimi OA, Kalinina J, Olsen SK, Eliseenkova AV, Xu C, Neubert TA, Zhang F, Linhardt RJ, Yu X, White KE, Inagaki T, Kliewer SA, Yamamoto M, Kurosu H, Ogawa Y, Kuro-o M, Lanske B, Razzaque MS, Mohammadi M 2007 Molecular insights into the klotho-dependent, endocrine mode of action of fibroblast growth factor 19 subfamily members. Mol Cell Biol 27:3417–3428 [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from The Journal of Clinical Endocrinology and Metabolism are provided here courtesy of The Endocrine Society

RESOURCES