Isolated C-Terminal Tail of FGF23 Alleviates Hypophosphatemia by Inhibiting FGF23-FGFR-Klotho Complex Formation

Isolated C-terminal tail of FGF23 alleviates hypophosphatemia by inhibiting FGF23-FGFR-Klotho complex formation

Regina Goetza,1, Yuji Nakadab,1, Ming Chang Huc,1, Hiroshi Kurosub, Lei Wangb, Teruyo Nakatanid, Mingjun Shic, Anna V. Eliseenkovaa, Mohammed S. Razzaqued,e, Orson W. Moec,2, Makoto Kuro-ob,2, and Moosa Mohammadia,2

aDepartment of Pharmacology, New York University School of Medicine, New York, NY 10016; bDepartment of Pathology and cDepartment of Internal Medicine, Department of Physiology, and Charles and Jane Pak Center of Mineral Metabolism, University of Texas Southwestern Medical Center, Dallas,TX 75390; dDepartment of Oral and Developmental Biology, Harvard School of Dental Medicine, Boston, MA 02115; and eDepartment of Pathology, Riyadh Armed Forces Hospital, Riyadh, Kingdom of Saudi Arabia

Edited by John T. Potts, Jr., Massachusetts General Hospital, Charlestown, MA, and approved November 5, 2009 (received for review February 23, 2009) Fibroblast growth factor (FGF) 23 inhibits renal phosphate reabsorp- and complications include “overshoot” hyperphosphatemia, hy- tion by activating FGF receptor (FGFR) 1c in a Klotho-dependent pocalcemia, and metastatic calcification (3, 7). In addition, pa- fashion. The phosphaturic activity of FGF23 is abrogated by proteo- renteral regimens are not practical for chronic disorders. Most lytic cleavage at the RXXR motif that lies at the boundary between importantly, replacement therapy alone is never adequate when the FGF core homology domain and the 72-residue-long C-terminal there is significant renal phosphate wasting. Therefore, novel tail of FGF23. Here, we show that the soluble ectodomains of FGFR1c strategies for the treatment of hypophosphatemia are needed. and Klotho are sufficient to form a ternary complex with FGF23 in Fibroblast growth factor (FGF) 23, originally identified as the vitro. The C-terminal tail of FGF23 mediates binding of FGF23 to a de mutated gene in patients with the phosphate wasting disorder novo site generated at the composite FGFR1c-Klotho interface. autosomal dominant hypophosphatemic rickets (ADHR) (11) Consistent with this finding, the isolated 72-residue-long C-terminal and as the causative factor of tumor-induced osteomalacia (12, tail of FGF23 impairs FGF23 signaling by competing with full-length 13), is an endocrine regulator of phosphate homeostasis. FGF23 ligand for binding to the binary FGFR-Klotho complex. Injection of inhibits reabsorption of phosphate in the renal proximal tubule the FGF23 C-terminal tail peptide into healthy rats inhibits renal by decreasing the abundance of the type II sodium-dependent phosphate excretion and induces hyperphosphatemia. In a mouse phosphate transporters NaPi-2A and NaPi-2C in the apical brush model of renal phosphate wasting attributable to high FGF23, the border membrane (14–16). FGF23 activity is regulated by a FGF23 C-terminal peptide reduces phosphate excretion, leading to an proteolytic cleavage at the 176RXXR179 motif, located at the increase in serum phosphate concentration. Our data indicate that boundary between the FGF core homology domain and the 72- proteolytic cleavage at the RXXR motif abrogates FGF23 activity by a residue-long C-terminal tail of FGF23 (17, 18). The proteolytic dual mechanism: by removing the binding site for the binary FGFR- cleavage generates an inactive N-terminal fragment (Y25 to Klotho complex that resides in the C-terminal region of FGF23, and R179, the FGF core homology domain) and a C-terminal frag- by generating an endogenous inhibitor of FGF23. We propose that ment (S180 to I251) (19) (Fig. 1A). Missense mutations of either peptides derived from the C-terminal tail of FGF23 or peptidomi- R176 or R179 of the 176RXXR179 motif inhibit this proteolytic metics and small-molecule organomimetics of the C-terminal tail can cleavage (17, 18) and lead to accumulation of full-length, bio- be used as therapeutics to treat renal phosphate wasting. active FGF23, thereby inducing renal phosphate wasting in ADHR patients (11). To exert its phosphaturic activity, FGF23 FGF23 antagonist | endogenous inhibitor of FGF23 | FGF23 C-terminal requires Klotho as an obligate coreceptor (20, 21), a protein first peptide | binary FGF receptor 1c-Klotho complex | composite FGF receptor described as an aging suppressor (22). The dependency on PHARMACOLOGY 1c-Klotho interface Klotho compensates for the poor binding affinity of FGF23 to both FGF receptor (FGFR) and heparan sulfate (19). Klotho norganic phosphate plays a key role in a myriad of biological constitutively binds the cognate FGFRs of FGF23, and the bi- Iprocesses, including bone mineralization, reversible regulation nary FGFR-Klotho complexes exhibit enhanced binding affinity of protein function by phosphorylation, and production of adeno- for FGF23 (20, 21). sine triphosphate. Plasma levels of phosphate range between 2.2 We have recently shown that in contrast to full-length FGF23, and 4.9 mg/dL (1, 2), and are primarily regulated by modifying renal the inactive N-terminal fragment of proteolytic cleavage fails to tubular reabsorption. Because of phosphate’s pleiotropic activity, coimmunoprecipitate with binary FGFR-Klotho complexes, imbalances in phosphate homeostasis adversely affect essentially suggesting that the 72-residue-long C-terminal tail of FGF23 every major tissue/organ. Hypophosphatemia is a common clinical mediates binding of FGF23 to its cognate FGFR-Klotho com- condition with an incidence ranging from 0.2% to 3.1% in all plexes (19). Here, we demonstrate that the C-terminal tail of hospital admissions to 21.5% to 80% in specific subgroups of hospitalized patients (3, 4). Clinical manifestations of hypophosphatemia include respiratory failure, cardiac arrhythmia, he- Author contributions: R.G., Y.N., M.C.H., H.K., L.W., M.S.R., O.W.M., M.K., and M.M. molysis, rhabdomyolysis, seizures, and coma acutely and myalgia designed research; R.G., Y.N., M.C.H., H.K., L.W., T.N., and M.S. performed research; R.G. and A.V.E. contributed new reagents/analytic tools; R.G., Y.N., M.C.H., H.K., L.W., T.N., M. and osteomalacia chronically (3). Hypophosphatemia originates S., M.S.R., O.W.M., M.K., and M.M. analyzed data; and R.G., O.W.M., M.K., and M.M. from diverse pathophysiologic mechanisms, most importantly from wrote the paper. renal phosphate wasting, an inherited or acquired condition in The authors declare no conflict of interest. which renal tubular reabsorption of phosphate is impaired (5, 6). This article is a PNAS Direct Submission. Oral or i.v. administration of inorganic phosphate salts cur- 1R.G., Y.N., and M.C.H. contributed equally to this work. rently is the mainstay for the management of hypophosphatemia. 2To whom correspondence may be addressed. E-mail: [email protected], Oral phosphate therapy requires high doses, which frequently lead [email protected], or [email protected]. to diarrhea or gastric irritation (7). For i.v. phosphate therapy, This article contains supporting information online at www.pnas.org/cgi/content/full/ the response to any given dose is sometimes unpredictable (8–10), 0902006107/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.0902006107 PNAS | January 5, 2010 | vol. 107 | no. 1 | 407–412 Downloaded by guest on September 30, 2021 Fig. 1. FGF23 C-terminal tail mediates binding of FGF23 to the binary FGFR-Klotho complex, and the isolated FGF23 C-terminal tail peptide competes with FGF23 for binding to FGFR-Klotho. (A) FGF23 proteins and peptides used in this study. Amino acid boundaries of each protein/peptide are labeled with residue letter and number. The FGF23 core region is shaded gray, and the position of the proteolytic cleavage site RXXR is indicated. (B) Representative SPR sensorgram of FGFR1c binding to Klotho, and ﬁtted saturation binding curve. Klotho ectodomain was immobilized on a biosensor chip, and increasing concentrations of

FGFR1c ectodomain were passed over the chip. The dissociation constant (KD) was calculated from the saturation binding curve. (C and D) Representative SPR sensorgrams illustrating binding of FGF2328-251 (C) and FGF23180-251 (D) to the binary FGFR1c-Klotho complex. FGF2328-251 and FGF23180-251 were immobilized on a biosensor chip, and increasing concentrations of FGFR1c-Klotho complex were passed over the chip. (E and F) Representative SPR sensorgrams illustrating inhibition by FGF23180-251 (E)orFGF23180-205 (F) of FGFR1c-Klotho binding to FGF2328-251 immobilized on a biosensor chip. Increasing concentrations of either FGF23180-251 or FGF23180-205 were mixed with a ﬁxed concentration of FGFR1c-Klotho complex, and the mixtures were passed over a FGF23 chip.

FGF23 mediates binding of FGF23 to a de novo site at the Klotho bound FGFR1c with high affinity (KD = 72 nM; Fig. 1B). composite FGFR-Klotho interface. We exploit this finding for Because Klotho harbors a high-affinity binding site for FGFR1c, therapeutic purposes and show that the isolated C-terminal tail we reasoned that Klotho might also possess a distinct high-af- of FGF23 can compete with full-length ligand for binding to the finity binding site for FGF23 and promote FGF23-FGFR1c FGFR-Klotho complex, and hence can antagonize the phos- binding by engaging FGF23 and FGFR1c simultaneously. To test phaturic activity of FGF23 in vivo, both in healthy rats and in a this, FGF2328-251 was coupled to a biosensor chip, and increasing mouse model of phosphate wasting disorders. concentrations of Klotho ectodomain were passed over the chip. As shown in Fig. S1C, Klotho bound poorly to FGF2328-251. Results These data demonstrate that the Klotho ectodomain contains a The Ternary FGF23-FGFR1c-Klotho Complex Can Be Reconstituted in high-affinity binding site for FGFR1c but not for FGF23. Solution by Using Recombinant Soluble Ectodomains of FGFR1c and Next, we measured binding of FGF23 to FGFR1c by injecting Klotho. To understand how FGF23, FGFR, and Klotho interact increasing concentrations of FGFR1c over the FGF23 chip. As to form a ternary complex, we decided to reconstitute the ternary shown in Fig. S1D, FGF2328-251 exhibited poor binding to complex in solution by using bioactive, full-length FGF23 FGFR1c. Thus, the SPR data show that FGF23 exhibits poor (FGF2328-251) (Fig. 1A) and the soluble ectodomains of FGFR1c binding affinity for both the Klotho ectodomain alone and the and Klotho. The binary complex of FGFR1c ectodomain with FGFR1c ectodomain alone. Together with the size-exclusion Klotho ectodomain was formed by capturing the Klotho ecto- chromatography data showing that FGF23 binds stably to the domain onto an FGFR1c affinity column from conditioned purified binary FGFR1c-Klotho complex, the data raised the media of a HEK293 cell line ectopically expressing the Klotho question of whether FGF23 binds to a de novo site generated at ectodomain (20). The FGFR1c-Klotho complex was further the composite FGFR1c-Klotho interface. To test this, we puri- purified by size-exclusion chromatography to remove excess fied FGFR1c-Klotho complex as described above and passed FGFR1c (Fig. S1A). Next, the FGFR1c-Klotho complex was increasing concentrations of the binary complex over the FGF23 mixed with FGF2328-251, and ternary complex formation was chip. As shown in Fig. 1C, FGF2328-251 bound to the FGFR1c- examined by size-exclusion chromatography. As shown in Fig. Klotho complex demonstrating that FGF23 interacts with a de S1B, FGF23 coeluted with the FGFR1c-Klotho complex, dem- novo site generated at the composite FGFR1c-Klotho interface. onstrating that the ectodomains of FGFR1c and Klotho are sufficient to form a stable ternary complex with FGF23. C-Terminal Tail of FGF23 Mediates Binding of FGF23 to a de Novo Site The size-exclusion data showing that Klotho and FGFR1c at the Composite FGFR1c-Klotho Interface. We then examined ectodomains form a stable binary complex (Fig. S1A) indicate whether the C-terminal tail of FGF23 mediates binding of FGF23 that Klotho must harbor a high-affinity binding site for FGFR1c. to the FGFR1c-Klotho complex. To test this, the C-terminal tail To further confirm this, we used surface plasmon resonance peptide of FGF23 (FGF23180-251; Fig. 1A) was coupled to a bi- (SPR) spectroscopy to determine the dissociation constant of the osensor chip, and increasing concentrations of FGFR1c-Klotho FGFR1c-Klotho interaction. Klotho ectodomain was immobi- complex were passed over the chip. As shown in Fig. 1D, lized on a biosensor chip, and increasing concentrations FGFR1c FGF23180-251 avidly bound to the binary complex. Size-exclusion ectodomain were passed over the chip. Consistent with the re- chromatography and coimmunoprecipitation experiments yiel- sults obtained by using size-exclusion chromatography (Fig. S1A), ded similar results supporting the SPR data (Fig. S2A–C).

408 | www.pnas.org/cgi/doi/10.1073/pnas.0902006107 Goetz et al. Downloaded by guest on September 30, 2021 To fully nail down that the C-terminal tail of FGF23 mediates FGF23 binding to the binary FGFR1c-Klotho complex, a fixed concentration of FGFR1c-Klotho was mixed with increasing concentrations of FGF23180-251, and the mixtures were passed over the FGF23 chip. Mixtures of FGF2328-251 with FGFR1c- Klotho were used as a control. As shown in Fig. 1E and Fig. S2D, FGF23180-251 competed, in a dose-dependent fashion, with FGF2328-251 for binding to the FGFR1c-Klotho complex. Half- maximum inhibition of FGFR1c-Klotho binding to FGF2328-251 was reached with a 3.3-fold molar excess of FGF23180-251 over Fig. 2. FGF23 C-terminal tail peptide inhibits tyrosine phosphorylation of FGFR1c-Klotho complex (Fig. S2D). As expected, less than an FRS2α and downstream activation of MAP kinase cascade induced by FGF23. equimolar amount of FGF2328-251 relative to FGFR1c-Klotho Shown is an immunoblot analysis for phosphorylation of FRS2α (pFRS2α)and complex already yielded 50% inhibition of binding of the binary 44/42 MAP kinase (p44/42 MAPK) in a HEK293 Klotho cell line, which had been complex to immobilized FGF2328-251 (Fig. S2D and E). A coim- stimulated with FGF proteins/peptide as denoted. Numbers above the lanes fi give the amounts of protein/peptide added in nanomolar. To control for equal munoprecipitation-based competition assay also con rmed that sample loading, the protein blots were probed with an antibody to Klotho. the C-terminal tail peptide of FGF23 can inhibit binding of FGF23 to its binary cognate FGFR-Klotho complex (Fig. S2F). Together, the data unambiguously demonstrate that the C-terminal tail of of the FGF23 antagonists, we examined the ability of the FGF23 harbors the binding site for the binary FGFR-Klotho FGF23180-251 peptide to inhibit signaling of FGF2, a prototypical complex and hence is essential for formation of the ternary paracrine FGF, which does not require Klotho for signaling. As FGF23-FGFR-Klotho complex. Importantly, the binding data shown in Fig. S4B, the FGF23 antagonist failed to inhibit tyrosine unveil that proteolytic cleavage at the RXXR motif abrogates phosphorylation of FRS2α and downstream activation of MAP FGF23 activity by removing the binding site for the binary FGFR- kinase cascade induced by FGF2. These data show that FGF23 C- Klotho complex that resides in the C-terminal tail of FGF23. terminal peptides specifically block FGF23 signaling. In renal proximal tubule epithelium, FGF23 signaling leads to FGF23 Residues S180 to T200 Comprise the Minimal Binding Epitope inhibition of phosphate uptake. To establish further that FGF23 for the FGFR-Klotho Complex. In follow-up studies, we found that C-terminal peptides block FGF23 action, we studied the effects of 28-200 FGF23 , which lacks the last 51 C-terminal amino acids, still the peptides on sodium-coupled phosphate uptake in a proximal retains the ability to coimmunoprecipitate with the binary tubular cell model. As shown in Fig. S5A, FGF23180-251 antago- FGFR-Klotho complex (Fig. S2C). The finding suggested that nized the inhibition of phosphate uptake by FGF2328-251 in a 28-200 FGF23 may have similar biological activity as the full-length dose-dependent fashion, with an IC of ∼21 nM. FGF23180-205 28-200 50 protein. To test this, we compared the ability of FGF23 and exhibited a similar, albeit less potent antagonistic effect (Fig. 28-251 FGF23 to induce tyrosine phosphorylation of FGF receptor S5B). As expected, neither of the two FGF23 C-terminal peptides substrate 2α (FRS2α) and downstream activation of MAP kinase altered phosphate uptake when applied alone (Fig. S5). cascade in Klotho-expressing cultured cells, and to induce 28-200 phosphaturia in mice. As shown in Fig. S3A, FGF23 induced FGF23 C-Terminal Peptides Antagonize Phosphaturic Activity of FGF23 phosphorylation of FRS2α and downstream activation of MAP in Vivo. These findings motivated us to conduct in vivo studies and kinase cascade at a dose comparable to that of FGF2328-251. The investigate whether the FGF23 C-terminal peptides antagonize truncated FGF23 was also nearly as effective as the full-length the phosphaturic effects of endogenous FGF23. I.v. injection of ligand in reducing serum phosphate concentration in healthy FGF23180-251 into healthy Sprague–Dawley rats led to renal C57BL/6 mice (Fig. S3B). These data show that deletion of the phosphate retention and hyperphosphatemia (Fig. 3), suggesting last 51 amino acids from the FGF23 C terminus has little effect on that FGF23 C-terminal peptides antagonize the phosphaturic FGF23 biological activity, narrowing down the epitope on the action of endogenous FGF23. As expected, injection of FGF2328- FGF23 C-terminal tail for the composite FGFR-Klotho interface 251 induced increases in excretion rate and fractional excretion of PHARMACOLOGY to residues S180 and T200. Indeed, a FGF23 peptide comprising phosphate, and led to a significant decrease in plasma phosphate the minimal binding epitope for FGFR-Klotho (FGF23180-205; compared to vehicle-treated animals (Fig. 3). Fig. 1A) was able to compete, in a dose-dependent fashion, with FGF23 exerts its phosphaturic activity by inhibiting phosphate FGF2328-251 for binding to the binary FGFR1c-Klotho complex uptake by renal proximal tubule epithelium. The effect has been F (Fig. 1 ). Half-maximum inhibition of FGFR1c-Klotho binding attributed to reduced transport activity of NaPi-2A and NaPi-2C, 28-251 to FGF23 was reached with a 5.7-fold molar excess of down-regulation of NaPi-2A and NaPi-2C proteins in the apical FGF23180-205 over FGFR1c-Klotho complex (Fig. S2D). Sim- brush border membrane, and upon more chronic exposure to ilarly, in a coimmunoprecipitation-based competition assay, the FGF23, repression of NaPi-2A and NaPi-2C genes (14–16, 24, 180-205 FGF23 peptide was able to inhibit binding of FGF23 to the 25). We examined the abundance of NaPi-2A and NaPi-2C binary complexes of its cognate FGFR and Klotho (Fig. S3C). proteins in brush border membrane vesicles isolated from the Our data also explain the finding by Garringer and colleagues kidneys of rats. I.v. injection of FGF23180-251 into healthy rats led (23) showing that residues P189 to P203 are required for to an increase in NaPi-2A protein expression in the apical brush FGF23 signaling. border membrane compared to vehicle treatment (Fig. S6A and B). The peptide exhibited similar effects on the NaPi-2C protein FGF23 C-Terminal Peptides Block FGF23 Signaling. Based on these (Fig. S6C). As expected, injection of FGF2328-251 led to a de- 180-251 180-205 data, we postulated that FGF23 and FGF23 should crease in NaPi-2A protein expression (Fig. S6A and B). These antagonize FGF23 signaling by competing with full-length findings establish that FGF23 C-terminal peptides counteract or FGF23 for binding to the FGFR-Klotho complex. To test this, we cancel out FGF23’s phosphaturic action mediated through NaPi- 28-251 stimulated cells stably overexpressing Klotho with FGF23 2A and NaPi-2C. alone or FGF2328-251 mixed with increasing concentrations of either FGF23180-251 or FGF23180-205. As shown in Fig. 2 and Fig. Therapeutic Potential of FGF23 C-Terminal Peptides. To evaluate the S4A, both peptides inhibited, in a dose-dependent fashion, therapeutic potential of FGF23180-251 for treating renal phos- FGF23-induced tyrosine phosphorylation of FRS2α and down- phate wasting, we analyzed the peptide’sefficacy in Hyp mice, a stream activation of MAP kinase cascade. To test the specificity mouse model of human X-linked hypophosphatemia (XLH)

Goetz et al. PNAS | January 5, 2010 | vol. 107 | no. 1 | 409 Downloaded by guest on September 30, 2021 Fig. 3. FGF23 C-terminal tail peptide antagonizes phosphaturic activity of Fig. 4. FGF23 C-terminal peptides alleviate renal phosphate wasting in Hyp − 180-251 180-205 μ FGF23 in vivo. FGF2328-251 (0.1 μg·kg body weight 1) or FGF23180-251 (0.1 μg·kg mice. FGF23 (1 mg), FGF23 (860 g), or vehicle were injected i.p. − body weight 1) were injected i.v. into Sprague–Dawley rats. Serum and urine into Hyp mice. Urine phosphate (urinary Pi) and creatinine levels and serum phosphate levels (serum Pi) were measured before and at the indicated time parameters were measured and calculated before and 3 h after injection. FE 180-205 P , fractional excretion of phosphate; U V, phosphate excretion rate; Cl , points after the injection. Urinary Pi of Hyp mice treated with FGF23 i Pi Cr < creatinine clearance. was not determined (ND). Bars and error bars are mean + SE. *, P 0.05 by ANOVA.

(26–29). XLH is an inherited phosphate wasting disorder asso- These tumors are often difficult to locate, however, or the tumors ciated with high FGF23, which is thought to be due to reduced fi clearance of FGF23 from the circulation. Excess FGF23 causes are found in locations that are dif cult to access, leaving most increased phosphate excretion resulting in hypophosphatemia. patients with tumor-induced osteomalacia also currently with no As shown in Fig. 4, i.p. injection of FGF23180-251 induced a de- options other than symptomatic therapy (30, 31). Because excess crease in renal phosphate excretion in Hyp mice compared to FGF23 is the pathogenic factor in phosphate wasting disorders, blocking its action with FGF23 C-terminal peptides holds promise vehicle treatment. The effect persisted for at least 4 h after in- fi jection. Concomitantly, serum phosphate levels were elevated by of providing the rst causative pharmacotherapy. In a mouse the FGF23 antagonist treatment (Fig. 4). Likewise, i.p. injection model of phosphate wasting disorders, we have shown that FGF23 of the FGF23180-205 peptide, which comprises the minimal C-terminal peptides are effective in counteracting the phospha- fi binding epitope for the composite FGFR-Klotho interface, turic action of FGF23. Our ndings warrant further evaluation of ’ fi caused an increase in serum phosphate in Hyp mice compared to the peptides ef cacy in nonhuman primates, and eventually, in vehicle-treated animals (Fig. 4). These results show that FGF23 humans. Neutralizing FGF23 activity with antibody provides an C-terminal peptides are effective in reducing renal phosphate alternative approach for treating renal phosphate wasting. Indeed, wasting caused by excess FGF23. Aono, Yamazaki, and colleagues (32, 33) have explored this approach and developed antibodies against FGF23 that effectively Discussion neutralize FGF23 activity in both healthy mice and Hyp mice. Our study provides firm evidence that proteolytic cleavage at the Although it has been conclusively demonstrated that the phos- RXXR motif abrogates FGF23 activity by a dual mechanism: by phaturic activity of FGF23 is Klotho-dependent (34), the possi- removing FGF23’s binding site for the binary FGFR-Klotho bility that FGF23 may have some Klotho-independent functions complex, and by generating an endogenous inhibitor of FGF23. has not yet been ruled out experimentally. In this regard, our in- We exploited this regulatory mechanism to develop an FGF23 hibitory peptide approach may offer a more targeted therapy for antagonist with therapeutic potential for renal phosphate wasting. hypophosphatemia than anti-FGF23 antibodies because these Patients with phosphate wasting disorders are, by and large, peptides specifically target the binary FGFR-Klotho complex and treated symptomatically, with oral phosphate supplementation hence only neutralize Klotho-dependent function of FGF23. In and 1,25-dihydroxyvitamin D3/calcitriol. As alluded to in the in- contrast, the antibody approach does not discriminate between troduction, oral phosphate therapy can be poorly tolerated, and in Klotho-dependent and -independent functions of FGF23. These certain circumstances can induce hyperparathyroidism and poses peptides can also serve as an experimental tool to dissect Klotho- risk of exacerbation of hypophosphatemia. In patients with XLH, dependent and -independent functions of FGF23. The ability of the persistent and even exaggerated renal phosphate wasting the FGF23 C-terminal peptides to specifically recognize the binary during therapy can cause nephrocalcinosis and nephrolithiasis. For receptor complex makes them a powerful tool to image tissues that patients with renal phosphate wasting from tumor-induced os- express the cognate FGFR-Klotho complexes of FGF23. teomalacia, a causative treatment option exists, which is resection Hypophosphatemia complicating recovery from kidney trans- of the tumor producing excess amounts of phosphaturic hormone. plantation and parenteral iron therapy has been associated with

410 | www.pnas.org/cgi/doi/10.1073/pnas.0902006107 Goetz et al. Downloaded by guest on September 30, 2021 increased plasma levels of FGF23 (35, 36). Thus, the FGF23 Materials and Methods antagonist discovered in this study may be of therapeutic value Purification of FGF23, FGFR, and Klotho Proteins. See SI Materials and Methods; for a broader collection of patients than inherited or tumor- Fig. S7. induced phosphate wasting disorders alone. Another indication for therapy with FGF23 C-terminal pep- Analysis of FGF23-FGFR1c-Klotho Interactions by SPR Spectroscopy. See SI tides might be chronic kidney disease, a condition with a growing Materials and Methods. incidence, currently affecting ∼20 million people in the United States alone. In patients with chronic kidney disease, plasma Analysis of FGF23 Protein/Peptide Binding to FGFR1c-Klotho Complex by Size- levels of FGF23 increase as kidney function declines (37), Exclusion Chromatography. Binding of FGF23 proteins/peptides to the 1:1 and the gradual increases in plasma FGF23 correlate with dis- binary complex of FGFR1c and Klotho ectodomains was analyzed by using a ease progression (38). Although the precise role of FGF23 in HiLoad 16/60 Superdex 200 prep grade size-exclusion column on an ÄKTA- fi the pathogenesis of chronic kidney disease and its sequelae re- puri er (GE Healthcare) (SI Materials and Methods). mains to be determined, blocking FGF23 action with FGF23 Pull-Down Assays of FGF23 Protein/Peptide Binding to FGFR-Klotho Complex. C-terminal peptides might prove effective in preventing or at- Binding of FGF23 proteins/peptides to FGFR-Klotho complexes isolated tenuating the occurrence of disease complications such as hy- from lysate of a HEK293 cell line expressing the membrane-spanning form of perparathyroidism. murine Klotho was analyzed as described in ref. 19 (SI Materials and Methods). Our identification of the FGF23 C-terminal tail as an FGF23 antagonist suggests that proteolytic cleavage not only removes Analysis of Phosphorylation of FRS2α and 44/42 MAP Kinase in Epithelial Cell the binding site on FGF23 for the FGFR-Klotho complex, but Lines. Phosphorylation of FRS2α and 44/42 MAP kinase in response to also generates an endogenous inhibitor of FGF23. A pathophy- treatment with FGF23 proteins/peptides was analyzed in HEK293 and CHO siological role of the latter mechanism is indicated by familial cell lines expressing the membrane-spanning form of murine Klotho (20, 45) tumoral calcinosis (FTC), an autosomal recessive metabolic (SI Materials and Methods). disorder with clinical manifestations mirroring those of phosphate wasting disorders. Missense mutations in either the UDP- Measurement of Phosphate Uptake by Opossum Kidney Cells. Phosphate N-acetyl-α-D-galactosamine:polypeptide N-acetylgalactosaminyl- uptake by the opossum kidney cell line OKP, which endogenously expresses transferase 3 (GALNT3) gene (39) or the FGF23 gene (23, 40) FGFR1-4 and Klotho (Fig. S8), was measured in response to cell stimulation have been associated with FTC. All FTC patients have abnor- with FGF23 proteins/peptides (SI Materials and Methods). mally high plasma levels of the C-terminal proteolytic fragment of FGF23 (39, 40). Our findings suggest that excess C-terminal Measurement of Phosphate in Serum and Urine of Rodents. The phosphaturic 28-200 ∼ FGF23 fragment may aggravate hyperphosphatemia and the activity of FGF23 was examined in 6-week-old C57BL/6 mice by using a resulting soft tissue calcification, by antagonizing the action of published protocol (19). The anti-phosphaturic activity of FGF23 C-terminal peptides was examined in normal Sprague–Dawley rats and in Hyp mice, a any residual, functional FGF23 ligand in these patients. Our – fi mouse model of XLH (27 29) (SI Materials and Methods). In addition, plasma ndings also call for the need to interpret full-length and frag- clearance of recombinant full-length FGF23 and C-terminal tail peptide of ments of FGF23 separately in chronic kidney disease. FGF23 was analyzed in normal Sprague-Dawley rats (SI Materials and There has been a conundrum surrounding the mechanism of Methods; Fig. S9). The experiments in mice were approved by the Harvard action of FGF23 in the kidney because Klotho is expressed in the University Animal Care and Research Committee board. The experiments in distal convoluted tubule (41–43), whereas FGF23 inhibits rats were approved by the Institutional Animal Care and Use Committee at phosphate reabsorption in the proximal tubule (14–16). A recent the University of Texas Southwestern Medical Center at Dallas. study suggested that FGF23 signaling initiates in the distal tubule

and its effects are then transmitted to the proximal tubule Analysis of NaPi-2A and NaPi-2C Protein Abundance in the Apical Brush Border through an unknown diffusible paracrine factor (44). In addition Membrane of Renal Proximal Tubule Epithelium. See SI Materials and Methods. to the membrane-bound isoform of Klotho, alternative splicing and proteolytic cleavage give rise to two soluble isoforms of Statistical Analysis. Data are expressed as the mean ± SE (n ≥ 6 or more).

Klotho found in the circulation (45–48). Importantly, the re- Statistical analysis was performed by using Student’s unpaired or paired t test PHARMACOLOGY combinant Klotho ectodomain that we used to reconstitute the or by using analysis of variance (ANOVA) when applicable. A value of P ≤ 0.05 ﬁ ternary FGF23-FGFR-Klotho complex corresponds to the was considered as statistically signi cant. complete ectodomain of Klotho that is shed into the circulation ACKNOWLEDGMENTS. This work was supported by National Institutes of by a proteolytic cleavage at the juncture between the ex- Health Grants DE13686 (to M.M.), AG19712 (to M.K.), AG25326 (to M.K.), tracellular domain and transmembrane domain (45, 46). Thus, DK48482 (to O.W.M.), DK20543 (to O.W.M.), and DK077276 (to M.S.R.), and our data points to the possibility that it is the shed soluble iso- by the Irma T. Hirschl Fund (M.M.), the Eisai Research Fund (M.K.), the Ellison form of Klotho that makes its way to the proximal tubule to Medical Foundation (M.K.), Ted Nash Long Life Foundation (M.K.), the promote formation of FGF23-FGFR-Klotho ternary complex, Simmons Family Foundation (O.W.M.), and a seed grant from the Pak Center and inhibition of phosphate reabsorption. of Mineral Metabolism and Clinical Research (to M.C.H.).

1. Dwyer K, Barone JE, Rogers JF (1992) Severe hypophosphatemia in postoperative 9. Charron T, et al. (2003) Intravenous phosphate in the intensive care unit: more patients. Nutr Clin Pract 7:279–283. aggressive repletion regimens for moderate and severe hypophosphatemia. Intensive 2. Alon US, et al. (2008) Calcimimetics as an Adjuvant Treatment for Familial Care Med 29:1273–1278. Hypophosphatemic Rickets. Clin J Am Soc Nephrol 3:658–664. 10. Rosen GH, Boullata JI, O'Rangers EA, Enow NB, Shin B (1995) Intravenous phosphate 3. Gaasbeek A, Meinders AE (2005) Hypophosphatemia: an update on its etiology and repletion regimen for critically ill patients with moderate hypophosphatemia. Crit – treatment. Am J Med 118:1094 1101. Care Med 23:1204–1210. 4. Brunelli SM, Goldfarb S (2007) Hypophosphatemia: clinical consequences and 11. Anonymous; ADHR Consortium (2000) Autosomal dominant hypophosphataemic – management. J Am Soc Nephrol 18:1999 2003. rickets is associated with mutations in FGF23. Nat Genet 26:345–348. 5. Imel EA, Econs MJ (2005) Fibroblast growth factor 23: roles in health and disease. JAm 12. White KE, et al. (2001) The autosomal dominant hypophosphatemic rickets (ADHR) – Soc Nephrol 16:2565 2575. gene is a secreted polypeptide overexpressed by tumors that cause phosphate 6. Negri AL (2007) Hereditary hypophosphatemias: new genes in the bone-kidney axis. wasting. J Clin Endocrinol Metab 86:497–500. – Nephrology 12:317 320. 13. Shimada T, et al. (2001) Cloning and characterization of FGF23 as a causative factor of 7. Shiber JR, Mattu A (2002) Serum phosphate abnormalities in the emergency tumor-induced osteomalacia. Proc Natl Acad Sci USA 98:6500–6505. – department. J Emerg Med 23:395 400. 14. Baum M, Schiavi S, Dwarakanath V, Quigley R (2005) Effect of ﬁbroblast growth 8. Bohannon NJ (1989) Large phosphate shifts with treatment for hyperglycemia. Arch factor-23 on phosphate transport in proximal tubules. Kidney Int 68:1148–1153. Intern Med 149:1423–1425.

Goetz et al. PNAS | January 5, 2010 | vol. 107 | no. 1 | 411 Downloaded by guest on September 30, 2021 15. Perwad F, Zhang MY, Tenenhouse HS, Portale AA (2007) Fibroblast growth factor 23 32. Yamazaki Y, et al. (2008) Anti-FGF23 neutralizing antibodies show the physiological impairs phosphorus and vitamin D metabolism in vivo and suppresses 25- role and structural features of FGF23. J Bone Miner Res 23:1509–1518. hydroxyvitamin D-1alpha-hydroxylase expression in vitro. Am J Physiol Renal Physiol 33. Aono Y, et al. (2009) Therapeutic effects of anti-FGF23 antibodies in hypo- 293:F1577–F1583. phosphatemic rickets/osteomalacia. J Bone Miner Res 24:1879–1888. 16. Larsson T, et al. (2004) Transgenic mice expressing fibroblast growth factor 23 under 34. Nakatani T, Ohnishi M, Razzaque MS (2009) Inactivation of klotho function induces the control of the alpha1(I) collagen promoter exhibit growth retardation, hyperphosphatemia even in presence of high serum fibroblast growth factor 23 osteomalacia, and disturbed phosphate homeostasis. Endocrinology 145:3087–3094. levels in a genetically engineered hypophosphatemic (Hyp) mouse model. FASEB J 23: 17. Shimada T, et al. (2002) Mutant FGF-23 responsible for autosomal dominant hypo- 3702–3711. phosphatemic rickets is resistant to proteolytic cleavage and causes hypophosphatemia in 35. Schouten BJ, Hunt PJ, Livesey JH, Frampton CM, Soule SG (2009) FGF23 elevation and vivo. Endocrinology 143:3179–3182. hypophosphatemia after intravenous iron polymaltose: a prospective study. J Clin 18. White KE, et al. (2001) Autosomal-dominant hypophosphatemic rickets (ADHR) Endocrinol Metab 94:2332–2337. mutations stabilize FGF-23. Kidney Int 60:2079–2086. 36. Bhan I, et al. (2006) Post-transplant hypophosphatemia: tertiary ‘hyper-phosphatoninism’? 19. Goetz R, et al. (2007) Molecular insights into the klotho-dependent, endocrine mode Kidney Int 70:1486–1494. of action of fibroblast growth factor 19 subfamily members. Mol Cell Biol 27: 37. Larsson T, Nisbeth U, Ljunggren O, Jüppner H, Jonsson KB (2003) Circulating 3417–3428. concentration of FGF-23 increases as renal function declines in patients with chronic 20. Kurosu H, et al. (2006) Regulation of fibroblast growth factor-23 signaling by klotho. kidney disease, but does not change in response to variation in phosphate intake in J Biol Chem 281:6120–6123. healthy volunteers. Kidney Int 64:2272–2279. 21. Urakawa I, et al. (2006) Klotho converts canonical FGF receptor into a specific receptor 38. Fliser D, et al. (2007) Fibroblast growth factor 23 (FGF23) predicts progression of for FGF23. Nature 444:770–774. chronic kidney disease: the Mild to Moderate Kidney Disease (MMKD) Study. J Am Soc 22. Kuro-o M, et al. (1997) Mutation of the mouse klotho gene leads to a syndrome Nephrol 18:2600–2608. resembling ageing. Nature 390:45–51. 39. Garringer HJ, et al. (2007) Two novel GALNT3 mutations in familial tumoral calcinosis. 23. Garringer HJ, et al. (2008) Molecular genetic and biochemical analyses of FGF23 Am J Med Genet A 143A:2390–2396. mutations in familial tumoral calcinosis. Am J Physiol Endocrinol Metab 295: 40. Araya K, et al. (2005) A novel mutation in fibroblast growth factor 23 gene as a cause E929–E937. of tumoral calcinosis. J Clin Endocrinol Metab 90:5523–5527. 24. Yamashita T, Konishi M, Miyake A, Inui K, Itoh N (2002) Fibroblast growth factor 41. Kato Y, et al. (2000) Establishment of the anti-Klotho monoclonal antibodies and (FGF)-23 inhibits renal phosphate reabsorption by activation of the mitogen-activated detection of Klotho protein in kidneys. Biochem Biophys Res Commun 267:597–602. protein kinase pathway. J Biol Chem 277:28265–28270. 42. Li SA, et al. (2004) Immunohistochemical localization of Klotho protein in brain, 25. Segawa H, et al. (2003) Effect of hydrolysis-resistant FGF23-R179Q on dietary kidney, and reproductive organs of mice. Cell Struct Funct 29:91–99. phosphate regulation of the renal type-II Na/Pi transporter. Pflugers Arch 446: 43. Tsujikawa H, Kurotaki Y, Fujimori T, Fukuda K, Nabeshima Y (2003) Klotho, a gene 585–592. related to a syndrome resembling human premature aging, functions in a negative 26. Anonymous; The HYP Consortium (1995) A gene (PEX) with homologies to endo- regulatory circuit of vitamin D endocrine system. Mol Endocrinol 17:2393–2403. peptidases is mutated in patients with X-linked hypophosphatemic rickets. Nat Genet 44. Farrow EG, Davis SI, Summers LJ, White KE (2009) Initial FGF23-mediated signaling 11:130–136. occurs in the distal convoluted tubule. J Am Soc Nephrol 20:955–960. 27. Beck L, et al. (1997) Pex/PEX tissue distribution and evidence for a deletion in the 3′ 45. Imura A, et al. (2004) Secreted Klotho protein in sera and CSF: implication for post- region of the Pex gene in X-linked hypophosphatemic mice. J Clin Invest 99:1200–1209. translational cleavage in release of Klotho protein from cell membrane. FEBS Lett 28. Eicher EM, Southard JL, Scriver CR, Glorieux FH (1976) Hypophosphatemia: mouse 565:143–147. model for human familial hypophosphatemic (vitamin D-resistant) rickets. Proc Natl 46. Kurosu H, et al. (2005) Suppression of aging in mice by the hormone Klotho. Science Acad Sci USA 73:4667–4671. 309:1829–1833. 29. Strom TM, et al. (1997) Pex gene deletions in Gy and Hyp mice provide mouse models 47. Matsumura Y, et al. (1998) Identification of the human klotho gene and its two for X-linked hypophosphatemia. Hum Mol Genet 6:165–171. transcripts encoding membrane and secreted klotho protein. Biochem Biophys Res 30. van Boekel G, et al. (2008) Tumor producing fibroblast growth factor 23 localized by Commun 242:626–630. two-staged venous sampling. Eur J Endocrinol 158:431–437. 48. Shiraki-Iida T, et al. (1998) Structure of the mouse klotho gene and its two transcripts 31. Jan de Beur SM (2005) Tumor-induced osteomalacia. JAMA 294:1260–1267. encoding membrane and secreted protein. FEBS Lett 424:6–10.

412 | www.pnas.org/cgi/doi/10.1073/pnas.0902006107 Goetz et al. Downloaded by guest on September 30, 2021