Fibroblast growth factor 23 (FGF23) is a critical, bone-derived mediator of phosphate homeostasis.1 In kidney proximal tubule epithelial cells, FGF23 signaling controls expression of the vitamin D metabolizing enzymes CYP27B1 and CYP24A1, resulting in decreased synthesis and elevated turnover of the active vitamin D metabolite 1,25(OH)2D3.2, 3 In addition, FGF23 impairs expression of the sodium-phosphate co-transporters NaPi-2a (SLC34A1) and NaPi-2c (SLC34A3) in the brush border membrane (BBM) of proximal tubular cells, which mediate the reabsorption of urinary phosphate.4, 5
FGF23 signaling is transduced by members of the FGF receptor (FGFR) family in conjunction with the essential co-receptor Klotho, which confers tissue-specificity for endocrine FGF23 signals owing to its predominant expression in kidney.6, 7Fgf23- and Klotho-deficient mice show largely overlapping phenotypes, resembling familial tumoral calcinosis (FTC), which is associated with hyperphosphatemia, increased or inappropriately normal levels of 1,25(OH)2D3, and ectopic calcifications.8–11
In contrast, excess levels of FGF23 result in hypophosphatemia and are associated with several hereditary hypophosphatemic disorders with skeletal abnormalities as a consequence of impaired bone mineralization and growth, including X-linked hypophosphatemic rickets (XLH), autosomal dominant hypophosphatemic rickets (ADHR), and autosomal recessive hypophosphatemic rickets (ARHR).12–16 In addition, in rare cases secretion of FGF23 by tumor cells has been identified to cause hypophosphatemia, resulting in tumor-induced osteomalacia (TIO).3 Whereas ADHR is characterized by gain-of-function mutations in FGF23 itself,17 XLH and ARHR are caused by inactivating mutations in the PHEX and DMP1 genes, respectively, leading to elevated expression of FGF23 in bone.13, 14, 18 Another form of ARHR is caused by loss-of-function mutations in the ENPP1 gene, which also cause increased expression of FGF23.15, 16 The disease-relevant function of FGF23 in XLH and ARHR has been elucidated in Phex- and Dmp1-deficient mice, in which targeted deletion of FGF23 not only rescues the hypophosphatemic conditions of these animal models, but resembles the FTC phenotype of single Fgf23-null mice.11, 19, 20 Likewise, disruption of Klotho overcorrects hypophosphatemia in the Hyp mouse model,21 further highlighting the functional role of the FGF23/Klotho pathway in XLH and related hypophosphatemic diseases.
In vitro, FGF23/Klotho signaling has been shown to be mediated by FGFR4 or the IIIc isoforms of FGFR1 and FGFR3.6, 7 In vivo, genetic depletion of FGFR family members only partially resembles the loss of Fgf23 or Klotho, owing to embryonic lethal events as well as potential redundancy or compensatory mechanisms within the FGFR family.22–25 However, pharmacological inhibition of FGFRs using a pan-specific FGFR kinase inhibitor effectively blocks FGF23 function in wild-type mice, emphasizing the essential role of FGFR signal transduction in phosphate and 1,25(OH)2D3 homeostasis.26 In Hyp mice, simultaneous deletion of FGFR3 and FGFR4 partially corrects the hypophosphatemic condition.22 Therefore, FGFRs constitute promising therapeutic targets for hypophosphatemic disorders caused by aberrant FGF23 expression.
Hereditary hypophosphatemic diseases typically present in early childhood with short stature and skeletal abnormalities, such as bending deformities of the legs and rickets.27 Current therapeutic approaches to these diseases are mainly limited to treatment with activated vitamin D analogues, such as 1,25(OH)2D3 (Rocaltrol) and 1α-hydroxyvitamin D (Alfacalcidol), and phosphate supplementation, resulting in partial correction of skeletal abnormalities, the extent of which depends on disease severity. Despite therapy, patients typically exhibit reduced adult height and require careful monitoring on treatment to avoid complications such as abdominal pain, diarrhea, secondary hyperparathyroidism, and ectopic calcifications.28
In this study, we preclinically assessed the potential use of FGFR inhibitors for the therapy of FGF23-mediated hypophosphatemic disorders using NVP-BGJ398, a novel selective, pan-specific FGFR inhibitor.29 Using two different mouse models of FGF23-mediated hypophosphatemic rickets, Hyp and Dmp1-null mice,30, 31 resembling the human diseases XLH and ARHR, we find that pharmacological inhibition of FGFRs efficiently suppresses aberrant FGF23 signaling and alleviates the hypophosphatemic and hypocalcemic conditions of these mice. Correspondingly, long-term FGFR inhibition in Hyp mice leads to the normalization of bone mineralization and a striking reorganization of the disturbed growth plate structure. We therefore propose FGFR inhibitor therapy as a potential approach for the treatment of FGF23-mediated hypophosphatemic diseases.
Wild-type C57BL/6 and Hyp (B6.Cg-PhexHyp/J) mice were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). Dmp1-null mice were generated by Feng and colleagues30 and were licensed from the University of Missouri–Kansas City (Kansas City, MO, USA). All mice were kept in cages under standard laboratory conditions with constant temperature of 20 to 24°C and a 12-hour–12-hour light-dark cycle. Mice were fed on a standard rodent diet (3302; Provimi Kliba SA, Penthalaz, Switzerland) containing 1.15% calcium, 0.85% phosphate, and 1000 UI/kg vitamin D with water ad libitum. Protocols, handling, and care of the mice conformed to the Swiss federal law for animal protection under the control of the Cantonal Veterinary Office Basel-Stadt, Switzerland.
FGFR inhibitor treatment
The FGFR inhibitor NVP-BGJ398 is a small molecular weight compound featuring an N-aryl-N′-pyrimidin-4-yl urea motif.29 NVP-BGJ398 (50 mg/kg body weight; Novartis, Basel, Switzerland) or vehicle only (PEG-300 [Applichem]/Glucose 5% [B. Braun, Melsungen, Germany], 2:1 mix) was administered by oral gavage. Mice were used at 5 to 7 weeks of age in the case of single-dose administrations. For long-term treatment over 8 weeks, dosing was initiated at 5 weeks of age with a schedule of three treatments per week (3qw). Mice were anesthetized by isoflurane inhalation and blood was collected from the caval vein. Mice were euthanized by exsanguination and kidney and tibial and femoral bones were obtained. Concentrations of NVP-BGJ398 in kidney at 7 hours and 24 hours posttreatment were determined by liquid chromatography/tandem mass spectroscopy (LC/MS-MS).
Serum was separated from whole blood using clot activator centrifugation tubes (Sarstedt, Nümbrecht, Germany). Serum (100 µL) was used for determination of phosphate and calcium levels using the VetScan diagnostic profiling system (Abaxis, Darmstadt, Germany). Serum concentrations of 1,25(OH)2D3 were determined using a radio receptor assay kit (Immundiagnostic, Bensheim, Germany). FGF23 serum levels were analyzed by an ELISA detecting intact FGF23 (Kainos, Tokyo, Japan). Determination of PTH levels was performed using a mouse PTH ELISA (Immutopics, San Clemente, CA, USA).
RNA purification and quantitative real-time PCR
For isolation of kidney RNA, approximately 60 mg of tissue was homogenized in 1.5 mL RTL buffer (Qiagen, Hilden, Germany) with a rotor-stator homogenizer (Digitana, Yverdon-les-Bains, Switzerland) and RNA was purified with the RNeasy Mini kit. Random hexamer primed cDNA was synthesized with 0.5 to 2 µg RNA and MultiScribe MuLV reverse transcriptase (Applied Biosystems, Carlsbad, CA, USA). Quantitative real-time PCR was performed in an iQ5 Real-Time PCR Detection System (BioRad, Hercules, CA, USA) using a quantitative PCR (qPCR) core kit for probe assay (Eurogentec, Seraing, Belgium) and an equivalent of 40 or 80 ng RNA of each sample. The data were normalized to Gapdh expression. TaqMan assays and primer sequences used are indicated in the Supplemental Material.
Radiography and micro–computed tomography analyses
Radiographs of femur and tibia were taken ex vivo using a high-resolution radiography system (Faxitron MX-20; Faxitron, Buffalo Grove, IL, USA). Micro–computed tomography (µCT) measurements were performed ex vivo using a Scanco vivaCT 40 system (voxel size 6 µm; high resolution; Scanco Medical, Brüttisellen, Switzerland). For cancellous and cortical bone analyses a fixed threshold of 200 was used to determine the mineralized bone fraction from 50 slices. A Gaussian filter was applied to remove noise (σ = 0.7; support = 1). Cancellous bone mineral density and bone volume per tissue volume and trabecular number, thickness, and separation were determined in the distal femur metaphysis. In addition cortical thickness and cortical bone mineral density and bone volume per tissue volume were determined.
Bone histomorphometric analyses
The left femur was fixed for 24 hours in 4% phosphate-buffered paraformaldehyde, dehydrated, defatted at 4°C, and embedded in methylmethacrylate resin. A set of 5 µm nonconsecutive longitudinal sections was cut in the frontal mid-body plane (Leica RM2155 microtome; Leica Microsystems, Heerbrugg, Switzerland). Osteoblast number and osteoid surface per bone surface and osteoid width were determined on Goldner-stained sections in the secondary spongiosa of the distal metaphysis and epiphysis using a Leica DM microscope fitted with a SONY DXC-950P camera and adapted Quantimet 600 software (Leica). Osteoid width was determined in addition at the endocortical surface of the distal metaphysis. Microscopic images were digitized and evaluated semiautomatically on screen (200-fold magnification). Sections were stained for tartrate-resistant acid phosphatase 5b (TRAP5b) activity for determination of osteoclast number per bone surface in the secondary spongiosa of the distal metaphysis and epiphysis using a Merz grid (200-fold magnification). Bone histomorphometric nomenclature was applied as recommended by Parfitt and colleagues.32
All data shown represent mean ± standard error of the mean (SEM). Statistical analyses were performed using Student's t test (two-tailed). The significance level is indicated by asterisks: *p < 0.05; **p < 0.01; ***p < 0.001.
FGF23 is the disease-causing factor in several hypophosphatemic conditions including X-linked hypophosphatemic rickets (XLH).1 We have recently provided evidence for a functional role of FGFRs in renal FGF23/Klotho signaling in vivo by means of pharmacological inhibition of FGFRs using the preclinical tool compound PD173074.26 Here, we aimed to determine whether FGFR inhibition also counteracts pathological FGF23 signaling and to provide preclinical proof of efficacy in hypophosphatemic rickets for NVP-BGJ398, a novel inhibitor in Phase I clinical trials for cancer patients with FGFR genetically altered tumors.33 To this end, we used the Hyp and Dmp1-null mouse models,18, 31, 34 and analyzed the effect of treatment with NVP-BGJ398, which inhibits the kinase activity of all four FGFR family members at nanomolar one-half maximal inhibitory concentration (IC50) values and displays high specificity for FGFRs in cellular kinase profiling assays.29
FGFR inhibition using NVP-BGJ398 suppresses renal FGF23 signaling
FGF23 exerts its hypophosphatemic functions in part by transcriptional regulation of the 1,25(OH)2D3-metabolizing enzymes CYP27B1 and CYP24A1 in the kidney.2, 3 Despite the elevated FGF23 levels present in Hyp mice, Cyp27b1 and Cyp24a1 expression and 1,25(OH)2D3 serum levels in Hyp mice were not significantly different compared to wild-type mice (Fig. 1A–C), in line with previous reports.35, 36 To initially demonstrate the inhibitory effect of NVP-BGJ398 on FGF23 signaling in wild-type and Hyp mice we performed a single-dose, short-term treatment study. Based on the pharmacokinetic profile of the compound (Supplemental Fig. S1) we analyzed renal FGF23 target gene expression at 7 hours postdosing to illustrate the immediate effects of FGFR inhibition before the onset of feedback regulations upon release of pathway inhibition.26 We found that in both wild-type and Hyp mice, NVP-BGJ398 treatment led to increased Cyp27b1 levels and an almost complete loss of Cyp24a1 expression (Fig. 1A, B). Accordingly, this resulted in a strong increase in 1,25(OH)2D3 serum levels in both wild-type and Hyp mice at 7 hours postdosing of NVP-BGJ398 (Fig. 1C). These results illustrate that pharmacological inhibition of FGFRs with NVP-BGJ398 counteracts FGF23 signaling in wild-type and Hyp mice.
In addition, we also observed effects of FGFR inhibitor treatment on PTH serum levels (Supplemental Fig. S2). In wild-type mice PTH levels were significantly increased after 7 hours of NVP-BGJ398 treatment, whereas reduced levels were observed at 24 hours postdosing, consistent with the effect observed with the FGFR inhibitor PD173074.26 In contrast, PTH levels were higher in Hyp mice but not affected by NVP-BGJ398 treatment. Furthermore, we noted a transient repression of FGF23 bone mRNA and serum levels upon NVP-BGJ398 treatment (Supplemental Fig. S3A, B), in line with the previously reported regulatory function of FGFR signaling on FGF23 expression.26, 37
Short-term FGFR inhibition did not affect NaPi-2a and NaPi-2c mRNA levels in the kidney (Supplemental Fig. S4A, B) and NaPi-2a expression in the brush border membrane (Supplemental Fig. S4C). Consequently, no significant changes in urinary phosphate levels were observed over a 24-hour time-course following FGFR inhibition (Supplemental Fig. S4D) and NVP-BGJ398 treatment did not impinge on fractional excretion of phosphate in Hyp mice, whereas it only mildly affected the phosphate filtration rate in wild-type mice (Supplemental Fig. S4E).
NVP-BGJ398 treatment ameliorates the hypophosphatemic conditions of Hyp and Dmp1-null mice
A single dose of NVP-BGJ398 induced elevated serum calcium and phosphate levels in both wild-type and Hyp mice at 24 hours postdosing, thus alleviating the severe hypocalcemia and hypophosphatemia observed in Hyp mice. Serum calcium levels of NVP-BGJ398-treated Hyp mice were indistinguishable from vehicle-treated wild-type mice (Fig. 1D), whereas serum phosphate concentrations in the inhibitor treated group were still significantly lower compared to wild-type mice (Fig. 1E).
A mouse model for ARHR, the genetically engineered Dmp1-null strain, was also examined following FGFR inhibition. As seen with the Phex-deficient Hyp model, renal expression of Cyp27b1 was increased whereas Cyp24a1 levels were repressed upon treatment with NVP-BGJ398 (Supplemental Fig. S5A, B). Moreover, as observed for Hyp mice, FGFR inhibition led to increased serum phosphate and calcium levels in Dmp1-null mice (Supplemental Fig. S5C, D).
Long-term FGFR inhibition enhances body weight and tail length development in Hyp mice
Because single-dose treatments with NVP-BGJ398 alleviated the hypocalcemic and hypophosphatemic phenotypes of Hyp and Dmp1-null mice, we aimed to monitor a potential amelioration of the rickets-like bone phenotypes upon long-term FGFR inhibition. Here, we focused on the Hyp model, given the substantial improvements of mineral ion homeostasis upon single-dose FGFR inhibitor treatment. Treatments were performed over a course of 8 weeks. Owing to the persistence of elevated calcium and phosphate levels for at least 48 hours post-NVP-BGJ398 administration (Supplemental Fig. S6A, B)—extending beyond the clearance of the compound from the kidney (Supplemental Fig. S1)—mice were treated only 3qw with NVP-BGJ398 (50 mg/kg body weight) or vehicle.
Before the onset of therapy treatment, at 5 weeks of age, Hyp mice displayed a reduced body weight compared to wild-type littermates. Although body weight of both vehicle and NVP-BGJ398-treated Hyp mice remained significantly lower compared to wild-type littermates during the course of treatment, pharmacological FGFR inhibition in Hyp mice led to a significant increase in body weight compared to the vehicle control group from day 31 of treatment on (Fig. 2A). Overall, the total body weight gain in NVP-BGJ398-treated Hyp mice was similar to vehicle-treated wild-type mice (Fig. 2B). A shorter tail is a pronounced feature of the hypophosphatemic rickets phenotype of Hyp mice, reflecting the impaired bone formation.36 Again, the tail length of both Hyp groups was significantly shorter compared to wild-type littermates throughout the treatment period. However, during the 8 weeks of treatment NVP-BGJ398-treated Hyp mice displayed a much stronger increase in tail length compared to control Hyp mice (Fig. 2C). Moreover, the tail length gain in Hyp mice treated with the FGFR inhibitor was also significantly higher compared to vehicle-treated wild-type littermates (Fig. 2D).
Long-term therapy with NVP-BGJ398 restores mineral ion homeostasis in Hyp mice
To examine the effect of continuous FGFR inhibition on phosphate and calcium homeostasis in Hyp mice, we analyzed serum calcium and phosphate concentrations at the end of the 8-week study. To distinguish immediate short-term responses to NVP-BGJ398 treatment from steady-state effects of continuous FGFR inhibition, serum was prepared at 24 hours after terminal dosing of the 8-week study, a time point when pharmacological inhibition was relieved based on the pharmacokinetic profile of NVP-BGJ398 in wild-type and Hyp mice (Supplemental Fig. S1). We found that in contrast to single-dose FGFR inhibitor administration (Fig. 1D, E), long-term therapy with NVP-BGJ398 led to a complete normalization of both calcium and phosphate levels in Hyp mice (Fig. 3A, B). Despite the transient repressive effect of FGFR inhibition on FGF23 expression (see Supplemental Fig. S3), long-term treatment with NVP-BGJ398 led to a further increase of FGF23 serum concentrations in Hyp mice (Fig. 3C), which was accompanied by a normalization of PTH levels (Fig. 3D), whereas 1,25(OH)2D3 was not significantly different among the treatment groups (Fig. 3E). Also, renal Klotho expression was not affected by FGFR inhibitor treatment (Supplemental Fig. S7). Taken together, these results illustrate the beneficial effect of pharmacological FGFR inhibition in the context of aberrant FGF23 signaling and point toward an alleviation of the bone formation deficiency of Hyp mice.
FGFR inhibition enhances longitudinal bone growth in Hyp mice
We therefore analyzed the effect of long-term FGFR inhibition on longitudinal growth of femur and tibia by radiography and found that NVP-BGJ398-treated Hyp mice displayed significant elongation of both femur (Fig. 4A, C) and tibia (Fig. 4B, D) compared to the vehicle-treated control group. Still, the enhanced bone growth did not result in femur or tibia sizes comparable to wild-type mice, but FGFR inhibition did partially alleviate the widening of both femoral and tibial growth plate areas, which is typically observed in rickets (Fig. 4A, B).
Long-term NVP-BGJ398 treatment ameliorates osteoid abundance and impaired matrix mineralization in Hyp mice
To determine the effect of FGFR inhibitor treatment on bone structure in more detail we performed µCT analyses of the distal femoral metaphysis. This analysis revealed impaired mineralization of the cortical bone area in Hyp mice, apparent given the gaps and holes within the Hyp femoral cortex structure (Fig. 5A, indicated by arrowheads). Consistent with this observation the relative bone volume within the cortical compartment, which approaches 100% in healthy rodents, was reduced in vehicle-treated Hyp mice compared to wild-type controls (Fig. 5B). Accordingly, cortical bone mineral density was decreased (Table 1). Moreover, animals presented with decreased average cortical thickness (Fig. 5C). In contrast, cortex of NVP-BGJ398-treated Hyp mice appeared intact (Fig. 5A) and relative cortical bone volume was indistinguishable from wild-type mice (Fig. 5B). Also, cortical bone mineral density was partially rescued (Table 1) and cortex thickness was significantly increased compared to vehicle-treated Hyp mice (Fig. 5C).
Table 1. Femoral Bone Structure and Histomorphometric Indices of Wild-Type and Hyp Mice Upon Long-Term Treatment With the FGFR Inhibitor NVP-BGJ398
Data are shown as mean ± SEM and were compared by unpaired Student's t test.
Hyp = hypophosphatemic; FGFR = fibroblast growth factor receptor; NVP-BGJ398 = a novel selective, pan-specific FGFR inhibitor; BV/TV = bone volume/tissue volume; BMD = bone mineral density; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular spacing; OS/BS = osteoid surface/bone surface; O.Wi = osteoid width; N.Obl/BS = number of osteoblasts/bone surface; N.Ocl/BS = number of osteoclasts/bone surface; Ct.Th = cortical thickness at the distal metaphysis.
In line with those observations, the histomorphometric analysis revealed a significant amelioration of the abnormal endocortical osteoid width in Hyp mice (Table 1). Metaphyseal cancellous bone volume and bone mineral density were markedly reduced in Hyp mice owing to a decrease in trabecular number and a concomitant increase in trabecular separation compared to wild-type controls, whereas trabecular thickness was unaltered. FGFR inhibition in Hyp mice did not correct these abnormalities as determined by µCT (Table 1). However, histomorphometric analysis demonstrated that NVP-BGJ398 treatment also significantly normalized matrix mineralization in the cancellous bone compartment as reflected in the reduction in osteoid width and surface present in Hyp mice (Table 1). Because in Hyp mice, irrespective of treatment, the amount of metaphyseal trabecular bone surfaces available for quantitative evaluation was low, we aimed to confirm our histomorphometric findings at a site with higher cancellous bone volume. Visual inspection suggested that this was the case in the distal femur epiphysis, where indeed bone volume, when evaluated on the sections as the sum of the mineralized and unmineralized bone matrix, was higher in all groups compared to the metaphyseal site and not different between groups (Fig. 6A, and data not shown). At this site, NVP-BGJ398 treatment also led to a substantial improvement of matrix mineralization as reflected by a significant decrease in the osteoid surface and width in Hyp mice (Table 1).
Osteoblast number was nonsignificantly elevated in the metaphysis and significantly increased in the epiphysis in Hyp mice. However, at both sites osteoblast number of NVP-BGJ398-treated Hyp mice was indistinguishable from wild-type littermates. Osteoclast count was comparable between all groups in the metaphysis. In the epiphysis, we observed an increased osteoclast number in vehicle-treated Hyp mice, whereas osteoclast count in NVP-BGJ398-treated Hyp mice was comparable to wild-type controls (Table 1). These data indicate that FGFR inhibitor treatment significantly reduced the mineralization defects present in Hyp mice at all skeletal envelopes, as well as any concomitant abnormalities in histomorphometric indices. Taken together, the radiography (Fig. 4), microtomography (Fig. 5, Table 1) and histomorphometric (Fig. 6, Table 1) analyses revealed a favorable effect of FGFR inhibition on longitudinal growth, structural integrity, and mineralization of bone in Hyp mice.
Treatment with NVP-BGJ398 corrects growth plate organization in Hyp mice
In addition, we found an ameliorative effect of NVP-BGJ398 treatment on growth plate organization in tibial histological sections of NVP-BGJ398-treated Hyp mice (Fig. 6A). In vehicle-treated Hyp mice the columnar organization and directional growth of chondrocytes was disturbed in contrast to the highly ordered structure in wild-type mice. In NVP-BGJ398-treated Hyp mice, however, we observed a striking reorganization of the growth plate area (Fig. 6A, left panels), and a reformation of the columnar stacks of chondrocytes along with an increased height of the proliferative zone (Fig. 6A, right panels).
In summary, our data indicate that pharmacological inhibition of FGFRs might be sufficient to inhibit aberrant FGF23 signaling and to alleviate the hypophosphatemic rickets phenotype of XLH and potentially other FGF23-related hypophosphatemic diseases, such as ARHR.
In this study we show that pharmacological inhibition of FGFRs using the novel, pan-specific FGFR inhibitor NVP-BGJ398 counteracts pathological FGF23 signaling, thereby depicting a potential novel therapeutic approach for the treatment of FGF23-mediated hypophosphatemic disorders. In particular, FGFR inhibition corrects hypophosphatemia and hypocalcemia in the Hyp mouse model of XLH. Consequently, long-term treatment with NVP-BGJ398 alleviates the rickets-like bone phenotype in this model and leads to enhanced bone mineralization, normalization of bone turnover and a striking restoration of the growth plate organization.
Pharmacological FGFR inhibition as a potential novel therapeutic approach for FGF23-mediated hypophosphatemic diseases
XLH and other FGF23-mediated hypophosphatemic diseases such as ADHR and ARHR commonly manifest clinically in early childhood with short stature and bowing deformities of the legs.28 Current medical therapy consists of phosphate supplementation and treatment with activated vitamin analogues from time of diagnosis until completion of growth. Although therapy improves growth and rickets in patients, correction is often limited and results in impaired postpubertal height.28 Owing to the persistence of FGF23 signaling—constituting a continuous counteractive force—the administration of high doses of phosphate and vitamin D analogues is required for medical therapy of XLH and other FGF23-mediated hypophosphatemic diseases, necessitating close monitoring and dose adjustments to avoid toxicity risks such as ectopic calcifications or secondary hyperparathyroidism.28, 38 Therefore, directly targeting pathological FGF23 signaling by blocking FGFR signal transduction might provide an advantageous therapeutic approach over the current standard of treatment.
Effects of single-dose short-term FGFR inhibition
To establish an efficacious dose regimen for long-term FGFR inhibitor therapy and to monitor short-term events of blocking renal FGF23 signaling we initially performed single-dose treatments with NVP-BGJ398 in wild-type and Hyp mice. Similar to our previous observation in wild-type mice using an FGFR inhibitor tool compound,26 we noticed immediate effects of FGFR inhibition by NVP-BGJ398 on renal FGF23 signaling. In line with the potent suppressive function of FGF23 on 1,25(OH)2D3 synthesis,2, 3 we observed increased 1,25(OH)2D3 serum levels after treatment with the FGFR inhibitor NVP-BGJ398 in Hyp mice. Pharmacological inhibition of FGFRs also resulted in increased serum calcium and phosphate levels in both Hyp and Dmp1-deficient mice. While FGF23 was reported to inhibit renal reabsorption of phosphate by decreasing the expression of the sodium-phosphate co-transporters NaPi-2a and NaPi-2c in the brush border membrane (BBM) of proximal tubule epithelial cells,4, 5 renal expression of NaPi-2a and fractional excretion of phosphate was not significantly affected by FGFR inhibition in Hyp mice, indicating that the correction of hypophosphatemia might be mediated via intestinal absorption of dietary phosphate in consequence of the increased 1,25(OH)2D3 synthesis.39 The pronounced effect of FGFR inhibition on 1,25(OH)2D3 levels is in line with a more rapid effect of recombinant FGF23 injection in mice on vitamin D metabolism compared to changes in NaPi-2a expression.2 Therefore, short-term pharmacological FGFR pathway inhibition might not be sufficient to induce changes in urinary phosphate excretion in contrast to more sustained FGF23 loss of function approaches, such as genomic depletion or treatment with FGF23-neutralizing antibodies.9, 36
FGF23 signaling directly inhibits PTH expression in the parathyroid gland.40 Consequently, single-dose FGFR inhibitor treatment transiently induced PTH serum levels in wild-type mice after 7 hours of treatment. However, upon clearance of the compound at 24 hours postdosing, PTH levels were reduced in wild-type mice, potentially promoting the hypercalcemic conditions observed in wild-type mice at this time point. Hyp mice showed higher PTH levels compared to wild-type littermates in line with previous observations.19 Interestingly, NVP-BGJ398 treatment did not affect PTH levels in Hyp mice, indicating that PTH does not directly contribute to the effects on mineral ion metabolism observed upon short-term FGFR inhibition in Hyp mice.
Long-term NVP-BGJ398 treatment alleviates the pathological effects of FGF23 in Hyp mice
Noteworthy, the increase in phosphate and calcium levels in Hyp mice persists for at least 48 hours after FGFR inhibitor treatment and thus exceeds the pharmacological inhibition of FGFRs by NVP-BGJ398, which is cleared from the kidney within 24 hours of administration. This illustrates that persistent inhibition of FGFRs is not required for the therapy of FGF23-mediated hypophosphatemia, allowing for intermittent dose regimens. Correspondingly, a 3qw dosing schedule was used in this study. During the treatment period of 8 weeks, we observed increased body weight gain in NVP-BGJ398-treated Hyp mice compared to the vehicle control group, indicating that intermittent FGFR inhibitor treatment was well tolerated. Moreover, long-term treatment with NVP-BGJ398 led to a complete normalization of hypophosphatemia and hypocalcemia and significantly enhanced longitudinal growth of the long bones in Hyp mice. Furthermore, the abundance of osteoid tissue observed in control Hyp mice was markedly reduced in the NVP-BGJ398-treated group owing to a normalization of bone mineralization, resulting in significant rescue of bone mass. Also, FGFR inhibitor led to increased cortex mineralization in Hyp mice and it would be interesting to address in future studies whether the observed improvement of bone geometry and mass translates into increased mechanical bone strength. Body weight and bone growth in Hyp mice receiving NVP-BGJ398 was, however, still lower compared to wild-type mice after 8 weeks of treatment. This is likely owing to the duration of dosing and the age when FGFR inhibitor therapy was initiated. Because the mineral ion defects in Hyp mice were corrected and growth plate organization was normalized by NVP-BGJ398 treatment, we hypothesize that earlier initiation and an extended treatment period could further alleviate or completely reverse the pathological phenotypes of FGF23-mediated hypophosphatemic diseases.
Effects of long-term pharmacological FGFR inhibition on growth plate organization in Hyp mice
A likely on-target effect of systemic FGFR inhibition is expected from the function of FGFR3 in the control of proliferation and differentiation of chondrocytes. Genetic depletion of FGFR3 causes skeletal overgrowth due to increased chondrocyte proliferation.41 Accordingly, we observed an enlargement of the proliferative zone of the growth plate in NVP-BGJ398-treated Hyp mice. Enhanced proliferation of chondrocytes in response to FGFR inhibition may therefore contribute to the increased longitudinal bone growth in Hyp mice treated with NVP-BGJ398. Also, normal phosphate levels are essential for terminal differentiation and subsequent apoptotic clearance of chondrocytes, and phosphate deficiency results in the expansion of hypertrophic chondrocytes in the growth plates of Hyp mice.42, 43 In addition, 1,25(OH)2D3 exerts a compensatory function in the maintenance of a normal growth plate phenotype in NaPi-2a-deficient mice with persistent hypophosphatemia.44 Therefore, the normalization of phosphate levels, the transient increase in serum 1,25(OH)2D3 concentrations, and the potential effect of FGFR inhibition on chondrocyte proliferation most likely cooperatively contribute to the reorganization of growth plate structure in Hyp mice treated with NVP-BGJ398.
Steady-state effects of persistent FGFR inhibitor treatment
We have previously reported that FGF signaling is essential for FGF23 expression in bone and activation of the FGFR pathway was recently linked to the elevated expression of FGF23 in Hyp and Dmp1-deficient mice.26, 37 Correspondingly, FGFR inhibition led to a decrease of both FGF23 mRNA expression and serum protein levels in Hyp mice, but the reduction of FGF23 levels was transient and closely correlated with the pharmacokinetic of FGFR inhibition by NVP-BGJ398. In contrast, Hyp mice treated with NVP-BGJ398 over a course of 8 weeks showed elevated FGF23 levels. As an indication of the steady-state effects of long-term FGFR inhibition, serum parameters at the end of the 8-week treatment period were determined 24 hours after final dosing. At this time point pharmacological inhibition of FGFRs is likely relieved owing to the clearance of the compound. Therefore, this analysis is indicative of the steady-state effect of long-term FGFR inhibition in contrast to the immediate changes observed upon single-dose short-term NVP-BGJ398 treatment. Hence, the increase in FGF23 levels might reflect a feedback regulation as a consequence of the correction of hypophosphatemia in Hyp mice by NVP-BGJ398 treatment. In a similar fashion, persistent inhibition of mitogen-activated protein kinase (MAPK) signaling downstream of FGF23 leads to elevated FGF23 expression and serum levels in Hyp mice.45 Likewise, current therapeutic approaches involving activated vitamin D analogues and phosphate supplementation further induce FGF23 serum levels in XLH patients.46 Because FGF23 directly impinges on PTH expression and secretion in the parathyroid gland,40 elevated levels of FGF23 presumably mediate the normalization of PTH serum concentrations observed in Hyp mice upon long-term NVP-BGJ398 treatment. This might provide a therapeutic benefit compared to phosphate administration, which is associated with the induction of hyperparathyroidism.38, 47 Also, Hyp mice receiving NVP-BGJ398 for 8 weeks showed normal 1,25(OH)2D3 serum concentrations at 24 hours after terminal dosing, indicating that the enhancing effect of FGFR inhibition on 1,25(OH)2D3 synthesis observed upon single-dose treatment does not lead to sustained hypervitaminosis D in continuous therapy, thus depicting another potential advantage compared to vitamin D analogue–based therapy.
Using a different approach, Aono and colleagues36 demonstrated similar therapeutic effects in the Hyp mouse model by applying an FGF23-neutralizing antibody. Compared to systemic inhibition of FGFR signaling, specifically blocking FGF23 function constitutes a more targeted approach for the therapy of XLH, but the persistent antibody-mediated inhibition of the FGF23 pathway raises the concern of inducing a physiological condition resembling FGF23 deficiency, resulting in hyperphosphatemia and associated toxicities.48, 49 In contrast, the transient nature of pharmacological pathway inhibition potentially facilitates the adjustment of phosphate/1,25(OH)2D3 homeostasis in FGF23-mediated hypophosphatemia patients with varying levels of aberrant FGF23 activity. Noteworthy, while FGF23-neutralizing antibodies induce transient hyperphosphatemia in Hyp mice,36, 50 FGFR inhibitor treatment in Hyp mice did not lead to elevations in serum calcium or phosphate beyond levels observed in vehicle-treated wild-type littermates.
In summary, our study indicates the use of pharmacological FGFR inhibition as a potential novel approach for the therapy of FGF23-mediated hypophosphatemic diseases. In particular, NVP-BGJ398 is already applied clinically for cancer indications and thus might hold promise for clinical use in hypophosphatemic disorders in the future. Whereas FGFR inhibition alone might be sufficient to alleviate the pathological effects of aberrant FGF23 signaling, a combination therapy including FGFR inhibitor treatment and phosphate/vitamin D analogue therapy could provide additional benefit and allow the reduction of drug doses. Furthermore, when concomitantly blocking FGF23 signaling, presumably lower doses of phosphate or vitamin D analogues are required, thereby decreasing the risk of adverse effects of therapy. Besides XLH, ADHR, ARHR, and TIO, for which the pathological role of FGF23 is well established, FGFR inhibitor treatment could be of therapeutic use in several other hypophosphatemic syndromes such as epidermal nevus syndrome, osteoglophonic dysplasia, McCune-Albright syndrome, and persistent post-renal transplant hypophosphatemia, which have been associated with increased FGF23 levels.1, 51, 52 In addition, elevated FGF23 levels have recently been reported as the putative causal factor for the development of left ventricular hypertrophy (LVH) and cardiovascular disease in patients with chronic kidney disease (CKD),53 revealing a novel and presumably Klotho-independent pathological effect of aberrant FGF23 signaling. Because concomitant pharmacological inhibition of FGFRs prevented the development of LVH in a rat model of CKD,53 FGFR inhibitor treatment might also be considered as preventive therapy for LVH in CKD patients in future clinical trials.
SW, CH, AT, NB, VG, WRS, FH, MK, and DGP are employees of Novartis Institutes for BioMedical Research. NEH and OB state that they have no conflicts of interest.
We thank R. Rebmann, F. Reimann, A. Studer, P. Ingold, M. Merdes, B. Bohler, D. Sterker, M. Sütterlin, and C. Stoudmann for excellent technical assistance. We are grateful to J. Feng and colleagues for kindly sharing the Dmp1-null mice under license and to I. Kramer for providing technical expertise and for the interpretation of histological data. We thank H. Schmid and B. Hänzi for helpful discussions and critical reading of the manuscript.
Authors' roles: Study design: SW, WRS, FH, MK, and DGP. Study conduct and data collection: SW, CH, OB, AT, and NB. Data analysis and interpretation: SW, CH, OB, VG, NEH, WRS, FH, MK, and DGP. Drafting manuscript: SW and DGP. Revising manuscript content: SW, OB, NEH, FH, MK, and DGP. All authors approved the final version of manuscript. SW and DGP take responsibility for the integrity of the data analysis.