Parathyroid hormone 1 receptor is essential to induce FGF23 production and maintain systemic mineral ion homeostasis

Animals

PTH1R^fl/fl mice, in which exon 1 of the PTH1R was flanked by loxP sites (kindly provided by T. K.), have been described (7). We first crossed hemizygous Prx1cre transgenic males (The Jackson Laboratory, Bar Harbor, ME, USA) with PTH1R^fl/fl females to generate Prx1cre;PTH1R^fl/+ mice. Prx1cre;PTH1R^fl/+ males were then bred with PTH1R^fl/fl female mice to obtain Prx1cre;PTH1R^fl/fl and PTH1R^fl/fl for further analyses. Total DNA was isolated from tail biopsies, and routine PCR was used to genotype various mice. The following primer sequences were used to perform PCR analyses: PTH1R-flox-forward, 5′-ATGAGGTCTGAGGTACATG GCTCTGA-3′, PTH1R-flox-reverse, 5′-CCTGCTGACCTCTCTGAAAGAATGT-3′; and Cre-forward, 5′-CGCGGTCTGGC AGTAAAAACTATC-3′, Cre-rev, 5′-CCCACCGTCAGTACGTGAGATATC-3′. The total body weight of each mouse was recorded weekly starting at birth [postnatal day (P)0]. All studies performed were approved by the Institutional Animal Care and Use Committee at the Harvard Medical School.

Biochemical analyses

Blood was obtained by cheek pouch puncture. Serum and urinary calcium, phosphorus, and creatinine levels were assayed with Stanbio kits (Stanbio Laboratory, Boerne, TX, USA). Serum intact PTH(1–84) and C-terminal FGF23 concentrations were measured with ELISA kits (Immutopics Inc., San Clemente, CA, USA). Serum concentration of intact (i)FGF23 was measured with commercial kits (Kainos Laboratories, Inc., Tokyo, Japan). The ELISA kits for 1,25-dihydroxy vitamin D and cross-linked C-telopeptide (CTX) were purchased from IDS (Fountain Hills, AZ, USA). Urinary cAMP was measured by radioimmunoassay (RIA).

Skeletal mineralization and bone histology

P0 mice were prepared and subjected to Alizarin Red S and Alcian blue staining to analyze the mineralization pattern of the skeleton (25). The hind limbs were fixed in 10% buffered formalin at 4°C overnight. After they were embedded in paraffin, 5 μm sections were cut on an HM360 microtome (Microm, Walldorf, Germany). The sections were stained with hematoxylin (VWR, Radnor, PA, USA) and eosin (Sigma-Aldrich, Carlsbad, CA, USA) (H&E). Von Kossa staining was performed on undecalcified sections of femurs at P0.

Micro-computed tomographic and histomorphometry analyses

Femurs from 3-wk-old animals were used for micro-computed tomographic (CT) analyses with a Scanco Medical micro-CT 35 system (Scanco, Southeastern, PA, USA), with an isotropic voxel size of 7 μm, according to published guidelines (26). The mutant bones were scanned with 300 three-micrometer slices for a total area of 0.9 mm, because the bones were smaller compared with the control littermates. Histomorphometric processing of bone specimens and cancellous bone was performed (27). The femur and L2 vertebrae were fixed in 70% ethanol at 4°C overnight and then embedded in methylmethacrylate. Three-micrometer-thick midsagittal sections were prepared on an HM 360 microtome (Microm). Histomorphometric measurements in the distal ends of femur were performed on sections stained with von Kossa/McNeal using a semiautomatic system (OsteoMeasure; OsteoMetrics, Decatur, GA, USA) and an Axioskop microscope (Zeiss, Oberkochen, Germany) with a drawing attachment. The area within 0.25 mm of the growth plate was excluded from the measurements. The region used for the cortical analysis is a 1 mm-long region from the growth plate (or articular surface region, since mutants lack a growth plate) down into the diaphysis on the anterior side. All histomorphometric parameters were calculated and expressed according to the suggestions made by the American Society of Bone and Mineral Research nomenclature committee (26).

Immunohistochemistry

Slides for FGF23 immunohistochemistry staining were deparaffinized in xylene and rehydrated in an alcohol series (100, 100, 95, and 95%). Sections were immersed in 3% H₂O₂ in methanol for 10 min, blocked in 5% goat serum for 30 min, and incubated with rat anti-FGF23 primary antibody (1:200; Amgen, Thousand Oaks, CA, USA) overnight at 4°C. Tissue was stained with biotinylated goat anti-rat secondary antibody (1:200; Vector Laboratories, Burlingame, CA, USA) for 1 h at room temperature followed by horseradish peroxidase (HRP) substrate (BD Pharmingen, San Jose, CA, USA) and developed with 3,3′-diaminobenzidine (DAB; Vector Laboratories). The slides were counterstained with hematoxylin.

Quantitative real-time PCR

Total RNA was extracted from the cortical region of long bones of the limbs, lumbar vertebrae (L1–L5) and kidney from 3-wk-old animals using Trizol (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. Long bones of the limbs were collected after flushing out the bone marrow. cDNA was generated with SuperscriptRT II (Invitrogen). Quantitative real-time PCR with Taqman and SYBR Green was performed for bone and kidney, respectively, on the StepOnePlus Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). Relative expression was calculated for each gene by the 2^−ΔΔCt method, normalized with β-actin housekeeping gene expression, and presented as fold changes relative to the control.

Western blot analysis

Whole-kidney protein was isolated from fresh kidneys of 3-wk-old mice and homogenized in RIPA buffer (Alfa Aesar, Ward Hill, MA, USA) with 80 mM sucrose and protease inhibitor cocktail tablets (Complete Mini, EDTA-free; Roche, Indianapolis, IN, USA). To assess Napi2a expression, we isolated protein from the brush-border membrane by a published Ca²⁺ precipitation method (28). Protein concentration was measured by the bicinchoninic acid protein assay (Pierce, Rockford, IL, USA), using bovine serum albumen as a standard. Protein samples were heated at 70°C for 10 min in sample buffer containing 2.5 μl NuPAGE LDS buffer and 1 μl NuPAGE Reducing Agent according to the manufacturer’s protocol. These were then subjected to 6% SDS-polyacrylamide gel electrophoresis (Invitrogen). The separated proteins in the gel were transferred electrophoretically to Hybond-P PVDF transfer membranes (GE Healthcare Life Sciences, Pittsburgh, PA, USA). After incubation in blocking solution for 1 h, the membranes were further treated with a rat anti-human Klotho monoclonal antibody (1:1000) (KM2076; Cosmo Bio, Tokyo, Japan), mouse anti-calbindinD_28k monoclonal antibody (1:1000; Sigma-Aldrich), rabbit anti-TRPV5 monoclonal antibody (1:1000; Abcam, Cambridge, MA, USA), and rabbit affinity-purified anti-type 2a NaP_i cotransporter antibody (1:4000; a generous gift of Dr. K. Miyamoto, Tokushima University, Tokushima, Japan). Mouse anti-actin monoclonal antibody (1:5000; Sigma-Aldrich) was used as an internal control. HRP–conjugated anti-rat, anti-mouse, or anti-rabbit IgG was the secondary antibody (Jackson ImmunoResearch Laboratories, West Grove, PA, USA), and signals were detected with the SuperSignal West Pico Chemiluminescent Substrate system (Pierce). The integrated optical density of bands was quantified with the ImageJ software. Each sample was normalized to β-actin.

Single and intermittent PTH(1–34) treatment

Human recombinant PTH(1–34) (Bachem, Torrance, CA, USA) (25 nmol/kg) was subcutaneously injected daily for a 2 wk period starting at P8 (Fig. 6A). Animals of the vehicle group were injected with an equal volume of sterile saline. For a single injection, 50 nmol·kg⁻¹ hPTH(1–34) was injected subcutaneously into 3-wk-old animals. At birth (P0), and 0.5, 1, and 2 h after the final injection, blood was collected by cheek puncture, and the mice were subsequently euthanized for tissue collection.

Statistics

Prism 5.0 (GraphPad Software, Inc., San Diego, CA, USA) was used for statistical analysis. Comparisons between groups were evaluated by unpaired 2-tailed Student’s t test between 2 groups or by 1-way ANOVA followed by Tukey’s test for multiple comparisons. Variables between PTH and vehicle injection groups were evaluated by 2-way ANOVA. All values are expressed as means ± sem. Significance was set at P < 0.05 for all analyses.

Generation of Prx1cre;PTH1R^fl/fl mice

We crossed PTH1R^fl/fl mice with transgenic mice expressing Cre recombinase under the control of the Prx1 promoter. Prx1 is expressed in the mesenchymal condensations that form the developing limbs and calvaria (24). We then interbred the resulting progeny to produce limb mesenchyme-specific PTH1R-knockout mice. PTH1R mRNA levels were significantly decreased in the long bones, but not in spine or kidneys, of 3-wk-old Prx1Cre;PTH1R^fl/fl mice when compared with PTH1R^fl/fl littermates, confirming the efficient and specific deletion of PTH1R in the targeted tissue (Fig. 1A).

Prx1Cre;PTH1R^fl/fl mice were born with the expected rate of Mendelian inheritance, were viable, and had shortened limbs at birth (P0) (Fig. 1B). The mutant mice were growth retarded after birth and remained significantly smaller than the controls (Fig. 1C). Furthermore, although there was a decrease in the survival rate of Prx1Cre;PTH1R^fl/fl animals compared with control littermates, some Prx1Cre;PTH1R^fl/fl mice survived up to 24 wk (Fig. 1D), in contrast to global and chondrocyte-specific PTH1R-knockout mice, in which the mutations are perinatally lethal (6, 7).

Skeletal mineralization and bone histology

We first examined skeletal mineralization by using Alizarin red/Alcian blue staining at P0, revealing a phenotype similar to that observed in global PTH1R-knockout mice (6). At P0, Prx1Cre;PTH1R^fl/fl mice exhibited greatly shortened long bones, reduced cartilaginous areas (blue) and enhanced mineralization (red) in limbs and calvaria (Fig. 2A).

We performed H&E and von Kossa staining at P0 to further characterize the skeletal phenotype of Prx1Cre;PTH1R^fl/fl mice. Mutant mice presented with reduced amounts of both trabecular and cortical bone in their limbs. The femurs were already shortened compared with those of the control littermates (Fig. 2B). Prx1cre;PTH1R^fl/fl mice developed chondrodysplasia resembling PTH1R-knockout mice and Col2cre;PTH1R^fl/fl mice. The femoral growth plate was irregular and shortened, with a lack of columnar chondrocytes (Fig. 2B), and von Kossa staining suggested less mineralized trabecular and cortical bone in the Prx1cre;PTH1R^fl/fl mice compared with control littermates (Fig. 2C). Femurs of 3-wk-old mice did not exhibit a growth plate or a secondary ossification center and had reduced trabecular bone volume (Fig. 2F). Our observations were supported by micro-CT analyses and reconstructed 3-dimensional (3-D) images of femoral trabecular and cortical bone (Fig. 2D, E). Quantification by micro-CT analysis further confirmed that the Prx1cre;PTH1R^fl/fl long bones had significantly reduced trabecular bone volume fraction (BV/TV) compared to control littermates (6.82 ± 1.10 vs. 1.14 ± 0.46%; P < 0.0001) (Fig. 2D). Analysis of the cortical bone at the mid diaphysis also showed a significant reduction in average cortical bone thickness (C.Th) (0.098 ± 0.01 vs. 0.053 ± 0.01 mm, P < 0.0001; Fig. 2Ea, Eb). micro-CT analysis of the cortical bone from the same region used for histomorphometric measurements showed reduced average cortical bone thickness (0.044 ± 0.006 mm) (Fig. 2Ec). Notably, the anterior side of the cortical bone was thicker than the posterior side of the mutant cortical bone where appeared as irregularly woven bone. Histomorphometric analyses were performed on the trabecular and cortical regions of the femur and lumbar spine of 3-wk-old Prx1cre;PTH1R^fl/fl and PTH1R^fl/fl mice. The reduction of trabecular bone in mutant long bones limited the accuracy of histomorphometric measurements, and therefore we focused on the cortical regions. Analysis of the mineralized cortical bone of mutant femurs showed no significant difference in bone area (B.Ar), number of osteocytes (N.Ot), or number of osteocytes per bone area (N.Ot/B.Ar) when compared to those measurements in the control littermates. However, there was a trend of decreased cortical thickness (Fig. 2F and Table 1). We also characterized the trabecular and cortical bone phenotype of lumbar spine of the same animals by histomorphometry as an internal control (Fig. 2G; Table 1). The data showed that there were no significant changes in the bone volume:tissue volume (BV/TV) ratio, trabecular thickness (Tb.Th), trabecular number (Tb.N), and trabecular separation (Tb.Sp). In addition, no changes of N.Ot and N.Ot/B.Ar were observed in the cortical regions of the spine. These results indicate that the observed phenotype is restricted to the long bones of the limbs.

Deletion of PTH1R leads to decreased local FGF23 protein and serum FGF23 levels, despite normal Fgf23 mRNA expression in Prx1cre;PTH1R^fl/fl bones

The main sources of FGF23 are osteoblasts and osteocytes (29). To test whether PTH1R deletion in long bones would alter FGF23 expression in vivo, we measured Fgf23 mRNA expression in cortical bone of mutant and control mice by quantitative (q)RT-PCR. No changes in Fgf23 mRNA expression were detected in the Prx1cre;PTH1R^fl/fl mice, when compared with the controls (Fig. 3A). Fgf23 mRNA expression levels, normalized by additional housekeeping genes (Gapdh, Ppib) and an osteocyte marker Dmp1, were consistent with this observation (Supplemental Fig. 1A). We then analyzed FGF23 protein expression in bone cells by immunohistochemistry (Fig. 3E). A significant decrease in the number of FGF23⁺ osteocytes was observed in Prx1cre;PTH1R^fl/fl when compared to the count in PTH1R^fl/fl mice (Fig. 3B, E). We validated this finding by assessing serum FGF23 levels by ELISA. Circulating iFGF23 and C-terminal FGF23 levels in Prx1cre;PTH1R^fl/fl animals were significantly lower than those in control mice at 3 wk of age (Fig. 3G, H). The decreased serum FGF23 levels and the reduced number of FGF23⁺ cells in mutant bones suggest that lack of PTH1R expression in long bones leads to a sharp reduction of FGF23 production. Fgf23 mRNA expression in the spine (internal control) was up-regulated in mutants, when normalized by β-actin, Gapdh, and Ppib, and showed a tendency to increase when normalized by Dmp1 (Fig. 3C; Supplemental Fig. 1B). However, no changes in FGF23 protein expression were observed in the mutants when compared to the controls (Fig. 3D, F).

Serum and urinary biochemistries

We then measured several serum and urinary parameters to determine whether lack of PTH1R signaling in long bones causes any changes in systemic mineral ion homeostasis. Prx1cre;PTH1R^fl/fl mice were normocalcemic, similar to control animals (Fig. 4A). Serum PTH levels were also unchanged in mutants, corresponding with the normal Ca²⁺ levels (Fig. 4E). Serum phosphorus (Fig. 4B) and 1,25(OH)₂D₃ (Fig. 4F) levels in Prx1cre;PTH1R^fl/fl mice were significantly lower when compared to levels in control littermates. Mutant animals were unexpectedly hypercalciuric and hyperphosphaturic, with a significantly higher urinary concentration of calcium/creatinine and phosphate/creatinine (Fig. 4C, D). Moreover, elevated serum CTX was detected in the mutant mice (Fig. 4G).

Effects on renal genes involved in mineral ion homeostasis

We then sought to identify the cause of the observed increases in wasting of urinary Ca²⁺ and inorganic phosphate (P_i). We analyzed the expression pattern of renal genes known to be involved in the regulation of mineral ion homeostasis, including calbindinD28k, Klotho, and transient receptor potential vanilloid 5 (Trpv5). qRT-PCR analysis showed a significant decrease in mRNA expression of calbindinD28k and Klotho in Prx1cre;PTH1R^fl/fl mice, whereas the expression of Trpv5 mRNA remained unchanged (Fig. 5A). Western blot analysis confirmed the markedly reduced calbindinD28k and Klotho protein levels, whereas TRPV5 protein levels were similar to control levels in Prx1Cre;PTH1R^fl/fl mice (Fig. 5B, C). P_i is primarily reabsorbed in the proximal tubule region by the sodium-dependent phosphate transporters Napi2a and Napi2c (30, 31). We found a significant decrease in Napi2a mRNA and protein levels in Prx1Cre;PTH1R^fl/fl mice (Fig. 5A–C).

Urinary P_i levels increased by 1.8-fold at 0.5 h and 3.4-fold at 1 h after a single injection of hPTH(1–34) into PTH1R^fl/fl mice. In Prx1cre;PTH1R^fl/fl mice, PTH injection induced a robust (2.9-fold) increase in urinary P_i, rising to 2.9-fold at 0.5h and 5.7-fold at 1h after injection. At 1 h, urinary P_i levels peaked earlier than in control animals and remained significantly higher 2 h after PTH injection (Fig. 5D). Furthermore, urinary cAMP levels of Prx1cre;PTH1R^fl/fl mice at basal level were significantly higher than those of control animals (Fig. 5E). hPTH(1–34) injection induced rapid elevation in urinary cAMP levels in both PTH1R^fl/fl and Prx1cre;PTH1R^fl/fl mice. The increase of cAMP in the mutant mice was significantly greater than that in control mice at 0.5 and 1 h after injection, but declined at 2 h to control levels (Fig. 5F).

Response to intermittent hPTH(1–34) injection

We next determined the effects of PTH1R deficiency on FGF23 production in vivo. Prx1Cre;PTH1R^fl/fl and control mice were challenged with daily subcutaneous injections of hPTH(1–34) or vehicle. Daily injections started at 1 wk of age and continued for 14 d. At 3 wk of age, the mice were euthanized, serum biochemistry was assessed, and skeletal specimens were collected (Fig. 6A). The PTH injections had no effect on body size or weight of mutants compared to vehicle-treated mice (Fig. 6B). Control mice injected with PTH had the expected increase in serum calcium and decrease in serum phosphate, confirming the efficacy of the PTH administration (Fig. 6C, D).Serum calcium did not change in the Prx1Cre;PTH1R^fl/fl mice, whereas serum phosphate decreased markedly (Fig. 6C, D). The reduced P_i levels were likely caused by the renal response to PTH, because PTH1R continues to be expressed in mutant kidneys. However, serum Ca²⁺ remained unchanged, suggesting that long bones are an important factor in modulating calcium in response to PTH signaling. PTH treatment significantly increased serum 1,25(OH)₂D₃ levels, in accordance with elevated renal Cyp27b1 and decreased Cyp24a1 in both Prx1Cre;PTH1R^fl/fl mice and controls (Fig. 6F, G).

Loss of PTH1R in long bone causes resistance in FGF23 induction after hPTH(1–34) administration

Previous reports showed that intermittent PTH administration increases FGF23 transcription and secretion (32) and also suggested that Nurr1 mediates PTH regulation of Fgf23 transcription (22). We tested whether PTH1R signaling in bone is necessary to induce FGF23 in vivo by measuring Fgf23 mRNA expression and serum levels in Prx1Cre;PTH1R^fl/fl and control mice after PTH administration. Fgf23 mRNA expression was robustly upregulated in control long bones, which was accompanied by an elevation of Nurr1 transcription, but Fgf23 and Nurr1 expression remained unchanged in Prx1Cre;PTH1R^fl/fl long bones (Fig. 7A, B). Both mutant and control mice had a significant increase in Fgf23 mRNA expression in spinal bone (Fig. 7C). These data indicate that local loss of PTH1R results in the failure of Fgf23 induction.

We then investigated the association of increased Fgf23 transcription with elevated serum FGF23 levels by ELISA. Injection of hPTH(1–34) led to significantly increased iFGF23 serum levels in the control mice, but failed to do so in Prx1Cre;PTH1R^fl/fl mice (Fig. 7D).Serum C-terminal FGF23 levels were significantly elevated in both control and mutant mice (Fig. 7E).The increase in C-terminal FGF23 levels in Prx1Cre;PTH1R^fl/fl mice occurred within 1 hour after the final injection (Fig. 7F), FGF23 levels did not increase until 2 hours after injection in controls (Fig. 7F,G).

Secretion and degradation of FGF23 protein is dependent on proper glycosylation by ppGalNAc-T3 (Galnt3), which protects iFGF23 from cleavage (16).We found that hPTH(1–34) treatment significantly down-regulated Galnt3 mRNA expression levels in PTH1R^fl/fl long bones, suggesting a role for PTH in balancing intact and C-terminal FGF23 levels (Fig. 7H). Galnt3 expression in bones of Prx1Cre;PTH1R^fl/fl mice was markedly reduced before injections and did not diminish further (Fig. 7H).