Conditional Deletion of Murine Fgf23: Interruption of the Normal Skeletal Responses to Phosphate Challenge and Rescue of Genetic Hypophosphatemia

Abstract

The transgenic and knockout (KO) animals involving Fgf23 have been highly informative in defining novel aspects of mineral metabolism, but are limited by shortened lifespan, inability of spatial/temporal FGF23 control, and infertility of the global KO. To more finely test the role of systemic and genetic influences in FGF23 production, a mouse was developed that carried a floxed ("f")-Fgf23 allele (exon 2 floxed) which demonstrated in vivo recombination when bred to global-Cre transgenic mice (eIIa-cre). Mice homozygous for the recombined allele ("Δ") had undetectable serum intact FGF23, elevated serum phosphate (p < 0.05), and increased kidney Cyp27b1 mRNA (p < 0.05), similar to global Fgf23-KO mice. To isolate cellular FGF23 responses during phosphate challenge, Fgf23(Δ/f) mice were mated with early osteoblast type Iα1 collagen 2.3-kb promoter-cre mice (Col2.3-cre) and the late osteoblast/early osteocyte Dentin matrix protein-1-cre (Dmp1-cre). Fgf23(Δ/f) /Col2.3-cre(+) and Fgf23(Δ/f) /Dmp1-cre(+) exhibited reduced baseline serum intact FGF23 versus controls. After challenge with high-phosphate diet Cre(-) mice had 2.1-fold to 2.5-fold increased serum FGF23 (p < 0.01), but Col2.3-cre(+) mice had no significant increase, and Dmp1-cre(+) mice had only a 37% increase (p < 0.01) despite prevailing hyperphosphatemia in both models. The Fgf23(Δ/f) /Col2.3-cre was bred onto the Hyp (murine X-linked hypophosphatemia [XLH] model) genetic background to test the contribution of osteoblasts and osteocytes to elevated FGF23 and Hyp disease phenotypes. Whereas Hyp mice maintained inappropriately elevated FGF23 considering their marked hypophosphatemia, Hyp/Fgf23(Δ/f) /Col2.3-cre(+) mice had serum FGF23 <4% of Hyp (p < 0.01), and this targeted restriction normalized serum phosphorus and ricketic bone disease. In summary, deleting FGF23 within early osteoblasts and osteocytes demonstrated that both cell types contribute to baseline circulating FGF23 concentrations, and that targeting osteoblasts/osteocytes for FGF23 production can modify systemic responses to changes in serum phosphate concentrations and rescue the Hyp genetic syndrome. © 2016 American Society for Bone and Mineral Research.

Keywords: CRE-RECOMBINASE; FGF-23; KLOTHO; OSTEOBLAST; OSTEOCYTE; PHOSPHATE; VITAMIN D; XLH.

Figure 1. Phenotypes following flox-Fgf23 recombination

(A) (upper panel): Schematic of Flox-Fgf23 allele showing Fgf23 exons 1–3, positions of genotyping and sequencing forward (F1) and reverse (R1, R2) PCR primers before (top) and after (bottom) Cre-mediated recombination; location of LoxP sites surrounding Fgf23 exon 2 (triangles) and remaining Flp recognition target (FRT) site (open arrowhead) shown; (lower panel): Gel electrophoresis of PCR amplification encompassing the proximal LoxP site (primers F1/R1) demonstrated PCR products discernable for wild type (‘+/+’), heterozygous (‘f/+’), or homozygous (‘f/f’) flox-Fgf23 alleles. (B) PCR amplification of genomic DNA (primers F1/R2) resulted in products of the expected sizes from Fgf23^f/f alleles (1978 bp) and the recombined Fgf23^Δ/Δ alleles (743 bp) (DNA markers are on left, in bp: lambda virus HindIII fragments and 100 bp DNA ladder); (C) Radiographs showed that compared to Fgf23^f/f (f/f eIIa-cre⁻) and Fgf23^Δ/+ (Δ/+ eIIa-cre⁻) mice, Fgf23^Δ/f/ella-cre⁺ (Δ/f ella-cre⁺) mice were smaller, and had: (D) undetectable serum FGF23, significantly elevated serum phosphate, similar serum calcium, and elevated kidney vitamin D 1α-OHase (*p<0.05 and ╪p<0.01 vs (f/f) and (Δ/+) respectively, n=3–5 mice per group).

Figure 2. Bone cell-specific targeting of Fgf23 and responses to phosphate challenge

(A) Serum intact FGF23 was tested in Fgf23^Δ/f/Col2.3-cre⁻ (Δ/f Cre⁻) and Fgf23^Δ/f/Col2.3-cre⁺ (Δ/f Cre⁺) mice receiving a control or a high phosphate (‘HighP’) diet. Baseline serum intact FGF23 was reduced in the Fgf23^Δ/f/Col2.3-cre⁺ mice at 8 weeks, and with 2 week HighP diet challenge was suppressed compared to Fgf23^Δ/f/Col2.3-cre⁻ mice (**p<0.01 vs control diet same genotype, and ^╪╪p<0.01 vs Cre⁻ same diet; n=3–9); (inset) Fgf23 mRNA was significantly reduced in Fgf23^Δ/f/Col2.3-cre⁺ mice versus Cre⁻ controls (*p<0.05); (B) Fgf23^Δ/f/Dmp1-cre⁺ mice had lower serum intact FGF23 at 8 weeks of age versus Fgf23^Δ/f/Dmp1-cre⁻ while receiving control diet; (inset) femur Fgf23 mRNA was not statistically different between Cre⁺ or Cre⁻ mice. Following high phosphate diet, the rise of serum intact FGF23 was blunted in Fgf23^Δ/f/Dmp1-cre⁺ mice (**p<0.01 vs control diet same genotype; ^╪╪p<0.01 vs Cre⁻ same diet; n=8–12); (C) Serum phosphate was elevated in the mice receiving HighP diet, and was significantly higher in Fgf23^Δ/f/Col2.3-cre⁺ versus all other groups; Fgf23^Δ/f/Dmp1-cre⁺ mice receiving HighP diet had a trend towards elevated serum phosphate and was significantly different from Fgf23^Δ/f/Dmp1-cre⁻ mice receiving the HighP diet (*p<0.05 and **p<0.01 vs control diet same genotype, ^╪p<0.05 and ^╪╪p<0.01 vs Cre- same diet; n=3–12); (D) Serum calcium was not different across genotypes or diets; (E) Alkaline phosphatase was reduced in the Fgf23^Δ/f/Col2.3-cre⁺ mice receiving HighP diet (*p<0.05 vs control diet same genotype; ^╪p<0.05 vs Cre⁻ same diet); (F) Serum intact PTH was elevated in Fgf23^Δ/f/Col2.3-cre⁻ mice receiving the HighP diet, whereas no change was observed for the Fgf23^Δ/f/Col2.3-cre⁺ mice. Only the Fgf23^Δ/f/Dmp1-cre⁻ mice receiving HighP diet had a significant increase in serum PTH (*p<0.05 vs control diet same genotype).

Figure 3. Crossing conditionally-targeted Fgf23 alleles onto the Hyp genetic background

(A) Gross phenotyping showed that the Hyp/Fgf23^Δ/f/Col2.3-cre⁺ were similar to Fgf23^Δ/+/Col2.3-cre⁺ (phenotypically normal mice) with normal body size and tail length (arrow) versus Hyp mice; (B) quantification confirmed Hyp/Fgf23^Δ/f/Col2.3-cre⁺ femur and tail lengths were normalized versus Hyp and not different from Fgf23^Δ/+/Col2.3-cre⁺ mice (n=3–7 males; **p<0.01 Hyp vs. Fgf23^Δ/+/Col2.3-cre⁺ and Hyp/Fgf23^Δ/f/Col2.3-cre⁺); (C) Hyp/Fgf23^Δ/f/Col2.3-cre⁺ mice had significantly reduced basal intact serum FGF23 versus Fgf23^Δ/+/Col2.3-cre⁺ and Hyp from weaning to 8 weeks of age (**p<0.01 vs Fgf23^Δ/+/Col2.3-cre^{+; ╪}p<0.05,^╪╪p<0.01 vs Hyp/Fgf23^Δ/f/Col2.3-cre⁺; n=5–9); (D) Serum phosphate in Hyp/Fgf23^Δ/f/Col2.3-cre⁺ mice was corrected compared to the hypophosphatemia in Hyp and not different from Fgf23^Δ/+/Col2.3-cre⁺ mice. Additionally, serum PTH was highly elevated in Hyp compared to Fgf23^Δ/+/Col2.3-cre⁺ mice. These levels were corrected in the Hyp/Fgf23^Δ/f/Col2.3-cre⁺ mice (*p<0.05, **p<0.01 vs Fgf23^Δ/+/Col2.3-cre⁺, ^╪╪p<0.01 vs Hyp; n=7–9).

Figure 4. Skeletal μCT and histomorphometry of Hyp/Fgf23^Δ/f/Col2.3-cre⁺ mice

(A) As assessed by μCT, compared to Fgf23^Δ/+/Col2.3-cre⁺ mice the metaphyses of Hyp/Fgf23^Δ/f/Col2.3-cre⁺ lacked marrow space or trabeculae definition due to the abundant bone. Hyp mice had lack of trabecular bone and misshapen femoral head. (B) Mid-diaphysis images showed reduced porosity and higher bone area relative to total area in Hyp/Fgf23^Δ/f/Col2.3-cre⁺ versus Hyp; (C) Von Kossa staining of femora qualitatively revealed that Hyp/Fgf23^Δ/f/Col2.3-cre⁺ mice had resolved the Hyp widened growth plate, distended bone, and osteomalacia characteristic of this model; and (D) Histomorphometry demonstrated Hyp/Fgf23^Δ/f/Col2.3-cre⁺ bone had dynamic bone formation rates similar to Fgf23^Δ/+/Col2.3-cre⁺ as determined by dual fluorescence labeling whereas Hyp mice had virtually no label uptake (upper panels). Enlarged regions of label are boxed and shown beneath their respective genotypes (lower panels).