Murine Fetal Serum Phosphorus is Set Independent of FGF23 and PTH, Except in the Presence of Maternal Phosphate Loading

Abstract

Fibroblast growth factor 23 (FGF23) appears to play no role until after birth, given unaltered phosphate and bone metabolism in Fgf23- and Klotho-null fetuses. However, in those studies maternal serum phosphorus was normal. We studied whether maternal phosphate loading alters fetal serum phosphorus and invokes a fetal FGF23 or parathyroid hormone (PTH) response.

C57BL/6 wild-type (WT) female mice received low (0.3%), normal (0.7%), or high (1.65%) phosphate diets beginning 1 week prior to mating to WT males. Fgf23^+/- female mice received the normal or high-phosphate diets 1 week before mating to Fgf23^+/- males. One day before expected birth, we harvested maternal and fetal blood, intact fetuses, placentas, and fetal kidneys.

Increasing phosphate intake in WT resulted in progressively higher maternal serum phosphorus and FGF23 during pregnancy, while PTH remained undetectable. Fetal serum phosphorus was independent of the maternal phosphorus and PTH remained low, but FGF23 showed a small nonsignificant increase with high maternal serum phosphorus. There were no differences in fetal ash weight and mineral content, or placental gene expression.

High phosphate intake in Fgf23^+/- mice also increased maternal serum phosphorus and FGF23, but there was no change in PTH. WT fetuses remained unaffected by maternal high-phosphate intake, while Fgf23-null fetuses became hyperphosphatemic but had no change in PTH, skeletal ash weight or mineral content.

In conclusion, fetal phosphate metabolism is generally regulated independently of maternal serum phosphorus and fetal FGF23 or PTH. However, maternal phosphate loading reveals that fetal FGF23 can defend against the development of fetal hyperphosphatemia.

In the adult, the key phosphate-regulating hormones are parathyroid hormone (PTH), calcitriol, and fibroblast growth factor-23 (FGF23), while the main target organs are the kidneys and intestines (1, 2). Parathyroid hormone lowers serum phosphorus by downregulating the expression and activity of sodium-phosphate transporters (NaPi2a and NaPi2c) in the kidney tubules, thereby causing increased phosphaturia. Calcitriol increases serum phosphorus by stimulating intestinal phosphate absorption (via increased NaPi2b expression) and reabsorption of phosphate by the kidney tubules (via increased NaPi2a and NaPi2c expression). FGF23 lowers serum phosphorus by downregulating renal expression of NaPi2a and NaPi2c to cause phosphaturia. FGF23 also inhibits the synthesis and stimulates the catabolism of calcitriol, thereby leading to reduced intestinal phosphate absorption.

The importance of these hormones in the adult is evident by the disturbances that occur when any of them is present in excessive amounts or absent. Excess PTH or FGF23 action cause hypophosphatemia and renal phosphate wasting (1, 3–5), whereas absence of either hormone leads to hyperphosphatemia and hypophosphaturia (3, 4, 6). Loss of calcitriol, the vitamin D receptor (VDR), and vitamin D deficiency each cause hypophosphatemia and renal phosphate wasting (aggravated by a compensatory increase in PTH), whereas excess calcitriol and hypervitaminosis D lead to hyperphosphatemia (7, 8).

Fetal phosphate metabolism is regulated differently from the adult (9). This is apparent in the fetal circulation by increased serum phosphorus relative to maternal values, very low concentrations of PTH and calcitriol, the low to normal concentrations of intact FGF23, and high levels of PTH-related protein (PTHrP) (9).

The kidneys and intestines are a minor circuit for fetal mineral metabolism, with only the urine (the main constituent of amniotic fluid) being available for swallowing and then absorption through the intestines (9). The placenta is the key target organ through its active transport of phosphate, such that over 85% of the skeletal phosphate content of a full-term fetus is delivered during the third trimester (9). Mammalian fetuses typically have high serum phosphorus, which in human babies is typically 0.5 mmol/l higher than the maternal concentration (9). High serum phosphorus contributes to endochondral bone development by inducing apoptosis of hypertrophic chondrocytes (10, 11) and by being laid down by osteoblasts in osteoid before calcium binds to it (10, 12, 13). Therefore, high or low serum phosphorus in the fetus can be anticipated to disrupt the formation and mineralization of the fetal skeleton.

What hormones or factors regulate fetal phosphate metabolism or placental phosphate transport? PTH and PTHrP have some influence in murine fetuses, because the absence of parathyroids, PTH, or PTHrP each lead to a similar increase in serum phosphorus above wild-type (WT) values (14–17). However, neither PTH nor PTHrP stimulate placental phosphate transport (9). Calcitriol is not required for fetal phosphate metabolism since serum phosphorus, PTH, skeletal morphology, and mineralization were all normal in murine fetuses that lacked either the VDR (18, 19) or the ability to synthesize calcitriol (20). Vitamin D deficiency and loss of VDR also do not disturb placental phosphate transport (9). Surprisingly, FGF23 is not required to regulate fetal phosphate metabolism either, because neither loss of FGF23 action (through genetic ablation of FGF23 or its co-receptor Klotho), nor excess FGF23 (in a mouse model of X-linked hypophosphatemic rickets), disturbed fetal phosphate parameters (21–23). It is only after birth that disturbances of FGF23 physiology cause altered phosphate metabolism (9, 22). Excess FGF23 causes hypophosphatemia, renal phosphate wasting, and altered expression of NaPi2a and NaPi2c within 12 hours of delivery, whereas loss of FGF23 or Klotho took at least 5 days to have the expected opposing effects (22).

Collectively, these findings imply that factors independent of FGF23, PTH, and calcitriol regulate fetal serum phosphorus, placental phosphate transport, and, thereby, accretion of phosphate into bone (9). The rate of phosphate delivery across the placenta may override any effects that the absence of PTH, calcitriol, FGF23, or Klotho and excess FGF23 might otherwise have on the fetal kidneys or intestines. It is in the hours to days after birth that these hormones become important phosphate regulators through their actions on kidneys and intestines.

However, in prior studies of the effects of the absence of FGF23 or Klotho on fetal phosphate metabolism, the mothers were heterozygous with a serum phosphorus that was not significantly higher than in WT mothers (21, 22). Conversely, in studies of the effects of excess FGF23, the mothers were hypophosphatemic compared with WT mothers (21–23). At no time were hyperphosphatemic mothers compared with normophosphatemic mothers to determine whether presenting fetuses with an increased phosphate load would cause a compensatory increase in fetal FGF23 or PTH to normalize fetal phosphate homeostasis. Increased dietary phosphate intake has been shown to increase FGF23 and PTH in adult mice (24), but the effect of such dietary manipulation has not been examined during pregnancy.

In the current studies, we examined whether low, normal, or high dietary phosphate intake would alter maternal or fetal serum phosphorus, FGF23, and PTH, or the placental expression of phosphate-relevant genes. A possible response of fetal FGF23 prompted us to examine whether genetic loss of FGF23 would disturb the ability of fetuses to maintain normal phosphate parameters as compared with their WT littermates when faced with maternal dietary phosphate loading.

In this report, “phosphorus” is used when referring to its presence in blood and amniotic fluid, consistent with the fact that the assays report the phosphorus concentration. Conversely, phosphate is used to refer to the content in the diet and discussion of overall physiology. Serum contains mainly inorganic phosphates (dihydrogen and monohydrogen phosphate) and bone contains phosphate largely in the form of hydroxyapatite, while the soft tissues and extracellular fluids contain organic phosphates complexed with carbohydrates, lipids, and proteins (25).

Materials and Methods

Animal husbandry

In the first series of experiments, thirty C57BL/6 female mice were separated into three groups to receive low (0.3%), normal (0.7%) or high (1.65%) phosphate diets (Teklad/Envigo, Indianapolis, Indiana). One week later, maternal blood was collected and the mice were mated to C57BL/6 males. One day before expected birth, maternal and fetal blood were collected, while intact fetuses and placentas were harvested for analysis.

The results of the first experiments prompted additional studies using the previously described Fgf23 ablation model (26). The colony has been maintained by breeding heterozygous-deleted mice together, and periodically back-crossing into the parent strain (C57BL/6). Genotyping of Fgf23 fetuses and weanlings was done by polymerase chain reaction (PCR) on DNA extracted from tail clips using previously described primer sequences (27, 28). Twenty Fgf23^+/- female mice were separated into 2 groups. Half were placed on the high-phosphate (1.65%) diet, whereas the other half were maintained on an ingredient-matched normal phosphate (0.7%) diet (Teklad/Envigo). One week later, prepregnancy maternal blood was collected, and the mice were mated. Subsequent collection of maternal and fetal blood and tissues (intact fetuses, placentas, fetal kidneys) were carried out 1 day before expected birth (embryonic day [ED] 18.5), except for amniotic fluid being collected a day earlier (ED 17.5).

The matings were timed overnight; the presence of a vaginal mucus plug on the morning after mating marked ED 0.5. Normal gestation is 19 days, and so the fetal analyses were done at ED 18.5 (except ED 17.5 for amniotic fluid).

The Institutional Animal Care Committee of Memorial University of Newfoundland approved all procedures involving live animals.

Chemical and hormone assays

Serum and amniotic fluid were collected using methods as previously described (14, 29). Serum and amniotic phosphorus were analyzed using a colorimetric assay (Sekisui Diagnostics PEI Inc, Charlottetown, Prince Edward Island). Enzyme-linked immunosorbent assays were used to measure PTH 1–34 (Immutopics, San Clemente, California) and intact FGF23 (Kainos, Bunkyo City, Japan). Any values below the assay sensitivity were reset to values that equaled the respective assay’s detection limit; similarly, any values above the range of the assay were reset to equal the upper limit.

Fetal ash and skeletal mineral assay

As previously described (15), intact fetuses were reduced to ash in a furnace at 500°C for 24 hours. A Perkin Elmer 2380 Atomic Absorption Flame Spectrophotometer assayed the calcium and magnesium content of the ash, while the colorimetric assay mentioned above (Sekisui) was used to measure phosphorus content.

Histology

Placentas were snap frozen, then thawed, fixed in 10% buffered formalin, and embedded in paraffin. Von Kossa staining for phosphate was performed on 5-micron deparaffinized sections using 3% aqueous silver nitrate solution, 1-hour exposure to bright light, followed by 5 minutes of incubation with 2.5% sodium thiosulfate, and counterstaining with 2% methyl green. Although often used as a means of detecting calcium, the von Kossa procedure directly detects phosphate by its creation of silver-phosphate complexes; the presence of calcium is inferred because calcium is normally bound to phosphate.

Images were taken at 5x and 20x magnification and converted to gray scale. Image J 1.53A (National Institutes of Health, Bethesda, Maryland) was used to analyze the staining. All images were analyzed under the same manual settings for upper and lower threshold and identical area. Data are expressed as the percentage of area stained black from von Kossa.

Ribonucleic acid extraction and real-time quantitative-PCR

Placentas and fetal kidneys were snap-frozen in liquid nitrogen. Total RNA was purified using the RNeasy Midi Lipid Kit (Qiagen, Toronto, Ontario). Ribonucleic acid quantity and quality were confirmed with the Agilent 2100 BioAnalyzer (Agilent Technologies, Santa Clara, California). We used TaqMan Gene Expression Assays (with the manufacturer’s predesigned primers and probes for optimal amplification), and Fast Advanced Master Mix from Applied BioSystems (ABI/Life Technologies, Burlington, Ontario) to determine expression of Cyp27b1, Cp24a1, Ca²⁺-ATPase (Pmca1), calbindin-D9k (S100g), Trpv6, Fgf23, Klotho, NaPi2a, NaPi2b, NaPi2c, Xpr1, Pit1, and Pit2. Details of conditions and cycle times have been previously reported (16, 30, 31). In brief, complementary DNA (cDNA) was synthesized using the Taqman High Capacity cDNA Reverse Transcription Kit (ABI), and multiplex real time quantitated PCR (qPCR) reactions were run in triplicate on the ViiA 7 Real-Time PCR System (ABI) (16, 18). The minimum sample size was 5. Relative expression was determined from the threshold cycle (C_T) normalized to the reference gene.

Statistical analysis

Data were analyzed using StatPlus:Mac Pro, Build 6.7.1.0 (AnalystSoft Inc, Vancouver, British Columbia). qPCR data were analyzed by the comparative C_T method (2^ΔCT) (32). Analysis of variance (ANOVA) was then used for analysis of all biochemical, hormonal, ash, quantified histological, and qPCR data, with Tukey’s post hoc test used to determine which pairs of means differed significantly. Two-tailed probabilities are reported as mean ± standard error (SE) for fetal data except mean ± standard deviation (SD) for qPCR results.

Results

Effect of low, normal, or high phosphate intake on normal mothers and fetuses

Increasing maternal phosphate intake resulted in progressively higher maternal serum phosphorus during both prepregnancy and pregnancy (Fig. 1A). Moreover, the low phosphate diet caused a nonsignificant reduction in serum phosphorus during pregnancy as compared with prepregnancy (Fig. 1A). Maternal intact FGF23 similarly showed a significant increase with high phosphate intake as compared with low phosphate intake, both during prepregnancy and pregnancy (Fig. 1B). Parathyroid hormone was unaffected during prepregnancy regardless of phosphate intake, and followed the expected pattern of becoming undetectable during pregnancy (Fig. 1C) (33). Overall, the results confirm that increased phosphate intake in turn increases maternal serum phosphorus during prepregnancy and pregnancy, and that this provokes a compensatory increase in FGF23 but no change in PTH.

Figure 1.

Biochemical and hormonal responses in WT mothers exposed to low, normal, and high-phosphate diets. Increasing phosphate intake caused an increase in maternal serum phosphorus during prepregnancy and pregnancy (A), accompanied by increased FGF23 (B). Maternal phosphate intake did not alter PTH, which became suppressed to below the detection limit in all as expected during pregnancy, hence no visible error bars (C). Note that low phosphate intake caused apparent maternal hypophosphatemia during pregnancy, although the value did not reach statistical significance (A). The numbers of observations are indicated in parentheses. Abbreviations: PTH, parathyroid hormone; WT, wild-type.

The fetal response to phosphate loading differed markedly from the maternal values. Fetal serum phosphorus was maintained at the same high value independent of the maternal serum phosphorus (Fig. 2A). Fetal FGF23 and PTH remained at low values and showed no significant changes in response to low or high maternal serum phosphorus (Fig. 2B and 2C). However, a trend for higher serum FGF23 was apparent in fetuses whose mothers were hyperphosphatemic as compared with fetuses of mothers that were hypophosphatemic, and this finding prompted the second set of experiments, below.

Figure 2.

Fetal response to dietary phosphate loading in WT mothers. Unlike their mothers, which achieved a progressive increase in serum phosphorus in response to increased dietary intake, the fetuses maintained the same high serum phosphorus value independent of the maternal serum phosphorus (A). Fetal FGF23 and PTH remained at low values and did not change significantly in response to the maternal serum phosphorus or dietary assignment (B,C). Similarly, the incorporation of mineral into the developing fetal skeleton was unaltered by maternal phosphorus intake, as shown by the normal ash weight (D), and the normal ash content of calcium (E) and phosphate (F). Error bars in B and C are too narrow to be discerned. The numbers of observations are indicated in parentheses. Abbreviations: PTH, parathyroid hormone; WT, wild-type.

Further assessment determined that there were no differences in fetal ash weight, or the mineral content of the ash (Fig. 2D and 2F). The low phosphate diet caused a reduction in placental expression of Trpv6; apart from this, there were no other changes in placental expression of genes known to be relevant to phosphate or calcium transport (Table 1). In particular, the high-phosphate diet did not cause any changes in placental gene expression compared to the normal phosphate diet. Genes examined included the known phosphate transporters NaPi2a, NaPi2b, NaPi2c, Pit1, Pit2, and Xpr1 (34–38).

Table 1.

Relative expression of placental phosphotropic and calciotropic genes in WT placentas obtained from pregnancies in which WT mothers consumed low, normal, or high-phosphate diets

Gene	Low P Diet	Normal P Diet	High P Diet
Fgf23	1.276 ± 1.153	1.000 ± 0.725	0.501 ± 0.213
Klotho	1.224 ± 0.238	1.000 ± 0.188	0.971 ± 0.171
NaPi2a	1.032 ± 0.301	1.000 ± 0.394	1.392 ± 0.408
NaPi2b	1.054 ± 0.268	1.000 ± 0.249	1.275 ± 0.163
NaPi2c	0.765 ± 0.260	1.000 ± 0.191	0.996 ± 0.201
Cyp24a1	0.645 ± 0.587	1.000 ± 0.601	1.465 ± 0.598
Cyp27b1	0.177 ± 0.215	1.000 ± 1.089	1.164 ± 1.834
Pmca1	0.965 ± 0.168	1.000 ± 0.110	0.994 ± 0.151
S100g	0.877 ± 0.216	1.000 ± 0.361	1.245 ± 0.484
Trpv6	0.607 ± 0.235a	1.000 ± 0.151	1.070 ± 0.090
Xpr1	1.133 ± 0.207	1.000 ± 0.033	0.943 ± 0.141
Pit1	0.785 ± 0.132	1.000 ± 0.255	0.911 ± 0.211
Pit2	0.797 ± 0.289	1.000 ± 0.314	0.751 ± 0.088

Abbreviations: P, phosphate; WT, wild-type.

All values have been normalized to the expression in placentas from the normal diet. Sample sizes were 5 per group.

^a High P vs low P diet, P < 0.003; normal P vs low P diet, P < 0.009.

Effect of normal or high phosphate intake on Fgf23^+/- mothers and Fgf23-null fetuses

To further investigate whether the nonsignificant rise of FGF23 in WT fetuses represented a real compensatory response that was physiologically important, we studied the effect of dietary phosphate loading in Fgf23^+/- mothers and their Fgf23-null and WT fetuses.

A high phosphate intake resulted in a nonsignificantly higher serum phosphorus in Fgf23^+/- mothers during both prepregnancy and pregnancy (Fig. 3A). Maternal intact FGF23 showed a significant increase in response to high phosphate intake (Fig. 3B), while PTH remained unaffected by phosphate intake during prepregnancy and became suppressed as expected during pregnancy (Fig. 3C). Overall, the results were similar to those in WT mothers, demonstrating that increased maternal phosphate intake increases the maternal serum phosphorus during prepregnancy and pregnancy, provoking a compensatory increase in FGF23 but not in PTH.

Figure 3.

Biochemical and hormonal responses in Fgf23^±/- mothers exposed to normal and high-phosphate diets. Increased phosphate intake caused a nominally higher maternal serum phosphorus during prepregnancy and pregnancy (A), which provoked a significant increase in FGF23 during pregnancy (B). Maternal phosphate intake did not alter PTH, which became suppressed as expected during pregnancy (C). The numbers of observations are indicated in parentheses. Abbreviation: PTH, parathyroid hormone.

Loss of fetal FGF23 altered the fetal response to changes in maternal serum phosphorus. Fgf23-null fetuses became hyperphosphatemic whereas once again, WT fetuses showed no change in serum phosphorus (Fig. 4A). Fetal FGF23 was undetectable in Fgf23-null fetuses and did not change in WT fetuses (Fig. 4B), while PTH remained at low values regardless of the maternal serum phosphorus or the fetal genotype (Fig. 4C). There was no change in the renal excretion of phosphorus into the amniotic fluid (Fig. 4D). Despite the higher serum phosphorus in Fgf23-null fetuses exposed to maternal phosphate loading, there were no differences in fetal ash weight or mineral content (Fig. 4E–4G).

Figure 4.

Responses of Fgf23-null fetuses to maternal dietary phosphate loading. Ffg23-null fetuses developed hyperphosphatemia in response to the increased maternal phosphate load, while the other 3 groups (Ffg23-null fetuses from the normal diet and WT fetuses from both diets) were indistinguishable by serum phosphorus (A). Fetal FGF23 was absent in Fgf23-null fetuses (hence no error bars) and did not increase in WT fetuses (B); PTH remained at low values in both (C). Error bars on the high dietary phosphate group in C are too narrow to be discerned. The amniotic fluid phosphorus concentration did not differ significantly among the 4 groups (D). The incorporation of mineral into the developing fetal skeleton was unaltered by maternal phosphate intake or genotype, as shown by the normal ash weight (E) and the normal ash content of calcium (F) and phosphate (G). The numbers of observations are indicated in parentheses. Abbreviations: PTH, parathyroid hormone; WT, wild-type.

Staining of placental sections to detect phosphate revealed a significant increase in perivascular phosphate deposition in WT placentas that had experienced an increased maternal phosphate load (Fig. 5). Fgf23-null placentas from the high-phosphate diet showed a trend for increased staining but did not reach the same intensity as the WT (Fig. 5).

Figure 5.

Mineral deposition in WT and Fgf23-null placentas. Phosphate deposition in placentas was visualized by von Kossa staining, in which silver ions react with phosphate to create a black precipitate. Panel A summarizes the digital analysis of all images. Phosphate deposition was significantly elevated in WT placentas from mothers consuming the high-phosphate diet compared with WT and Fgf23-null placentas from the normal phosphate diet. Fgf23-null placentas from the high-phosphate diet showed a trend for increased deposition but were not significantly different from any value. The numbers of observations are indicated in parentheses. Panel B shows representative images of placentas from the different genotypes and diets. The scale bar in the upper panel indicates 500 microns, while the bar in the lower panel indicates 200 microns. Abbreviations: PTH, parathyroid hormone; WT, wild-type.

With respect to placental gene expression (Table 2), Fgf23 was absent from Fgf23-null placentas, confirming that only fetal cells contribute to its expression there. Fgf23-null placentas from mothers on the high-phosphate diet displayed increases in the expression of Pmca1, Pit1, and Pit2, although the increase in Pit1 expression was marginal, while the increase in Pit2 was also present in Fgf23-null placentas from the normal phosphate diet. Cyp24a1 expression was significantly decreased in Fgf23-null placentas on the high-phosphate diet, but to a similar magnitude on the normal phosphate diet. S100g increased significantly in WT fetuses from mothers on the high-phosphate diet. There were no other changes in placental expression of genes relevant to phosphate or calcium transport.

Table 2.

Relative expression of placental phosphotropic and calciotropic genes in WT and Fgf23-null placentas obtained from pregnancies in which Fgf23^+/- mothers consumed normal or high-phosphate diets

Gene	WT	Fgf23 null	WT	Fgf23 null
	Normal P Diet	Normal P Diet	High-P Diet	High-P Diet
Fgf23	1.000 ± 0.998	Undetectable	1.208 ± 0.975	Undetectable
Klotho	1.000 ± 0.191	0.924 ± 0.229	1.038 ± 0.150	1.176 ± 0.270
NaPi2a	1.000 ± 0.097	0.931 ± 0.193	0.884 ± 0.175	1.145 ± 0.218
NaPi2b	1.000 ± 0.315	0.773 ± 0.161	1.052 ± 0.224	0.869 ± 0.267
NaPi2c	1.000 ± 0.299	0.776 ± 0.211	1.003 ± 0.238	0.699 ± 0.257
Cyp24a1	1.000 ± 0.580	0.367 ± 0.300a	0.362 ± 0.174a	0.202 ± 0.200b
Cyp27b1	1.000 ± 1.037	0.875 ± 0.450	0.415 ± 0.158	0.391 ± 0.104
Pmca1	1.000 ± 0.098	1.094 ± 0.089	1.227 ± 0.298	1.518 ± 0.134c,d
S100g	1.000 ± 0.286	0.923 ± 0.183	1.501 ± 0.298e	1.191 ± 0.332
Trpv6	1.000 ± 0.133	0.828 ± 0.197	0.835 ± 0.267	0.912 ± 0.195
Xpr1	1.000 ± 0.133	1.001 ± 0.129	0.986 ± 0.086	1.131 ± 0.217
Pit1	1.000 ± 0.135	0.911 ± 0.112	1.138 ± 0.071	1.199 ± 0.280f
Pit2	1.000 ± 0.156	1.337 ± 0.190g	1.169 ± 0.195	1.367 ± 0.209h

Abbreviations: P, phosphate; WT, wild-type.

All values have been normalized to the expression in WT placentas from the normal phosphate diet. Sample size was 5 per group.

^a P < 0.06 vs WT from normal diet; ^b< 0.02 vs WT from normal diet; ^cP < 0.008 vs Fgf23^-/- from normal diet; ^d0.002 vs WT from normal diet; ^eP < 0.03 vs Fgf23^-/- from normal diet; ^fP < 0.07 vs Fgf23^-/- from normal diet; ^gP < 0.05 vs WT from normal diet; ^hP < 0.03 vs WT from normal diet.

Regarding gene expression within fetal kidneys, Fgf23 nulls displayed an increase in renal Pmca1 expression compared with WT that was independent of the maternal diets, and an increase in Napi2a that also appeared to be independent of maternal diet (significantly increased on the high-phosphate diet and nonsignificantly increased on the normal diet) (Table 3). Pit2 expression was significantly increased in Fgf23-null fetal kidneys on the high-phosphate diet. There were no other changes in expression of genes relevant to phosphate or calcium transport.

Table 3.

Relative expression of renal phosphotropic and calciotropic genes in WT and Fgf23-null kidneys obtained from fetuses in which Fgf23^+/- mothers consumed normal or high-phosphate diets

Gene	WT	Fgf23 null	WT	Fgf23 null
	Normal P Diet	Normal P Diet	High-P Diet	High-P Diet
Klotho	1.000 ± 0.125	1.115 ± 0.315	0.990 ± 0.094	1.142 ± 0.226
NaPi2a	1.000 ± 0.054	1.290 ± 0.104	0.900 ± 0.200	1.295 ± 0.240a
NaPi2b	1.000 ± 0.117	1.154 ± 0.051	0.867 ± 0.222	1.067 ± 0.291
NaPi2c	1.000 ± 0.174	1.113 ± 0.266	0.894 ± 0.134	0.960 ± 0.219
Cyp24a1	1.000 ± 0.507	1.082 ± 0.629	1.242 ± 0.392	0.774 ± 0.318
Cyp27b1	1.000 ± 0.208	1.016 ± 0.276	0.838 ± 0.305	0.910 ± 0.213
Pmca1	1.000 ± 0.140	1.315 ± 0.088 b	1.153 ± 0.132	1.256 ± 0.123c
S100g	1.000 ± 0.115	0.934 ± 0.172	0.942 ± 0.177	0.933 ± 0.139
Trpv6	1.000 ± 0.156	0.993 ± 0.104	1.056 ± 0.336	1.058 ± 0.088
Pit1	1.000 ± 0.118	1.066 ± 0.225	1.080 ± 0.095	1.203 ± 0.194
Pit2	1.000 ± 0.090	1.094 ± 0.278	1.294 ± 0.234	1.621 ± 0.280d,e

Abbreviations: P, phosphate; WT, wild-type.

All values have been normalized to the expression in WT kidneys from the normal phosphate diet. Sample size was 5 per group.

^a P < 0.02 vs WT from high-P diet; ^bP < 0.005 vs WT from normal P diet; ^cP < 0.03 vs WT from normal P diet; ^dP < 0.003 vs WT from normal P diet; ^eP < 0.02 vs Fgf23^-/- from normal diet.

Discussion

In these studies, we hypothesized that increasing maternal dietary phosphate intake from low to normal to high would alter maternal phosphate homeostasis during pregnancy and, in turn, provide fetuses with an increased phosphate load. If FGF23 or PTH are essential for fetal phosphate regulation, then lowering or increasing maternal serum phosphorus should induce compensatory responses in fetal FGF23 and PTH in the same way that maternal hypocalcemia and hypercalcemia will respectively increase or suppress fetal PTH (9). Furthermore, if fetal FGF23 or PTH defend against hyperphosphatemia, increased maternal serum phosphorus should have an even more pronounced effect to raise serum phosphorus and increase PTH in Fgf23-null fetuses.

We found that in both WT and Fgf23^+/- mothers, increasing maternal phosphate intake caused an increase in maternal serum phosphorus and FGF23, but not in PTH. However, in WT mothers bearing only WT fetuses, this increase in maternal serum phosphorus had no significant effect on the fetal serum phosphorus, FGF23, or PTH. Nor did it affect skeletal ash weight or mineral (calcium and phosphate) content, or the placental expression of genes relevant to fetal phosphate metabolism.

On closer scrutiny, fetal serum FGF23 showed a nonsignificant increase in WT fetuses across the low, normal, and high-phosphate maternal diets, which might represent a true compensatory response. To determine whether this was a real and physiological important increase in FGF23, we investigated whether the absence of fetal FGF23 would affect the response to maternal dietary phosphate loading. We found that Fgf23-null fetuses became hyperphosphatemic when their mothers consumed the high-phosphate diet. Despite their increased serum phosphorus, such Fgf23 nulls (and their WT littermates) had no change in PTH, skeletal ash weight, or mineral content. Phosphate deposition was significantly increased in WT placentas from the high-phosphate diet, whereas Fgf23-null placentas showed a nonsignificant trend for higher phosphate deposition. This may indicate that WT fetuses were more capable of defending against the phosphate load by reducing forward flow into the fetal circulation, or increasing back-flow from the fetal circulation to the placenta. There were only minor changes in placental expression of known genes relevant to fetal phosphate metabolism in response to the high-phosphate diet, including a decrease in expression of Cyp24a1 in WT and Fgf23-null placentas, and increases in the expression of Pmca1, Pit1, and Pit2 in Fgf23-null placentas. There were increases in the renal expression of Pmca1, Napi2a, and Pit2, but these increases were dependent on the fetal genotype and not the maternal diet for Pmca1 and likely also for Napi2a.

Phosphate is abundant in the diet, but certain foods have a very high phosphate content and can lead to high intake. These include dairy products, beef liver, and other organ meats, oysters, sardines, fish roe, beer and ale, sodas and other bottled beverages, cocoa and chocolate, and most processed foods (39). Consequently, a high-phosphate diet models many aspects of the modern Western diet that can be rich in processed foods. The hyperphosphatemia may also model maternal hyperphosphatemia that occurs in the setting of renal failure in pregnancy, although in these studies renal function was normal.

The first set of studies, in WT mothers, confirmed that fetal serum phosphorus is set independently of the maternal serum phosphorus concentration, which ranged from low to high. This may imply the presence of a sensor that sets the fetal serum phosphorus concentration, perhaps by acting within the placenta to regulate the inflow of phosphorus. A similar result has been shown previously with the fetal ionized calcium concentration, which is set independent of the maternal serum calcium and not by the known calcium sensing receptor (9, 40). Thus, both fetal calcium and phosphorus concentrations may be set at their high prenatal levels by sensors that are distinct from those identified to be active in the child and adult. The lack of any change in the expression of phosphotropic genes in the placentas of WT fetuses from the high-phosphate diet suggests that they are not principle genes involved in placental phosphate transport or setting the fetal blood phosphorus level. The recently described phosphate exporter Xpr1 is highly expressed in the placenta, and its global absence leads to fetal hypophosphatemia, reduced skeletal mineral content, and marked phosphate deposition in the placenta (41); however, in our study, its expression is not invoked by WT placentas in response to maternal phosphate loading.

The second set of studies, in Fgf23^+/- mothers, confirmed that fetal FGF23 does play a role in defending the fetus against a maternal phosphate load, since the absence of FGF23 resulted in fetal hyperphosphatemia when the mothers had a high phosphate intake. Thus, fetal FGF23 is 1 of the effectors that helps maintain the fetal serum phosphorus independent of the maternal concentration, but only in the face of maternal phosphate loading. This confirms that the trend for a rise in FGF23 seen in WT fetuses on a high-phosphate diet (in the first set of experiments) was a true physiological response. Wild-type fetuses from Fgf23^+/- mothers did not show an increase in serum FGF23, but it is conceivable that a true increase was obscured by variability of the assay. Genetic background is another source of variability, which was mixed but predominantly C57BL/6 in the Fgf23 colony, compared with pure C57BL/6 mice in the first set of experiments. Furthermore, the mothers were heterozygous for loss of Fgf23 in the second set of experiments, as opposed to WT in the first set, and that may affect the fetal response. Alternatively, it may be that the normal ambient concentration of FGF23 in the fetal circulation was sufficient to defend against high maternal serum phosphorus in the second set of experiments, without requiring an increase in fetal FGF23 concentration.

The second set of studies also confirmed our previous observation that fetal serum phosphorus is normally set independently of maternal serum phosphorus and does not require fetal FGF23 or Klotho (21, 22). This explains why fetal serum phosphorus was not different among Fgf23-null fetuses from the normal diet, WT fetuses from the normal diet, and WT fetuses from the high-phosphate diet (Fig. 4A). The high-phosphate diet disrupted this relationship only when fetal FGF23 was absent.

In addition to fetal serum phosphorus being normal in our prior studies of Fgf23-null and Klotho-null fetuses when compared with their respective WT littermates (21, 22), we found that Phex-null male fetuses, which have 10-fold or higher levels of FGF23, also had normal serum phosphorus (21). Moreover, serum phosphorus was normal in Phex-null male fetuses regardless of whether the mother was WT and normophosphatemic versus Phex^+/- and hypophosphatemic (21). However, in those studies, the respective heterozygous and WT mothers consumed a normal phosphate diet and so the fetuses were not challenged to defend against a greater-than-expected influx of phosphate. Furthermore, we found in those prior studies that FGF23 does not regulate placental phosphate transport, since the rate was normal in both Ffg23-, Klotho-, and Phex-null fetuses compared with their respective WT littermates.

Why did Fgf23-null fetuses become hyperphosphatemic when their mothers consumed a high-phosphate diet? Dietary phosphate loading increased maternal serum phosphorus, which in turn increased the rate of flow across the placenta, as it similarly occurs when maternal serum calcium is increased (9). There was no change in the renal excretion of phosphate into amniotic fluid in response to maternal phosphate loading, which is consistent with our prior studies showing that excess FGF23 and absent FGF23 did not affect phosphate handling by the fetal kidneys (21, 22). Taken all together, the current data are consistent with FGF23’s role being that of stimulating the return of phosphate to the maternal circulation (back-flux), without stimulating the forward flow of phosphate from mother to fetus. This role is only invoked in the presence of maternal hyperphosphatemia. Alternatively, fetal FGF23 may play a role in regulating or responding to a putative phosphate sensor, such that in its absence serum phosphorus rises to a higher level.

Parathyroid hormone is normally suppressed during pregnancy by increased calcitriol, intestinal calcium absorption, and other factors (33). Parathyroid hormone is also normally maintained at low levels in the fetal circulation through the actions of the parathyroid calcium sensing receptor in response to the high fetal serum calcium level (9). This suppression of PTH was maintained in mothers and fetuses during both sets of experiments, which indicates that PTH is not responsive to dietary phosphate loading at either life stage.

Fgf23-null placentas from the high-phosphate diet displayed increased placental expression of Pmca1, which is conceivably a compensatory response to increased fetal serum phosphorus. This may occur to increase placental calcium transport to accompany the increased entry of phosphate. Pit1 and Pit2 were both increased in Fgf23-null placentas on the high-phosphate diet, but the increase in Pit1 expression was marginal at 1.199-fold, while the increase in Pit2 was due to the Fgf23-null genotype and not the maternal diet. The placental expression of Pit1 and Pit2 may be regulated by fetal FGF23.

Although fetal serum phosphorus was increased in Fgf23-null fetuses from the high-phosphate diet, this did not lead to any change in skeletal mineral content. The increment was evidently not large enough to cause increased mineralization, or of sufficient duration to overtly affect endochondral bone development. The observed changes in placental and renal gene expression don’t evidently explain the increased serum phosphorus of the Fgf23 nulls on the high-phosphate diet.

We did not measure placental phosphate transport, which was expected to be normal given that Fgf23-, Klotho-, and Phex-null fetuses all had normal phosphate transport in prior experiments (21, 22). Moreover, if increased placental phosphate transport did explain the increase in fetal serum phosphorus despite no change in whole body phosphate content, it would require an increase of <1%. Such a small difference cannot be resolved with the technique. This is based on an average fetal blood volume of 100 microliters, the 0.5 mmol/L increment in serum phosphorus, which represents an increase of 0.0465 mg of phosphate, and the 2.52 mg of phosphate present in each fetus (derived from ash weight and mineral content in Fig. 5). Moreover, fetuses accrete about 1 mg of phosphate during ED 18.5.

In summary, increasing dietary phosphate intake—from low to high—had predictable effects in WT mice to increase maternal serum phosphorus and FGF23 during prepregnancy and pregnancy, while PTH was unresponsive and became undetectable as expected during pregnancy. Despite these effects on the mothers, fetal serum phosphorus, PTH, skeletal phosphate content, and placental phosphotropic gene expression remained largely unaffected. A small but nonsignificant rise in fetal FGF23 in WT fetuses proved to be physiologically important when, in subsequent experiments, Fgf23-null fetuses developed hyperphosphatemia in response to the same maternal phosphate loading. Whether the absence of fetal FGF23 causes hyperphosphatemia in this setting by allowing increased placental phosphate transport, decreased placental back-flux to the maternal circulation, or alters of the set-point of a putative phosphate sensor cannot be determined. The absence of major changes in known placental genes suggests that FGF23 does not act through these pathways to prevent fetal hyperphosphatemia.

In conclusion, maternal FGF23 protects against the development of fetal hyperphosphatemia, while fetal FGF23 plays no role in phosphate and bone homeostasis during fetal development, except in the unusual circumstance of defending against maternal phosphate loading. Fetal serum phosphorus, and phosphate delivery and incorporation into bone, are otherwise regulated independent of maternal serum phosphorus and fetal FGF23. A phosphate sensor may set fetal serum phosphorus, with factors other than FGF23 and PTH regulating placental transport of phosphorus.

Additional Information

Disclosure Summary: The authors have nothing to disclose.

Data Availability

All data generated or analyzed during this study are included in this published article.