Upregulation of calcitriol during pregnancy and skeletal recovery after lactation do not require parathyroid hormone

Authors


Address correspondence to: Christopher S Kovacs, MD, Memorial University, Faculty of Medicine–Endocrinology, Health Sciences Centre, 300 Prince Philip Drive, St. John's, Newfoundland A1B 3V6, Canada. E-mail: ckovacs@mun.ca

ABSTRACT

Pregnancy invokes a doubling of intestinal calcium absorption whereas lactation programs skeletal resorption to provide calcium to milk. Postweaning bone formation restores the skeleton's bone mineral content (BMC), but the factors that regulate this are not established. We used Pth-null mice to test whether parathyroid hormone (PTH) is required for postweaning skeletal recovery. On a normal 1% calcium diet, wild-type (WT) and Pth-null mice each gained BMC during pregnancy, declined 15% to 18% below baseline during lactation, and restored the skeleton above baseline BMC within 14 days postweaning. A 2% calcium diet reduced the lactational decline in BMC without altering the gains achieved during pregnancy and postweaning. The hypocalcemia and hyperphosphatemia of Pth-null mice normalized during lactation and serum calcium remained normal during postweaning. Osteocalcin and propeptide of type 1 collagen (P1NP) each rose significantly after lactation to similar values in WT and Pth-null. Serum calcitriol increased fivefold during pregnancy in both genotypes whereas vitamin D binding protein levels were unchanged. Absence of PTH blocked a normal rise in fibroblast growth factor-23 (FGF23) during pregnancy despite high calcitriol. A 30-fold higher expression of Cyp27b1 in maternal kidneys versus placenta suggests that the pregnancy-related increase in calcitriol comes from the kidneys. Conversely, substantial placental expression of Cyp24a1 may contribute significantly to the metabolism of calcitriol. In conclusion, PTH is not required to upregulate renal expression of Cyp27b1 during pregnancy or to stimulate recovery from loss of BMC caused by lactation. A calcium-rich diet in rodents suppresses skeletal losses during lactation, unlike clinical trials that showed no effect of supplemental calcium on lactational decline in BMC.

Introduction

Reproduction invokes adaptive mechanisms that enable mammalian females to meet the calcium requirements of the developing fetus and suckling infant. During pregnancy the main adaptation is that intestinal calcium absorption doubles,[1, 2] driven in part by a substantial increase in serum calcitriol that begins early in pregnancy.[3] However, because increased intestinal calcium absorption occurs during pregnancy despite severe vitamin D deficiency, absence of vitamin D receptor (VDR), or absence of 1α-hydroxylase,[4-7] other factors, such as placental lactogen, prolactin, and estradiol, have been proposed to stimulate intestinal calcium absorption through pathways that are independent of calcitriol and VDR.[8, 9]

During lactation mammals provide calcium for milk production largely by resorbing the skeleton and, thereby, reducing bone mineral content (BMC).[1, 2, 10, 11] These losses are greatest in the trabecular content of the spine, and less so from cortical bone and the appendicular skeleton.[1, 2, 10, 11] Women typically lose 5% to 10% of spine BMC during 3 to 6 months of exclusive lactation, whereas teenaged mothers and women lactating twins may lose even more.[1, 2, 11] Rodents supply a proportionately greater amount of calcium for a litter of six to 12 pups, and they accomplish this by resorbing 20% to 25% of spine (trabecular) BMC and 10% of whole-body (cortical) BMC during 3 weeks of lactation, and by maintaining an upregulated rate of intestinal calcium absorption.[1, 2, 6, 12-15] Larger litters or dietary calcium restriction increase the amount of bone resorbed.[1] Mice lacking the gene for calcitonin and calcitonin gene-related peptide-α lose twice the amount of bone as their wild-type (WT) littermates.[12, 16]

The mechanisms that regulate skeletal resorption during lactation have been the subject of numerous recent investigations.[6, 12-14, 17-23] Histomorphometric, radiological, and bone marker assessments show that osteoclast-mediated bone resorption increases, whereas more recently osteocytic osteolysis has been shown to resorb bone as well.[1, 12, 13, 22, 24] A brain-breast-bone circuit coordinates the promotion of a bone resorptive state during lactation.[12] The pituitary releases prolactin and oxytocin in response to suckling; the gonadotropin-releasing hormone (GnRH) release in the hypothalamus is inhibited by suckling and prolactin, in turn causing suppression of estradiol production by the ovaries; mammary tissue releases parathyroid hormone–related protein (PTHrP) in response to suckling and high prolactin; and osteoclasts proliferate and upregulate in response to PTHrP, low estradiol, and possibly oxytocin and other factors.[2, 25] Most of the PTHrP is secreted at high concentrations into milk but suckling forces some of it to enter the maternal circulation.[2, 18, 19, 26] Both PTHrP and low estradiol are required for normal lactational bone loss to occur, because treatment with supraphysiological doses of estradiol,[20] or ablation of the PTHrP gene from mammary tissue,[17] blunted but did not obliterate the bone loss. Other hormones may also regulate bone metabolism during lactation.[3, 27-30]

After weaning, osteoclasts undergo apoptosis whereas osteoblasts rapidly lay down new osteoid.[1, 14, 23] Osteocytic lacunae also show tetracycline labeling, confirming that new bone is being laid down at these sites.[31, 32] Typically within 6 to 12 months in humans, and 2 to 3 weeks in mice, the BMC returns to the prepregnancy baseline value or above,[1, 33, 34] even in mice that have lost over 50% of spine BMC.[12] The speed and completeness of recovery differ by skeletal site and technique used; tibias and femurs of mice show incomplete recovery of trabecular bone by micro–computed tomography (µCT).[21] Clinical studies show that there is no long-term adverse effect of lactation on bone mass, bone density, or hip fracture risk; in several studies a history of lactation conferred a lower risk of osteoporosis.[1, 11, 35] A significant postweaning increase in bone mass occurs even in those who fractured during lactation.[36, 37]

The remarkable nature of postweaning bone recovery is that it is rapid and complete, unlike the slow and partial recovery that the adult skeleton achieves after other causes of bone loss, including bed rest, weightlessness, glucocorticoid therapy, and estrogen deprivation.[1, 38-43] Repletion of ovarian hormones facilitates but is not the explanation for this potent interval of bone formation, especially because the bone loss caused by estrogen deprivation therapy in reproductive-aged women is not fully restored after a year of normal ovarian function.[1] Furthermore, estradiol suppresses both bone formation and resorption when used to correct estrogen deficiency in perimenopausal or postmenopausal women.[44, 45] We have found that mice lacking calcitonin, VDR, or osteoblast-derived PTHrP are able to stimulate bone formation postweaning and fully restore BMC that was lost during lactation.[6, 12, 13]

PTH is a logical candidate for stimulating bone formation after lactation. When humans or rodents have adequate dietary calcium intake, PTH is normally suppressed to 20% to 30% of prepregnancy values during pregnancy and lactation, and then it rises to prepregnancy levels or higher during the postweaning period in both humans and rodents.[12, 13, 46, 47] PTH directly stimulates bone formation by acting on the PTH/PTHrP receptor in osteoblasts, and PTH analogs such as teriparatide are now approved anabolic treatments for osteoporosis.[48-51] Therefore, the ability of PTH to stimulate bone formation, and its postweaning increase from the low levels reached during lactation, make it plausible that restoration of normal PTH synthesis and release stimulates bone formation after lactation.

We hypothesized that PTH is essential for mineral homeostasis during pregnancy and to stimulate skeletal recovery after lactation, and we used Pth-null mice to test this hypothesis.

Subjects and Methods

Animal husbandry

The engineering of global knockout Pth-null mice has been previously described.[52] The original strain was back-crossed into Black Swiss (Taconic, Germantown, NY, USA) for more than 10 generations. WT and Pth-null sisters were selected for experiments. Genotyping was done by PCR on DNA extracted from tail clips of weaned pups.

A vaginal mucus plug on the morning after timed mating indicated day 0.5 of 19 days. Mice had ad libitum access to water and a standard rodent diet containing 1% calcium and 0.75% phosphorus. In separate studies, from 3 weeks of age matched WT and Pth-null sisters were given an enriched diet containing 2% calcium, 1.25% phosphorus, and 20% lactose (TekLad TD96348; Harlan Teklad, Madison, WI, USA). The Institutional Animal Care Committee of Memorial University of Newfoundland approved all procedures involving live animals.

Reproductive cycles and data collection time points

These studies were completed according to approximate 75-day reproductive cycles. Beginning at 10 weeks of age the prepregnancy interval consisted of two baseline BMC scans and blood/urine collections done over 5 to 10 days prior to first successful mating. Additional BMC scans and sample collections were done at day 18.5 of pregnancy, day 21 of lactation (39.5 days after start of pregnancy, which equals the day of weaning), and weekly during postweaning recovery (days 7, 14, and 21 postweaning). Day 7 of postweaning was used for postweaning serum and urine data in this manuscript, whereas the highest achieved postweaning BMC value (which usually occurs at day 14) was used to indicate the magnitude of recovery after lactation. Mice had an expected mean age of 23 weeks by the end of postweaning recovery.

BMC

As described,[6, 12, 13] we measured BMC with the PIXImus 2 Bone Densitometer (GE Lunar, Madison, WI, USA) and analyzed data with PIXImus software version 2.1. Anesthetized mice were immobilized prone on holding trays with the spine straightened; the head was excluded in order to reduce variability. Mice fed the enriched chow had to be fasted overnight to avoid radio-opaque stools that artifactually inflate the apparent whole-body and lumbar spine BMC. Prior quality control studies determined that fetal skeletons contribute <1% of apparent maternal BMC at the end of pregnancy and are, therefore, negligible.[12, 53] Whole-body and regional (spine, hindlimb) BMC measurements for each mouse were normalized to the mean value of their respective nonpregnant baseline measurements.

Chemical and hormone assays

Morning samples of urine were collected by having mice void into a clean cage; after this, blood was taken from tail veins. Serum or urine calcium, phosphorus, and creatinine were measured with colorimetric assays (Sekisui Diagnostics PEI Inc., Charlottetown, Prince Edward Island, Canada). Calcitriol was analyzed by Enzyme-linked Immunosorbent Assay (EIA) (Immunodiagnostic Systems Ltd., Boldon, Tyne and Wear, UK). Bone markers included osteocalcin assessed by a two-site immunoradiometric assay (Immutopics, San Clemente, CA, USA), propeptide of type 1 collagen (P1NP) by EIA (Immunodiagnostic Systems), and deoxypyridinoline by the METRA DPD EIA (Quidel Corporation, San Diego, CA, USA). Vitamin D binding protein was measured by EIA (Cusabio Biotech, Wuhan, China). Intact fibroblast growth factor-23 (FGF23) was measured using an EIA that has been extensively validated in mice (Kainos, Japan). Urinary deoxypyridinoline, calcium, and phosphorus were expressed relative to creatinine.

RNA extraction

Maternal kidneys of WT and Pth-null mice were removed at baseline and end of pregnancy, and then snap-frozen in liquid nitrogen. Archived mutant and matching WT littermate term (embryonic day [ED] 18.5) placentas were retrieved for three different knockout models that we had previously studied (Pth-null, Hoxa3-null, and Gcm2-null[54-56]); all had been back-crossed into Black Swiss mice. Total RNA was purified using the RNeasy Midi Kit (Qiagen, Toronto, ON, Canada). RNA quality was confirmed with the Agilent 2100 BioAnalyzer (Agilent Technologies, Santa Clara, CA, USA).

Real-time quantitative RT-PCR

We used TaqMan Gene Expression Assays and Fast Advanced Master Mix from Applied BioSystems (ABI)/Life Technologies (Burlington, ON, Canada) to determine expression of Cyp27b1 (1α-hydroxylase), Cp24a1 (24-hydroxylase), Fgf23, Klotho, and FGF receptors 1 through 4 (Fgfr1Fgfr4), with predesigned primers and probes for optimal amplification. Details of conditions and cycle times have been previously reported.[12, 13, 54] Briefly, cDNA was synthesized using the Taqman High Capacity cDNA Reverse Transcription Kit (ABI), and single-plex qPCR reactions were run in triplicate on the ViiA 7 Real-Time PCR System (ABI).[54, 57] Relative expression was determined from the threshold cycle (CT) normalized to the reference gene.

Statistical analysis

Data were analyzed using StatPlus:Mac Professional 2009, Build 5.8.3.8 (AnalystSoft Inc, Vancouver, BC, Canada). ANOVA was used for analysis of BMC and biochemical data, whereas a post hoc test determined which pairs of means differed significantly. Two-tailed probabilities are reported as mean ± SEM. qPCR data were analyzed by the Comparative CT Method (ΔΔCT)[58] and are reported as mean ± SD.

Results

BMC excursion in Pth-null mice

We examined serial changes in BMC during complete reproductive cycles of Pth-null mice and their WT sisters while each consumed a standard 1% calcium diet. At prepregnancy baseline the whole-body BMC was 0.498 ± 0.014 g in WT and 0.501 ± 0.016 g in Pth-null (p = NS). When expressed as relative changes to baseline for each mouse, whole-body BMC (which represents largely cortical bone) increased 12% to 15% during pregnancy in WT and Pth-null mice, declined about 20% during lactation to reach a value 5% to 8% below baseline, and recovered postweaning to approximately 10% above baseline (Fig. 1A).

Figure 1.

Effects of the reproductive cycle on BMC in Pth-null and WT mice consuming a normal versus enriched calcium diet. Relative changes in BMC on a standard 1% calcium diet are shown for whole body (A), spine (B), and hindlimb (C), and expressed as a percentage of individual prepregnancy baseline value. Pth-null and WT mice experienced similar changes in BMC during the reproductive cycle, including gains during pregnancy, losses in lactation, and increases during postweaning to the prepregnancy baseline level or above. BMC excursions on an enriched 2% calcium with 20% lactose diet are shown for whole body (D), spine (E), and hindlimb (F). In general the enriched calcium diet blunted the decline in BMC during lactation but did not alter the gains in BMC during pregnancy and postweaning. Braces indicate statistically significant relative differences between time points but within each genotype, whereas asterisks indicate values that are statistically significant versus prepregnancy baseline. Time points depicted are day 18.5 of pregnancy, day 21 of lactation, and day 7 of postweaning recovery. The number of observations are indicated in parentheses.

The largely trabecular content of the spine normally experiences the greatest losses of BMC during lactation. As we have previously seen in the Black Swiss background,[6, 12] BMC of the lumbar spine did not change significantly during pregnancy, but fell 18% below baseline before recovering postweaning to 15% above baseline in both WT and Pth-null (Fig. 1B). The hindlimb contains an intermediate content of trabecular and cortical bone and displayed the largest increase in BMC during pregnancy, fell during lactation, and recovered postweaning to a value that was significantly greater than baseline in Pth-nulls only (Fig. 1C).

Overall, both WT and Pth-nulls experienced significant relative changes in BMC at all three skeletal sites during the reproductive cycle, but there were no significant differences between genotypes. Pth-null mice clearly demonstrated that recovery of bone mass after lactation does not require PTH.

Few Pth-null mice survive to maturity due to sudden deaths that are presumed to be the result of hypocalcemic arrhythmias or tetany. Pth-null mice are also more susceptible to prolonged unconsciousness and death after anesthesia, which is problematic because the mice must be anesthetized for the BMC measurements. An earlier attempt at these studies resulted in few Pth-null mice lactating because many died after anesthesia or cannibalized their litters.[59] We adjusted the protocol to minimize the dose of anesthetic, duration of unconsciousness, and frequency of BMC measurements. Even with these adjustments the experiments took a long time to complete due to limited availability of sexually mature Pth-null females. However, in mice that entered the study at age 10 weeks, WT and Pth-null mice showed no differences in the ability to conceive, litter size, or sudden deaths during the reproductive cycle.

While the studies on the standard diet were underway, we also studied WT and Pth-null mice that had been raised from weaning on an enriched diet that increases passive absorption of calcium and has been used to rescue hypocalcemic Vdr-null mice.[6, 60] On the enriched diet the baseline whole-body BMC was 0.505 ± 0.018 g in WT and 0.495 ± 0.010 g in Pth-null, no different between genotypes or from the baseline values attained on the standard 1% calcium diet. BMC excursions during the reproductive cycles are shown in Fig. 1DF. The enriched diet blunted the magnitude of skeletal losses during lactation at the whole body and spine. The enriched diet did not enhance the BMC gains during pregnancy or postweaning; instead, BMC gains were unchanged at whole body or hind limb, and appeared blunted at the lumbar spine during postweaning. These results indicate that increased availability of dietary calcium blunts skeletal resorption during lactation, especially at the lumbar spine.

Mineral metabolism during lactation and recovery

As expected from their known phenotype, Pth-null mice were hypocalcemic and hyperphosphatemic at baseline on the standard 1% calcium diet (Fig. 2A, B). The hypocalcemia and hyperphosphatemia persisted during pregnancy, but both values normalized during lactation. The serum calcium remained normal during postweaning skeletal recovery whereas the mean serum phosphorus increased to become no different from the prepregnancy value. Urine calcium showed an apparent decline during pregnancy in both genotypes, and a possible increase during postweaning in Pth-nulls only, but these differences were not statistically significant (Fig. 2C). Urine phosphorus declined nonsignificantly during pregnancy but showed a marked and significant increase during lactation in both genotypes (Fig. 2D). This is consistent with an increase in renal filtered load of phosphorus due to increased bone resorption during lactation. Urine phosphorus did not differ between genotypes at any time point.

Figure 2.

Flux in serum and urine calcium and phosphorus during reproductive cycles. (A) Serum calcium remained normal at all time points in WT. Pth-null mice were hypocalcemic at baseline and during pregnancy, but the serum calcium became normal during lactation and post-weaning. (B) Serum phosphorus was normal in WT at all time points. Pth-null mice were hyperphosphatemic at baseline and during pregnancy, but the serum phosphorus reduced to normal during lactation and returned toward the baseline value during post-weaning. (C) Urine calcium showed no significant differences between WT or Pth-null at any time point, but a downward trend during pregnancy was evident. (D) Urine phosphorus increased significantly during lactation, likely as a consequence of enhanced bone resorption and possibly PTHrP-stimulated phosphaturia. There were no significant differences between WT or Pth-null at any time point. Time points depicted are baseline, day 18.5 of pregnancy, day 21 of lactation, and day 7 of postweaning recovery. The number of observations are indicated in parentheses.

A several-fold increase in serum calcitriol occurs during pregnancy and may contribute to a doubling of the efficiency of intestinal calcium absorption. Although PTH is the major stimulator of calcitriol synthesis, PTH levels usually decline whereas calcitriol levels are increasing during pregnancy, and so it has remained uncertain as to whether or not PTH is required for the normal pregnancy-related increase in calcitriol.[1, 2] We found that calcitriol increased over fivefold in both WT and Pth-null mice during pregnancy before declining during postweaning recovery to baseline (Fig. 3A). The assay requires a large sample volume, and due to limited blood samples it was not possible to measure calcitriol during lactation. The achieved peak value of 695 ± 132 pmol/L in pregnant WT mice was not significantly higher than the value of 449 ± 123 pmol/L in pregnant Pth-null sisters. Additional samples from unrelated pregnant WT mice were run simultaneously on the assay, and those showed a peak value of 536 ± 85 pmol/L[13] for a combined WT value of 616 ± 79 pmol/L that was still no different than the Pth-null value. The results clearly indicate that the large pregnancy-related increase in calcitriol does not require PTH, but a blunting of the maximal peak value in Pth-null mice cannot be excluded. Vitamin D binding protein levels did not differ between genotypes or time points (Fig. 3B), confirming that free calcitriol levels increase during pregnancy.

Figure 3.

Changes in calcitriol, vitamin D binding protein, and FGF23 during the reproductive cycle. (A) Calcitriol increased over fivefold during pregnancy in WT and Pth-null mice with no significance difference between genotypes. (B) Vitamin D binding protein (VDBP) did not change during reproductive cycles and there was no difference between genotypes. (C) FGF23 increased twofold only in WT during pregnancy, and showed no other significant changes within or between genotypes. Time points depicted include prepregnancy baseline, day 18.5 of pregnancy, day 21 of lactation, and day 7 of postweaning recovery. The number of observations are indicated in parentheses.

FGF23 is thought to regulate serum phosphorus by inhibiting renal expression of the sodium-phosphate co-transporter IIa (NaPi2a), thereby enhancing renal phosphorus excretion. FGF23 is also stimulated by calcitriol, PTH, and high serum phosphorus, and inhibited by hypocalcemia and hypophosphatemia.[61-63] Because WT and Pth-null mice displayed marked differences in serum calcitriol, PTH, calcium, and phosphorus during the reproductive cycle, we measured intact FGF23 at each time point (Fig. 3C). Serum FGF23 levels were no different at baseline between WT and Pth-null, indicating that absence of PTH or hyperphosphatemia in Pth-nulls are not sufficient to perturb basal levels of FGF23. FGF23 doubled during pregnancy in WT but remained unchanged in Pth-null mice at any time point, despite a fivefold increase in calcitriol during pregnancy, normalization of hypocalcemia and hyperphosphatemia during lactation, and reversal of hypocalcemia alone during postweaning recovery. Reduced renal expression of Klotho (the coreceptor) in WT only during pregnancy may account for blunting of responsiveness to the high levels of FGF23 during pregnancy (Fig. 4). Fgfr1 through Fgfr4 showed no substantial changes in expression, although a likely trivial increase in Fgfr2 occurred in Pth-nulls only (Table 1). Fgf23 was undetectable by qPCR in WT and Pth-null placentas, ruling out the placenta as a source of the twofold higher FGF23 in pregnant WT (not shown).

Figure 4.

Relative expression of Klotho in maternal kidney during pregnancy versus baseline. Renal Cyp27b1 expression of Kotho declined slightly during pregnancy in both genotypes but the result was statistically significant only in WT. The brace indicates p < 0.01 and the number of observations are indicated in parentheses. All qPCR values are expressed relative to the WT baseline.

Table 1. Relative mRNA Expression Determined by qPCR of FGFR1, FGFR2, FGFR3, and FGFR4 for WT and Pth-Null Kidneys at Prepregnancy Baseline and End of Pregnancy
 WTPth-null
Baseline (n = 3)Pregnancy (n = 3)Baseline (n = 3)Pregnancy (n = 3)
  1. Values have been normalized to the WT baseline.
  2. FGFR = fibroblast growth factor receptor; WT = wild-type; Pth = parathyroid hormone gene.
  3. ap < 0.05 between Pth-null pregnancy and Pth-null baseline.
FGFR11.00 ± 0.061.17 ± 0.141.11 ± 0.141.21 ± 0.05
FGFR21.00 ± 0.091.14 ± 0.061.02 ± 0.071.30 ± 0.08a
FGFR31.00 ± 0.021.03 ± 0.240.88 ± 0.201.18 ± 0.14
FGFR41.00 ± 0.040.96 ± 0.160.97 ± 0.191.08 ± 0.09

Bone turnover markers

The postweaning increase in BMC has been shown in prior studies to result from increased osteoblast-mediated bone formation, as well as remineralization of osteocytic lacunae. We measured the bone formation markers osteocalcin and P1NP and found that both were significantly increased during postweaning as compared to prepregnancy baseline, compatible with equal upregulation of osteoblast activity (Fig. 5A, B). The bone resorption marker deoxypyridinoline showed a small increase in late lactation without any significant differences between genotypes at any time point (Fig. 5C). Notably, the greatest increase in bone resorption markers occurs during mid-lactation[64, 65] but this time point was not assessed.

Figure 5.

Changes in bone turnover markers during the reproductive cycle. The bone formation markers osteocalcin (A) and P1NP (B) increased equally in WT and Pth-null mice during postweaning recovery as compared to prepregnancy baseline. (C) The bone resorption marker deoxypyridinoline increased during lactation but returned to normal during postweaning recovery. Measurements were not done at mid-lactation when bone resorption markers reach peak levels. Time points depicted include prepregnancy baseline, day 18.5 of pregnancy, day 21 of lactation, and day 7 of postweaning recovery. The number of observations are indicated in parentheses.

Renal and placental expression of Cyp27b1 and Cyp24a1

Because calcitriol increases several-fold during pregnancy whereas PTH declines in normal rodents and humans, it is conceivable that placental production is responsible for the higher blood level of calcitriol. We investigated this by examining the renal and placental expression of the genes that regulate the formation and catabolism of calcitriol.

Cyp27b1 converts 25-hydroxyvitamin D (25(OH)D) into calcitriol. We compared renal expression of Cyp27b1 in WT and Pth-null kidneys harvested at prepregnancy baseline and late pregnancy. In keeping with the known importance of PTH to stimulate Cyp27b1 expression and activity in nonpregnant adults, basal expression of Cyp27b1 was reduced 50% in Pth-nulls compared to WT (Fig. 5A). During pregnancy Cyp27b1 expression increased significantly in both genotypes to more than 1.5 times the WT basal level, remaining slightly lower in the Pth-null (Fig. 6A). Conversely, the expression of Cyp24a1, the 24-hydroxylase that converts calcitriol and 25(OH)D into inactive metabolites, showed an opposing pattern of increased expression in the Pth-null at baseline (Fig. 6B). During pregnancy Cyp24a1 expression declined to 0.66-fold in WT and increased further to 3.4-fold in Pth-nulls (Fig. 6B).

Figure 6.

Relative expression of Cyp27b1 and Cyp24a1 in maternal kidney during pregnancy versus baseline. Cyp27b1 converts 25(OH)D into calcitriol and PTH is normally the dominant stimulator for its expression. Conversely Cyp24a1 converts 25(OH)D and calcitriol into inactive 24-hydroxylated forms. (A) Renal Cyp27b1 expression was reduced 50% in Pth-null mice at baseline, but increased during pregnancy to 1.5-fold the WT baseline value. The peak value in the Pth-null during pregnancy was slightly but significantly lower than the simultaneous WT value of 1.7-fold. (B) Renal Cyp24a1 expression was twofold higher in Pth-null versus WT at baseline. During pregnancy Cyp24a1 expression increased to 3.5-fold in Pth-null mice while declining to 0.66-fold in WT. Braces indicate p < 0.01 and the number of observations are indicated in parentheses. All qPCR values are expressed relative to the WT baseline.

We next compared the expression of these genes at the same time point of late pregnancy in maternal kidneys and placentas. We used matched pairs of WT and Pth-null, Gcm2-null, and Hoxa3-null placentas from the same Black Swiss genetic background in order to examine different fetal models of hypoparathyroidism. The results among the three null placental genotypes were nearly identical, and so the qPCR results shown in Fig. 7 involve one sample from each of the hypoparathyroid models. Cyp27b1 showed a low level of placental expression that did not differ between WT and hypoparathyroid placentas; this level of expression was dwarfed by >30-fold higher values in maternal kidneys from pregnant WT or Pth-null mothers (Fig. 7A). Conversely, Cyp24a1 expression was reduced in hypoparathyroid placentas compared to WT, the opposite of pregnant Pth-null adults, which showed higher expression of Cyp24a1. Comparing maternal kidney to placental expression of Cyp24a1 revealed threefold higher expression in WT kidneys and 15-fold higher values in Pth-null kidneys (Fig. 7B). Thus, while the placenta appears to be a trivial source of Cyp27b1 mRNA compared to the maternal kidneys, placental expression of Cyp24a1 was about one-third that of the maternal kidneys during normal pregnancy. Absence of PTH appears to markedly increase renal Cyp24a1 expression but had little effect to blunt renal Cyp27b1 expression during pregnancy.

Figure 7.

Relative expression of Cyp27b1 and Cyp24a1 during pregnancy in maternal kidney versus placenta. We compared the relative expression of Cyp27b1 and Cyp24a1 in maternal kidneys during pregnancy to placentas obtained from three genetic models of fetal hypoparathyroidism and their matched WT littermates. (A) Placental Cyp27b1 expression was no different between WT and hypoparathyroid placentas, whereas maternal renal expression was 36-fold higher in WT and 33-fold higher in Pth-null. (B) Cyp24a1 expression was reduced 90% in hypoparathyroid versus WT placentas. Maternal renal expression of Cyp24a1 was 2.7-fold of the placental value in WT, but rose to 14.2-fold in Pth-null kidneys. Braces indicate p < 0.01 and the number of observations are indicated in parentheses. All qPCR values are expressed relative to the WT placenta baseline.

Discussion

Enhanced intestinal calcium absorption during pregnancy, and increased skeletal resorption during lactation, are physiologically important adaptations that enable the mother to provide needed calcium to her offspring. Perhaps the most exciting adaptation is the postweaning interval of bone formation that restores the skeleton to its previous strength and mineral content. Evidence cited earlier indicates that PTHrP, calcitriol/VDR, calcitonin, and recovery of estradiol levels to normal do not explain the speed or completeness of postweaning recovery. In the present study we add to this list by showing that PTH is also not required.

Although many Pth-null mice die spontaneously during the first 10 weeks of life, we did not notice any increase in deaths during pregnancy. This may be attributable to the increased intestinal calcium absorption; moreover, the normalization of serum calcium during lactation and postweaning should reduce the likelihood of spontaneous hypocalcemia. Clinical experience with hypoparathyroidism during human pregnancy and lactation is similar. Many but not all case reports have documented that hypoparathyroid women have fewer hypocalcemic symptoms and reduced requirement for supplemental calcium or calcitriol during pregnancy.[1, 25, 66] During lactation the serum calcium normalizes under the influence of PTHrP-mediated skeletal resorption, and hypercalcemia will occur if the supplemental calcium and calcitriol are not reduced or discontinued at parturition.[1, 2, 66]

Pth-null mice lactated normally, lost the same amount of BMC as their WT sisters, and restored the BMC completely after weaning. The increase in osteocalcin and P1NP at day 7 postweaning is consistent with stimulation of bone formation and the restoration of the depleted BMC despite absence of PTH. We also found that the 2% calcium diet blunted the lactational decline in BMC, which indicates that the skeleton can be spared by increased dietary intake of calcium. Whereas this result may seem self-evident, randomized clinical trials and observational studies have found that higher intakes of calcium do not reduce lactational loss of bone in women.[67-70] Lactating mice rely on both upregulated intestinal calcium absorption and increased skeletal resorption to meet the calcium demand of larger litters, whereas in humans skeletal resorption and a normal rate of intestinal calcium absorption are adequate. Our results differ from a prior study in rats, which found that a 1.4% calcium diet did not reduce the amount of bone lost during lactation; the 2% calcium and 20% lactose content of the diet we used may explain why bone loss was minimized in our study.[71]

In rodents and humans, total calcitriol levels increase at least twofold to threefold beginning early in pregnancy.[1, 72] In nonpregnant adults, PTH is the major stimulator of the renal Cyp27b1; without PTH, low calcitriol and hypocalcemia result. Because PTH usually declines whereas calcitriol increases early in pregnancy, it seemed plausible that PTH is not responsible for the higher calcitriol levels. Our finding that calcitriol increases fivefold in both Pth-null and WT mice confirms that calcitriol's rise during pregnancy is not driven by PTH. Estradiol, prolactin, and placental lactogen conceivably stimulate the renal Cyp27b1 during pregnancy.[1] PTHrP is unlikely to explain the early-pregnancy rise in calcitriol because it does not usually become detectable in the maternal circulation until late in pregnancy, and it is a weak stimulator of Cyp27b1 compared to PTH.[73, 74]

The placenta has often been assumed to be the source of the increased production of calcitriol during pregnancy. In the 1970s it was established that trophoblasts and maternal decidua convert 25(OH)D to calcitriol; more recent studies have confirmed that Cyp27b1 is expressed by trophoblasts.[75-79] Additional functional studies suggest that the placenta and other extrarenal tissues may contribute some calcitriol to the maternal circulation. Nephrectomized pregnant rats given tritiated 25(OH)D had measurable tritiated calcitriol appear in the circulation, whereas none appeared in nonpregnant nephrectomized dams.[75, 80] In 5/6-nephrectomized virgin rats, basal levels of PTH were high whereas calcitriol was low; during pregnancy, PTH suppressed to normal whereas calcitriol increased markedly.[81] The authors of that study concluded that extrarenal sources of Cyp27b1 (especially placenta) explained the pregnancy-related increase in calcitriol; alternatively, the data may indicate the residual capacity of the remnant kidney to produce more calcitriol in response to stimulation during pregnancy.

Our findings that renal Cyp27b1 expression increases over 1.5-fold during pregnancy, and is 35-fold higher in maternal kidneys than placenta, are compatible with the kidneys being the main source of calcitriol. Other pregnancy studies have shown renal expression and activity of Cyp27b1 to be upregulated twofold to fivefold over nonpregnant values.[1] Data from Cyp27b1-null pigs also suggest that the fetus and placenta do not contribute substantial calcitriol to the maternal circulation, because maternal calcitriol levels are very low and comparable to nonpregnant values despite bearing heterozygous fetuses that have normal placental production of calcitriol.[82] In a published report of an anephric woman who became pregnant, calcitriol levels were very low before and during pregnancy, confirming that it is not the placenta but functioning kidneys that are required for the normal pregnancy-related increase in calcitriol.[83] Overall, it appears likely that most of the calcitriol comes from upregulated synthesis in the maternal kidneys in humans and rodents.

Cyp24a1 is also expressed by kidneys, trophoblasts, yolk sac, decidua, and other tissues; it converts 25(OH)D and calcitriol into 24- and 23-hydroxylated forms that are targeted for excretion.[78, 84-86] The significance of this enzyme's expression in placenta is uncertain. In human placentas, Cyp24a1 is methylated,[87] and its expression was lower than in adjacent decidua, whereas Vdr and Cyp27b1 were expressed at higher levels in placenta than in decidua.[88] Investigators concluded from these data that trophoblast synthesis of calcitriol is “unfettered” by degradation to 24-hydroxylated forms because Cyp24a1 may be comparatively silenced by methylation.[87, 89] However, numerous functional studies have shown that human and rodent placentas preferentially metabolize 25(OH)D to 24,25-dihydroxyvitamin D and not to calcitriol[90-92]; consequently, serum levels of 24-hydroxylated forms are up to 40-fold higher than calcitriol in human, rodent, and sheep fetuses.[91-95] Our comparison of placentas to maternal kidneys, rather than decidua, may give a better indication as to the relative contribution of the placenta to synthesis and catabolism of calcitriol. Because Cyp24a1 expression in WT placenta was about one-third the level of expression in maternal kidney, the placenta may contribute significantly to the catabolism of calcitriol and 25(OH)D. Moreover, because Cyp24a1 was upregulated 3.5-fold during pregnancy in Pth-null kidneys compared to 0.66-fold in WT kidneys, increased catabolism may contribute to the observed calcitriol levels being slightly but not significantly lower in pregnant Pth-nulls as compared to WT.

Cyp24a1 expression is increased by its substrate calcitriol,[86] and so fivefold higher calcitriol during pregnancy might be expected to increase Cyp24a1 expression in the maternal kidneys. PTH normally counteracts the stimulatory effect of calcitriol on Cyp24a1 expression,[96-98] and it accomplishes this in part by destabilizing and degrading Cyp24a1 mRNA.[99] Pth-null mice lack this physiological inhibition and displayed increased Cyp24a1 expression at baseline, and even higher levels during pregnancy. Conversely, the WT kidney showed modestly decreased Cyp24a1 expression, which may in part contribute to the higher calcitriol levels. We found that hypoparathyroid placentas had the opposite phenotype of decreased expression of Cyp24a1 compared to WT. The normal fetal milieu is characterized by high serum phosphorus, low calcitriol, and low PTH as compared to simultaneous maternal values; hypoparathyroid fetuses have even high serum phosphorus and likely have even lower calcitriol levels.[54-56] The reduced expression of Cyp24a1 in hypoparathyroid placentas may simply reflect lack of stimulation by fetal calcitriol.

FGF23 regulates serum phosphorus mainly by increasing renal phosphorus excretion and decreasing calcitriol synthesis[62, 63, 100]; it also inhibits PTH expression.[101] More specifically, it suppresses NaPi2a and NaPi2c expression in proximal kidney tubules,[102] decreases expression of renal Cyp27b1, and enhances expression of Cyp24a1.[102] FGF23 release is potently stimulated by calcitriol and hyperphosphatemia with smaller effects from PTH and hypercalcemia; conversely, it is suppressed by hypophosphatemia and hypocalcemia.[62, 63, 100] These findings predict that FGF23 should show significant alterations in response to the fivefold higher calcitriol levels during pregnancy, absence of PTH, correction of hyperphosphatemia and hypocalcemia during the reproductive cycle, and increased bone resorption with higher renal filtered load of phosphorus during lactation.

However, although intact FGF23 increased twofold in pregnant WT mice, we found no significant changes within or between WT and Pth-nulls at any other time point in the reproductive cycle. Much of the data explaining the role of FGF23 has come from the studying the extreme physiological situations of absence of FGF23 (tumor calcinosis in humans; or ablation of Fgf23, Galtn3, or Klotho in mice)[62, 100, 103-107] or excess FGF23 (X-linked hypophosphatemic rickets, autosomal dominant hypophosphatemic rickets, autosomal recessive hypophosphatemic rickets, and tumor-induced osteomalacia in humans; or Fgf23 transgenics and Hyp mice).[62, 100, 103, 108] These extreme situations confirm that deficiency or excess of FGF23 will perturb phosphorus and calcium regulation, whereas our results suggest that FGF23 does not respond to increases in calcitriol or other biochemical changes that occur during pregnancy and lactation. The fivefold increase in calcitriol during pregnancy should have been a potent stimulus to increase FGF23, but no change in serum FGF23 was observed in Pth-nulls, nor did the serum or urine phosphorus change in either genotype. The increase in urine phosphorus during lactation is likely due to increased renal filtered load of phosphorus and not increased FGF23 action. The high serum FGF23 in pregnant WT is not contributed by the placenta because Fgf23 was undetectable by qPCR in both WT and Pth-null placentas.

Our findings differ somewhat from an earlier report that used the same assay but found that serum FGF23 was significantly reduced in male Pth-nulls.[109] Possible explanations for the difference include the age of the mice (6 weeks versus 10 to 23 weeks in our study), the genetic background (C57BL/6J versus Black Swiss), the reduced calcitriol levels in Pth-nulls in the earlier study versus the normal baseline levels in our study, and that male mice were exclusively examined in the earlier study whereas we studied female mice. The normal calcitriol values in Pth-null females at baseline may be the proximate explanation for why serum FGF23 was also normal, but gender differences in calcitriol synthesis and metabolism may be implied by these results.

Strengths of this study include that the mice were studied longitudinally throughout full reproductive cycles, that WT and Pth-null sisters were compared, and that three different hypoparathyroid placental models were examined. Limitations include that gastrointestinal calcium content on the enriched diet could have obscured whole-body and spine BMC losses during lactation; however, the hindlimb is free of such confounding. Renal and placental Cyp27b1 and Cyp24a1 mRNA expression may not reflect the actual activity of the respective enzymes. However, given the comparative tissue volume of two kidneys versus the placenta, and the 35-fold higher expression of Cyp27b1 in the kidneys, it seems likely that most of the pregnancy-related increase in calcitriol derives from the maternal kidneys. Other extrarenal sources of calcitriol may contribute calcitriol to the circulation during pregnancy. We did not measure plasma PTHrP as a result of limited availability of blood and the large volumes required for the assay; consequently, we cannot exclude that PTHrP may have compensated for loss of PTH. However, PTHrP rapidly disappears from the maternal circulation as the mammary tissue involutes, so it seems unlikely that this would be a persistent source of PTHrP during postweaning.[1, 2, 110] Moreover, we have separately shown that PTHrP is not required for skeletal recovery after lactation, and PTH does not compensate for lack of PTHrP in those mice.[13]

In conclusion, our results confirm that PTH is not required for the upregulation of Cyp27b1 and increased calcitriol during pregnancy, or for the upregulation of osteocalcin and P1NP, and the achievement of skeletal recovery after lactation. Our results also suggest that FGF23 is not an important regulator of phosphorus metabolism during the reproductive cycle in WT and Pth-null mice. Taken together with our previous investigations, these results emphasize that factors other than the known calciotropic hormones are controlling intestinal calcium absorption during pregnancy and recovery of bone mass after lactation. Understanding how this interval of rapid and substantial bone formation is regulated may lead to the discovery of novel factors that stimulate bone formation and which might be exploited to pharmacologically treat disorders of low bone mass and skeletal fragility.

Disclosures

All authors state that they have no conflicts of interest.

Acknowledgments

This work was supported by an operating grant to CSK from the Canadian Institutes of Health Research (MOP-84253).

Authors' roles: Study design and conduct: CSK and BJK. Data collection: BJK, YM, HMM, KLBF. Data analysis: CSK and BJK. Data interpretation: BJK, ACK, CSK. Drafting manuscript: BJK and CSK. Revising manuscript content: All authors. Approving final version of manuscript: All authors. CSK takes responsibility for the integrity of the data analysis.

Ancillary