Murine Placental-Fetal Phosphate Dyshomeostasis Caused by an Xpr1 Deficiency Accelerates Placental Calcification and Restricts Fetal Growth in Late Gestation

Animals

The Xpr1-knockout mice with a C57BL/6N background that were used for the experiments were originally produced by the CAM-SU Genomic Resource Center. All animal experiments were approved by the Ethics Committee of Animal Experiments of Huazhong University of Science and Technology and were conducted according to the standards for the care and use of laboratory animals established by the US National Institutes of Health. All mice were fed with a regular diet (1.05% calcium, 0.71% phosphate) and intercrossed between 8 and 12 weeks of age.

Genotyping

Mouse DNA was purified from the toe or tail of offspring and used for genotyping by polymerase chain reaction (PCR). Primers for the wild-type (WT) mouse genotype (forward primer: 5′-CCAGCTCAGCCTAAAGAGT-3′; reverse primer: 5′-GAATGTATTCAGGACTTCGGGTA-3′) and the knockout mouse genotype (forward primer: 5′-CCAGCTCAGCCTAAAGAGT-3′; reverse primer: 5′-GGCAAGAACATAAAGTGACC-3′) were used, and the WT and knockout products were 399 bp and 301 bp, respectively.

In situ hybridization

Fresh placentas were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) (prepared with diethyl pyrocarbonate [DEPC]-water) for 2 hours and kept in 30% sucrose solution (prepared with DEPC-water) for 8 hours before frozen sections (8 μm thickness) were cut. A digoxigenin-labeled antisense probe was synthesized by a MerMade 192E oligonucleotide synthesizer (BioAutomation, Inc., Irving, TX, USA). The probes were hybridized to frozen sections at 37°C overnight according to a standard protocol. Anti–digoxigenin–horseradish peroxidase (anti-DIG-HRP) and diaminobenzidine (DAB) were used for the color reaction.

Real-time PCR

We extracted RNA from the fresh placentas of mice using TRIzol (TaKaRa Bio, Otsu, Japan), following the manufacturer's protocol. Real-time PCR (RT-PCR) was performed with SYBR® Green Master Mix (Vazyme Biotech Co. Ltd., Nanjing, China) on a StepOnePlus™ Real-Time PCR Instrument (Life Technologies, Inc., Grand Island, NY, USA). Gapdh was used as an internal reference for each sample.

Histology

Fresh placentas and fetal femurs were harvested and imaged at different periods of gestation and then fixed in 4% paraformaldehyde for over 24 hours prior to the preparation of paraffin sections or frozen sections. Then, tissues for the paraffin sections were dehydrated by a graded series of alcohol and dimethylbenzene, immersed in paraffin wax, and embedded for over 12 hours at room temperature before sectioning. Tissues for the frozen sections were kept in 30% sucrose at 4°C until they sank and were embedded in optimal cutting temperature compound (OCT) for 20 min at −24°C before sectioning.

For Alizarin red (AR) staining, the tissue sections were dewaxed by dimethylbenzene and a series of graded alcohol, treated with Alizarin red for 20 min, washed with distilled water, and dried at 50°C for 30 min.

Silver staining of dewaxed tissue sections was performed following the same method as that used for AR staining. Component A consisted of 10% formaldehyde and 0.002% nitric acid. Component B consisted of 15% silver nitrate, 10% potassium nitrate, and 0.05% glycine. Component C consisted of 1% pyrogallol, 55% alcohol, and 0.002% nitric acid. The sections were soaked in component A for 5 min and washed with distilled water three times. Next, the sections were soaked in component B (37°C) for 5 min, taken up to remove component B, soaked in component C (45°C) for 5 to 10 s, and washed with distilled water.

Von Kossa (VK) staining of dewaxed tissue sections was performed following the same method as that used for AR staining. The sections were soaked in 5% sodium thiosulfate for 30 min and washed with distilled water three times. Then, 5% AgNO₃ was added and the sections were incubated under ultraviolet radiation for 30 min. The sections were then washed with distilled water three times and stained with eosin for 5 min.

μCT analyses

Fresh E18.5 fetuses were fixed in 4% paraformaldehyde for over 48 hours. Then, the fetuses were imaged with a high-resolution in vivo μCT scanner (SkyScan 1176; Bruker, Kontich, Belgium) with a 9-μm spatial resolution (58 kV, 431 μA, 0.5-mm Al filter). For femoral and tibial diaphysis analysis, the area starting from the midshaft diaphysis and extending 1 mm (100 slices) was selected. A calcium-hydroxyapatite standard was provided for calibrations of bone mineral density (BMD).

Morphometry of placental calcification and mineralized bone volume

The placentas were prepared for serial paraffin sectioning (5 μm thickness) with AR staining. Eight-micrometer-thick (8-μm-thick) frozen sections of fetal femurs were stained with VK stain. The sections were imaged by an Olympus BX53 microscope (Olympus, Waltham, MA, USA), and placental calcification and the mineral content of bone were quantified by Image-Pro Plus 6.0 software (Media Cybernetics, Inc., Shanghai, China). To quantitatively assess placental calcification, we evaluated the placental calcification grade of all AR-stained transverse serial sections based on pixel counts of the calcified area and defined the quantified calcification index (CCI) as follows:

(1)

The assessment of placental calcification was improved as described.26

Pi and calcium concentration measurement

The amniotic fluid was acquired by piercing the amniotic membrane with a glass capillary tube to measure the Pi concentration. Fetal serum was collected from fetal blood vessels after clipping the umbilical cord and was then quantitatively added into normal saline (blood:saline = 1:149) and centrifuged. After centrifugation, the upper portion of the liquid was retained as the fetal serum sample. The amniotic fluid and fetal serum Pi concentrations were detected by molybdate-malachite green spectrophotometry. Component A consisted of 0.045% malachite green. Component B consisted of 4.2 g ammonium molybdate dissolved in 100 mL 4 N HCl. Then, 2 μL amniotic fluid with 48 μL distilled water or a 25-μL fetal serum sample with 25 μL distilled water was added to 50 μL of standard Pi solution samples (0, 3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, and 36μM). Next, 75 μL component A and 25 μL component B were added, and each sample was mixed for several seconds before the absorbance was measured by a microplate reader (λ = 620 nm). Calcium concentration was measured using a Calcium Colorimetric Assay Kit (BioVision Inc., Milpitas, CA, USA). A total of 24 WT and 28 Xpr1^+/− female mice were used for collecting fetal serum and amniotic fluid. No adverse events appeared.

Placental Pi transfer

Pregnant mice were given 50 μCi ³²P (HTA Co., Ltd., Beijing, China) via an intracardiac injection, and the ³²P level of fetuses was measured with a liquid scintillation counter, as described.27 Each fetal ³²P value was normalized to the mean value of WT fetuses within each litter. The WT fetuses were chosen as the baseline for comparison within each litter due to their normal viability. A total of 10 WT and 15 Xpr1^+/− female mice were used for this experiment. No adverse events appeared.

Fetal ash and skeletal mineral measurement

Entire fetuses (E18.5) were placed into lidded crucibles and reduced to ash by a furnace (500°C) for 24 hours, as described.28 The Pi and calcium content of the ash was measured by the same method as that used for measurement of the mineral in amniotic fluid. Because the fetal size is influenced by the number of fetuses from each litter, the data obtained from each fetus were normalized to the mean value of WT fetuses within each litter.

Skeletal preparations

The skin and viscera were removed from E18.5 fetuses, and the fetuses were then fixed in 100% ethanol for 2 days and stained with Alcian blue 8GX and Alizarin red S, as described.29

RNA sequencing

Total RNA was extracted from E17.5 placentas. mRNA was isolated from total RNA by magnetic beads with oligo-dT (a synthetic single-stranded oligonucleotide with 5′- and 3′-hydroxyl ends) and randomly fragmented. After reverse transcription, end-repair, adaptor ligation, and PCR amplification, ligation products were purified to remove self-ligated adapters and incompletely ligated fragments. Libraries were constructed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) and quantified using a Qubit 2.0 fluorometer (Thermo Fisher Scientific, Waltham, MA, USA). The DNA was sequenced according to Illumina's standard protocol (Illumina, San Diego, CA, USA).

Immunohistochemistry

Eight-micrometer-thick (8-μm-thick) frozen placenta sections were heated in a microwave for antigen retrieval, blocked with 3% bovine serum albumin, and incubated with anti-granzyme B antibody (1/100; Abcam, Cambridge, MA, USA; ab4059). Next, the color reaction was performed with HRP-labeled antibody (Wuhan Servicebio Technology Co., Ltd., Wuhan, China; G1210) and DAB, and the sections were counterstained with hematoxylin.

Detection of alkaline phosphatase

Alkaline phosphatase (ALP) staining of frozen tissue sections was performed following the Gomori modified calcium-cobalt staining method using an ALP kit (Wuhan Servicebio Technology Co., Ltd., Wuhan, China) according to the manufacturer's instructions. The density of ALP staining was quantified by Image-Pro Plus 6.0 software.

Statistical analysis

All results are presented as the mean ± standard error of the mean (SE). Differences between two groups were determined by Mann-Whitney test. The comparisons among multiple groups were determined by ANOVA followed by Tukey's test. A value of p < 0.05 was considered significant. The above statistical analyses were performed using GraphPad Prism software (GraphPad Software, Inc., La Jolla, CA, USA).

In the RNA sequencing (RNA-Seq) analysis, read quality was performed for all samples by using FastQC version 0.11.2 (Andrews.2010) software (Babraham Bioinformatics, Cambridge, UK; http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) with the default parameters. Expression estimation of transcript was performed by using RSEM (V1.2.29).30 Raw read counts of gene expression were applied to estimate the differentially expressed genes by using edgeR package.31 The p values were obtained by edgeR using an exact test through the negative binomial distribution. The significant difference was identified with a p value <0.01 and with a fold change greater than two. The mRNA levels were calculated and presented by using fragments per kilobase of exon per million fragments mapped (FPKM) values. Volcano plot was generated by Microsoft Excel (Microsoft Corp., Redmond, WA, USA). Gene ontology (GO) analysis was performed using DAVID 6.8 (Laboratory of Human Retrovirology and Immunoinformatics [LHRI], Frederick National Laboratory for Cancer Research, Frederick, MD, USA; https://david.ncifcrf.gov/home.jsp). Heatmap was generated by MeV 4.9.0 software. Venn diagram was created by using Venny 2.1.0 (https://bioinfogp.cnb.csic.es/tools/venny/).

Expression of Xpr1 in the placenta

XPR1 is highly expressed in the human placenta.23 We aligned the mouse XPR1 against the sequences in the human protein database and identified a human XPR1 with 95% identity to mouse XPR1 (Supplemental Fig. S1). We found that Xpr1 is also widely expressed in mouse placenta by in situ hybridization analysis. Specifically, the spongiotrophoblast layer had a higher Xpr1 expression level than the labyrinthine layer, chorionic plate, and maternal decidual tissue (Fig. 1 A). Xpr1 mRNA expression in the placenta at E12.5, E15.5, and E18.5 was assessed by RT-PCR. Xpr1 mRNA expression levels obviously increased during gestation (Fig. 1 B). These results suggested that Xpr1 might be required for normal placental function during the process of placental development. What roles does XPR1 play in the placenta and is XPR1 involved in placental-fetal Pi homeostasis? In this study, we generated Xpr1 knockout mice and performed genotyping by PCR (Supplemental Fig. S2 A-B). RT-PCR analysis showed that a significantly lower Xpr1 mRNA expression level was detected in heterozygous mouse placentas at E12.5, E15.5, and E18.5 than in the WT placentas at the same time points (Fig. 1 B).

Xpr1 knockout disturbs placental-fetal Pi homeostasis

To determine whether the absence of Xpr1 affects normal placental-fetal Pi homeostasis, we measured the concentration of Pi in amniotic fluid, which is mainly composed of fetal metabolites. At the E15.5 stage, the amniotic sac contained abundant amniotic fluid. Compared with the WT fetuses, Xpr1^+/− and Xpr1^−/− fetuses presented significantly reduced Pi levels in amniotic fluid in Xpr1^+/− mouse intercrosses (Fig. 2 A). To further verify fetal Pi homeostasis, we detected the Pi concentration in fetal serum and found that the Xpr1^+/− and Xpr1^−/− fetuses had a lower Pi level in serum than the WT fetuses at both E15.5 and E18.5 (Fig. 2 B,C). In male Xpr1^+/− and female WT mouse intercrosses, Xpr1^+/− fetuses displayed a similar reduction in the Pi level in the amniotic fluid at E15.5 and in the serum at both E15.5 and E18.5 compared with the levels in littermate controls (Supplemental Fig. S3 A-C). In addition, there were no differences in the Pi and calcium content in maternal serum and the calcium content of amniotic fluid and fetal serum between WT and Xpr1-deficient mice (Supplemental Fig. S3 D-K).

Placental Pi transport, measured by the level of ³²P accumulation in fetuses, was abnormal in Xpr1-deficient placentas compared with WT placentas. Interestingly, the ³²P levels in the Xpr1^+/− and Xpr1^−/− fetuses were significantly reduced at the E15.5 stage but increased at the E18.5 stage in Xpr1^+/− mouse intercrosses (Fig. 2 D,E). In male Xpr1^+/− and female WT mouse intercrosses, the ³²P level in the Xpr1^+/− fetuses displayed a similar reduction at the E15.5 stage and an increase at the E18.5 stage (Supplemental Fig. S4 A-B), implying these abnormal placental Pi transport have no close relationship with the genotype of the maternal portion of placenta. In addition, no significant differences were observed in ³²P accumulation in placentas between WT and Xpr1-deficient placentas (Supplemental Fig. S4 C-F).

Histological analysis of fetal femurs by VK staining and quantification revealed a significant reduction in the mineral content in the Xpr1^−/− fetuses compared with that in littermate WT fetuses at E18.5 (Fig. 3 A,B). Reduced bone mineralization in Xpr1-deficient fetuses was also confirmed by μCT analysis of fetal skeletons, which exhibited decreased femoral and tibial BMD (~10% and >40% loss in Xpr1^+/− and Xpr1^−/− fetuses, respectively) and length (~5% and ~25% loss in Xpr1^+/− and Xpr1^−/− fetuses, respectively) (Fig. 3 C,D), and by the decreased number of Alizarin red S–stained caudal vertebra (Fig. 3 E,F). The Pi and calcium content in fetal skeletons were assayed by reducing entire fetuses to ash and then measuring the Pi and calcium content of the ash. In both the Xpr1^+/− mouse intercrosses and male Xpr1^+/− and female WT mouse intercrosses, Xpr1^−/− fetuses showed significantly reduced ash weight and Pi and calcium content compared with WT fetuses, and Xpr1^+/− fetuses showed a similarly reduced trend (Fig. 3 G,H and Supplemental Fig. S4 G-H). The reduction in the Pi and calcium content of the ash was disproportionate to the reduced fetal weight in the Xpr1-deficient fetuses, as shown by the significantly reduced ratios of the ash Pi and calcium content to fetal weight in Xpr1^−/− fetuses and by a similarly reduced trend in Xpr1^+/− fetuses (Fig. 3 I and Supplemental Fig. S4 I), which might suggest a reduced overall growth rate with decreased net skeletal mineralization in the Xpr1-deficient fetuses.

Loss of XPR1 expression results in severe placental calcification

Considering that XPR1 plays an important role in placental-fetal Pi homeostasis and is expressed at high levels in the placenta, we carried out a pathological examination to investigate whether XPR1 deficiency results in placental dysfunction and/or abnormal development of the placenta. In comparison to WT placentas, the Xpr1^+/− and Xpr1^−/− placentas showed severe calcification in AR-stained transverse and horizontal sections at E15.5 and E18.5, and the severe calcification was mainly concentrated in the labyrinthine layer (Fig. 4 A–F, A′–F′ and Supplemental Fig. S5 A-F, A′-F′). A comparison of the degree of calcification among the WT, Xpr1^+/−, and Xpr1^−/− placentas from E15.5 to E18.5 showed the following average CCI values: 0.05%, 0.26%, and 0.83% for WT, Xpr1^+/−, and Xpr1^−/−, respectively, at E15.5; and 0.23%, 2.52%, and 6.51% for WT, Xpr1^+/−, and Xpr1^−/−, respectively, at E18.5. The degree of Xpr1^+/− and Xpr1^−/− placental calcification was increased with age and greater than that in the WT placentas, and the placental calcification was more severe for the Xpr1^−/− placentas than for the Xpr1^+/− placentas (Fig. 4 G,H). Similarly, in male Xpr1^+/− and female WT mouse intercrosses, compared with the WT placentas, the Xpr1^+/− placentas also showed an increasing degree of calcification with age (Supplemental Fig. S6 A-F, A′-D′). To further discern the patterns of initial calcification formation, horizontal sections of E15.5 Xpr1^+/− placentas with slight Xpr1-associated calcification were stained with silver stain. The mineral crystal deposits were positively stained by black staining. Crystals were small in size and widely distributed in the labyrinthine layer. Amounts of crystals were observed in the trophoblasts and endothelial cells, indicating that Xpr1-associated calcification initially occurred inside the cells (Fig. 4 I).

Growth restriction occurred in Xpr1 knockout fetuses

As Xpr1 deficiency disturbs placental-fetal Pi homeostasis and results in severe placental calcification in late gestation, these might increase the risk of adverse outcomes in the offspring. Mendelian inheritance statistics was performed for offspring at the time of birth. On rare occasions Xpr1^−/− pups were born, but all of them were found dead within 1 hour after birth (P0 [1 hour]) (100% loss), indicating that the rare Xpr1^−/− pups died in utero or immediately after birth. Xpr1^+/− pups were born at lower rates (~36% loss), and a small proportion (~20%) of those pups died prior to postnatal day 7 (P7) (Supplemental Tables S1 and S2). These results indicate that Xpr1 deficiency likely results in severe fetal growth restriction during gestation. Therefore, different stages of gestation were examined to characterize the embryonic phenotypes. In Xpr1^+/− mouse intercrosses, the number of genotypes (WT:Xpr1^+/−:Xpr1^−/− = 1:2:1) was in accordance with Mendelian inheritance at E15.5 (Supplemental Table S1), but the Xpr1^−/− fetuses were smaller and had a lower body weight than the WT fetuses, whereas no significant differences were observed between the WT and Xpr1^+/− fetuses (Fig. 5 A,C and Supplemental Fig. S7 A,C). At the E18.5 stage, the number of Xpr1^−/− embryos was largely reduced (75% loss) (Supplemental Table S1), and the Xpr1^−/− fetuses were smaller and had lower body weights than their WT littermates (Fig. 5 B,D). Both Xpr1^+/− mouse intercrosses and male Xpr1^+/− and female WT mouse intercrosses generated Xpr1^+/− embryos, and these were always accompanied by dead fetuses and reduced embryo numbers at E18.5 (~34% loss). At that stage, Xpr1^−/− fetuses exhibited more severe abnormal features than Xpr1^+/− fetuses (Fig. 5 B,D, Supplemental Fig. S7 B,D and Supplemental Tables S1 and S2).

Identification of potential factors related to the loss of XPR1 expression

To determine the potential factors associated with placental dysfunction resulting from Xpr1 deficiency, we performed RNA-Seq with four WT and four Xpr1^+/− littermate placentas at E17.5 that were derived from male Xpr1^+/− and female WT mouse intercrosses. A total of 425 transcripts showed changes (p < 0.01) in expression in the Xpr1^+/− placentas compared with their expression in the WT placentas, including 250 (58.8%) upregulated and 175 (41.2%) downregulated transcripts (Fig. 6 A). GO analysis showed that the upregulated transcripts were enriched for cytolysis, proteolysis, collagen fibril organization, cartilage development, angiogenesis, blood vessel development, immune response, and other biological processes (Fig. 6 B).

Notably, a set of granzymes (Gzmb, Gzmc, Gzmd, Gzme, Gzmf, and Gzmg), which play crucial roles in the biological processes of cytolysis, proteolysis, and immune response, showed significantly increased mRNA level (p < 0.01, fold change greater than two) in the Xpr1^+/− placentas. In addition, the expression level (p < 0.01, fold change greater than two) of the perforin Prf1, which is functionally related to granzymes, was also significantly higher (Fig. 6 D,E). In addition, granzymes and perforin showed a similarly increased trend in the Xpr1-deficient placentas at E18.5 but no differences between WT and Xpr1-deficient placentas at E15.5 (Supplemental Fig. S8 C-D), implying that granzymes and perforin might be associated with the placental calcification in late gestation. To assess the roles that granzymes play in the Xpr1-deficient placentas, immunohistochemical analysis of granzyme B was performed with WT, Xpr1^+/−, and Xpr1^−/− placentas at E18.5. We found significantly increased granzyme B accumulation at the sites of placental calcification, suggesting that the granzymes were associated with the formation of placental calcification (Fig. 7 A–F, D′–F′).

In contrast, a series of transcripts with decreased expression levels (p < 0.01) were markedly enriched for phospholipid catabolic process (Pla2g4d), cell adhesion (Stab1), postembryonic development (Mapk8ip3), RNA splicing (Srsf5), regulation of phosphorylation of RNA polymerase (Ccnl2), and mRNA processing (Pabpn1), which implied a delay in placental development and function (Fig. 6 C,D,F). Interestingly, the expression level (p < 0.01) of Ip6k2, which encodes the mammalian enzyme inositol pyrophosphate that is associated with eukaryotic phosphate homeostasis, was also decreased in the Xpr1^+/− placentas (Fig. 6 D,F).

Although previous studies have shown that the Na⁺/Pi transporters Slc34a2, Slc20a1, and Slc20a2 are highly expressed in the placenta and might play important roles in maintaining maternal-fetal Pi homeostasis, RNA levels of these three genes showed no marked differences between the WT and Xpr1-deficient placentas at E15.5, E17.5, and E18.5 (Fig. 6 G and Supplemental Fig. S8 E-F). In addition, no significant differences were found in RNA levels of genes known to be involved in physiological or pathological calcification, including the osteoblastic and chondrocytic differentiation molecules (Runx2, Bmp2, Msx2, and Sox9), smooth muscle cells markers (Tagln and Acta2), osteochondrogenic markers (Runx1, Spp1, Bglap2, and Alpl), and tissue calcification inhibitor (Mgp) (Supplemental Fig. S9 A-C). However, ALP staining for E18.5 placentas showed that ALP was mainly concentrated in the labyrinthine layer and significantly increased in Xpr1-deficient placentas (Supplemental Fig. S10 A-D).