The Paracrine Feedback Loop Between Vitamin D₃ (1,25(OH)₂D₃) and PTHrP in Prehypertrophic Chondrocytes

Abstract

The endocrine feedback loop between vitamin D₃ (1,25(OH)₂D₃) and parathyroid hormone (PTH) plays a central role in skeletal development. PTH-related protein (PTHrP) shares homology and its receptor (PTHR1) with PTH. The aim of this study was to investigate whether there is a functional paracrine feedback loop between 1,25(OH)₂D₃ and PTHrP in the growth plate, in parallel with the endocrine feedback loop between 1,25(OH)₂D₃ and PTH. This was investigated in ATDC5 cells treated with 10⁻⁸ M 1,25(OH)₂D₃ or PTHrP, Col2-pd2EGFP transgenic mice, and primary Col2-pd2EGFP growth plate chondrocytes isolated by FACS, using RT-qPCR, Western blot, PTHrP ELISA, chromatin immunoprecipitation (ChIP) assay, silencing of the 1,25(OH)₂D₃ receptor (VDR), immunofluorescent staining, immunohistochemistry, and histomorphometric analysis of the growth plate. The ChIP assay confirmed functional binding of the VDR to the PTHrP promoter, but not to the PTHR1 promoter. Treatment with 1,25(OH)₂D₃ decreased PTHrP protein production, an effect which was prevented by silencing of the VDR. Treatment with PTHrP significantly induced VDR production, but did not affect 1α- and 24-hydroxylase expression. Hypertrophic differentiation was inhibited by PTHrP and 1,25(OH)₂D₃ treatment. Taken together, these findings indicate that there is a functional paracrine feedback loop between 1,25(OH)₂D₃ and PTHrP in the growth plate. 1,25(OH)₂D₃ decreases PTHrP production, while PTHrP increases chondrocyte sensitivity to 1,25(OH)₂D₃ by increasing VDR production. In light of the role of 1,25(OH)₂D₃ and PTHrP in modulating chondrocyte differentiation, 1,25(OH)₂D₃ in addition to PTHrP could potentially be used to prevent undesirable hypertrophic chondrocyte differentiation during cartilage repair or regeneration.

Longitudinal bone growth occurs at the growth plate, a highly organized cartilage structure that contains proliferating chondrocytes. These cells undergo a maturation process involving hypertrophy followed by apoptosis, thereby facilitating bone formation (Nilsson et al., 2005; Brochhausen et al., 2009). Some changes that occur in cartilage after injury or osteoarthritis (OA) resemble the processes that occur during the differentiation of growth plate chondrocytes (Dreier, 2010; Zhang et al., 2012). In healthy articular cartilage, chondrocytes resist proliferation and terminal differentiation. In contrast, cartilage damage caused by injuries or OA reactivates chondrocyte hypertrophy as part of a repair mechanism, accompanied by acquisition of an autolytic phenotype and cartilage degradation (Dreier, 2010; van der Kraan and van den Berg, 2012; Zhang et al., 2012). Ultimately, the hypertrophic chondrocytes undergo apoptosis to enable bone deposition (van der Kraan and van den Berg, 2012). The inferior quality of the repaired cartilage suggests that the inhibition of chondrocyte hypertrophy could be a target of treatment to improve cartilage repair (Zhang et al., 2012).

Chondrocyte proliferation and differentiation at the growth plate is regulated through the interaction of systemic hormones (endocrine level) and locally produced growth factors (autocrine and/or paracrine level). The endocrine feedback loop between the active metabolite of vitamin D₃ (1,25(OH)₂D₃) and parathyroid hormone (PTH) plays a central role in calcium and phosphate homeostasis during skeletal growth (Nilsson et al., 2005). Vitamin D₃ is hydroxylated in the liver to 25-hydroxycholecalciferol (25(OH)D₃), which is thereafter hydroxylated in various target cells into 1,25(OH)₂D₃ by the enzyme 1α-hydroxylase (St-Arnaud and Naja, 2011). 1,25(OH)₂D₃ in turn can be deactivated and catabolized by the enzyme 24-hydroxylase (Akeno et al., 1997; Tryfonidou et al., 2003). It has been shown that 1,25(OH)₂D₃ exerts its genomic effects by binding to its nuclear receptor (VDR), and that this complex then binds to vitamin D₃ response elements (VDREs) in the promoter region of various target genes (Healy et al., 2003, 2005b; St-Arnaud and Naja, 2011).

Both 1,25(OH)₂D₃ and PTH are also active at the growth plate and play an important autocrine and/or paracrine role during chondrocyte proliferation and/or differentiation (Kato et al., 1990; Klaus et al., 1991; Drissi et al., 2002). Growth plate chondrocytes express the VDR and the enzymes 1α- and 24-hydroxylase in vitro as well as in vivo (Boyan et al., 2002; Hugel et al., 2004; Nilsson et al., 2005; Naja et al., 2009; St-Arnaud and Naja, 2011). PTH-related protein (PTHrP) resembles PTH in genetic sequence and structure and both PTH and PTHrP share the same receptor: PTHR1 (Schipani and Provot, 2003; Zhang et al., 2012). PTHR1 is expressed in low levels by proliferating chondrocytes and in high levels by pre/early hypertrophic chondrocytes (Kronenberg, 2003; Mak et al., 2008; Zhang et al., 2012). PTHrP is produced by proliferative growth plate chondrocytes and prevents proliferative cells from leaving the proliferating pool. In this way, hypertrophic chondrocyte differentiation is delayed (Kronenberg, 2003; Mak et al., 2008; Hirai et al., 2011; Zhang et al., 2012).

Understanding the processes behind chondrocyte differentiation is crucial, not only from a developmental point of view. Regenerative strategies for bone and cartilage make use of mesenchymal (stem) cells undergoing chondrogenic differentiation. The growth plate can be used as a model to study these processes, mainly because it has a highly organized structure, with chondrocytes undergoing differentiation in an orderly fashion (Nilsson et al., 2005; Brochhausen et al., 2009; Denison et al., 2009). Therefore, the main aim of this study was to investigate whether there is a functional paracrine feedback loop between 1,25(OH)₂D₃ and PTHrP in prehypertrophic growth plate chondrocytes, in parallel to the well-known endocrine 1,25(OH)₂D₃-PTH feedback loop.

We hypothesized that PTHrP increases the sensitivity of growth plate chondrocytes to 1,25(OH)₂D₃ either by increasing 1,25(OH)₂D₃ production by upregulating 1α-hydroxylase, and/or decreasing the catabolism of 1,25(OH)₂D₃ by downregulating 24-hydroxylase, and/or by upregulating VDR expression (Supplementary File 1). The feedback loop is closed by the inhibition of PTHrP and/or PTHR1 transcription by the binding of 1,25(OH)₂D₃ to a VDRE located in the promoter region of (one of) these target genes.

Materials and Methods

In vitro studies with the ATDC5 cell line

Cell culture and experimental design

The mouse chondrogenic ATDC5 cell line was kindly provided by Dr. T. Welting (UMC Maastricht, The Netherlands). Cell culture was performed as described previously (Caron et al., 2012). Standard differentiation culture medium was supplemented with 0.2 mM l-ascorbic acid 2-phosphate (AsAP, A8960, Sigma-Aldrich, Saint Louis) to reduce the time in culture until prehypertrophic differentiation (Altaf et al., 2006). Cells grown in a standard differentiation medium were compared with cells grown in a standard differentiation medium supplemented with 10⁻⁸ M PTHrP (pTH-Related Protein (1–34) amide, H-9095, Bachem, Bubendorf, Switzerland) or 10⁻⁸ M 1,25(OH)₂D₃ (kindly provided by Dr. C. Veldhuizen, Dishman, The Netherlands). Stripped fetal bovine serum (FBS) was used, which is devoid of vitamin D₃ metabolites and growth factors. ATDC5 cells were plated on 24-well plates (Greiner Cellstar®) at a density of 6,400 cells/cm². Six hours later, cell differentiation was induced (day 0). The three different culture groups were studied from differentiation day 7 until day 10 (prehypertrophic phase) at the following time points: T₀, T₁, T₂, T₄, T₈, T₂₄, T₂₈, T₄₈, and T₇₂ (the digits indicate the number of hours after PTHrP or 1,25(OH)₂D₃ was first added to the cell culture medium). A T₀, T₂₄, and T₄₈, 10⁻⁸ M PTHrP or 1,25(OH)₂D₃ was added to the experimental plates. The experiment was repeated at least six times for each time point and culture condition.

RNA isolation and cDNA synthesis

Total RNA was extracted with the aid of the RNeasy® Mini kit (74104, Qiagen, Valencia, CA), according to the manufacturer's protocol. An additional DNA digestion step with DNase (RNAse-Free DNase Set, 79254, Qiagen) was included to ensure DNA removal. Total RNA was quantified with a Nanodrop® ND-1000 spectrophotometer (Thermo Scientific, Wilmington, DE). cDNA was synthesized using the iScript™ cDNA Synthesis Kit (170-8891, Bio-Rad, Veenendaal, The Netherlands), according to the manufacturer's instructions.

Quantitative determination of the expression of target genes

Primer nucleotide sequences for all reference genes were obtained from the Eccles Lab Reference Gene List (http://openwetware.org/wiki/Eccles:QPCR_reference_genes). Most primer nucleotide sequences for the target genes were obtained from PrimerBank (http://pga.mgh.harvard.edu/primerbank/index.html). For the other target genes, primer sequences were designed using PerlPrimer (http://perlprimer.sourceforge.net). Subsequently, M-fold version 3.2 was used to check for secondary structure formation (Zuker, 2003). Primer uniqueness and specificity was determined using BLAST (Altschul et al., 1997). Annealing temperatures were established by performing a temperature gradient PCR on a 16-fold dilution series. Amplicons were validated by sequence analysis using an ABI Prism 3130XL genetic analyzer (Applied Biosystems, Foster City, CA). All primers were purchased from Eurogentec (Maastricht, The Netherlands). An overview of the primer pairs used is given in Supplementary File 2. In order to normalize the relative expression of the target genes, a set of 10 reference genes was evaluated: Hspca, Rpl32, Rps19, Ywhaz, B2m, Gapdh, Hbms­, Hprt-1, Sdha, and Tbp. The geNorm program was used to evaluate the stability of the housekeeping genes (Vandesompele et al., 2002). The three most stably expressed reference genes in the ATDC5 cell line, Rpl32, Rps1, and Sdha, were chosen to normalize gene expression. However, Sdha was not used as a reference gene in the experiment in which the VDR was silenced, because Sdha gene expression was not sufficiently stable under the experimental conditions used. RT-qPCR was performed using the iQ™ SYBR Green Supermix Kit (Bio-Rad, Veenendaal, The Netherlands) and the MyiQ™ single color Real-Time PCR Detection System (Bio-Rad).

Protein isolation and Western blot

Semi-quantitative protein expression of the VDR and collagen type X was determined using Western blot. Protein was extracted from the ATDC5 cells with 50 μl RIPA buffer per well, and protein concentration was determined with a Lowry assay (500-0116, Bio-Rad). Thereafter, 30 μg protein was subjected to 12% SDS–PAGE, and electroblotted onto a Hybond-C nitrocellulose membrane (90RPN203C, GE Healthcare Life Sciences, Diegem, Belgium). Only one SDS–PAGE gel and membrane was used per target protein. The membrane was blocked for 60 min, followed by overnight incubation at 4°C with the first antibody (Supplementary File 3). Thereafter, the membrane was washed and incubated for 60 min with a horseradish peroxidase (HRP)-conjugated second antibody. Protein was detected by enhanced chemiluminescence (Molecular Imager ChemiDoc XRS System, Bio-Rad). Control experiments were included in which the first antibody was omitted. After completion of Western blotting of the target proteins, the membranes were stripped using Stripping Buffer (Restore™ Western Blot Stripping Buffer, 21059, Thermo Scientific), and β-actin protein expression was determined. The mean volume of the protein bands was determined with Quantity One software using volumetric band analysis. The mean volume of the target gene was divided by the mean volume of β-actin (target gene/β-actin ratio), to correct for different protein concentrations applied to the membranes. Western blot images were prepared using Adobe Photoshop CS5.1. Linear adjustment of brightness, contrast, and color balance was applied to every pixel in the image.

PTHrP ELISA

The concentration of PTHrP in the cell lysate and culture media (which was stored at −70°C) was measured with an ELISA (ELISA kit for mouse PLP, USCN E90819Mu), according to the manufacturer's instructions.

mRNA superinduction

RNA and protein samples were taken from the control cultures and the 10⁻⁸ M 1,25(OH)₂D₃-treated cultures on differentiation day 7 at T₀, T₁, T₂, T₄, T₈, T₁₂, and T₂₄. RNA isolation, cDNA synthesis, RT-qPCR for Pthrp and a PTHrP ELISA were performed as described previously.

DNA quantification

DNA was quantified using the Quant-iT™ PicoGreen® dsDNA Assay Kit (P11496, Invitrogen, Paisley, UK).

Silencing RNA

Stealth RNAi™ siRNA Duplex (10620312, Invitrogen) was used to silence the ATDC5 VDR. The siRNA oligonucleotide sequence used for the mouse VDR (nm_009504) was: 5′ CCCUUCAAUGGAGAUUGCCGCAUCA 3′ (ORF, GC percentage estimated at 52%). A Stealth RNAi™ Control Duplex (Invitrogen) sequence 5′ CCCUAACGAGGGUUA CGCCAUUUCA 3′ (scrambled mock, GC percentage estimated at 52%) was used to determine the effect of Stealth RNAi™ siRNA Duplex versus background. Lipofectamine™ RNAiMAX (13778-075, Invitrogen) was used as transfection reagent. ATDC5 cells were plated on 24-well plates (Greiner Cellstar®) at a standard density of 6,400 /cm². Six hours later, ATDC5 cell differentiation was induced (day 0). On differentiation day 4, RNA in the approximately 100% confluent ATDC5 cells was silenced (Lipofectamine™ RNAiMAX concentration of 5.3 µl/ml, siRNA VDR-oligo (Stealth RNAi™ siRNA Duplex)/scrambled mock (Stealth RNAi™ siRNA Control Duplex) concentration of 50 nM). After siRNA transfection medium was added to the experimental plates, the ATDC5 cells were incubated for 48 h at 37ºC in 5% CO₂ before the siRNA transfection medium was replaced by normal differentiation medium. On day 7 of differentiation (which was 72 h after siRNA initiation), 10⁻⁸ M PTHrP or 1,25(OH)₂D₃ was added to determine whether the effects of PTHrP and 1,25(OH)₂D₃ on the ATDC5 cells could be prevented by VDR silencing. From differentiation day 7 until day 10 (i.e., 72–144 h after siRNA initiation), the different culture groups were studied at the time points T₀, T₂₄, T₄₈, and T₇₂ (the digits indicate the number of hours after PTHrP or 1,25(OH)₂D₃ was first added to the cell culture medium). 10⁻⁸ M PTHrP or 1,25(OH)₂D₃ was added at T₀, T₂₄, and T₄₈. Nine different cell culture conditions (control, control + scrambled mock, control + VDR-oligo, PTHrP, PTHrP + scrambled mock, PTHrP + VDR-oligo, 1,25(OH)₂D₃, 1,25(OH)₂D₃ + scrambled mock, and 1,25(OH)₂D₃ + VDR-oligo) were studied and compared.

Chromatin immunoprecipitation (ChIP) assay

The sequence and location of the VDREs in the PTHrP promoter region have only been determined in the rat (Falzon, 1996; Kremer et al., 1996a). The Mus musculus and Rattus norvegicus PTHrP promoter regions were BLASTed (Altschul et al., 1997) to determine their homology (Supplementary File 4). Primers were designed and validated for the two VDRE regions of the mouse PTHrP promoter (Supplementary File 4, Fig. 3A). As a negative control, primers were designed at 1,000 bp upstream of VDRE2.

Figure 3.

Identification of a functional vitamin D₃ response element (VDRE) in the PTHrP and PTHR1 promoter region. A: Schematic presentation of the mouse PTHrP promoter region with location of the vitamin D₃ response elements (VDREs). Primers were designed for VDRE1 and VDRE2. The location of VDRE1 and VDRE2 upstream from TSS is 913 and 571 bp, respectively. bp, base pairs; TSS, transcription start site. B: VDR ChIP assay performed on ATDC5 cells in the prehypertrophic phase of differentiation. Cells were fixed after 24 h of treatment with 10⁻⁸ M 1,25(OH)₂D₃. IgG was used as normalization control and a sequence 1,000 bp upstream of VDRE2 was used as a negative control. Enrichment is given as mean ± SD; n = 8. The results indicate functional binding of 1,25(OH)₂D₃ and its receptor, the VDR, to the PTHrP promoter region. C: Overview of expected VDREs in the PTHR1 promoter region. The PTHR1 gene has two promoter regions, P1 and P2. In bone and cartilage, the P2 promoter controls PTHR1 expression, whereas expression of P1 is restricted mainly to the kidney. The location of the expected VDRE1 (AGGTGA) and VDRE2 (GGTTGA) is 2,092; and 2,027 bp upstream from TSS, respectively. The arrows indicate the position of the primers used in this study. bp, base pairs; TSS, transcription start site. D: VDR ChIP assay performed on ATDC5 cells in the prehypertrophic phase of differentiation. The VDRE combination contains both the expected VDRE1 and VDRE2 sequence. The same conditions apply as described with B. *P < 0.05, **P < 0.01.

To date, there is no VDRE reported for the PTHR1 gene. The PTHR1 gene has two promoter regions, P1 and P2 (Fig. 3C). In bone and cartilage, the P2 promoter controls Pthr1 gene expression, and therefore we searched for possible VDREs in this P2 promoter with the aid of the core binding motif consensus sequence RGKTSA (R=A or G, K=G or T, S=C or G) (Toell et al., 2000). This motif was included in two six-nucleotide sequences in the region upstream of U3: AGGTGA and GGTTGA, which are 2,092 and 2,027 bp upstream of the transcription start site (TSS), respectively. The distance between these sequences was 104 bp. The consensus is that a VDRE has normally three to six nucleotides between the two motif sequences (Carlberg, 1995). We cannot exclude that the area between these two sequences loops back to bring the sequences close together, enabling the VDR to bind. For this reason, primers were designed for each six-nucleotide sequence (VDRE1 and VDRE2), but also for the region containing these two core binding motif consensus sequences (VDRE combination). As negative and positive controls, primers were designed at 1,000 bp upstream of the expected VDREs and for one of the VDREs of the 24-hydroxylase promoter, respectively (Supplementary File 4). VDREs for 24-hydroxylase have been identified in the rat (Zierold et al., 1995), and therefore the rat promoter region was BLASTed against the mouse 24-hydroxylase promoter region. Both VDREs revealed 100% alignment. Primers were designed for the VDRE 5′ CGCACCCGCTGAACC 3′.

The ChIP assay was performed as described previously (Pandit et al., 2012), with minor modifications. Briefly, ATDC5 cells were seeded in Falcon Primaria petri dishes (353803, BD Biosciences, Breda, The Netherlands) at a density of 6,400 cells/cm² and cultured for 7 days (approximately 22 × 10⁶ cells per dish), as described earlier in this section. Cells were treated with 10⁻⁸M 1,25(OH)₂D₃ for 24 h. For immunoprecipitation, rat anti-VDR (MA1-710, Affinity Bioreagents, Golden, CO) was used, whereas an equal amount of rat IgG (IgG2b, CBL606, Chemicon, Billerica, MA) was used as a normalization control. For more detailed information on the ChIP assay, see Supplementary File 5.

Statistical analysis

For determination of relative quantitative gene expression, the corrected ΔCt method was used (Pfaffl, 2001). ΔCt-values were determined for each time point by subtracting the mean reference gene Ct-value at the time point of interest from the target gene Ct-value at the same time point: ΔCt = Ct_target − Ct_{mean ref}. Subsequently, all values were related to the T₀ time point by subtracting the ΔCt-value for T₀ from the ΔCt-value for T_x, the time point of interest: ΔCt_Tx − ΔCt_T0. For each individual experiment, target gene expression per time point of interest (n-fold change) was determined separately. Afterwards, for each target gene, the mean n-fold changes and standard deviations in gene expression per time point of interest were calculated. In the silencing study, VDR knockdown percentages in the siRNA VDR-oligo cultures were determined by subtracting the VDR Ct-value from the mean reference gene Ct-value for each time point: Ct_ref − Ct_vdr. Subsequently, this value was subtracted from the value for the siRNA scrambled mock cultures or the value for the control cultures, to obtain ΔCt-values. VDR knockdown percentages were calculated using the formula: 100 × 1−(1/E_vdr^ΔCt).

Statistical analysis was performed using R Studio (version 0.96, http://www.rstudio.com) and R (version 2.15.2) (R Core Team, 2012). To determine whether the enrichment in the ChIP experiments was statistically significant, the data were examined for normal distribution, and a one-way ANOVA with Benjamini–Hochberg correction was used. For the rest of the data (target gene and PTHrP protein production), a mixed model for dependent data was used. This mixed model was optimized per target gene/protein and comparison. After it was determined whether the data were normally distributed, the random part of the model was determined (e.g., with random slopes and/or random intercepts). Thereafter, the fixed part of the model was optimized. Interaction of time and treatment (culture condition) appeared necessary in all cases. In the above mentioned tests, a P-value <0.05 was considered significant.

In vivo studies

Animals

The animal studies were approved by the institutional animal care committee (DEC 2008.III.03.024), as required by Dutch law. The colony of the transgenic Col2-pd2EGFP reporter mice was maintained at the SPF facilities with approval from the Dutch ministry of Infrastructure and Environment (DEM/SAS IG 99-057/03). The Col2-pd2EGFP transgenic mouse is appropriate for visualizing Col2a1 expression, for monitoring chondrocyte differentiation, and for isolating murine growth plate chondrocytes by fluorescence activated cell sorting (FACS) (Tryfonidou et al., 2011).

Diets

Vitamin D₃ sufficient (control, TD 07370, 0.47% Ca, 0.3% P, 2,200 IU/kg vitamin D₃) and deficient (TD 89123, 0.47% Ca, 0.3% P, 0 IU/kg vitamin D₃) diets were purchased from Teklad Harlan SD (Indianapolis, IN). The vitamin D₃ content of these diets was re-analyzed by an independent laboratory (TNO Nutrition and Food Research, Zeist, The Netherlands); the vitamin D₃-sufficient diet contained 1,900 IU/kg and the vitamin D₃-deficient diet less than 20 IU/kg.

Experimental design

The mice were kept under standard conditions in the experimental animal facility of the University of Utrecht. Control offspring were obtained from dams maintained on the vitamin D₃-sufficient diet. Vitamin D₃-deficient pups were obtained by feeding the dams a vitamin D₃-deficient diet, from 2 weeks prior to mating until weaning. The weaned offspring were given the vitamin D₃-deficient diet until euthanasia at 6 weeks of age. In order to ensure vitamin D₃ deficiency, the pregnant females and their offspring were housed in filter-top cages, in which all fluorescent light was shielded, thereby preventing the endogenous production of vitamin D₃. Only those pups that had a weight within 2 SD of the mean at 3 weeks of age were included. The pups were weaned at approximately 3 weeks of age, depending on whether they could feed independently. Weaned pups were housed in groups according to diet and gender (in order to prevent mating): vitamin D₃ sufficient (VitD+), vitamin D₃ deficient (VitD−), and vitamin D₃ deficient supplemented with 1,25(OH)₂D₃ (VitD−; +1,25D). The animals were weighed every week at fixed times. At 3 weeks of age, the vitamin D₃-deficient pups were given either 50 ng 1,25(OH)₂D₃ (VitD−; +1,25D, intraperitoneal (IP), 1 ng/µl in sterilized peanut oil; mean dose 5 ng/g body weight, BW) or placebo (VitD−, sterilized peanut oil). Thereafter, the dose of vitamin D₃ metabolite was adjusted weekly, based on the mean BW of the respective group. The vitamin D₃-sufficient mice (VitD+) received placebo (sterilized peanut oil) IP. The IP administration was performed five times a week, lege artis, by the animal caretakers. At 6 weeks of age, blood (as much as possible) was obtained by cardiac puncture under general anesthesia, followed by cervical dislocation.

Serum biochemistry

Blood samples for the measurement of calcium (Ca), inorganic phosphate (P), and vitamin D₃ metabolites were collected in heparin-coated mini-collection-tubes (450479, Greiner Bio-One, Monroe, NC). For the measurement of PTH, blood was collected in EDTA-coated mini-collection tubes (REF 450475, Greiner Bio-One). Samples were immediately placed on ice until centrifugation and plasma was stored at −20ºC until further analysis. Ca and P levels were measured according to standard procedures. Plasma 25(OH)D₃ levels were measured to verify vitamin D₃ deficiency in the respective groups. 25(OH)D₃ was extracted from 25 μl plasma with the Bligh and Dyer method (Bligh and Dyer, 1959) and quantified with the aid of a competitive binding assay. Thereafter, plasma samples for pairs of mice were pooled within the ascribed group (due to sample volume limitations) and vitamin D₃ metabolites were extracted from the pooled plasma using acetonitrile followed by a two-step phase extraction (C18 and Silicagel cartridge) and separated by straight phase HPLC. 25(OH)D₃ was quantitatively determined using a competitive protein binding assay with a sensitivity of 2 nmol/L and 1,25(OH)₂D₃ was quantitatively determined with a radioreceptor assay (Reinhardt et al., 1984), with a sensitivity of 40 pmol/L. Levels were corrected for procedural losses (recovery) with the aid of the specific 3H-labeled vitamin D₃ metabolite. PTH was determined according to the manufacturer's instructions (intact PTH mouse EIA, Alpco Diagnostics, Salem, NH).

Growth plate histology

After euthanasia, the right tibia was removed and fixed in 4% formaldehyde (pH = 7.4, 4ºC) for 24 h and decalcified in 0.5 M EDTA in Ca–Mg free Hanks solution (pH = 7.8, 4ºC) for 7 days. After demineralization, the bones were washed and bisected in the sagittal plane. One half was embedded in Tissue Tek (O.C.T. compound, Sakura Fine Technical Co. Ltd., Tokyo, Japan) and stored at −70ºC until further processing. The other half of the tibia was embedded in paraffin and stored at 4ºC until further processing.

Quantitative determination of the expression of target genes

From each mouse, growth plate chondrocytes were isolated as described previously (Tryfonidou et al., 2011). Briefly, after euthanasia, the long bones and rib cartilage (except for the right tibia) were removed and submerged in cold (4ºC) Hanks + 2% pen/strep. The growth plate and adjacent tissue from each animal was dissected with the aid of a stereoscope and scalpel blade, pooled, and digested overnight at 37ºC 5% CO₂ with collagenase II (4176, Worthington, Lakewood, NJ). After digestion, cells were suspended in DMEM/F12 + glutamine without phenol red (21041, Gibco, Bleiswijk, The Netherlands) with 10% FCS (PAA) and 2% pen/strep. The cell solution was individually run through a FACS (BD Influx cell sorter, BD Biosciences) with a nozzle of 100 µm. After collection, the cells were lysed in 350 µl RLT buffer (Qiagen) containing 1% β-mercaptoethanol and stored at −70°C until further processing. RNA isolation, cDNA production, and RT-qPCR were performed as described for the in vitro studies. The relative level of gene expression of Pthrp, Pthr1, Vdr, and ColX was determined (Supplementary File 2).

Immunofluorescent staining

Immunofluorescent staining of bone for collagen type X was performed as described previously (Tryfonidou et al., 2011) on 10-μm cryosections, in order to define the pattern of GFP expression in relation to the phase of differentiation of growth plate chondrocytes. Briefly, after sections were rinsed with PBS, antigen retrieval with 4 mg/ml bovine hyaluronidase (450 IU/mg, Sigma-Aldrich, St. Louis, MO) was performed. After blockade of aspecific binding sites with 10% goat serum, sections were incubated with rabbit polyclonal anti-mouse collagen type X (PXNC2, 1:100) overnight at 4°C (Lunstrum et al., 1999). Goat anti-rabbit ALEXA 568 (1:100, Invitrogen) was used as second antibody. Nuclear counterstaining was performed with TO-PRO®-3, and the slides were mounted in Prolong Gold anti-fade reagent (Invitrogen). Confocal microscopy was performed using a Leica TCS SP confocal laser scanning microscope. Immunofluorescent staining images were prepared using Adobe Photoshop CS5.1. Linear adjustment of brightness, contrast, and color balance was applied to every pixel in the image.

Immunohistochemistry

Paraffin sections of 4 μm were cut and mounted on Microscope KP plus slides (Klinipath B.V., Duiven, The Netherlands). Each slide contained three sections: one section of each group in an ad random order. Slides were deparaffinized through xylene (two times 5 min) and graded ethanol (96%, 80%, 70%, 60%, 30%; 5 min each), followed by two rinses with PBS. Mid-sagittal sections of the three samples representing each group were included in the same Pap-pen circle and were thus incubated under identical conditions. Thereafter, antigen retrieval was performed (Supplementary File 3). After inhibition of endogenous peroxidase for 5 min, sections were preincubated with blocking buffer for 30 min at room temperature (RT). The sections were incubated overnight with first antibody at 4°C. For further processing, depending on the first antibody, either the EnVision™-HRP detection system (Dako), the goat ImmunnoCruz™ system (sc-2053, Santa Cruz Biotechnology, Inc., Heidelberg, Germany), or specific second antibody was applied for 30 min at RT, followed by incubation with streptavidin-conjugated with HRP for 30 min at RT. All antibodies were visualized with the liquid DAB+ substrate chromogen system (K3468, Dako, Glostrup, Denmark). In control experiments, the first antibody was omitted and, depending on the antibody, either substitution of the first antibody with its respective serum, and/or competition of the first antibody with corresponding peptides available for PTHR1 (sc-12777 P) was performed. Raw images were made using a Colorview IIIU digital camera (Olympus, Zoeterwoude, The Netherlands) mounted to a BX-40 microscope (Olympus). Histomorphometry was performed on the raw images with Image J software package (Rasband NIH, Bethesda, ML). The mean width of the growth plate (GPl.Th.) was calculated from the width of the growth plate at 20 set locations. The mean width of the proliferative (GPl.Th._Pr) and hypertrophic zone (GPl.Th._Hp) was determined in a similar way. The proliferative zone comprised the region containing columnar chondrocytes of constant size, whereas the hypertrophic zone was defined by collagen type X staining.

A custom-made color range selection optimized for each immunohistochemical stain was used to determine the area within the growth plate that stained positive. For each growth plate section, the proportion of the surface area of the growth plate that stained positively for the respective protein was calculated, as were the mean gray value and integrated density. Quantitative image analysis of the nuclear VDR area that stained positive was performed using the Image J plugin ImmunoRatio, which calculates the percentage of positively stained nuclear area (labeling index) by using a color deconvolution algorithm for separating the staining components (diaminobenzidine and hematoxylin) and adaptive thresholding for nuclear area segmentation.

In vitro studies with primary growth plate chondrocytes

Cell culture and experimental design

The isolation of primary growth plate chondrocytes from the limbs of three 9-day-old Col2-pd2EGFP mice was performed as described previously (Tryfonidou et al., 2011). Approximately 26% of the selected population was positive for GFP. On differentiation day 0 (T₀), droplets of 10 μl (containing approximately 20,000 cells) were placed on a Falcon Primaria petri dish (353803, BD Biosciences), not touching each other to create multiple high-density micro-cultures. After 2 h, 10 ml differentiation medium was added as describe previously, and after 2 days (T₂), 10⁻⁸ M 1,25(OH)₂D₃ (Dishman BV, Veenendaal, The Netherlands) was daily added to the 1,25(OH)₂D₃-treated cultures. In a similar manner, high-density microcultures on chamber slides (Lab-Tek®) were studied from differentiation day 0 until day 7. At the time points T₀, T₃, T₅, and T₇ (the digits indicate the number of days after differentiation culture medium was first added to the wells), medium for PTHrP ELISA and cells for RNA isolation were obtained, and formalin-fixed slices were stained for collagen type X (as described previously). Confocal microscopy was performed using a Leica SPEII confocal laser-scanning microscope (Leica Microsystems, Wetzlar, Germany). The concentration of PTHrP in culture media was measured with an ELISA (ELISA kit for mouse PLP, USCN E90819Mu), according to the manufacturer's instructions.

Chromatin immunoprecipitation (ChIP) assay

The ChIP analysis of the cultured primary growth plate chondrocytes (Falcon Primaria petri dish) was performed as described for the in vitro studies.

Quantitative determination of the expression of target genes

RNA isolation, cDNA production, and RT-qPCR were performed as described for the in vitro studies. Ultimately, the relative gene expression of Pthrp, Pthr1, Vdr, and ColX was determined (Supplementary File 2).

Statistical analysis

After it was determined that the data had a normal distribution, a one-way ANOVA with Benjamini–Hochberg correction was used to analyze the significance of differences in the mean (±SD) width of the growth plate (including the proliferative and hypertrophic zone), plasma parameters, target gene expression, and ChIP assay results between the different groups. For body weight and PTHrP protein levels, a mixed model for dependent data was used as described at the in vitro studies. In the above-mentioned tests, a P-value <0.05 was considered significant.

Results

In vitro studies with the ATDC5 cell line

The effect of PTHrP on the vitamin D₃ pathway

Protein and mRNA expression of 1α- and 24-hydroxylase was stable during culture and appeared not affected by treatment with 10⁻⁸ M PTHrP (data not shown). At all time points, Vdr gene expression was significantly higher in the PTHrP-treated cultures than in the control cultures (P < 0.01, Fig. 1A). After the first addition of PTHrP at T₀, Vdr gene expression increased significantly to reach a maximum at T₄ and declined thereafter. At T₂₈, 4 h after the second addition of PTHrP, Vdr gene expression was significantly higher than that at T₂₄ (P < 0.01), but significantly lower than that at T₄ (P < 0.001). Vdr gene expression did not change substantially after T₂₈. VDR protein levels were also significantly higher at time points T₂₄ and T₄₈ in the PTHrP-treated cultures than in the control cultures (P < 0.01, Fig. 1B). Altogether, these data indicate that PTHrP treatment increased ATDC5 chondrocyte sensitivity for 1,25(OH)₂D₃ by upregulating VDR expression and not by influencing 1α- and/or 24-hydroxylase expression.

Figure 1.

VDR and PTHrP expression in 1,25(OH)₂D₃- and PTHrP-treated ATDC5 chondrocytes in the prehypertrophic phase of differentiation. ATDC5 cells were treated starting from day 7 of differentiation (T₀) with 10⁻⁸ M PTHrP or 1,25(OH)₂D₃ at T₀, T₂₄, and T₄₈. A. Relative Vdr gene expression (mean ± SD; n = 8). T₀ in the control culture was set at 1. B: VDR protein expression expressed as VDR/β-actin ratio (mean ± SD; n = 6). T₀ in the control culture was set at 1. c = control culture, P = 10⁻⁸ M PTHrP-treated culture, d = 10⁻⁸ M 1,25(OH)₂D₃-treated culture. C: Relative Pthrp gene expression (mean ± SD; n = 8). T₀ in the control culture was set at 1. D: PTHrP protein levels corrected for DNA content (pg PTHrP/ng DNA) in ATDC5 culture media of control cultures and 10⁻⁸ M 1,25(OH)₂D₃-treated cultures, determined using a PTHrP ELISA (mean ± SD; n = 4). *P < 0.05, **P < 0.01, ***P < 0.001.

The effect of vitamin D₃ on the PTHrP pathway

Treatment of ATDC5 chondrocytes with 10⁻⁸ M 1,25(OH)₂D₃ increased the activity of the vitamin D₃ pathway (Armbrecht and Boltz, 1991), as evidenced by a more than 1,000 times increased expression of the 24-OHase gene compared with control cultures (data not shown). 10⁻⁸ M 1,25(OH)₂D₃ treatment also significantly increased VDR gene and protein expression in ATDC5 chondrocytes in the prehypertrophic phase of differentiation (P < 0.05, Fig. 1A,B) (Davoodi et al., 1995; Klaus et al., 1998; Healy et al., 2005a). Pthrp gene expression was significantly higher in the 10⁻⁸ M 1,25(OH)₂D₃-treated cultures than in the control cultures at all time points (P < 0.001, Fig. 1C). In contrast, the PTHrP protein content of the medium from 1,25(OH)₂D₃-treated cultures was significantly lower than that of medium from the control cultures at T₄₈ and T₇₂ (P < 0.001, Fig. 1D). This seemingly contradictory result could be attributed to a phenomenon called “mRNA superinduction.” It has been reported that the inhibition of PTHrP protein synthesis leads to the induction of Pthrp mRNA expression (“mRNA superinduction”) in a number of cell lines (Ikeda et al., 1990). We explored this possibility in the ATDC5 cell line by measuring PTHrP mRNA and protein levels (in culture medium and cell lysate, corrected for DNA content) in the first 24 h after treatment with 1,25(OH)₂D₃. PTHrP protein production decreased by 10⁻⁸ M 1,25(OH)₂D₃ treatment from T₄/T₈ onward (P < 0.01, Fig. 2B,C) and remained lower in the 10⁻⁸ M 1,25(OH)₂D₃-treated cultures than in the control cultures until T₇₂ (Fig. 1D). In contrast, Pthrp mRNA expression significantly increased from T₂₄ onward by 10⁻⁸ M 1,25(OH)₂D₃ treatment (Figs. 1A and 2C). Given the time line of these events, where the increase in Pthrp mRNA expression (T₂₄) is preceded by an initial significant decrease in PTHrP protein production (T₄/T₈), it seems reasonable to assume that the increased Pthrp mRNA levels were due to the “mRNA superinduction” phenomenon. To determine whether this observed effect was physiological or pharmacological, we conducted an additional experiment investigating the time line of PTHrP mRNA and protein expression in the presence of 10⁻⁸, 10⁻¹⁰, and 10⁻¹² M 1,25(OH)₂D₃. PTHrP protein levels were dose-dependently influenced by 1,25(OH)₂D₃ treatment (Fig. 2B,C). Moreover, the increased Pthrp mRNA expression observed in the 10⁻⁸M 1,25(OH)₂D₃-treated ATDCs chondrocytes (at T₂₄), was not observed in the 10⁻¹⁰ and 10⁻¹² M 1,25(OH)₂D₃-treated cells. This indicates that the “Pthrp mRNA superinduction” phenomenon, only observed in the 10⁻⁸ M 1,25(OH)₂D₃-treated ATDC5 chondrocytes, is pharmacological and can be avoided with lower physiological 1,25(OH)₂D₃ regimes.

Figure 2.

Pthrp mRNA superinduction phenomenon. 10⁻⁸, 10⁻¹⁰, and 10⁻¹² M 1,25(OH)₂D₃-treated ATDC5 cell cultures were compared with control cultures. At T₀ (on day 7 of differentiation, i.e., the prehypertrophic phase of the ATDC5 chondrocytes), 1,25(OH)₂D₃ was added to the culture medium and the respective PTHrP gene (A) and protein content of cell lysates (B) and culture media (C) were determined at T_1–24 after 1,25(OH)₂D₃ treatment (mean ± SD; n = 6 for control and 10⁻⁸ M 1,25(OH)₂D₃ cultures and n = 4 for 10⁻¹⁰ and 10⁻¹² M 1,25(OH)₂D₃ cultures). Relative Pthrp gene expression at T₀ in the control culture was set at 1. The pharmacological “Pthrp mRNA superinduction” phenomenon is only observed in the 10⁻⁸ M 1,25(OH)₂D₃-treated ATDC5 chondrocytes, and can be avoided with lower physiological regimes. *P < 0.05, **P < 0.01, ***P < 0.001.

Functional binding of 1,25(OH)₂D₃ to the promoter region of the PTHrP gene

Computational analysis of the PTHrP promoter revealed the presence of two VDR binding elements (Fig. 3A). ChIP experiments showed a significant 2-fold enrichment of VDRE1 (P < 0.01) and a 1.6-fold enrichment of VDRE2 (n.s.) in the PTHrP promoter (Fig. 3B). No enrichment was observed in the PTHrP negative control (sequence 1,000 bp upstream of VDRE2) and a significant 2.5-fold enrichment was observed in the positive control (24-hydroxylase promoter, P < 0.05, Fig. 3B). Altogether, this indicates that in ATDC5 chondrocytes in the prehypertrophic phase of differentiation, 1,25(OH)₂D₃ binds to its nuclear receptor, the VDR, and together they bind to a 1,25(OH)₂D₃-responsive region (VDRE1) in the PTHrP promoter. In this way, 1,25(OH)₂D₃ directly regulates Pthrp expression via genomic effects.

Computational analysis of the PTHR1 promoter revealed the presence of two VDREs (Fig. 3C). ChIP experiments showed a 1.4-fold enrichment of both VDRE1 and VDRE2, and a 2.1-fold enrichment of the VDRE combination (region containing both the expected VDRE1 and VDRE2) in the PTHR1 promoter (Fig. 3D). However, these results were not significant and the PTHR1 negative control (sequence 1,000 bp upstream of the expected VDREs) also revealed a 1.5-fold enrichment of the VDR antibody. Altogether, this indicates that there is no functional binding of 1,25(OH)₂D₃ and the VDR to a VDRE in the PTHR1 promoter of ATDC5 chondrocytes in the prehypertrophic phase of differentiation.

Determination of the role of the VDR in the paracrine feedback loop between PTHrP and 1,25(OH)₂D₃

To clarify the role of the VDR in the paracrine feedback loop between PTHrP and 1,25(OH)₂D₃, we successfully silenced the VDR in 100% confluent ATDC5 chondrocytes in the prehypertrophic phase of differentiation; significant VDR knockdown was seen at all time points in the control + VDR-oligo cultures and the 1,25(OH)₂D₃ + VDR-oligo cultures (P < 0.05, Fig. 4A,C).

Figure 4.

The role of VDR in the paracrine feedback loop between PTHrP and 1,25(OH)₂D₃. A,B: Relative Vdr and Pthrp gene expression in ATDC5 cells in the prehypertrophic phase of differentiation (mean ± SD; n = 6). In the VDR-oligo groups, silencing of the VDR was performed on differentiation day 4. At T₀ (day 7 of differentiation), T₂₄, and T₄₈, 10⁻⁸ M 1,25(OH)₂D₃ was added to the three 1,25(OH)₂D₃-treated culture groups. Relative target gene expression at T₀ in the control culture was set at 1. C: Western blot analysis confirming the successful siRNA transfection in 10⁻⁸ M 1,25(OH)₂D₃-treated prehypertrophic ATDC5 cells with approximately 50% downregulation of VDR production. Time is shown in hours after 1,25(OH)₂D₃ was first added and the VDR/β-actin ratio (mean ± SD; n = 3) in the 1,25(OH)₂D₃ culture at T₂₄ was set at 1. d = 10⁻⁸ M 1,25(OH)₂D₃-supplemented culture, ds = 1,25(OH)₂D₃ + scrambled mock culture, do = 1,25(OH)₂D₃+ VDR-oligo culture. D: PTHrP protein levels (mean ± SD; n = 3) corrected for DNA content in ATDC5 culture media, determined using a PTHrP ELISA. *P < 0.05, **P < 0.01, ***P < 0.001.

Up- and downregulated PTHrP protein production in the scrambled mock cultures compared with the control cultures was observed at several time points (both with and without 1,25(OH)₂D₃ supplementation, Fig. 4D), which can be explained by off-target effects of the scrambled mock siRNA sequence (Persengiev et al., 2004). The culture medium PTHrP protein content increased over time in the control cultures, whereas it decreased over time in the 1,25(OH)₂D₃-treated cultures, with the difference being significant at T₂₄–T₇₂ (P < 0.05, Fig. 4D). At T₄₈ and T₇₂, PTHrP protein levels in the 1,25(OH)₂D₃+ VDR-oligo cultures were significantly higher than in the 1,25(OH)₂D₃ cultures (P < 0.05), indicating that VDR silencing counteracted the 1,25(OH)₂D₃-mediated inhibitory effect on PTHrP protein production in ATDC5 chondrocytes in the prehypertrophic phase of differentiation.

The role of the paracrine PTHrP-1,25(OH)₂D₃ feedback loop in hypertrophic chondrocyte differentiation

1,25(OH)₂D₃ treatment of the prehypertrophic ATDC5 cultures decreased the DNA content by about 40% whereas the addition of PTHrP increased the DNA content by about 30% compared with the control cultures (data not shown). Gene expression of the (pre) hypertrophic differentiation markers Col9, Pthr1, and ColX increased significantly over time in the control cultures (P < 0.05), but not in the PTHrP- and 1,25(OH)₂D₃-treated cultures (Fig. 5A). Generally, Pthr1, Col9, and ColX gene expression was higher in the control cultures than in the PTHrP- or 1,25(OH)₂D₃-treated cultures at all time points. However, Western blot analysis indicated that only on differentiation day 14 (T₁₆₈), collagen type X protein expression was significantly lower in the PTHrP- and 1,25(OH)₂D₃-treated cultures than in the control cultures (P < 0.001, Fig. 5B), but not on differentiation day 10 (T₇₂). This indicates that both 1,25(OH)₂D₃ and PTHrP treatment inhibited hypertrophic differentiation of the ATDC5 chondrocytes, which was noticed earlier at the mRNA than at the protein level.

Figure 5.

The role of the paracrine PTHrP-1,25(OH)₂D₃ feedback loop in hypertrophic chondrocyte differentiation. A: Relative Col9, ColX, and Pthr1 gene expression in PTHrP- or 1,25(OH)₂D₃-treated ATDC5 cell cultures compared with control cultures (mean ± SD; n = 8). 10⁻⁸ M PTHrP or 1,25(OH)₂D₃ was added daily from T₀, which was day 7 of differentiation, that is, the prehypertrophic phase of the ATDC5 chondrocytes. Relative target gene expression at T₀ in the control culture was set at 1. B: Collagen type X protein expression expressed as Collagen type X/β-actin ratio (mean ± SD; n = 3). Time is shown in hours after PTHrP/1,25(OH)₂D₃ was first added to the culture medium on differentiation day 7. T₇₂ in the control culture was set at 1. c = control culture, P = 10⁻⁸ M PTHrP-treated culture, d = 10⁻⁸ M 1,25(OH)₂D₃-treated culture. *P < 0.05, **P < 0.01, ***P < 0.001.

To clarify the role of the paracrine feedback loop between PTHrP and 1,25(OH)₂D₃ in growth plate chondrocyte differentiation, we silenced the VDR in ATDC5 chondrocytes in the prehypertrophic phase of differentiation. In the control + scrambled mock and the control + VDR-oligo cultures, ColX gene expression did not increase over time, in contrast to what was observed in the control cultures (Fig. 6A). For the PTHrP- and 1,25(OH)₂D₃-treated ATDC5 cultures, no significant differences in collagen type X expression between the non-silenced (PTHrP/1,25(OH)₂D₃) and the VDR silenced (PTHrP/1,25(OH)₂D₃+VDR-oligo) cultures were observed (Fig. 6A,B), indicating that VDR silencing did not counteract the PTHrP/1,25(OH)₂D₃-mediated inhibitory effect on collagen type X gene and protein expression in ATDC5 chondrocytes in the prehypertrophic phase of differentiation. In contrast, Pthr1 expression was significantly upregulated in the 1,25(OH)₂D₃ + VDR-oligo cultures at T₄₈ and T₇₂ (P < 0.01, Fig. 6), indicating that VDR silencing counteracted the 1,25(OH)₂D₃-mediated inhibitory effect on Pthr1 expression in ATDC5 chondrocytes in the prehypertrophic phase of differentiation.

Figure 6.

The role of the VDR in the paracrine PTHrP-1,25(OH)₂D₃ feedback loop during hypertrophic chondrocyte differentiation. A: Relative ColX and Pthr1 gene expression corrected for reference genes in ATDC5 cells in the prehypertrophic differentiation phase (mean ± SD; n = 6). In the VDR-oligo groups, silencing of the VDR was performed on differentiation day 4. 10⁻⁸ M PTHrP or 1,25(OH)₂D₃ was added daily, starting at T₀, which was day 7 of differentiation. Relative target gene expression at T₀ in the control culture was set at 1. B: Collagen type X protein expression in 10⁻⁸ M PTHrP- and 1,25(OH)₂D₃-treated ATDC5 cells silenced for the VDR. Time is shown in hours after PTHrP/1,25(OH)₂D₃ was first added to the culture medium on differentiation day 7. The collagen type X/β-actin ratio (mean ± SD; n = 3) in the PTHrP/1,25(OH)₂D₃ culture at T₂₄ was set at 1. p = PTHrP culture, pm = PTHrP + scrambled mock culture, po = PTHrP + VDR-oligo culture, d = 1,25(OH)₂D₃ culture, dm = 1,25(OH)₂D₃ + scrambled mock culture, do = 1,25(OH)₂D₃ + VDR-oligo culture. *P < 0.05, **P < 0.01, ***P < 0.001.

Generally, ColX and Vdr gene expression was significantly higher in the scrambled mock cultures than in the non-silenced control cultures, whereas the scrambled mock treatment did not affect Pthr1 and Pthrp expression (Figs. 4A,B and 6A). The difference in ColX and Vdr gene expression was, however, not accompanied with differences in VDR and collagen type X protein expression between the scrambled mock and the non-silenced control cultures (Figs. 4C and 6B), indicating that the scrambled mock-induced upregulated mRNA expression was not translated into increased protein expression. Most probably, the upregulated ColX and Vdr mRNA expression can be attributed to off-target effects of the scrambled mock sequence (Persengiev et al., 2004).

In vivo studies

Animals and biochemistry

From 4 weeks of age onward, the control vitamin D₃-sufficient (VitD+) mice weighed significantly more than the VitD− mice (P < 0.001, Supplementary file 6A). Plasma 25(OH)D₃ levels confirmed vitamin D₃ deficiency in VitD− mice (P < 0.001, Supplementary file 6A). As expected, the plasma concentration of 1,25(OH)₂D₃ was significantly higher in VitD−; +1,25D mice than in VitD+ and VitD− mice (P < 0.01). Plasma calcium (Ca) levels were significantly lower in VitD− mice than in VitD+ mice (P < 0.01), but were significantly higher in VitD−; +1,25D mice than in VitD− and VitD+ mice (P < 0.01). Inorganic phosphate (P) plasma levels were not significantly different between groups, but PTH levels were significantly higher in VitD− mice than in VitD+ and VitD−; +1,25D mice (P < 0.001).

Growth plate histomorphometry

The mean growth plate height (GPl.Th) was significantly higher in VitD− mice than in VitD+ and VitD−; +1,25D mice (P < 0.001, Fig. 7A). The mean GPl.Th._Pr did not differ between groups, whereas the GPl.Th._Pz/GPl.Th ratio was significantly lower in VitD− mice than in the other groups (P < 0.05). The mean GPl.Th._Hyp and the ratio of GPl.Th._Hyp/GPl.Th were significantly higher in VitD− mice than in VitD+ or VitD−; +1,25D mice (P < 0.05). The standard deviation of GPl.Th._Hyp, a measure of the irregularity of the hypertrophic zone, was also significantly higher in VitD− mice than in VitD+ and VitD−; +1,25D mice (P < 0.01, Fig. 7A). The Col2-pd2EGFP vitamin D₃-deficient mice thus showed the classical signs of low body weight, hypocalcemia, hyperparathyroidism, and rickets with an enlarged growth plate. The latter was mainly due to an increased and irregular hypertrophic zone (Donohue and Demay, 2002). Their vitamin D₃-deficient phenotype was successfully reversed by 1,25(OH)₂D₃ supplementation.

Figure 7.

Results of the in vivo study (A + B) and the primary cell culture (C + D) of Col2-pd2EGFP transgenic mice growth plate chondrocytes. A: Histomorphometry (H&E) and immunofluorescent staining for collagen type X (red) of tibial growth plates from Col2-Col2-pd2EGFP transgenic mice (6 weeks of age) indicating that vitamin D₃ deficient mice (VitD−) developed rickets. Supplementation with 1,25(OH)₂D₃ (VitD−; +1,25D) reversed the rachitic phenotype as compared with controls (VitD+). Note that Col2-pd2EGFP fluorescence (green) is native (mean ± SD; n = 7). B: Relative ColX, Vdr, Pthrp, and Pthr1 gene expression corrected for reference genes in sorted Col2-pd2EGFP positive growth plate chondrocytes after termination of the study, at 6 weeks of age (mean ± SD; n = 6). Relative target gene expression in the growth plate of control (VitD+) mice was set at 1. C: Relative ColX, Vdr, Pthrp, and Pthr1 gene expression corrected for reference genes in control or 1,25(OH)₂D₃-treated cultured primary Col2-pd2EGFP growth plate chondrocytes. Time is shown in days after the start of differentiation (T₀, T₃, T₅, and T₇). From T₂ onward, 10⁻⁸ M 1,25(OH)₂D₃ was daily added to the culture medium. Relative target gene expression at T₀ in the control culture was set at 1. These data were not subjected to statistical analysis due to the small sample size (n = 1 experimental replicate). D: From left to right: Immunofluorescent staining for collagen type X and native expression of Col2-pd2EGFP in control and 10⁻⁸ M 1,25(OH)₂D₃-treated cultured primary Col2-pd2EGFP growth plate chondrocytes. Transcription of collagen type 2 (green) and hypertrophic differentiation of chondrocytes (red) were detected in control cultures, whereas the 10⁻⁸ M 1,25(OH)₂D₃-treated cells expressed no Col2-pd2EGFP and only occasionally collagen type X immunofluorescence in their pericellular matrix. A ChIP assay revealed enrichment of PTHrP VDRE1, but not the VDRE2. A PTHrP ELISA performed on culture media revealed decreased PTHrP production in the 10⁻⁸ M 1,25(OH)₂D₃-treated primary growth plate chondrocytes compared with control cultures (mean ± SD; n = 4). *P < 0.05, **P < 0.01, ***P < 0.001.

Gene and protein expression of primary growth plate chondrocytes

Col2-pd2EGFP-positive growth plate chondrocytes were sorted and processed for RNA isolation on an individual basis. The number of positive Col2-pd2EGFP chondrocytes was 5.1 × 10⁵± 3.1 × 10⁵, 1.4 × 10⁶± 8.1 × 10⁵, and 3.3 × 10⁵± 2.1 × 10⁵ for the VitD+, VitD−, and VitD−; +1,25D mice, respectively. (Pre) hypertrophic differentiation marker (ColX and Pthr1) gene expression was significantly higher in VitD− mice than in mice from the two other groups (P < 0.01, Fig. 7B). Furthermore, Vdr gene expression was significantly higher in VitD− mice than in VitD+ mice (P < 0.05) and Pthrp gene expression was significantly higher in VitD−; +1,25D mice than in VitD+ mice (P < 0.05, Fig. 7B).

Confocal imaging of tibial growth plates from Col2-pd2EGFP transgenic mice indicated that VitD+ and VitD−; +1,25D mice had less Col2-pd2EGFP than VitD− mice (Fig. 7A). Nuclear VDR protein expression was highest in VitD− mice and lowest in VitD+ mice (Supplementary file 6B). Within the proliferative and hypertrophic zones, the percentage area staining for PTHrP (over the total surface of the growth plate, and thus corrected for the larger surface of the growth plate in VitD− mice), the integrated density, and mean gray value did not differ between groups (Supplementary file 6B). However, in accordance with gene expression levels, PTHR1 staining was present in all growth plate zones and seemed to be in greater amounts in VitD− mice than in mice from the other two groups (Supplementary file 6B).

In vitro studies with primary growth plate chondrocytes

As other signaling pathways may interfere with the interpretation of the in vivo results, an in vitro study was performed with primary Col2-pd2EGFP growth plate chondrocytes. Vitamin D₃-deficient primary growth plate chondrocytes (controls) were compared with 10⁻⁸ M 1,25(OH)₂D₃-treated primary growth plate chondrocytes over 7 days. ColX gene expression decreased with time and was lower in 1,25(OH)₂D₃-treated than in control primary growth plate chondrocytes at all times points (Fig. 7C). In line with our hypothesis, Pthrp gene expression was lower and Vdr gene expression was higher in 1,25(OH)₂D₃-treated compared with control primary growth plate chondrocytes at all time points (Fig. 7C).

Confocal imaging of native Col2-pd2EGFP fluorescence and staining for collagen type X revealed transcription of collagen type II and hypertrophic differentiation of the control cultures at T₇, whereas 1,25(OH)₂D₃-treated primary growth plate chondrocytes showed no Col2-pd2EGFP fluorescence and collagen type X staining in their pericellular matrix (Fig. 7D), which was in line with our hypothesis and confirms the reported gene expression profiles.

ChIP assays showed no enrichment of VDRE1, VDRE2, or the negative control in the PTHR1 promoter (data not shown), indicating that there is no functional binding of 1,25(OH)₂D₃ and the VDR to a VDRE in the PTHR1 promoter of primary Col2-pd2EGFP growth plate chondrocytes. However, in parallel with what was found in the chondrogenic ATDC5 cell line, ChIP assays on sorted Col2-pd2EGFP growth plate chondrocytes showed a significant 2.6-fold enrichment of VDRE1 (P < 0.05), and no enrichment of VDRE2 or the negative control in the PTHrP promoter (Fig. 7D). Furthermore, PTHrP production in 1,25(OH)₂D₃-treated primary growth plate chondrocytes was significantly lower than in control cultures at all time points (P < 0.01, Fig. 7D). These results indicate that also in primary Col2-pd2EGFP growth plate chondrocytes, 1,25(OH)₂D₃ and the VDR together bind to a 1,25(OH)₂D₃-responsive region (VDRE1) in the PTHrP promoter. In this way, 1,25(OH)₂D₃ directly inhibits PTHrP production via the VDR.

Discussion

The regulation of chondrocyte differentiation is a key event in skeletal development. Regenerative strategies for cartilage engineering use mesenchymal stem cells (MSCs), but are hampered by the inherent capacity of chondrogenically differentiating MSCs to undergo hypertrophic differentiation (Hellingman et al., 2012). Hence, understanding the processes that regulate chondrocyte differentiation is crucial to further fine-tune regenerative strategies for cartilage and bone engineering. In order to do so, we used growth plate chondrocytes, which undergo differentiation in an orderly fashion. We used complementary in vitro (ATDC5 and primary growth plate chondrocytes) and in vivo (Col2-pd2EGFP mice) models and demonstrated that there is a functional paracrine feedback loop between 1,25(OH)₂D₃ and PTHrP in prehypertrophic growth plate chondrocytes (Fig. 8).

Figure 8.

The paracrine feedback loop between PTHrP and 1,25(OH)₂D₃ in prehypertrophic growth plate chondrocytes. 1,25(OH)₂D₃ inhibits PTHrP production through a VDRE in the PTHrP promoter, and PTHrP increases chondrocyte sensitivity to 1,25(OH)₂D₃ via increased VDR expression.

The effect of PTHrP on the vitamin D₃ pathway

We hypothesized that in growth plate chondrocytes, PTHrP treatment would increase vitamin D₃ activity, either by interference with the key enzymes of vitamin D₃ metabolism (1α- and 24-hydroxylase) or by a direct effect on the VDR (Supplementary File 1). In the kidney, PTH stimulates 1α-hydroxylase and inhibits 24-hydroxylase (Zierold et al., 2003; Xue et al., 2005). The subcutaneous administration of PTHrP to healthy women increased plasma 1,25(OH)₂D₃ levels, but did not affect endogenous PTH levels, indicating that—at an endocrine level—PTHrP stimulates 1α-hydroxylase in the kidney in the same way that PTH does (Henry et al., 1997). However, in the current study, PTHrP did not significantly affect the expression of 1α- and 24-hydroxylase in ATDC5 chondrocytes in the prehypertrophic phase of differentiation. Although we did not determine the actual activity of the respective enzymes, it is tempting to speculate that PTH and PTHrP influence 1α- and 24-hydroxylase expression in a tissue/cell-specific manner.

We found that PTHrP significantly increased VDR expression (Fig. 1A,B). Vdr gene expression was upregulated 4 h after initiation of PTHrP treatment (T₄), whereas repeated treatment resulted in a lower increase in Vdr gene expression at T₂₈ (Fig. 1A). Desensitization of the PTH-induced cAMP response (Jongen et al., 1996) might explain the decreased response of the ATDC5 cells to PTHrP after the second addition of PTHrP (at T₂₄) compared with the first addition (at T₀). Previous studies found a PTH/PTHrP-mediated decreased VDR expression in renal, intestinal, and osteoblast-like cells (Reinhardt and Horst, 1990; Sriussadaporn et al., 1995; Healy et al., 2005b), but increased VDR expression in growth plate chondrocytes and osteoblast-like cells (Pols et al., 1988; Klaus et al., 1997). A possible explanation for these contradictory results is again a cell-line/tissue specific effect.

The effect of vitamin D₃ on the PTHrP pathway

We hypothesized that treatment with 1,25(OH)₂D₃ would decrease PTHrP production, because Pthrp expression is reported to be suppressed by 1,25(OH)₂D₃ in several other cell types (Kremer et al., 1991; Falzon, 1996; Sepulveda et al., 2006).

In contrast to our hypothesis, in the in vivo studies with the Col2-pd2EGFP mouse model, Pthrp gene expression was increased in Col2-pd2EGFP growth plate chondrocytes from VitD−; +1,25D mice compared with VitD− or VitD+ mice (Fig. 7B). However, these in vivo studies are limited by several factors. First, growth plate chondrocytes were isolated on the basis of FACS of Col2-pd2EGFP-positive chondrocytes and, hence, may not be representative of the total population of growth plate chondrocytes. Second, the altered calcium homeostasis in the vitamin D₃-deficient mice treated with 1,25(OH)₂D₃ resulted in hypercalcemia (Supplementary file 6A), which could have attributed to the increased Pthrp and Vdr expression in VitD−; +1,25D mice (Fig. 7B). Calcium has been shown to regulate PTHrP secretion (Kremer et al., 1996b; Chattopadhyay et al., 2000) and to have an additive effect on the homologous upregulation of VDR expression during treatment with 1,25(OH)₂D₃ (Davoodi et al., 1995; Klaus et al., 1998; Healy et al., 2005a). To exclude the effects of interfering factors, we also performed in vitro studies with ATDC5 and primary Col2-pd2EGFP growth plate chondrocytes and studied the pattern of expression of target genes and proteins in the presence and absence of 1,25(OH)₂D₃. In line with our hypothesis, PTHrP gene and protein expression was lower in the 1,25(OH)₂D₃-treated primary Col2-pd2EGFP growth plate chondrocytes compared with control cultures (Fig. 7C,D). Also in prehypertrophic ATDC5 chondrocytes, PTHrP protein production was decreased by 1,25(OH)₂D₃-treatment (Figs. 1D and 2B,C). The counterintuitive increase of Pthrp gene expression in 10⁻⁸ M 1,25(OH)₂D₃-treated ATDC5 cultures compared with control cultures from 24 h of treatment onward (Figs. 1C and 2A) could be explained by “Pthrp mRNA superinduction” (Ikeda et al., 1990), which is only provoked by the preceding decreased PTHrP protein expression in 10⁻⁸ M 1,25(OH)₂D₃-treated prehypertrophic ATDC5 chondrocytes (Fig. 2B,C), and not at lower dosages.

We furthermore silenced the VDR to clarify its role in the regulation of PTHrP expression by 1,25(OH)₂D₃. VDR silencing counteracted the inhibitory effect of 1,25(OH)₂D₃ on PTHrP protein production (Fig. 4D). In addition, we observed significant enrichment of the PTHrP VDRE1 in 1,25(OH)₂D₃-treated ATDC5 and primary growth plate chondrocytes in the prehypertrophic phase of differentiation (Figs. 3B and 7D). This is the first study to report functional binding of 1,25(OH)₂D₃ through its receptor to the promoter region of the PTHrP gene in growth plate chondrocytes. The 1,25(OH)₂D₃-responsive region (VDRE) in the PTHrP promoter has already been characterized in other tissues (Kremer et al., 1991, 1996a; Falzon, 1996). Taken together, these findings prove that in the growth plate, the functional paracrine feedback loop between PTHrP and 1,25(OH)₂D₃ is closed by the inhibition of PTHrP transcription by the binding of 1,25(OH)₂D₃ to a VDRE located in the PTHrP (and not the PTHR1) promoter region (Fig. 8).

The role of the paracrine PTHrP-1,25(OH)₂D₃ feedback loop in hypertrophic chondrocyte differentiation

Having established that there is a functional paracrine feedback loop between PTHrP and 1,25(OH)₂D₃ in growth plate chondrocytes (Fig. 8), we wanted to define the role of this feedback loop in hypertrophic chondrocyte differentiation. Both 1,25(OH)₂D₃ and PTHrP affected the proliferation and differentiation of growth plate chondrocytes. 1,25(OH)₂D₃ treatment resulted in a decreased DNA content of the prehypertrophic ATDC5 chondrocytes, indirectly indicating that 1,25(OH)₂D₃ had an anti-proliferative effect, which is in line with previous reports (Klaus et al., 1991). Moreover, 1,25(OH)₂D₃ had an inhibitory effect on chondrocyte hypertrophy, based on the reduced Col9, ColX, and Pthr1 gene expression and collagen type X protein expression in 1,25(OH)₂D₃-treated ATDC5 and primary Col2-pd2EGFP positive growth plate chondrocytes (Figs. 5 and 7C,D). This is by no means a new finding, since 1,25(OH)₂D₃ has been shown to inhibit terminal chondrocyte differentiation both in vitro and in vivo (Kato et al., 1990; Drissi et al., 2002; Idelevich et al., 2011; Castillo et al., 2012). In addition, VDR silencing in 1,25(OH)₂D₃-treated ATDC5 cells only partially reversed the anti-hypertrophic effects of 1,25(OH)₂D₃: it prevented the inhibitory effect of 1,25(OH)₂D₃ on Pthr1 gene expression, but it did not affect collagen type X expression (Fig. 6). In unpublished studies, we found that treatment with 10⁻¹⁰ and 10⁻¹² M 1,25(OH)₂D₃ decreased the expression of (pre) hypertrophic differentiation markers, and hence VDR silencing may have been ineffective in counteracting the strong inhibitory effect of 10⁻⁸ M 1,25(OH)₂D₃ on hypertrophic differentiation. An alternative explanation is that 1,25(OH)₂D₃ also exerts effects by binding to a membrane-associated receptor PDIA3 (Boyan et al., 2006; St-Arnaud and Naja, 2011).

PTHrP is a well-known suppressor of hypertrophic chondrocyte differentiation (Ballock and O'Keefe, 2003; van der Eerden et al., 2003; Hoogendam et al., 2007; Brochhausen et al., 2009). Accordingly, in the control ATDC5 cultures, Col9 gene and collagen type X gene and protein expression increased with time, but this increase was prevented by the addition of PTHrP (Fig. 5). Furthermore, the addition of PTHrP to ATDC5 cells in a prehypertrophic differentiation phase significantly downregulated the expression of the gene for the receptor of PTHrP, Pthr1 (Fig. 5A), which is also in line with previous reports (Weisser et al., 2002; Wealthall, 2009). This homologous downregulation of Pthr1 is possibly a measure to prevent overstimulation, but another explanation lies in the physiological role of PTHrP, namely, to prevent proliferative cells leaving the proliferating pool. In this way, hypertrophic chondrocyte differentiation—and thus PTHR1 production—is delayed (Kronenberg, 2003; Mak et al., 2008; Hirai et al., 2011; Zhang et al., 2012). Taken together, the results of our study indicate that PTHrP could be used clinically to inhibit undesirable hypertrophic chondrocyte differentiation. Accordingly, in vitro work has already demonstrated that PTHrP successfully inhibited hypertrophic differentiation of articular chondrocytes (Wang et al., 2011; Zhang et al., 2013) and cartilage constructs engineered from bone marrow-derived mesenchymal stem cells (BMSCs), without losing cartilage-specific matrix proteins (US patent 20080318859, Kafienah et al., 2007). Moreover, in vivo, intra-articular PTHrP injection together with collagen-silk scaffold implantation (4–6 weeks post-injury) inhibited terminal differentiation and enhanced chondrogenesis in induced osteochondral defects in rabbits (Zhang et al., 2013). In contrast, in chondrogenically differentiated BMSC pellets, PTHrP could not diminish the T₃-induced enhancement of hypertrophy (Mueller et al., 2013). Despite significant reduction of some hypertrophic markers, the absolute level of expression was still high compared with articular chondrocyte-based cartilage constructs (Hellingman et al., 2012). Furthermore, PTHrP has even been reported to suppress chondrogenic differentiation of BMSC pellets (Weiss et al., 2010). Noteworthy is the fact that above-mentioned studies started the PTHrP treatment at different time points, that is, before or after manifestation of hypertrophy. Thus, the use of PTHrP needs to be further investigated with regard to the inhibition of hypertrophic chondrocyte differentiation, articular cartilage repair, and the generation of stable engineered cartilage from MSCs (Zhang et al., 2012).

In order to further study how the interaction between 1,25(OH)₂D₃ and PTHrP influences hypertrophic chondrocyte differentiation, we evaluated whether PTHrP prevents the differentiation of chondrocytes through a VDR-dependent mechanism. For this purpose, the VDR was successfully silenced in PTHrP-supplemented cultures, but the expression of Pthr1 gene and collagen type X protein was hardly influenced (Fig. 6). As it is not possible to discuss these results in the light of earlier reports, we can only postulate that the anti-hypertrophic effect of PTHrP is independent of the VDR. This indicates that 1,25(OH)₂D₃ and PTHrP individually influence hypertrophic chondrocyte differentiation and thus may have a synergistic effect in suppressing hypertrophic differentiation.

Conclusions

Taken together, the data obtained using an integrative approach involving in vitro studies with ATDC5 and primary growth plate chondrocytes and in vivo studies with Col2-pd2EGFP transgenic mice led us to conclude that there is a functional paracrine feedback loop between PTHrP and 1,25(OH)₂D₃ in prehypertrophic growth plate chondrocytes. 1,25(OH)₂D₃ inhibits PTHrP production through a functional binding place (VDRE) in the PTHrP promoter, and PTHrP increases chondrocyte sensitivity to 1,25(OH)₂D₃ by increasing VDR production (Fig. 8). The results of this study furthermore indicate that 1,25(OH)₂D₃ and PTHrP individually influence chondrocyte hypertrophy and, hence, may have the potential to inhibit undesirable hypertrophic chondrocyte differentiation during cartilage repair or engineering. To our knowledge, the effect of 1,25(OH)₂D₃ and the synergistic effect of a combination of PTHrP and 1,25(OH)₂D₃ on MSC-based cartilage regeneration has not yet been evaluated and might provide leads for new strategies to improve the quality of engineered cartilage.

Glossary

1α-OHase

1α-hydroxylase

24-OHase

24-hydroxylase

AsAP

ascorbic acid 2-phosphate

bp

base pairs

BMSC

bone marrow-derived stem cells

BW

body weight

Ca

calcium

ChIP

chromatin immunoprecipitation

Col2a1

collagen type 2

Col9

collagen type 9

ColX

collagen type X

FACS

fluorescence-activated cell sorting

FBS

fetal bovine serum

GPl.Th

mean growth plate height

GPl.Th._Pr

mean height of the proliferative growth plate zone

GPl.Th._Hp

mean height of the hypertrophic growth plate zone

HRP

horseradish peroxidase

Mab

monoclonal antibody

MSC

mesenchymal stem cell

n.s.

not significant

OA

osteoarthritis

P

inorganic phosphate

Pab

polyclonal antibody

PTH

parathyroid hormone

PTHrP

PTH-related protein

PTHR1

PTH/PTHrP receptor

RT

room temperature

SD

standard deviation

siRNA

silencing RNA

Tm

melting temperature

TSS

transcription start site

VDR

nuclear vitamin D3 receptor

VDRE

vitamin D3 response element

Supporting Information

Additional supporting information may be found in the online version of this article at the publisher's web-site.