Heterozygous inactivation of Gnas in adipose‐derived mesenchymal progenitor cells enhances osteoblast differentiation and promotes heterotopic ossification

Abstract

Human genetic disorders sharing the common feature of subcutaneous heterotopic ossification (HO) are caused by heterozygous inactivating mutations in GNAS, a gene encoding multiple transcripts including two stimulatory G proteins, the α subunit of the stimulatory G protein (G_sα) of adenylyl cyclase, and the extralong form of G_sα, XLαs. In one such disorder, progressive osseous heteroplasia (POH), bone formation initiates within subcutaneous fat before progressing to deeper tissues, suggesting that osteogenesis may involve abnormal differentiation of mesenchymal precursors that are present in adipose tissues. We determined by immunohistochemical analysis that GNAS protein expression is limited to G_sα in bone‐lining cells and to G_sα and XLαs in osteocytes. By contrast, the GNAS proteins G_sα, XLαs, and NESP55 are detected in adipocytes and in adipose stroma. Although Gnas transcripts, as assessed by quantitative RT‐PCR, show no significant changes on osteoblast differentiation of bone‐derived precursor cells, the abundance of these transcripts is enhanced by osteoblast differentiation of adipose‐derived mesenchymal progenitors. Using a mouse knockout model, we determined that heterozygous inactivation of Gnas (by disruption of the G_sα‐specific exon 1) abrogates upregulation of multiple Gnas transcripts that normally occurs with osteoblast differentiation in wild‐type adipose stromal cells. These transcriptional changes in Gnas^+/− mice are accompanied by accelerated osteoblast differentiation of adipose stromal cells in vitro. In vivo, altered osteoblast differentiation in Gnas^+/− mice manifests as subcutaneous HO by an intramembranous process. Taken together, these data suggest that Gnas is a key regulator of fate decisions in adipose‐derived mesenchymal progenitor cells, specifically those which are involved in bone formation. © 2011 American Society for Bone and Mineral Research

GNAS/Gnas, Heterotopic ossification, Osteoblast differentiation, Progressive osseous heteroplasia (POH), Albright hereditary osteodystrophy (AHO), Pseudohypoparathyroidism (PHP), Osteoma cutis (OC)

Introduction

Related genetic disorders that are associated with heterozygous inactivating mutations of GNAS include progressive osseous heteroplasia (POH), Albright hereditary osteodystrophy/pseudopseudohypoparathyroidism (AHO/PPHP), pseudohypoparathyroidism (PHP), and osteoma cutis (OC).1–6 This spectrum of GNAS inactivation disorders has the common feature of superficial/dermal ossification with the de novo formation of islands of heterotopic bone in skin and subcutaneous fat.

Based on our previous evaluation,7GNAS‐based disorders of HO can be broadly divided into those presenting with stable superficial bony lesions and those in which superficial lesions progress into deep connective tissue. Among the nonprogressive forms are OC, AHO/PPHP, and PHP‐1a/c. The progressive types of GNAS‐based HO are POH and the POH‐related syndromes. POH presents clinically with superficial HO that progresses to deeper tissues in the absence of multiple other AHO features and without hormone resistance. POH and progressive HO syndromes can be distinguished from other GNAS‐based disorders by the single parameter of HO extension from superficial to deep tissue.

The main product of the GNAS gene is the α subunit of the stimulatory G protein (G_sα). However, multiple mRNAs are transcribed from different promoters at the GNAS locus, and the GNAS gene locus is imprinted, showing differential mRNA expression patterns of these transcripts from maternally and paternally inherited alleles.5, 6, 8 HO in POH is not limited to the dermis and subcutaneous tissues,1, 9, 10 and GNAS mutations in POH patients are paternally inherited. Therefore, the bony patches that coalesce into plaques and later progress to deeper connective tissues (including fascia, skeletal muscle, tendons, and ligaments) correlate progression of heterotopic ossification with inactivating mutations carried on the paternally inherited allele and, by extension, to paternal allele‐specific transcripts.

In addition to G_sα, multiple transcripts are produced using different GNAS promoters, including XLαs, A/B (1A in mouse), and Nesp. The imprinted GNAS gene locus shows maternal, paternal, and biallelic expression of mRNAs.5, 6, 8 For example, G_sα, mRNA is biallelically expressed in most tissues, but the XLαs transcript which encodes an extralong form of the G protein G_sα, is synthesized only from the paternally inherited allele. The Nesp55 transcript is maternally expressed.

Functionally, heterotrimeric G proteins, composed of α, β, and γ subunits, couple extracellular signals from specific cell surface receptors to intracellular effectors.5, 11 G proteins bind guanine nucleotides and are defined by the α subunit of the complex. G_sα is expressed ubiquitously and couples multiple receptors to stimulation of adenylyl cyclase, PKC, and specific ion channels.

The tissue distribution of HO lesions in GNAS‐inactivation disorders such as POH suggests that pathogenesis involves abnormal differentiation of mesenchymal stem cells (MSCs) and/or more committed precursor cells that are present in skin, subcutaneous fat, and muscle, tendon, and ligament tissue. Considerable evidence supports that tissues contain multipotential progenitor cells that can give rise to osteoblasts and adipocytes.12–18 Intramembranous bone formation occurring in subcutaneous fat in POH patients suggests a close, perhaps reciprocal relationship between adipogenesis and osteogenesis in peripheral tissues that is perturbed in patients with this disease. Given this relationship, we sought to uncover a role for GNAS in osteoblast differentiation in adipose‐derived mesenchymal progenitor cells.

Materials and Methods

Animals

Gnas^+/− mice having a heterozygous deletion in exon 119, 20 were bred as F₁ SvEv × CD1 crosses maintaining the deletion on the paternal allele. Analyses were carried out on 3‐month‐old animals unless otherwise indicated. All animal handling and protocols were approved by the University of Pennsylvania Institutional Animal Care and Use Committee.

Isolation and culture of bone marrow stromal cells (BMSCs)

Murine BMSCs were harvested from wild‐type and Gnas^+/− mice according to methods described previously.21 Briefly, metaphyses were removed, and bone marrow plugs were expelled by insertion of a 21‐gauge needle into the marrow cavity and flushing the cavity with 10 mL of α‐minimal essential medium (α‐MEM) with nucleosides (Gibco/Invitrogen, Carlsbad, CA, USA). The bone marrow cells (marrow plugs) from femurs of the same animals were collected, pooled, and dispersed by repeat pipeting before centrifugation at 300g for 10 minutes. The supernatant was removed and the pellet resuspended in α‐MEM containing 10% fetal calf serum (FCS; Hyclone, Rockford, IL, USA). Cells were plated at a density of two marrow plugs per 25 cm² of tissue culture growth surface and refed every 3 days until confluent. Adherent cell strains were derived from individual animals and used at passage 3 or fewer for all experiments. The ability of these cells to differentiate along osteogenic and adipogenic lineages was confirmed.

Isolation and culture of adipose soft tissue stromal cells (STSCs)

Owing to the paucity of subcutaneous fat in Gnas^+/− mice, we isolated STSCs from fat pads that overlie the pelvis and proximal femurs; this adipose tissue is more intimately associated with cutaneous tissue and underlying fascia than abdominal fat. Murine STSCs were harvested from wild‐type and Gnas^+/− mice as described previously.22 Briefly, fat pads were excised, washed in 1× PBS, and minced using the double‐scalpel method. Minced fat tissue was digested with type 2 collagenase (Sigma, St Louis, MO, USA) for 1 hour (37 °C) with shaking. Following digestion, an equal volume of growth medium (DMEM/F12 containing 10% FCS and antibiotics) was added, and the cell suspension was filtered through a 100‐µM cell strainer (BD, Falcon, Franklin Lakes, NJ, USA). Cells in the filtrate were recovered by centrifugation (300g, 10 minutes) and plated in growth medium containing DMEM/F12, 10% FCS, and antibiotics. Adherent cell strains were established from individual animals and used at passage 3 or fewer for all experiments. The ability of these cells to differentiate along osteogenic and adipogenic lineages was confirmed.

Osteogenic differentiation in vitro

For osteogenic differentiation, cells were plated at a density of 10,000/cm² and allowed to attach overnight in growth medium. The following day, the cells were treated with osteogenic differentiation medium containing growth medium supplemented with 100 ng/mL of recombinant bone morphogenetic protein 2 (rBMP‐2; R&D Systems, Minneapolis MN, USA), 10 mM β‐glycerophosphate, and 50 µg/mL of ascorbic acid that was replenished every 3 days. Cells were harvested at specified time points for whole‐cell RNA isolation, detection of alkaline phosphatase (ALP), and detection of mineralization.

Detection of ALP and mineralization

ALP activity was determined histochemically. In brief, cells were washed three times with ddH₂O, and ALP was detected with BCIP/NBT‐plus substrate (Moss Substrates, Pasadena, MD, USA) at 37 °C for 20 minutes. For detection of mineralization, cells were washed three times with ddH₂O and stained with 1% alizarin red S solution (Ricca Chemical Company, Arlington, TX, USA) for 15 minutes and then washed four times with ddH₂O (or until washes were colorless). Images were taken using a Nikon Eclipse TS100 microscope (Nikon Instruments, Inc., Melville, NY, USA) at ×100 magnification.

To quantify mineralization, alizarin red S–stained cells from duplicate wells at each time point were solubilized in 0.5 N HCL and 5% SDS for 30 minutes, and absorbance was measured at 405 nm using a BioRad microplate reader (BioRad, Hercules, CA, USA). Results were normalized to protein content using the BCA assay (Pierce Scientific, Rockford, IL, USA).

Immunohistochemistry

Immunohistochemistry (IHC) was performed for the identification of Gnas‐XLαs, NESP, and G_sα in bone and fat tissue using specific antisera and a horseradish peroxidase (HRP)–conjugated universal secondary antibody detection system (Lab Vision Corporation, Fremont, CA, USA), as described by the manufacturer. Diaminobenzidine (DAB) served as the chromogen/substrate. Briefly, deparaffinized and rehydrated tissue sections were incubated with a hydrogen peroxide–blocking solution for 15 minutes to reduce nonspecific staining owing to endogenous peroxidase and then with preimmune serum at a dilution of 1:100 for 30 minutes. The primary antisera were incubated with sections at a dilution of 1:100 for 1 hour. The addition of HRP‐conjugated secondary antibody and DAB was performed as described by the manufacturer. All washes were performed with PBS, except after chromogen/DAB substrate incubation, which substituted deionized water. Sections were counterstained with hematoxylin. Antiserum against G_sα was purchased commercially (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA). Antigenic peptides for NESP (C‐KGPIPIRRH) and XLαs (C‐HLRPPSPEIQAAD) were synthesized with a Cys residue at the N‐terminus. Peptides were coupled with maleimide‐activated keyhole limpet hemocyanin (KLH) carrier protein (Pierce/Thermo‐Fisher, Rockford, IL, USA) following the manufacture's protocol.23 Peptide‐KLH complexes were used to immunize rabbits. Antibody production was carried out by Cocalico Biologicals (Reamstown, PA, USA). Antibody specificity was confirmed by reactivity with protein species of the estimated molecular weight on Western blotting of crude mouse pituitary lysates and by reactivity on IHC of mouse pituitary tissue sections.

Histology

Tissue samples were fixed in neutral buffered formalin, decalcified, infiltrated and embedded in paraffin, and sectioned at a thickness of 6 µm. Cut samples then were deparaffinized, stained with Harris hematoxylin solution, and counterstained with hematoxylin and eosin (H&E) by standard procedures.

Animal imaging

Whole‐body radiographic images of the frozen or PFA‐fixed mice were performed with a modified FFDM system (Senographe 2000D, General Electric Medical Systems, Milwaukee, WI, USA). The data were analyzed using ImageJ software, http://rsb.info.nih.gov/ij.

RNA isolation and relative quantitative RT‐PCR

RNA was isolated from day 14 cultures treated with and without osteogenic factors using the RNeasy mini kit (Cat. No. 74104; Qiagen, Valencia, CA, USA) according to the manufacturer's intructions. For quantitative RT‐PCR (qRT‐PCR), 5 µL of a 1:20 diluted cDNA and 5.0 µM of each forward and reverse primer in 25 µL of reaction volume was used for amplification of β₂‐microglobulin mRNA; 5 uL of a 1:5 diluted cDNA and 6.0 µM of each forward and reverse primer in 25 µL of reaction volume was used for amplification of G_sα mRNA; 5 µL of a 1:5 diluted cDNA and 7.5 µM of each forward and reverse primer in 25 µL of reaction volume was used for amplification of 1A, Nesp, and XLαs mRNAs. qPCR was performed using the ABI 7500 Fast Real Time PCR System (Applied BioSystems, Foster City, CA, USA), Fast SYBR Green Master Mix (Cat. No. 4385612; Applied BioSystems), and the primer pairs listed in Table 1. Transcript levels normalized to β₂‐microglobulin mRNA were calculated by the delta‐delta method.24 At least three biologic replicates were performed for each transcript, with at least four experimental replicates per biological isolate.

1

Primer Sets Used for qPCR

Gene	5'‐Forward‐3'	5'‐Reverse‐3'
Gsα	GCGCGAGGCCAACAAAAAGAT	TGCCAGACTCTCCAGCACCCAG
1A	GCGTGTGAGTGCGTCTCACT	GATCCTCATCTGCTTCACAATGG
Nesp	CGTCCAGATTCTCCTTGTTTTCA	GATCCTCATCTGCTTCACAATGG
Xlαs	CGAGAGCAGAAGCGCGCA	AGACTCTCCAGCACCTAGA
β2MG	AGTATGGCCGAGCCCAAGA	GCTTGATCACATGTCTCGATCC
ALP	TGAGCGACACGGACAAGA	GGCCTGGTAGTTGTTGTGAG
OCN	CTGACAAAGCCTTCATGTCCAA	GCGGGCGAGTCTGTTCACTA
OPN	GCACTCCAACTGCCCAAGA	TTTTGGAGCCCTGCTTTCTG
Osx	ATGGCGTCCTCTCTGCTTG	TGAAAGGTCAGCGTATGGCTT
Runx2	GTGCGGTGCAAACTTTCTCC	AATGACTCGGTTGGTCTCGG
GAPDH	CAAGGTCATCCATGACAACTTTG	GGCCATCCACAGTCTTCTGG

For other experiments, RNA was isolated at specified time points using TRIzol Reagent (Invitrogen) according to the manufacturer's instructions. For quantitative RT‐PCR (qRT‐PCR), 2.5 µL of a 1:10 diluted cDNA and 5.0 µM of each forward and reverse primer in 12.5 µL of final reaction volume was used for amplification of alkaline phosphatase (ALP), osteocalcin (OCN), osteopontin (OPN), osterix (OSX), and Runx2 mRNA. qPCR was performed using the ABI 7500 Fast Real Time PCR System (Applied BioSystems), Fast SYBR Green Master Mix (Cat. No. 4385612; Applied BioSystems), and the primer pairs listed in Table 1. Transcript levels normalized to Gapdh mRNA were used to calculate relative changes in mRNA expression. At least three biologic replicates were performed for each transcript, with three experimental replicates per biologic isolate.

BrdU labeling

STSCs (passage 2) were seeded in duplicate into a 24‐well plate with cover slips at a density of 0.75 × 10⁴ cells/well. The following day, BrdU was added (diluted 1:100) and incubated in normal growth medium for 16 hours. Cells were fixed in 4% paraformaldehyde for 30 minutes and washed three times with wash buffer (0.1 M PBS [pH 7.4] with 1% Triton X‐100). Following washes, cells were incubated in HCl (1 N) for 10 minutes on ice, followed by incubation with HCl (2 N) for 10 minutes at room temperature and then at 37 °C for 20 minutes. Immediately after acid washes, cells were buffered with the addition of borate buffer (0.1 M) for 12 minutes and washed three times with wash buffer. Cells were blocked in goat serum (0.1 M PBS [pH 7.4] plus Triton X‐100 + glycine [1 M] + 5% normal goat serum) for 1 hour. Following two washes in PBS, mouse anti‐BrdU (Alexa Fluor‐488 conjugate) diluted 1:400 was added and incubated for 2 hours at room temperature in the dark. After incubation with BrdU antibody, 4,6‐diamidino‐2‐phenylindole (DAPI) was added for 5 minutes for the detection of nuclear staining and washed twice with PBS, then cover slips were mounted on slides using Flouromount‐G (SouthernBiotech, Birmingham, AL, USA), and fluorescence visualized using a Nikon Eclipse 90i microscope (Nikon Instruments, Inc.). STSCs isolated from three wild‐type and three Gnas^+/− mice were plated separately in duplicate. Percent BrdU incorporation was calculated as BrdU⁺ cells/DAPI⁺ cells × 100. At least 100 DAPI⁺ cells were scored per cover slip.

Statistical analysis

The Student's t test (two‐sided and paired) was used to determine whether the mean value for relative Gnas transcript expression in cells with osteoblast differentiation differed significantly from that in cells without osteoblast differentiation. Similarly, the Student's t test (two‐sided and paired) was used to determine whether the mean value for transcript levels of osteogenic markers, BrdU incorporation, or quantitative mineralization of cells derived from Gnas^+/− mice differed significantly from wild‐type cells.

Results

Gnas is a complex genetic locus that encodes multiple transcripts. In most cells, G_sα is expressed biallelically, whereas other transcripts show a parent‐of‐origin‐dependent imprinted pattern of transcript expression. XLαs, Nespas, and 1A are specifically expressed from the paternally inherited allele, whereas Nesp is expressed only from the maternally inherited allele.5, 6, 8, 25

To initiate investigation of potential roles of the protein products encoded by the Gnas gene (ie, G_sα, XLαs, and Nesp55) in bone and adipose tissue, we detected these proteins by IHC. We found that the major protein product of Gnas, G_sα, is expressed in bone‐lining cells and osteocytes (Fig. 1). The XLαs protein is also expressed in osteocytes but not in bone‐lining cells. Nesp55 is undetectable in osteocytes and bone‐lining cells (Fig. 1). In contrast, all three Gnas proteins are expressed in adipocytes and in adipose stroma (Fig. 1).

1

Detection of Gnas protein products (ie, G_sα, XLαs, Nesp) in bone and fat tissue from wild‐type mice by immunohistochemical analysis. Arrowheads = bone‐lining cells; arrows = osteocytes.

Heterotopic bone formation initiates within subcutaneous fat in patients with POH,9, 26 suggesting that osteogenesis involves abnormal differentiation of adipose MSCs and/or more committed STSCs present in peripheral tissues such as fat. We therefore isolated murine STSCs from adipose tissue and bone marrow MSCs, two well‐characterized osteogenic cell systems, in order to examine the expression of Gnas transcripts during osteoblast differentiation from these mesenchymal precursor cells.

Using qRT‐PCR, we found no significant changes in any Gnas transcript on osteoblast differentiation of mesenchymal precursor cells derived from wild‐type bone marrow (Fig. 2A). However, the abundance of 1A, Nesp, and XLαs transcripts is statistically significantly enhanced with osteoblast differentiation in wild‐type STSCs (Fig. 2B). The transcript for G_sα showed a statistically nonsignificant trend toward greater expression with osteoblast differentiation (Fig. 2B). By contrast, heterozygous inactivation of the G_sα transcript‐specific exon 1 in Gnas abrogated the upregulation of multiple Gnas transcripts in STSCs compared with wild‐type animals (Fig. 2B).

2

Expression of Gnas transcripts in (A) bone marrow stromal cells (BMSCs) and (B) adipocyte soft tissue stromal cells (STSCs). Specific Gnas transcripts were quantified by qRT‐PCR after 14 days in culture in the presence or absence of osteogenic differentiation factors. Relative expression of each transcript was normalized to day 1 levels. Ob = osteoblast.

The transcriptional changes in Gnas^+/− mice are accompanied by accelerated osteoblast differentiation of STSCs by day 20, as measured by the accumulation of mineralized matrix (Fig. 3). No differences in ALP expression throughout the 20‐day time course were detected (data not shown). Although ALP expression precedes mineralization in our culture system, there is also an overlap of the two that potentially could obscure differences in ALP (because ALP‐expressing STSCs from Gnas^+/− mice mineralize more quickly). In support of this possibility, ALP transcript expression is increased in Gnas^+/− mice compared with wild‐type mice (Fig. 4). To test whether increased expression of matrix proteins may be facilitating mineralization, we also evaluated other osteogenic markers and found that osteocalcin and osteopontin transcripts are upregulated in Gnas^+/− mice (Fig. 4). No differences in early markers, including Runx2 and osterix, were detected between Gnas^+/− and wild‐type cells (data not shown). To determine whether enhanced mineralization is a result of recruitment of a greater number of precursor cells, we determined cell proliferation rates of wild‐type and Gnas^+/− cells, but we did not detect any differences in STSC proliferative potential by BrdU pulse labeling (%BrdU⁺ wild‐type STSCs 65.1% ± 7.4%; %BrdU⁺Gnas^+/− STSCs 56.2 ± 2.2%, p = 0.16).

3

Accelerated osteoblast differentiation of adipocyte soft tissue stromal cells (STSCs) from Gnas^+/− mice compared with wild‐type (WT) animals. (A) Mineralization of wild‐type and Gnas^+/− STSC cultures by alizarin red S staining on day 20. (B) Quantification of mineralization at days 9, 17, and 20 during osteoblast differentiation in wild‐type and Gnas^+/− STSC cultures. *Statistically significant (p < 0.05).

4

Expression of osteogenic markers during differentiation of STSCs derived from wild‐type and Gnas^+/− mice. Expression of (A) alkaline phosphatase (ALP), (B) osteopontin (OPN), and (C) osteocalcin (OCN) mRNAs were quantified by RT‐PCR. Transcript levels were normalized to Gapdh mRNA. Relative expression of the indicated transcripts was compared between genotypes. *Statistically significant (p < 0.05).

In vivo, altered osteoblast differentiation in Gnas^+/− mice manifests phenotypically as subcutaneous HO that is detected by 9 months of age (Fig. 5A–B, D). No HO was detected in wild‐type mice. The presence of bony spicules in fibroproliferative tissue, in the absence of cartilaginous elements, suggests that the ectopic ossification occurs through an intramembranous process (Fig. 5C). In contrast to POH patients but similar to AHO/PHP patients, observation of Gnas^+/− mice up to 24 months of age revealed no progression of HO to deeper connective tissues, such as fascia, skeletal muscle, tendons, and ligaments. The appearance of heterotopic bone in the context of adjacent adipocytes is illustrated in Fig. 5C.

5

Gnas^+/− mice develop subcutaneous heterotopic ossification. (A) X‐ray showing subcutaneous ossification in a 12‐month‐old Gnas^+/− male mouse. (B) Area of interest from panel A at higher magnification. White arrowheads indicate sites of heterotopic ossification. (C) Hematoxylin and eosin staining of a histologic section showing intramembranous ossification. Note the appearance of heterotopic bone in the context of adjacent adipocytes. (D) Onset of heterotopic ossification (HO) with age. Male Gnas^+/− mice were euthanized and imaged at the ages indicated (3 to 24 months). Numbers of unaffected animals are indicated by gray bars and mice with HO by black bars. By 12 months of age, all Gnas^+/− mice demonstrate HO. B = bone; F = fat.

Discussion

Here we have demonstrated that Gnas transcripts in cells from wild‐type mice show no significant changes on osteoblast differentiation of bone marrow stromal cells but are upregulated with osteoblast differentiation of adipose mesenchymal progenitor cells. In contrast, heterozygous inactivation of Gnas results in dysregulation and decreased expression of multiple Gnas transcripts during osteoblast differentiation of soft tissue (adipose) stromal cells. These transcriptional changes in cells from Gnas^+/− mice are correlated with accelerated osteoblast differentiation in vitro. By 9 months of age, aberrant osteoblast differentiation in Gnas^+/− mice results in the formation of subcutaneous HO. These findings suggest that the formation of HO in genetic disorders that are associated with heterozygous inactivating mutations of GNAS, including progressive osseous heteroplasia (POH), Albright hereditary osteodystrophy (AHO), pseudohypoparathyroidism (PHP), and osteoma cutis (OC), is the result of decreased GNAS/Gnas expression.

Gnas knockout mice used in this study have a heterozygous deletion of exon 1 (homozygous knockouts are not viable) that is predicted to affect only the G_sα transcript.19, 20 This knockout mouse line forms subcutaneous heterotopic bone,27 as shown in Fig. 5. Thus, in addition to providing an in vivo model for HO, these Gnas knockout mice are a source of multiple cell types that can be examined ex vivo and tested for cell differentiation capabilities. Given that HO formation in POH and in the Gnas knockout mice appears to be initiated in subcutaneous adipose tissue, our current studies use adipose‐derived stromal cells that allow investigation of stem cells from adult/postnatal tissue that may be relevant to heterotopic bone induction.

The GNAS gene is transcriptionally complex, initiating mRNA synthesis from multiple promoters and unique first exons that splice into common exons 2 to 13.5, 11 The most abundant product of the GNAS gene is G_sα mRNA, which is expressed biallelically in most tissues. However, other GNAS transcripts are imprinted (showing parent‐of‐origin monoallelic expression).28–30 The GNAS Nesp55 transcript is expressed maternally, and the GNAS XLαs and 1A transcripts are expressed paternally. Our data show that statistically significant increases in Gnas XLαs and 1A transcripts occur with osteoblast differentiation in wild‐type but not Gnas^+/−‐derived adipose stromal cells. This suggests that the clinical effects of GNAS‐inactivating mutations in patients is not solely dependent on the level of G_sα expression but may be influenced by the expression levels of other GNAS transcripts. Studies from other investigators31 have shown that XLαs, like G_sα, activates cAMP signaling, suggesting that paternally inherited mutations may result in a larger deficit of cAMP signaling (owing to reduction of G_sα plus the absence of XLαs) compared with maternally inherited mutations (which do not affect expression of XLαs). The roles of Nesp and 1A are less clear, but given its low transcript abundance, it is likely that Nesp plays a minor, if any, role in bone.

The differences in Gnas transcript levels between adipose and bone tissues may be related to their responses to differentiation; in fat, transcript levels are upregulated with osteoblast differentiation; in bone, they are essentially unchanged. This may be related to the role of Gnas in preventing bone formation in tissues where bone should not form. Thus reduction of G_sα (and cAMP signaling) could mediate the formation of superficial HO, but a further deficit in cAMP signaling, perhaps by inactivation of XLαs, may be required for more progressive HO to occur (Fig. 6). This possibility could explain why POH and progressive HO syndromes are at the severe end of a phenotypic spectrum of GNAS‐inactivating conditions associated with extraskeletal ossification.7

6

Possible mechanisms of heterotopic bone formation regulated by GNAS/Gnas. In tissues where bone doses not ordinarily form, Gnas may have a role in preventing extraskeletal ossification in wild‐type animals. Reduction in G_sα (and cAMP signaling) could promote superficial bone formation as seen in Gnas^+/− mice. In POH, where progressive heterotopic ossification occurs, a further deficit in cAMP signaling may be required, perhaps by inactivation of XLαs.

We previously defined a set of clinical characteristics that serve as minimum diagnostic criteria for POH.7 POH can be distinguished from other GNAS‐based disorders of heterotopic ossification (HO) by the presence of superficial HO that progresses to deeper tissues in the absence of multiple other features of Albright hereditary osteodystrophy (AHO) and without hormone resistance.7 It is important to note in the context of clinical HO that the exon 1 deletion in Gnas^+/− mice results in only superficial (subcutaneous) HO that does not progress into deeper tissues, and to date, no exon 1 mutations have been found in POH patients.4 Since the exon 1 deletion in Gnas^+/− mice reduces basal expression of all Gnas transcripts except XLαs, it is interesting to speculate that preservation of XLαs mRNA levels prevents progression of HO in these mice7, 32, 33 (Fig. 6).

We are intrigued by the unexpected observation that dysregulation of multiple Gnas transcripts occurs in response to Gnas exon 1 disruption. One possibility is that there is coordinated regulation of other Gnas transcripts with G_sα expression, although this is purely speculative. In support of such a possibility, an imprinting control region has been identified in the Gnas cluster that affects the expression of multiple transcripts.34 In this case, a paternally derived deletion of the germ line differentially methylated region was associated with the antisense Nespas transcript, but a point mutation in exon 6 also has been described that causes diverse phenotypes depending on whether the mutation is maternally or paternally transmitted.35 It thus is possible that the heterozygous deletion in exon 1 eliminates a control region that also affects expression of other Gnas transcripts.

Although the specific cell targets of heterotopic ossification in genetic conditions such as POH are unknown, investigation of the enhanced osteogenic potential of osteoblast/adipocyte progenitors (such as soft tissue stromal cells) with inactivating Gnas mutations may have direct implications for understanding heterotopic ossification and its progression. Adipocyte STSC strains derived from individual Gnas mutant mice show an enhanced osteogenic potential compared with wild‐type cells, which is demonstrated in vivo as subcutaneous HO that becomes more prevalent with increasing age. Heterotopic ossification in Gnas^+/− mice could proceed by recruitment of a greater number of precursor cells, by the greater predisposition of precursor cells toward an osteoblastic phenotype (ie, lineage switching between adipogenic and osteogenic fates), or by upregulation of matrix proteins necessary for mineralization to occur. Our data support the latter two possibilities because ossification in vivo tends to occur in the context of adjacent adipocytes, and later markers of osteogenesis are upregulated during differentiation of Gnas^+/− derived adipose stromal cells in vitro.

Taken together, these data support the concept that Gnas is a key factor in the regulation of cell fate decisions, particularly those which are involved in bone formation. The role of GNAS inactivation in POH and similar genetic conditions may be related to downstream cAMP signaling and regulation of lineage switching between osteoblast and adipocyte fates. Maternal versus paternal inheritance of GNAS mutations affects the phenotypic consequences in patients, and this study confirms the hypothesis that altered GNAS/Gnas expression affects the regulation of osteoblast differentiation. We propose that in POH and related GNAS‐inactivation disorders, HO results from decreased GNAS/Gnas expression to promote osteogenesis and suppress adipogenesis in multipotent connective tissue progenitor cells.

Disclosures

All the authors state that they have no conflicts of interest.

Acknowledgements

We wish to acknowledge Deyu Zhang and Kevin P Egan for excellent technical assistance, Salin Chakkalakal and Andrew Maidment for whole‐body radiographic imaging, and Michael Levine and Emily Germain‐Lee for the generous gift of Gnas^+/− mice. This work was supported by the National Institutes of Health (R01‐AR046831 to EMS, RJP), the John A Hartford Foundation (RJP), the Progressive Osseous Heteroplasia Association (POHA), the Italian POH Association, the International Fibrodysplasia Ossificans Progressiva Association (IFOPA), the University of Pennsylvania Center for Research in FOP and Related Disorders, the Penn Center for Musculoskeletal Disorders, and the Isaac and Rose Nassau Professorship of Orthopaedic Molecular Medicine (FSK).

Authors' roles: Conception and design (RJP, EMS), acquisition of data (RJP, MX, ER, AR, JK, and PCB), analysis and interpretation of data (RJP, EMS, and FSK), manuscript drafting/revising (RJP, JK, EMS, and FSK), and approval of the final version of the submitted manuscript (all authors).