Stimulatory G protein directly regulates hypertrophic differentiation of growth plate cartilage in vivo

Abstract

Stimulatory heterotrimeric G protein (Gs) transduces signals from various cell-surface receptors to adenylyl cyclases, which generate cAMP. The α subunit of Gs (Gsα) is encoded by GNAS (Gnas in mice), and heterozygous Gsα inactivating mutations lead to Albright hereditary osteodystrophy. The in vivo role of Gsα in skeletogenesis is largely unknown, because of early embryonic lethality of mice with disruption of Gnas exon 2 (Gnas^E2–/E2–) and the absence of easily detectable phenotypes in growth plate chondrocytes of heterozygous mutant mice (Gnas^+/E2–). We generated chimeric mice containing wild-type cells and either Gnas^E2–/E2– or Gnas^+/E2– cells. Gnas^E2–/E2– chondrocytes phenocopied PTH/PTHrP receptor (PPR)^–/– cells by prematurely undergoing hypertrophy. Introduction of a transgene expressing Gsα, one of several gene products that include Gnas exon 2, into Gnas^E2–/E2– cells prevented premature hypertrophy. Gsα mRNA expression detected by real-time RT-PCR analysis was reduced to approximately half that of the wild-type in both paternal and maternal Gnas^+/E2– growth plate chondrocytes, indicating biallelic expression of Gsα in these cells. Hypertrophy of Gnas^+/E2– chondrocytes was modestly but significantly premature in chimeric growth plates of mice containing wild-type and Gnas^+/E2– cells. These data suggest that Gsα is the primary mediator of the actions of PPR in growth plate chondrocytes and that there is haploinsufficiency of Gsα signaling in Gnas^+/E2– chondrocytes.

Parathyroid hormone (PTH)-related protein (PTHrP) is a paracrine factor important for regulation of chondrocyte differentiation during endochondral bone formation. Actions of PTHrP are mediated through the PTH/PTHrP receptor (PPR), which can couple to both stimulatory heterotrimeric G protein (Gs) (which activates adenylyl cyclase) and Gq/11 (which activates phospholipase C) in cultured cells (1, 2). Growth plate chondrocytes in PTHrP^–/– or PPR^–/– mice show accelerated hypertrophy (3, 4). In contrast, mice carrying a mutant PPR selectively deficient in signaling through Gq show a mild delay in hypertrophy of growth plate chondrocytes (5). Thus, opposing actions of Gs and Gq signaling pathways may be required for normal endochondral bone formation.

The α subunit of Gs (Gsα) is encoded by GNAS (Gnas in mice), a complex gene locus leading to multiple imprinted transcripts through the use of different first exons and promoters (6). Heterozygous mutations in GNAS exons 1–13 encoding Gsα (Gnas exons 1–12 in mice, because the intron between exons 9 and 10 is absent in mice) are associated with skeletal defects as part of a constellation of physical features termed Albright hereditary osteodystrophy (AHO), including short stature and brachydactyly (6, 7). In some patients, the same GNAS mutations additionally lead to resistance to some, but not all, hormones that act via Gsα, including PTH and thyroid stimulating hormone, a condition termed pseudohypoparathyroidism (PHP) type Ia (6–8). Whereas maternal inheritance of a Gsα mutation leads to PHP type Ia, paternal transmission of the same mutation results in pseudopseudohypoparathyroidism (PPHP), a disorder characterized by AHO in the absence of hormone resistance. Consistent with this imprinted mode of inheritance of hormone resistance, Gsα expression occurs predominantly from the maternal allele in certain human tissues, including the thyroid gland (9–11), the ovary (9), and the pituitary gland (12). In the renal cortex, there is evidence for both biallelic (13) and maternal expression of Gsα (14). AHO, unlike hormone resistance, develops after both paternal and maternal transmission of Gsα mutations. These mutations lead to ≈50% reduction of Gsα protein level/activity in erythrocytes and skin fibroblasts of patients with PHP type Ia and PPHP. It has therefore been hypothesized that haploinsufficiency of Gs signaling in various tissues may be responsible for the development of the AHO phenotype (6, 15).

Consistent with the maternal inheritance of hormonal resistance in PHP type Ia, mice heterozygous for disruption of maternal Gnas exon 2 (Gnas^matE2–/+), but not those heterozygous for disruption of paternal Gnas exon 2 (Gnas^+/patE2–), have reduced Gsα levels in renal cortex along with PTH resistance (16). Also consistent with the short stature observed in AHO, both Gnas^matE2–/+ and Gnas^+/patE2– mice are shorter than wild-type littermates. However, the growth plates of these animals appear normal, making it difficult to assess whether the short stature is due to deficiency of Gsα in growth plate cartilage or due to systemic effects. Hence, although the mouse model with disrupted Gnas exon 2 provided important insights into the mechanisms underlying hormonal resistance in PHP type Ia, it did not explain the skeletal phenotypes of AHO. Moreover, homozygous disruption of Gnas exon 2 leads to early embryonic lethality during the postimplantation stage of development, and it has therefore remained unclear whether total loss of Gsα in the growth plate results in a phenotype similar to that observed in PPR^–/– or PTHrP^–/– mice or one that is unique, perhaps due to disruption of other undefined regulatory pathways that also use Gsα signaling in the growth plate. Furthermore, XLαs, a large variant of Gsα with a distinct amino terminus encoded by transcripts using a unique first exon, has “Gs-like” signaling properties in vitro (6, 17, 18), and it thus has to be clarified whether this paternally expressed protein may have a distinct role in the growth plate.

To better evaluate in vivo roles of Gsα in endochondral bone formation as well as to address whether haploinsufficiency of Gsα signaling contributes to the AHO skeletal phenotype, we generated chimeric mice containing both wild-type cells and cells with a null mutation in the Gnas exon 2, and compared both cell types side by side in vivo.

Methods

Generation of Embryonic Stem (ES) Cell Lines de Novo. ES cells were generated as described in ref. 19. To generate Gnas^+/patE2– and Gnas^matE2–/+ ES cell lines, male and female Gnas^+/E2– mice in a CD1 background were mated with wild-type C57BL/6 female and male mice, respectively, for blastocyst collection. To generate Gnas^E2–/E2– ES cell lines, Gnas^+/E2– mice in a CD1-C57BL/6 F₁ hybrid background (16) were mated with each other for blastocyst collection. ES cells homozygous for the null mutation in the PPR gene were isolated as described in ref. 20.

Stable Transfection of Gnas^E2–/E2– ES Cells with a Plasmid Encoding Gsα. Rat Gsα cDNA with a hemagglutinin (HA) tag (a generous gift from A. Federman, University of California, San Francisco) was subcloned into XhoI–HindIII sites of the pcDNA3.1/hyg(–) vector (Invitrogen), which uses the cytomegalovirus (CMV) promoter to drive expression. We call the resultant expression vector pCMV-rGsα; note that rat and human Gsα proteins are virtually identical with only a single amino acid difference (residue 139 is asparagine in rat and aspartic acid in human). The expression vector was linearized with MfeI. Gnas^E2–/E2– ES cells were plated onto a hygromycin-resistant feeder layer, and they were transfected with 1 μg of the linearized pCMV-rGsα by Effectene transfection reagent (Qiagen, Hilden, Germany) at subconfluency. Forty-eight hours after transfection, hygromycin was added to the medium at 200 μg/ml. Sensitive cells started dying 2 days after treatment. Seven days after treatment, resistant colonies were picked. These colonies were cultured without leukemia inhibitory factor (LIF) to allow differentiation into fibroblast-like cells before Western blot and assessment of cAMP production. To detect integration of the expression vector, PCR for the rat Gsα gene (sense primer, 5′-ggcaacagtaagaccgagga-3′; antisense primer, 5′-ccttggcatgctcatagaattc-3′) was performed (annealing, 60°C; extension, 72°C; denaturing, 94°C).

Generation of Chimeric Mice. Chimeras were generated by blastocyst injection as described in ref. 21. Gnas^+/patE2–, Gnas^matE2–/+, and Gnas^E2–/E2– ES cells were injected into CD1-C57BL/6 F₁ hybrid blastocysts. To distinguish cells derived from host blastocysts and cells derived from ES cells, a β-galactosidase transgene was introduced into all wild-type hosts by mating C57BL/6 male mice carrying one copy of a β-galactosidase transgene (22) with wild-type CD1 females. The presence of the transgene was confirmed by staining for β-galactosidase activity. To produce chimeras with differing levels of ES cell contribution, the number of ES cells injected into the blastocele cavity was varied between 5 and 15. At least two independently established ES cell lines of each genotype yielded an identical phenotype. Resultant chimeric mice were killed at various ages from embryonic day (E) 10 through the neonatal period. To illustrate changes in chondrocyte differentiation, examples of bones from E17.5 are illustrated here. All of the observations in the results were verified in at least five different chimeric mice. The degree of chimerism was estimated by staining for β-galactosidase activity or by in situ hybridization for Gsα mRNA.

Chondrocyte Isolation, RNA Extraction, and Real-Time PCR. Chondrocytes were isolated from wild-type, Gnas^+/patE2–, or Gnas^matE2–/+ newborn mice as described in ref. 23. Briefly, limbs were dissected and placed in Hanks' balanced salt solution (HBSS) (Invitrogen). Soft tissue was removed, and epiphyses of the long bones were microdissected, followed by 0.25% trypsin-EDTA (Invitrogen) digestion for 30 min. Isolated cartilage was then digested with 195 units/ml collagenase type II (Worthington) in HBSS for 2 h at 37°C. Total RNA was extracted by using the RNeasy Mini RNA isolation kit (Qiagen). Reverse transcription of 1 μg of total RNA was performed by using the SuperScript First-Strand Synthesis System for RT-PCR (Invitrogen). Real-time PCR was carried out on a DNA Engine Opticon 2 system (MJ Research, Waltham, MA). The QuantiTect SYBR Green PCR kit (Qiagen) was used with cycling conditions as recommended by the manufacturer; annealing was performed at 60°C. Forward PCR primers for amplification of Gsα and XLαs transcripts were 5′-gcagaaggacaagcaggtct-3′ and 5′-ctcatcgacaagcaactgga-3′, respectively. Both reactions used the same reverse primer, 5′-ccctctccgttaaacccatt-3′. Primers for amplification of β-actin transcript were 5′-gatctggcaccacaccttct-3′ (forward) and 5′-ggggtgttgaaggtctcaaa-3′ (reverse). Normalized gene expression relative to β-actin was calculated with q-gene software (24). Amplification efficiencies by using plasmid templates were equal for all transcripts. Genotypes of mice were confirmed by PCR with tail DNA and the following primers for amplification of the neomycin-resistance gene inserted in Gnas exon 2 as part of the knockout strategy (16): 5′-aaggtgagatgacaggagatc-3′ (forward) and 5′-gatcggccattgaacaagatg-3′ (reverse).

Western Blot Analysis. Western blot analysis was performed on cell extracts from ES cells. Whole cell lysates (10 μg) were separated by 10% SDS/PAGE and transferred onto poly(vinylidene difluoride) filters. The filters were incubated with anti-HA tag antibody (1:500, Wako Biochemicals, Osaka), anti-actin (1:1,000, Santa Cruz Biotechnology), and anti-Gsα antibody (1:1,000, Santa Cruz Biotechnology). Antigen–antibody complexes were detected with horseradish peroxidase-conjugated secondary antibodies and visualized by using ECL Plus (Amersham Biosciences).

cAMP Assay. Accumulation of cAMP in hygromycin-resistant ES cells was measured in the presence of 2 mM isobutyl methyl xanthine (IBMX, Sigma). Control cells in 24-well plates were treated at room temperature for 1 h before termination of the reaction by adding 50 mM HCl. For PTH-stimulated cAMP accumulation, a 1-h period incubation with 10^–8 M synthetic PTH analog, [Y34]hPTH(1–34)amide (synthesized at the MGH Biopolymer Core Facility, Boston), was performed at room temperature. The PTH-containing medium was then removed, and the stimulation was terminated by adding 50 mM HCl. The amount of cAMP in each well was determined by using RIA as described in ref. 25.

Southern Blot Analysis. Southern blot analysis was performed as described in ref. 26. To detect disrupted alleles of Gnas exon 2, 10 μg of genomic DNA was digested with XmnI, Southern blotted, and probed with a NotI–SmaI fragment encoding the 5′ promoter region of the Gnas gene (16). The sizes of genomic DNA fragments expected from the normal and disrupted alleles are 11 and 9 kb, respectively (12).

Histological Analysis. Chimeric mice were killed at various ages, dissected, and fixed in 4% paraformaldehyde/PBS at 4°C for 4 h. For detection of β-galactosidase activity, tissues were stained with X-Gal (5-bromo-4-chloro-3-indolyl β-d-galactoside) as described in ref. 27. Subsequently, they were processed, embedded in paraffin, and cut. Sections were stained with hematoxylin/eosin (H&E) or nuclear fast red for morphological study.

To measure and compare the distances of hypertrophic chondrocytes from the articular surface, growth plates were obtained from five different chimeras for each genotype, stained for β-galactosidase activity plus H&E, and sectioned in the median plane. The distance of the earliest hypertrophic chondrocytes (detected based on morphology) from the articular surface was measured under the microscope. The earliest hypertrophic chondrocytes were defined as those located closest to the articular surface. The measurements were performed by three of the investigators blinded with regard to the genotypes. The significance of difference was assessed by using Student's t test.

In Situ Hybridization. Tissues were fixed in 4% paraformaldehyde/PBS overnight at 4°C, processed, embedded in paraffin, and sectioned. In situ hybridization was performed as described in ref. 28 by using complementary ³⁵S-labeled riboprobes for rat Gsα (full-length cDNA), rat XLαs (nucleotides 380–1154, Gen-Bank accession no. X84047), mouse type X collagen, mouse Patched1 (Ptc1), mouse Ihh, mouse osteopontin, rat PPR, and mouse PTHrP (29).

Image Acquisition. An Axioskop 2 Plus (Zeiss) was used for microscopic observation (bright fields and dark fields at 100-fold magnification). Pictures were taken with an Axiocam HRc (Zeiss) camera, and images were acquired with axiovision 3.0 software (Zeiss).

Results

Isolation of Gnas-Mutant ES Cells and Generation of Chimeras. A schematic representation of the Gnas locus is shown in Fig. 1A. Gnas^E2–/E2– ES cells were generated de novo from inner cell masses of blastocysts derived by mating Gnas^+/E2– mice with each other. Two of six isolated ES cell lines were Gnas^E2–/E2–. To obtain Gnas^+/patE2– ES cells, Gnas^+/E2– males were mated with wild-type females. Two of four isolated ES cell lines were Gnas^+/patE2–. To obtain Gnas^matE2–/+ ES cells, Gnas^+/E2– females were mated with wild-type males. Two of five isolated ES cell lines were Gnas^matE2–/+. Genotypes were confirmed by Southern blot (Fig. 1B). To visually distinguish between cells derived from host blastocysts and cells derived from ES cells, one copy of a β-galactosidase transgene, engineered to be widely expressed in mouse tissues (22), was introduced into host blastocysts by appropriate mating. Generated ES cells were then injected into host blastocysts tagged with β-galactosidase. Thus, chimeric mice contain β-galactosidase-tagged wild-type cells (derived from blastocysts) and nontagged mutant cells (derived from ES cells).

Fig. 1.

Isolation of Gnas-mutant ES cells and generation of chimeras. (A) Schematic diagram of the Gnas locus. S, translation start site; X, termination codon. 1, 2, and 3–12 denote the number of the exons. Exons of the antisense and the 1A transcripts are not shown. (B Left) Southern blot analysis of isolated ES cells for the mutation in Gnas exon 2. Wt, wild-type; Het, heterozygous mutant; Hom, homozygous mutant. (B Right) The restriction map of the Gnas locus. XmnI denotes the restriction site by XmnI endonuclease.

Embryos containing both wild-type and Gnas^E2–/E2– cells exhibited severe patterning defects resembling the overactivation of hedgehog signaling. If the chimerism was very high (>90%), the embryos were absorbed by E10, as Gnas^E2–/E2– embryos were (data not shown). The embryos with chimerism between 10% and 90% exhibited severe patterning defects, including exencephaly, and did not survive beyond birth. Embryos containing wild-type and Gnas^E2–/+ cells at any ratio did survive beyond birth and grew into adulthood. Unlike Gnas^E2–/+ mice, these chimeric mice did not show significant growth defects, probably because of the predominant contribution of wild-type cells. For these reasons, we focused on the study of embryos. To illustrate hypertrophic differentiation of growth plate chondrocytes, we show representative data from E17.5, because the phenotypes of chondrocyte differentiation were similar at different ages, and embryos at E17.5 contain chondrocytes at various differentiation stages.

Gnas^E2–/E2– Chondrocytes Undergo Ectopic Hypertrophy and Mimic PPR^–/– Chondrocytes. The wild-type fetal growth plate consists of four major layers of chondrocytes with distinct morphology and gene expression profiles: from the end of bone, periarticular proliferating, columnar proliferating, prehypertrophic, and hypertrophic layers (29). The PTHrP signal directly prevents the switch from proliferation to hypertrophy of columnar proliferating chondrocytes (20). In the growth plate of chimeric mice containing wild-type and PPR^–/– cells, mutant chondrocytes ectopically and prematurely adopted the hypertrophic phenotype (detected by their characteristic morphology and type X collagen expression), as they progress from the surrounding layer of wild-type periarticular proliferating chondrocytes to the surrounding layer of wild-type columnar proliferating chondrocytes (Fig. 2A). In the growth plates of chimeric mice containing wild-type cells and Gnas^E2–/E2– cells, mutant chondrocytes similarly adopted the hypertrophic phenotype at an ectopic location (Fig. 2A). Based on the observations of stained and nonstained cells for β-galactosidase activity, all ectopically hypertrophied cells were mutant (Fig. 2B, arrowheads), whereas all wild-type cells maintained their expected morphology. Gnas^E2–/E2– ectopic hypertrophic chondrocytes expressed other markers of hypertrophy such as Indian hedgehog (Ihh) and osteopontin, and induced Ihh's transcriptional target Patched 1 (Ptc1) in surrounding wild-type cells, as did PPR^–/– ectopic hypertrophic chondrocytes. PTHrP mRNA expression was up-regulated in the periarticular layer (Fig. 2C). Thus, Gnas^E2–/E2– cells phenocopied PPR^–/– cells with respect to premature hypertrophic differentiation. Unlike PPR^–/– ectopic hypertrophic chondrocytes, however, Gnas^E2–/E2– ectopic hypertrophic chondrocytes expressed PPR mRNA (Fig. 2C).

Fig. 2.

Gnas^E2–/E2– cells phenocopy PPR^–/– cells in cartilage. (A) In situ hybridization for type X collagen mRNA of tibial sections from E17.5 wild-type, PPR^–/–/wild-type chimera, and Gnas^E2–/E2–/wild-type chimera embryos. pp, Periarticular proliferating chondrocytes; cp, columnar proliferating chondrocytes; ph/h, prehypertrophic and hypertrophic chondrocytes. (Scale bars, 100 μm.) (B) Staining for β-galactosidase activity plus H&E staining of sections from E17.5 Gnas^E2–/E2–/wild-type chimeric embryo. Wild-type cells were stained blue because of the presence of the β-galactosidase transgene. Arrowheads indicate ectopic hypertrophic chondrocytes. (Scale bar, 100 μm.) (C) In situ hybridization for Indian hedgehog (Ihh), Patched 1 (Ptc1), osteopontin (OP), PTHrP, and PPR mRNAs of tibial sections from E17.5 wild-type and Gnas^E2–/E2–/wild-type chimera embryos. (Scale bar, 100 μm.)

Gnas^+/E2– Chondrocytes also Undergo Premature Hypertrophy, but to a Lesser Extent Than Gnas^E2–/E2– Chondrocytes. AHO is associated with heterozygous inactivating mutations in any one of the GNAS exons encoding Gsα and a large portion of XLαs (except for exon 3, which can be alternatively spliced out and still produce functional Gsα) (6). Although abnormal chondrocyte development may contribute to the AHO phenotype, there has been no clear in vivo evidence that Gnas^+/E2– chondrocytes are functionally defective. The chimeric strategy provided us with an opportunity to identify even subtle differences between wild-type and Gnas^+/E2– chondrocytes by comparing them side by side in vivo. Wild-type cells derived from ES cells differentiated in a way that was indistinguishable from that of wild-type cells derived from host blastocysts (Fig. 3A). In the growth plates of chimeric mice containing wild-type and Gnas^+/E2– cells, both Gnas^+/patE2– chondrocytes and, to a lesser extent, Gnas^matE2–/+ chondrocytes adopted a hypertrophic morphology at a location closer to the ends of bone than wild-type cells (Fig. 3A). The distances of hypertrophic chondrocytes from the articular surface in the ulnae and radii were 78 ± 3% for Gnas^+/patE2– cells and 91 ± 2% for Gnas^matE2–/+ cells compared with that of wild-type cells derived from host blastocysts within the same growth plates (n = 5, P < 0.05 by Student's t test); the difference in the distances from the articular surface of Gnas^matE2–/+ and Gnas^+/patE2– hypertrophic chondrocytes was also significant (n = 5, P < 0.05 by Student's t test). These findings suggested that Gs signaling in the growth plate is subject to haploinsufficiency.

Fig. 3.

Abnormal hypertrophy of Gnas^+/E2– chondrocytes. (A) H&E staining of the radii from E17.5 wild-type/wild-type (Left), Gnas^matE2–/+/wild-type (Center), and Gnas^+/patE2–/wild-type (Right) chimeric embryos. Cells derived from host blastocysts were stained blue with 5-bromo-4-chloro-3-indolyl β-d-galactoside (X-Gal). The earliest hypertrophic chondrocytes (detected by their characteristic morphology) derived from host blastocysts and ES cells were marked with blue and white arrowheads, respectively. (Scale bar, 100 μm.) (B) In situ hybridization for Gsα mRNA and XLαs mRNA of tibial sections from E17.5 wild-type embryos. Upper panels, bright fields; lower panels, dark fields. pp, Periarticular proliferating chondrocytes; cp, columnar proliferating chondrocytes; ph/h, prehypertrophic and hypertrophic chondrocytes. (Scale bar, 100 μm.) (C) Real-time RT-PCR analysis of Gsα and XLαs mRNAs expressed in limb chondrocytes isolated from wild-type (Wt), Gnas^+/patE2– (P–), or Gnas^matE2–/+ (M–) newborn mice.

Expression of Gsα and XLαs in the Growth Plate Cartilage. Based on the data presented above, it appears that Gsα plays a major physiologic role in regulation of hypertrophic differentiation of growth plate chondrocytes. However, other gene products from the Gnas locus may be responsible for the phenotypes in Gnas^E2–/E2– chondrocytes, because exon 2 is common to several different transcripts. Particularly, XLαs, encoded by transcripts that use a unique first exon and exons 2–12 of Gsα, is identical to the latter over a long stretch of C-terminal amino acids (6). Moreover, XLαs has been shown to have “Gs-like” signaling properties in vitro (17, 18). To determine the spatial distribution and relative expression of XLαs and Gsα mRNA in the growth plate, in situ hybridization and real-time RT-PCR analysis were performed. In wild-type mice, Gsα mRNA was expressed in the entire growth plate, with its highest expression in prehypertrophic chondrocytes; in contrast, XLαs mRNA appeared to be expressed at lower levels than Gsα mRNA, and its highest expression was in early hypertrophic chondrocytes (Fig. 3B). Mean normalized expression levels of Gsα and XLαs mRNA relative to β-actin mRNA in wild-type chondrocytes were 8.76 ± 1.37% (n = 4) and 0.37 ± 0.02% (n = 4), respectively, thus indicating a >20-fold difference between the expression levels of XLαs and Gsα in this tissue. Previously, monoallelic expression of Gsα has been shown in several different tissues. Based on the phenotype observed in chimeric growth plates (Fig. 3A), Gsα does not appear to be imprinted in growth plate chondrocytes. To confirm biallelic expression of Gsα in vivo, limb chondrocytes were isolated from wild-type, Gnas^+/patE2–, or Gnas^matE2–/+ newborn mice for real-time RT-PCR analysis. Gsα mRNA expression was reduced to approximately half that of the wild type in both Gnas^+/patE2– (46 ± 11%, n = 5) and Gnas^matE2–/+ (57 ± 10%, n = 5) chondrocytes (Fig. 3C), indicating biallelic expression of Gsα in these cells. In contrast, XLαs expression in Gnas^matE2–/+ chondrocytes was comparable to that of the wild type (78 ± 10%, n = 5), and XLαs expression in Gnas^+/patE2– chondrocytes was reduced dramatically (10 ± 1%, n = 5) (Fig. 3C), consistent with the paternal expression of XLαs transcripts documented previously in a number of different tissues.

Introduction of Gsα into Gnas^E2–/E2– Chondrocytes Rescues Ectopic Hypertrophy. To determine whether Gsα protein alone can reverse the phenotype seen in Gnas^E2–/E2– chondrocytes, Gnas^E2–/E2– cells were transfected with rat Gsα cDNA placed under the control of the CMV promoter. Stable transfectants were selected in the presence of hygromycin, and the integration of the transgene was confirmed by PCR of genomic DNA isolated from these transfectants (Fig. 4A). We refer to these ES cell lines as Gnas^E2–/E2–; pCMV-rGsα. After expansion of 54 resistant colonies and differentiation into fibroblastic cells, we measured cAMP production in response to PTH. Expression of rat Gsα protein was assessed by Western blot for the HA tag (Fig. 4A).

Fig. 4.

Rescue of Gs signaling in Gnas^E2–/E2– cells by rat Gsα. (A) Integration of pCMV-rGsα transgene was detected by PCR of genomic DNA for rat Gsα (Top), and expression of the transgene was detected by Western blot analysis for HA tag (Middle), in ES cell lines differentiated into fibroblast-like cells. (B) cAMP response of various differentiated ES cell lines to PTH. Absolute basal and stimulated cAMP levels are shown on the left, and fold-stimulation is shown on the right. Clones 21, 6, 12, and 44 were Gnas^E2–/E2– ES cell lines stably transfected with pCMV-rGsα. (C) Comparison of expression levels of Gsα and endogenous mouse Gsα proteins detected by Western blot for Gsα. Arrowhead indicates the longer variant.

Among 54 ES cell clones isolated, we chose four representative Gnas^E2–/E2–; pCMV-rGsα clones based on their responses to PTH treatment: clone 21 showed robust PTH-induced cAMP accumulation similar in extent to that in wild-type ES cells; clones 6 and 12 showed PTH-induced cAMP responses that were less pronounced than in wild-type cells but similar to that in Gnas^+/E2– ES cells; and clone 44 failed to respond to PTH, just as did Gnas^E2–/E2– cells (Fig. 4B). Gsα protein levels roughly corresponded to their respective responses to PTH (Fig. 4C). By injecting these transfected ES cells into wild-type blastocysts, we generated chimeric mice to assess the functional consequences. In contrast to Gnas^E2–/E2– chondrocytes, Gnas^E2–/E2–; CMV-rGsα (clone 21) chondrocytes did not undergo ectopic hypertrophy, whereas clones 6 and 12 partially did. The distances of hypertrophic chondrocytes from the articular surface in the ulnae and radii were 82 ± 3% for clone 6 and 88 ± 2% for clone 12, compared with that of wild-type cells derived from host blastocysts within the same growth plates (n = 5, P < 0.05 by Student's t test). Clone 44 chondrocytes, on the other hand, behaved similarly to the Gnas^E2–/E2– chondrocytes (Fig. 5). These data further suggest that Gsα, rather than the other Gnas gene products, is the major signaling molecule at the Gnas locus controlling chondrocyte hypertrophy.

Fig. 5.

Rescue of Gnas^E2–/E2– chondrocytes by rat Gsα. Shown is staining for β-galactosidase activity plus H&E staining of a section of the radii from an E17.5 Gnas^E2–/E2–; pCMV-rGsα/wild-type chimeric embryo. Cells derived from host blastocysts were stained blue with 5-bromo-4-chloro-3-indolyl β-d-galactoside (X-Gal). The earliest hypertrophic chondrocytes (detected by their characteristic morphology) derived from host blastocysts and ES cells were marked with blue and white arrowheads, respectively. The lower portions of the panels for clones 21 and 44 show in situ hybridization for type X collagen mRNA in serial sections. (Scale bar, 100 μm.)

Discussion

In this study, we investigated the in vivo roles of Gsα on differentiation of growth plate cartilage. Although haploinsufficiency of Gsα signaling in growth plate chondrocytes was suspected as being important for the development of AHO features, there has been no clear in vivo evidence for this hypothesis. In addition, although the allelic pattern of expression of Gsα in several other tissues have been studied, the pattern of expression in chondrocytes had not previously been investigated. Taking advantage of the chimeric technique, we discerned a difference between wild-type and Gnas^+/E2– chondrocytes in intact animals and showed that Gnas^+/E2– chondrocytes differentiate prematurely. Using chondrocytes from either Gnas^+/patE2– or Gnas^matE2–/+ mice, we also demonstrated that Gsα expression in these cells is biallelic, i.e., transcripts are derived from both parental alleles. Previously, a growth plate abnormality could not be shown in Gnas^+/E2– mice in vivo (16). There are two possible explanations for the apparent lack of a growth plate phenotype in those animals. First, the abnormality may be too subtle to be detected by conventional techniques. The chimeric technique successfully enhances the sensitivity of detection by comparing wild-type and mutant cells side by side in vivo. Second, other interacting signaling pathways may compensate for the abnormality in Gsα signaling. For example, PTHrP-PPR-Gsα signaling interacts with hedgehog signaling to form a negative feedback loop (29). The acceleration of hypertrophic differentiation due to impaired Gsα signaling results in a concomitant increase in Hh signaling and PTHrP expression that may partially compensate for the loss of Gsα signaling and may thereby obscure the Gsα haploinsufficiency. With the chimeric technique, these compensating mechanisms are not a major influence, as long as the abnormality is cell-autonomous, because cells of different genotypes are placed in the same milieu.

The difference between Gnas^+/patE2– and Gnas^matE2–/+ chondrocytes is modest but statistically significant. The reason for this difference may be other imprinted transcripts derived from the Gnas locus. For example, Nesp55 is derived only from the maternal allele, whereas XLαs is derived only from the paternal allele. Although the Nesp55 gene uses exons 2–12 of Gsα, these exons are in the 3′ UTR, and the disruption of exon 2 does not affect the expression of Nesp55 (T. Xie and L.S.W., unpublished observations). Therefore, it is unlikely that the disruption of NESP55 contributes to the difference in phenotypes between Gnas^+/patE2– and Gnas^matE2–/+ chondrocytes. On the other hand, XLαs shares exons 2–12 with Gsα, which are translated, and XLαs couples to several G protein-coupled receptors and stimulates adenylyl cyclases in vitro with an efficiency similar to that of Gsα (18). Moreover, XLαs and Gsα mRNAs show only a slightly different pattern of expression in the growth plate, although the level of expression of XLαs is much lower. Thus, the difference between Gnas^+/patE2– and Gnas^matE2–/+ chondrocytes may reflect the modest additional effect of the paternally derived XLαs on regulation of hypertrophy, suggesting that there may be redundancy between Gsα and XLαs in the growth plate.

The PPR is known to couple with Gq/11 in vitro as well as Gs (1, 2). Mice carrying a mutant PPR selectively deficient in signaling through Gq/11 show a modest delay in hypertrophic differentiation of growth plate chondrocytes (5), whereas the Gnas^+/E2– chondrocytes are expected to normally activate Gq/11 in response to PTHrP. Acceleration of hypertrophy as a result of disproportionate signaling through Gq/11 may thus contribute to the chimeric phenotype. Nevertheless, the close resemblance between mutant chondrocytes in the PRR^–/– chimeras and Gnas^E2–/E2– chimeras suggests that Gsα is the dominant signaling system downstream of the PPR in chondrocytes. It is noteworthy that the columns of wild-type proliferating chondrocytes in the growth plates of the Gnas^E2–/E2–/wild-type chimeras are not elongated. This observation contrasts with the elongation of wild-type columnar proliferating layer of the growth plate seen in PPR^–/–/wild-type chimeras. This elongation seen in PPR^–/–/wild-type chimeras is probably due to the effect of an increased level of PTHrP expression on wild-type cells. The difference between the two chimeric growth plates presumably derives from the difference between the two kinds of mutations studied. Gnas^E2–/E2– chondrocytes express normal PPRs, whereas PRR^–/– chondrocytes lack PPRs. Thus, the PPR of the Gnas^E2–/E2– chondrocytes can be expected to signal through Gq. This Gq signal from Gnas^E2–/E2– chondrocytes may lead to suppression of elongation of adjacent wild-type columns. Alternatively, the PPR may sequester PTHrP and constrain PTHrP from diffusing. PRR^–/– chondrocytes cannot bind PTHrP through PPR, whereas Gnas^E2–/E2– chondrocytes still can. Through these mechanisms or perhaps others, the columns of wild-type proliferating chondrocytes in the growth plates of the Gnas^E2–/E2–/wild-type chimeras did not elongate.

Mice carrying a null mutation in Gnas exon1 have been generated (¶ and M. Chen and L.S.W., unpublished results). Exon 1, unlike Gnas exons 2–12, is specific to Gsα. The preliminary analysis of the mutant mice has shown that homozygous mice display embryonic lethality and in that way resemble Gnas^E2–/E2– mice. However, heterozygous mice with exon 1 disruption on the paternal allele have a distinct phenotype from Gnas^+/patE2–. These data are in contrast to the findings in humans with AHO, for whom no correlation between severity of phenotypes and the location of affected exons, including exon 1, has been noticed thus far. These differences may reflect species-specific differences in the physiological and developmental roles of alternative GNAS/Gnas gene products. Generation of chimeric mice containing Gnas^+/E1– and comparison of these mice with Gnas^+/E2– chimeric mice will provide further insights into specific functions of Gsα and the other transcripts.

In summary, our results indicate that Gsα plays an essential role in differentiation of growth plate chondrocytes in vivo and that it is the primary mediator of the actions of PPR in this tissue. In addition, we show that, unlike in several other tissues where Gsα expression occurs predominantly from the maternal allele, expression of Gsα mRNA in chondrocytes occurs from both parental alleles. Furthermore, we provide strong evidence suggesting that haploinsufficiency of Gsα signaling in growth plate chondrocytes contributes to the skeletal phenotypes of AHO.