Requirement for RAR-mediated gene repression in skeletal progenitor differentiation

Abstract

Chondrogenesis is a multistep process culminating in the establishment of a precisely patterned template for bone formation. Previously, we identified a loss in retinoid receptor-mediated signaling as being necessary and sufficient for expression of the chondroblast phenotype (Weston et al., 2000. J. Cell Biol. 148:679-690). Here we demonstrate a close association between retinoic acid receptor (RAR) activity and the transcriptional activity of Sox9, a transcription factor required for cartilage formation. Specifically, inhibition of RAR-mediated signaling in primary cultures of mouse limb mesenchyme results in increased Sox9 expression and activity. This induction is attenuated by the histone deacetylase inhibitor, trichostatin A, and by coexpression of a dominant negative nuclear receptor corepressor-1, indicating an unexpected requirement for RAR-mediated repression in skeletal progenitor differentiation. Inhibition of RAR activity results in activation of the p38 mitogen-activated protein kinase (MAPK) and protein kinase A (PKA) pathways, indicating their potential role in the regulation of chondrogenesis by RAR repression. Accordingly, activation of RAR signaling, which attenuates differentiation, can be rescued by activation of p38 MAPK or PKA. In summary, these findings demonstrate a novel role for active RAR-mediated gene repression in chondrogenesis and establish a hierarchical network whereby RAR-mediated signaling functions upstream of the p38 MAPK and PKA signaling pathways to regulate emergence of the chondroblast phenotype.

Figure 1.

Inhibition of RAR activity enhances Sox9 activity and expression. Activation of the retinoid receptors in primary limb mesenchymal cultures with either at-RA or the RARα-specific agonist AGN193836 (836) attenuates activity of the pGL3(4X48) Sox9 reporter (A); however, at-RA appears to be a more effective inhibitor. In contrast, inhibition of RAR activity by the RARα-specific antagonist, AGN194301 (301), or with the RAR pan-antagonist, AGN194310 (310), enhances reporter activity with 310 being more potent than 301. The effects of these compounds on activity of pGL3(4X48) are inversely proportional to their ability to activate a luciferase reporter containing an RA response element (pW1-βRARE₃-Luc) (B). The concentration-dependent effects of each compound on Sox9 reporter activity also correspond with the ability of each compound to enhance or inhibit cartilage formation in vitro as indicated by alcian blue staining (C). In response to AGN194301, there is an increase in Sox9 mRNA as early as day 2, but this increase appears to be transient (D). Bar, 1.5 mm. ANOVAs (A and B), P < .0001; Bonferroni post-tests indicate significant differences in Sox9 reporter activity at concentrations ≥ 5 × 10⁻⁹ M for 310, 1 × 10⁻⁷ M for 301, 5 × 10⁻⁸ M for 836, and 5 × 10⁻⁹ M for at-RA. Significant changes in activity of the RARE reporter are induced by concentrations ≥ 5 × 10⁻⁹ for 310, 1 × 10⁻⁷ M for 301, 5 × 10⁻⁹ M for at-RA, and 5 × 10⁻⁹ M for 836, with all P values at least < 0.05.

Figure 2.

Sox9 transactivation of Col2a1 is inversely associated with RAR activity. To further study the influence of retinoid receptor activity on chondrogenesis, constructs containing constitutively active receptors (RARαVP16 and RXRαVP16) or dominant negative versions of the receptors (dnRARα and dnRXRα) were used. These constructs substantially alter the activity of the pW1-βRARE₃-Luc reporter (A and B). RARαVP16 and RXRαVP16, which enhance activity of the RARE reporter (A), attenuate activity of the pGL3(4X48) reporter (C). In contrast, the dnRARα and dnRXRα, which suppress RARE reporter activity (B) substantially activate pGL3(4X48), albeit the dnRARα has a more dramatic effect (D). ANOVAs (A–D), P < .0001; Bonferroni post-tests (A–D), *P < .001.

Figure 3.

Sox9 binding sites are essential for dnRARα-induced reporter activity. To examine the contribution of Sox9 to the effects of RAR inhibition, reporters with varying sensitivities to Sox9 were used to follow their response to the dnRARα. All reporter constructs were cotransfected with pcDNA3-hSox9 (A) or with dnRARα (B). Of four reporters analyzed, 4X48-p89 and pGL3(4X48) are most sensitive to Sox9 (A) and also exhibit the greatest response to dnRARα (B). In contrast, pGL3(−89+6), containing only the minimal Col2a1 promoter with no Sox9 binding sites, exhibits no activity in response to Sox9 (A) and is unaffected by the dnRARα (B). A reporter (pW1-Col2-Luc) containing two tandem repeats of a larger intron-1 segment of Col2a1 (including Sox9 binding sites) along with a promoter fragment is only mildly sensitive to Sox9 (A) and is activated to a much weaker extent by dnRARα (B) compared with the 4X48-containing reporters. All reporter inductions by hSox9 or dnRARα were normalized to basal levels of respective reporters. ANOVAs (A and B), P < .0001; Bonferroni post-tests, *P < .001.

Figure 4.

Induction of Sox9 activity by dnRARα is observed in other chondrogenic cells. The induction of Sox9 reporter activity by dnRARα in different cells compared with vector-transfected (−) controls is shown. The effect of dnRARα on Sox9 reporter activity is consistent in chondrogenic cells, since considerable increases in pGL3(4X48) are induced not only in limb mesenchymal cells but also in rat articular chondrocytes and in C5.18 cells, which both have chondrogenic capacity. In contrast, no noticeable change in reporter activity is induced in the nonchondrogenic COS P7 cells. Student's t tests, *P < .001.

Figure 5.

Histone deacetylase-mediated gene repression is required for chondrogenesis. The effects of TSA on Sox9 reporter activity in the presence or absence of AGN194310 (A) and on pW1-βRARE₃tkLuc (B) were analyzed. TSA attenuated Sox9 reporter activity in a concentration-dependent manner and inhibited the effects of AGN194301 (A). In contrast, TSA enhanced activity of the pW1-βRARE₃tkLuc reporter in a concentration-dependent manner (B). The inhibition in Sox9 reporter activity correlates with the decrease in the number of cartilage nodules forming in response to TSA as seen in day 4 alcian blue-stained cultures (C). The increases in Sox9 reporter activity induced by cotransfection with dnRARα or by treatment with AGN194301 or AGN194310 are attenuated by coexpression of pCMX-GAL4/N-CoR (D). Bar, 1.5 mm. ANOVAs (A and B), P < 0.0001 for all cases; Bonferroni post-tests (A and B), *P < .01, **P < .001, and ***P < .0001, all versus respective control cultures; Student's t tests (D), **P < .001 and ***P < .0001.

Figure 6.

The p38 MAPK pathway and the PKA pathway are activated in response to RAR inhibition. Reporters containing a cAMP response element (pCRE-TA-Luc) or activator protein-1 response element (pAP-1-TA-Luc) are both activated in response to cotransfection with dnRARα (A and B). Cotransfection with dnRARα also induces transactivation of a GAL4 reporter (pG5-Luc) by the transcription factors ATF2 and CREB, both of which are fused to the DNA binding domain of GAL4 (FA-ATF2 and FA-CREB) (C and D). The ability of PKAc and MKK6E to activate FA-CREB and FA-ATF2, respectively, was tested for positive control purposes. Student's t tests (A and B), *P < .0005; ANOVAs (C and D), P < .0001; Bonferroni post-tests (C and D), ^# P < .0001, **P < .001, and ***P < .01.

Figure 6.

The p38 MAPK pathway and the PKA pathway are activated in response to RAR inhibition. Reporters containing a cAMP response element (pCRE-TA-Luc) or activator protein-1 response element (pAP-1-TA-Luc) are both activated in response to cotransfection with dnRARα (A and B). Cotransfection with dnRARα also induces transactivation of a GAL4 reporter (pG5-Luc) by the transcription factors ATF2 and CREB, both of which are fused to the DNA binding domain of GAL4 (FA-ATF2 and FA-CREB) (C and D). The ability of PKAc and MKK6E to activate FA-CREB and FA-ATF2, respectively, was tested for positive control purposes. Student's t tests (A and B), *P < .0005; ANOVAs (C and D), P < .0001; Bonferroni post-tests (C and D), ^# P < .0001, **P < .001, and ***P < .01.

Figure 7.

Inhibition of p38 and PKA prevents chondrogenesis. In the presence of 5 or 10 μM SB202190, there is a decrease in Sox9 reporter activity compared with untreated controls (A). SB202190 also attenuates the chondrogenic response to AGN194301 and the dnRARα (A). This inhibition is reflected by a lack of cartilage formation in vitro, since almost no cartilage nodules form in response to 10 μM SB202190 (D) in contrast to the presence of numerous nodules in untreated control cultures (C). Similar to SB202190, the PKA inhibitor H89 reduces Sox9 reporter activity both in the presence or absence of AGN194301 or dnRARα (B). In H89-treated cultures (10 μM) (F), fewer nodules are detected compared with untreated cultures (E); these nodules are much smaller and stain only weakly with alcian blue. Bar, 1.5 mm. ANOVAs (A and B), P < .0001; Bonferroni post-tests (A and B), *P < .05, **P < .01, ***P < .001, and ^# P < .005, all versus respective non-SB202190 or non-H89–treated controls.

Figure 8.

Activation of ATF2 or CREB induces Sox9 transactivation response. The effects of different components of the p38 signaling cascade on ATF2 induction were analyzed. Transient transfection with MKK6E induces ATF2 activity greater than twofold; however, transient expression of p38α or p38β either alone or in combination has no appreciable effect on FA-ATF2 activity (A). However, when cotransfected with MKK6E, p38α or p38β can enhance FA-ATF2 activity considerably, and when both isoforms are transfected together along with MKK6E there is an even greater induction of FA-ATF2 (A). The effects of MKK6E, p38α, and p38β on Sox9 reporter activity corresponds with their ability to activate FA-ATF2 (B). MKK6E induces an almost 2.5-fold increase in activity of pGL3(4X48), whereas p38α and p38β alone have no noticeable effect. When cotransfected with MKK6E, p38α and p38β each enhance Sox9 reporter only slightly, but when transfected together along with MKK6E the increase in Sox9 transactivation is greater than fourfold (B). PKAc induces FA-CREB activity by >30-fold (C) but only enhances Sox9 reporter activity slightly (by <2-fold) (D). ANOVAs (A and B), P < .0001; Bonferroni post-tests, *P < .001 (A) and *P < .001 (B); Student's t test, **P < .0005 (C), *P < .001 (D).

Figure 9.

Regulation of Sox9 expression and transcriptional activity by PKA. Sox9 expression and activity were measured in response to manipulation of the PKA signaling pathway. In the absence of exogenous Sox9, addition of pCPT-cAMP (500 μM) or cotransfection with dnRARα increases Sox9 reporter activity, whereas H89 (10 μM) represses reporter activity in primary limb mesenchymal cells (A). In the presence of cotransfected Sox9, reporter activity is elevated >100-fold, and the addition of pCPT-cAMP or cotransfection with dnRARα only has a small stimulatory effect (<1.5 fold), whereas the addition of H89 slightly decreases reporter activity. The response of the mutant Sox9-181A is similar to wild-type Sox9 (A). To demonstrate that the mutation functioned as expected, COS P7 cells were transfected with wild-type Sox9 or the mutant Sox9 in the presence or absence of PKA (B). The presence of PKA leads to a >1.5-fold induction in reporter activity, whereas the mutant Sox9 exhibits little increase with cotransfected PKA. Modulation of PKA signaling influences Sox9 and Col2a1 expression in primary limb mesenchymal cells (C). Real-time quantitative PCR was used to demonstrate that activation of PKA leads to an increase in Sox9 and Col2al mRNA abundance, whereas treatment with H89 suppresses their expression. ANOVAs (A–C), P < .0001; Bonferroni post-tests (A–C), *P < .05, **P < .01, and ***P < .001, all versus respective nontreated controls.

Figure 10.

Activation of ATF2 or CREB can rescue the effects of RARαVP16. The ability of ATF2 and CREB to reverse the effects of RARαVP16 on Sox9 reporter activity was analyzed. MKK6E can partially prevent the inhibitory response of RARαVP16 whereas p38α and p38β alone or in combination had little if any effect (A). Cotransfection with MKK6E and p38α or p38β was able to partially restore Sox9 activity, whereas cotransfection of MKK6E with both p38α and p38β can completely restore the effects of RARαVP16 (A). Activation of PKA by transient transfection with PKAc was also able to restore Sox9 reporter activity to basal levels in RARαVP16- transfected cultures. ANOVA (A and B), P < .0001; Bonferroni post-tests, *P < .001 (A); *P < .001 (B).