L-Sox5 and Sox6 Proteins Enhance Chondrogenic miR-140 MicroRNA Expression by Strengthening Dimeric Sox9 Activity*

Background: miR-140 is a critical regulator of cartilage development and homeostasis. Results: The proximal upstream region of miR-140 has in vivo chondrogenic promoter activity and an L-Sox5/Sox6/Sox9 (Sox trio) response element. Conclusion: We reveal that L-Sox5 and Sox6 control miR-140 expression together with Sox9. Significance: Uncovering molecular mechanisms of chondrogenesis has implications for cartilage repair and restoration of tissue function. Sox9 plays a critical role in early chondrocyte initiation and promotion as well as repression of later maturation. Fellow Sox family members L-Sox5 and Sox6 also function as regulators of cartilage development by boosting Sox9 activation of chondrocyte-specific genes such as Col2a1 and Agc1; however, the regulatory mechanism and other target genes are largely unknown. MicroRNAs are a class of short, non-coding RNAs that act as negative regulators of gene expression by promoting target mRNA degradation and/or repressing translation. Analysis of genetically modified mice identified miR-140 as a cartilage-specific microRNA that could be a critical regulator of cartilage development and homeostasis. Recent findings suggest Sox9 promotes miR-140 expression, although the detailed mechanisms are not fully understood. In this study we demonstrate that the proximal upstream region of pri-miR-140 has chondrogenic promoter activity in vivo. We found an L-Sox5/Sox6/Sox9 (Sox trio) response element and detailed binding site in the promoter region. Furthermore, detailed analysis suggests the DNA binding and/or transactivation ability of Sox9 as a homodimer is boosted by L-Sox5 and Sox6. These findings provide new insight into cartilage-specific gene regulation by the Sox trio.

Chondrogenesis is executed in multiple steps during endochondral ossification. Mesenchymal cells condense, become immature chondroblasts, transform into prehypertrophic chondrocytes, and finally differentiate into hypertrophic chondrocytes (1). Extracellular matrix genes such as Col2a1 and Agc1 are expressed in the early stages; in the prehypertrophic and hypertrophic chondrocyte stages, however, their expression decreases, whereas Col10a1, Ihh, and osteopontin expression increases (2).
Two other Sox family proteins play a critical role in cartilage development: L-Sox5, a long product of Sox5, and Sox6. L-Sox5 or Sox6 single knock-out mice exhibit a mild skeletal phenotype, whereas double knock-out mice show significant cartilage defects similar to Sox9 knock-out mice (4,13). These proteins have a coiled-coil DNA-independent homodimerization domain but no other known functional domain (14). L-Sox5 and Sox6 were found to boost Sox9 activity in regulating Col2a1 expression; however, the regulatory mechanism is largely unknown as well as transcription targets beyond Col2a1 and Agc1 (8,14).
MicroRNAs (miRNAs) 3 are a class of short (20 -23 nucleotides), non-coding RNAs generated from primary transcripts (pri-miRNAs) by the action of Drosha and Dicer; they negatively regulate gene expression by promoting mRNA degradation and/or repressing translation through formation of RNAinduced silencing complexes and sequence-specific interaction with primarily 3Ј untranslated regions on the target mRNA (15)(16)(17). Many miRNAs have tissue-and time-specific expression patterns determined at the pri-miRNA transcription level (18 -20). Among known miRNAs, Tuddenham et al. (21) showed cartilage-specific expression of miR-140 in mouse embryos. We previously found that miR-140 expression was reduced in human osteoarthritis cartilage or in response to IL-1 stimulation, and miR-140-deficient mice exhibited short stature and age-related osteoarthritis symptoms (22,23). These observations suggest that miR-140 plays a critical role in cartilage development and homeostasis. Recent findings indicate that Sox9 promotes miR-140 expression (24,25), although detailed regulatory mechanisms are not fully understood.
We demonstrate in this study that the proximal upstream region of miR-140 has chondrogenic promoter activity in vivo and that cartilage-specific expression of miR-140 is generated from its specific transcript. We also reveal that L-Sox5 and Sox6 control miR-140 expression together with Sox9 through a response element in the promoter. Furthermore, detailed analysis suggests that the DNA binding and/or transactivation ability of Sox9 in its homodimer form is boosted by L-Sox5 and Sox6. The findings provide new insights into cartilage-specific gene regulation by this Sox trio.

EXPERIMENTAL PROCEDURES
Cell Culture, Transfection, and Adenovirus Infection-The human kidney cell line 293T and primary mouse chondrocytes were cultured in DMEM with 10% FBS at 37°C. Primary chondrocytes were prepared from mouse embryo ribs (E16.5) and digested with collagenase. The 293T cell line was transfected using FuGENE HD transfection reagent (Promega). Sox9-expressing recombinant adenovirus was prepared for mouse chondrocytes using the adenovirus expression vector kit (Takara), and infection was performed according to manufacturer's instructions.
Reverse Transcription and Quantitative PCR-Total RNA was extracted with ISOGEN (Nippon Gene) according to the manufacturer's protocol and reverse-transcribed with Super-Script II (Invitrogen) and oligo(dT). Quantitative gene expression analysis was performed via real-time PCR using TaqMan Universal Master Mix reagents and TaqMan Probes (Applied Biosystems) on an ABI PRISM 7900HT thermal cycler (Applied Biosystems). Col2a1 and Actb were measured using the mouse TaqMan probes Mm00491889_m1 and Mm00607939_m1, respectively (Applied Biosystems). Data were normalized to Actb gene expression for each experiment. Quantitative miRNA expression analysis was performed using the TaqMan MicroRNA reverse transcription kit and TaqMan MicroRNA assay (Applied Biosystems). miR-140 expression was measured using the TaqMan probe TM001187, and snoRNA202 (TM001232) expression was used as an internal control to normalize differences in each sample.
Reporter Assay-The pGL4.12 vector (Promega), including indicated cloned genome regions, and the indicated gene expression vector were transfected into 293T cells. The Renilla luciferase reporter pRL-TK (Promega) was co-transfected as a control to evaluate transfection efficiency. Cells were lysed, and luciferase activity was measured with the Dual-Glo TM Luciferase Assay System (Promega). Data were normalized to Renilla luciferase activity for each experiment. Mutations were introduced with QuikChange site-directed mutagenesis (Stratagene) according to the manufacturer's instructions.
Electrophoretic Mobility Shift Assay (EMSA)-DNA-protein binding was assayed with DNA probes 32 P-radiolabeled by endfilling with Klenow fragment and luciferase or Sox9 with or without anti-Sox9 antibodies. Reactions were carried out at 25°C for 30 min in binding buffer (20 mM HEPES (pH 7.9),10% glycerol, 50 mM KCl, 0.05% Nonidet P-40, 0.5 mM EDTA, 0.5 mM DTT, and 1 mM PMSF) and 0.5 g of poly(dG-dI), a nonspecific competitor. Anti-Sox9 antibody was preincubated with Sox9 for 30 min at 25°C before the addition of the radiolabeled probes. Binding reactions were separated by PAGE on a 4% gel for 3 h at 100 V. Proteins were synthesized in vitro with the TNT T7 Quick Coupled Transcription/Translation System (Promega) and each expression plasmid.

Chondrogenic Expression of miR-140 Is Regulated by Sox9-
miR-140 is located in intron 16 of the WW domain containing E3 ubiquitin protein ligase 2 (Wwp2 gene) and consists of an N-terminal C2 domain, a C-terminal HECT domain, and four WW domains in the center (Fig. 1A). To determine the transcriptional start site (TSS) and 3Ј end of pri-miR-140, we performed 5Ј-and 3Ј-RACE using mRNA from cultured mouse chondrocytes. We detected three different transcripts with the miR-140 sequence (Fig. 1B). The shortest was 2.1 kb long, starting inside intron 16. The two longer transcripts were splice variants 4.2 and 4.3 kb long, with the same TSS inside intron 10. All three isoforms had the same 3Ј end. miR-140 expression has been previously detected in the primordia of future bones and across the autopod, zeugopod, and stylopod of E11.5 mouse embryo forelimbs and hind limbs with a mature miR-140 LNA oligonucleotide probe (21). Still, precise temporal and spatial miR-140 expression patterns have not been characterized during chondrogenesis. To explore miR-140 expression further, we performed in situ hybridization using an RNA probe for pri-miR-140 common to all three isoforms (Fig. 1B). The expression pattern in the cartilage of E11.5 embryos was similar to the chondrocyte differentiation markers Sox9 and Col2a1, a target gene of Sox9 (Fig. 2, A-F). Sox9 and miR-140 expressions were also similarly detected in the cartilage of digits from E11.5 to E14.5 (Fig. 2, G and H).
Our in situ hybridization analysis found that miR-140 expression resembled Sox9 and one of its target genes, Col2a1 expression in limb buds, which suggests the possibility that Sox9 regulates miR-140 expression in limb development. To test this hypothesis, we examined the effect of repressing Sox9 expression on miR-140 expression. Sox9flox/flox;Ck19-Cre embryos abolish Sox9 expression before mesenchymal condensation, and chondrogenic cell lineage commitment is profoundly impaired as a consequence (28). In E12.5 wild type and Sox9flox mice without Cre expression, the miR-140 expression pattern overlapped with the Sox9 pattern (Fig. 3A, left). Expression of pri-miR-140 in Sox9 mutant cells at E12.5 was completely absent, however, as was Sox9 expression (Fig. 3A, right). These results suggest that Sox9 is a regulator of miR-140, but there is a possibility that the absence of miR-140 expression in the Sox9flox/flox;Ck19-Cre embryo limb buds was the result of absent chondrogenic lineage cells from impaired chondrogenic mesenchymal condensation. We thus examined Sox9flox/flox; Col2a1-Cre embryos, where Sox9 was inactivated in con-densed mesenchymal cells and differentiated chondrocytes through Col2a1-Cre-mediated recombination (4). In E16.5 wild type embryos, pri-miR-140 was expressed in the proliferating chondrocyte zone but absent in the hypertrophic zone (Fig. 3B, top). In contrast, miR-140 was completely absent in condensed chondrogenic mesenchymal cells and differentiated chondrocytes from Sox9flox/flox; Col2a1-Cre embryos (Fig. 3B, bottom). This shows that Sox9 dominates regulation of miR-140 expression in differentiating chondrocytes in vivo.  We also studied the effect of Sox9 overexpression on miR-140 expression through quantitative PCR analysis for cultured chondrocytes. Adenovirus infection-mediated Sox9 overexpression increased miR-140 expression as well as Col2a1 expression and correlated with increased multiplicity of infection; adenovirus infection-mediated LacZ overexpression did not affect miR-140 expression (Fig. 4, A and B). Taken together, the results from all these experiments indicate that miR-140 expression is up-regulated by Sox9 in chondrocytes and developing limbs.
Upstream Region of miR-140 Is Critical for Its Chondrogenic Expression and Sox9 Regulation-To determine a direct regulatory mechanism for miR-140 expression by Sox9, we constructed a reporter plasmid with a 3-kb region upstream from the miR-140 TSS and a luciferase gene (miR140 -3k). With the reporter plasmid and increasing amounts of Sox9-expressing plasmids, luciferase activity in 293T cells increased in a dose-dependent manner (Fig. 5). Sox9 thus regulates miR-140 expression by activating a promoter region 3 kb upstream of its TSS.
Another reporter plasmid with the 3-kb region upstream of miR-140 and a LacZ gene (Fig. 6A) was prepared and used to generate transgenic (Tg) mice (miR140 -3k-LacZ). We generated three lines of transgenic mice, and all of those transgenic E12.5 and E15.5 embryos revealed LacZ expression in forelimbs, hind limbs, and other cartilage tissue such as ribs and vertebrae with an expression pattern similar to miR-140 (Fig. 6,  B-E). These in vitro and in vivo results suggest that miR-140 expression in limb development and cartilaginous tissues is regulated by a 3-kb region upstream of its TSS and promoted by Sox9.

Sox9 Binds to Upstream Region of miR-140 and Enhances Its
Expression in Concert with L-Sox5 and Sox6-Previous studies have clarified that L-Sox5 and Sox6 are critical regulators of . miR-140 proximal promoter activity is up-regulated by Sox9. A, shown is a luciferase reporter construct containing a 3-kb proximal upstream region of pri-miR-140. B, 293T cells were co-transfected with pGL4 as a negative control or the miR140 -3k reporter plasmid and increasing amounts of Sox9-expressing plasmid (0, 50, 100, and 400 ng). Luciferase activity is presented as the mean Ϯ S.D.; n ϭ 3. Statistical differences were calculated using t test. *, p Ͻ 0.05; **, p Ͻ 0.01. chondrogenesis and function as Sox9 co-operators in chondrocytes (4,14). It has also been reported that expression of chondrogenic genes regulated by Sox9, such as Col2a1 and Agc1, is enhanced by L-Sox5 and Sox6 (8,14). Therefore, we explored whether L-Sox5 and Sox6 were involved with Sox9 in regulation of miR-140 expression.
We first examined promoter activity from the miR140 -3k reporter plasmid coupled with overexpression of L-Sox5, Sox6, and/or Sox9 in 293T cells. Luciferase activity was greater with overexpression of the Sox trio than with Sox9 alone, and L-Sox5 and/or Sox6 did not enhance activity without Sox9 overexpression (Fig. 7A). L-Sox5 and Sox6 can thus promote miR-140 expression through activating Sox9.
A series of deletion constructs in the miR-140 upstream region was then created to identify the Sox trio response element (Fig. 7B). Luciferase activity up-regulated by the Sox trio decreased with a deletion from Ϫ936 to Ϫ827 bp upstream of the miR140 TSS and was almost abolished with a deletion from Ϫ739 to Ϫ652 bp (Fig. 7C). Remarkably, an additional increase in activity by L-Sox5 and Sox6 co-transfected with Sox9 was not shown with deletion of the Ϫ739 to Ϫ652 bp upstream region (Fig. 7C).
Our luciferase assay analysis indicates that the Ϫ739 bp upstream region from miR-140 TSS is minimally required for its regulation by L-Sox5, Sox6, and Sox9 (Sox trio). We then constructed a LacZ reporter plasmid containing Ϫ739 bp upstream region from miR-140 TSS (Fig. 8A) and generated Tg mice (miR140 -0.7k-LacZ) to examine whether the minimal region is essential for its chondrogenic expression. We observed that the four other miR140 -0.7k-LacZ mice expressed LacZ in cartilaginous tissues that expression was similar to miR140 -3k-LacZ mice (Fig. 8, B and C). These results indicate that the Ϫ739-bp proximal upstream region of miR-140, which is minimally required for its regulation by Sox trio, is essential for its chondrogenic expression.
The Ϫ739 to Ϫ652-bp upstream region was then explored in detail because up-regulation by the Sox trio was abolished with its deletion. We looked for consensus Sox9 binding sequences because Sox trio up-regulation was still Sox9-dependent and found a putative imperfect palindromic sequence in the binding site (Fig. 9A). Previous reports indicated that Sox9 homodimers bind to enhancer regions that contain inverted Sox9 binding sites separated by 3-4 bp and represent a palindromic motif for chondrocyte-specific gene regulation (30 -32). To assess  whether Sox9 homodimer binding was critical for miR-140 regulation, a miR140 -0.7k reporter construct and Sox9 A76E expression plasmid that does not form homodimers were used (30 -32). There was no obvious up-regulation of luciferase activity by Sox9 A76E alone or with co-expression in the Sox trio, in contrast to the Sox9 wild type plasmid (Fig. 9B). To confirm that Sox9 binds to the putative palindromic sequence, we performed EMSA using oligonucleotide probes for the putative Sox binding (PSB) sequences (Fig. 9C). miR140-PSB probes detected a band when incubated with recombinant Sox9 protein, and additional incubation with anti-Sox9 antibody reduced the band with an upward supershift (Fig. 9D). Utilizing probes with point mutations in one or both Sox9 binding sequences (Fig. 9C) resulted in band reduction, where miR140-PSB mut1 almost abolished the band, and mutations in both motif sites completely abolished the band (Fig. 9D). The same results occurred with the supershift band produced by incubation with anti-Sox9 antibody (Fig. 9D). To further confirm Sox9 binding and promoter activity in the Ϫ739 ϳ Ϫ652-bp region upstream of the miR-140 TSS, another luciferase reporter assay was performed with a vector containing point mutations in one or both Sox9 motif sites. Dual mutations and a single Sox9 motif mutation diminished up-regulation of promoter activity by Sox trio expression (Fig. 9E).
Finally, the ability of Sox9 to bind to the endogenous miR-140 promoter was examined in E13.5 mouse limbs by a ChIP FIGURE 9. The L-Sox5 and Sox6 cooperative effect for miR-140 promoter activation is repressed by blockage of Sox9 homodimer formation. A, shown is a sequence of the region Ϫ739 and Ϫ652 bp upstream from the pri-miR-140 transcription start site. A putative Sox9 binding site is indicated in red; arrows indicate site orientation. B, 293T cells were co-transfected with the miR140 -0.7k reporter plasmid displayed in Fig. 7B and the indicated expressing plasmids. Wild type Sox9 (left) or Sox9 with a dimerization missense mutation (A76E) (right) were used as expression plasmids. Luciferase activity is presented as the mean Ϯ S.D.; n ϭ 3. C, shown are wild type and mutant oligonucleotide probe sequences corresponding to Sox9 binding sites in the miR-140 promoter. D, binding of oligonucleotide probes with wild type and mutated PSB consensus sequences complexed to recombinant Sox9 protein (rSox9) was detected by electrophoretic mobility shift assay. Mutations in one (mut1, mut2) or both (mut3) Sox9 binding motifs were compared with the wild type (WT) in the presence or absence of anti-Sox9 antibody. E, 293T cells were co-transfected with reporter plasmids containing the Ϫ739and Ϫ652-bp pri-miR-140 wild type upstream region (miR140-PSB) or the mutations displayed in C and the indicated expressing plasmids. Luciferase activity is presented as the mean Ϯ S.D.; n ϭ 3. assay using far upstream (approximately Ϫ5 kb) (miR140 Far Ups) and putative Sox9 binding site (miR140 PSB) primer sets as shown in Fig. 10A. The enrichment value of the putative Sox9 binding site was confirmed in Sox9 chromatin-immunoprecipitated DNA, although the far upstream site did not show enrichment (Fig. 10B). Additionally, the use of antibodies showed that L-Sox5 and Sox6 also bound to the putative Sox9 binding site but not the far upstream site (Fig. 10, C and D). Taken together, these results suggest that Sox9 enhances miR-140 expression by cooperating with L-Sox5 and Sox6 to bind Ϫ739 ϳ Ϫ652 bp upstream from the miR-140 TSS.

DISCUSSION
Most miRNAs, which frequently are located within the intron region of genes, originate from the spliced intronic RNA of a host gene transcript (33). The cartilage-specific intronic miRNA miR-140 is located in intron 16 of Wwp2, a host gene that functions as a critical regulator of craniofacial development and is also expressed in cartilaginous tissues (34,35). Wwp2 expression is directly regulated by the binding of Sox9 to intron 3 (35), which suggests that miR-140 expression can be similarly generated and regulated from the Wwp2 transcript. In this study we demonstrated that the proximal upstream region of pri-miR-140 located in intron 10 of Wwp2 has in vivo promoter activity. The results suggest that miR-140 may be derived from its own specific transcript during chondrogenesis.
Previous studies found miR-140 to be specifically expressed in chondrocytes and a critical regulator of cartilage development and homeostasis (21)(22)(23). Our in situ hybridization analysis confirmed that pri-miR-140 expression was chondrocytespecific in mouse embryos and mirrored Sox9 expression patterns as shown in limb buds. Also, miR-140 expression was abolished in Sox9-deficient limb buds and chondrocytes and enhanced in cultured chondrocytes with overexpressed Sox9. These findings show that Sox9 is a direct regulator of miR-140 as well as Wwp2. A recent study indicated that Sox9 binds to intron 10 of Wwp2 and up-regulates miR-140 expression (25). By further exploring this mechanism, we found that Sox9 acti-vates a promoter region upstream of miR-140 on intron 10 with in vivo chondrogenic activity. This strongly suggests that miR-140 in vivo chondrogenic expression is directly regulated in part through activation of its promoter region by Sox9 instead of splicing itself out of the host transcript.
Most previously known Sox9 targets were extracellular matrix genes, and only a few, such as Bapx1, S100A1, and S100B, were shown to be direct regulators of it except for extracellular matrix genes (29,36). Our in vivo and in vitro analyses demonstrated that miR-140 is also a downstream target of Sox9. Previous studies reported that miR-140 repressed Hdac4 to block chondrocyte maturation, Adamts-5 to promote cartilage development and homeostasis, and Sp1 to maintain chondrocyte proliferation (21,23,25). These diverse roles are similar to Sox9, which also promotes early chondrogenesis but represses chondrocyte maturation (2). Furthermore, Sox9 expression was significantly lower in osteoarthritis cartilage compared with normal cartilage (37), whereas miR-140 knockout mice exhibited age-related, osteoarthritis-like phenotypes (23). These studies strongly support miR-140 as both a downstream and a critical target of Sox9 due to its diverse roles in chondrogenesis. In fact, miR-140-deficient mice have a mild skeletal phenotype with a short stature (23).
In this study we showed that L-Sox5 and Sox6 were also involved in regulation of miR-140 expression with Sox9. Results showed that expression levels did not change in the presence of L-Sox5 and/or Sox6 without Sox9 co-expression, making control of miR-140 expression by L-Sox5 and Sox6 completely Sox9-dependent. L-Sox5 and Sox6 participate in the regulation of cartilage-specific genes such as Col2a1 and Agc1, although they have no known transactivation or transrepression domains (8,14). A recent study suggested that L-Sox5 and Sox6 may facilitate Sox9 DNA binding by an unknown mechanism (8). A comparable regulatory event by the Sox trio may also occur with the miR-140 promoter.
Our detailed luciferase, EMSA, and ChIP analyses identified the precise Sox trio response element and Sox9 binding site. Sox9 was previously reported as bound to intron 10 of Wwp2 to regulate miR-140 expression (25). However, the Sox9 binding site exhibited in this study is different and suggests the possibility that miR-140 expression is controlled by multiple Sox9 and/or the Sox trio binding in its promoter region, as shown with Col11a2 expression that is directed by Sox9 at least three cartilage-specific enhancer elements (38).
Sox9 is critical to chondrogenesis because it regulates cartilage-specific genes such as Col2a1, Col11a2, and Agc1 (7,10,11). In addition, Sox9 plays a critical role in testicular development by regulating targets such as Amh and Ptgds in Sertoli cells (39 -41). However, the regulatory mechanism of tissuespecific target distribution is less understood. Previous reports indicated a human SOX9 missense mutation (A76E) in an XY patient exhibiting skeletal abnormalities without sex reversal; this mutation perturbed Sox9 dimerization, resulting in dysregulation of chondrocyte-specific genes (30 -32). Sox9 homodimers are believed to bind to enhancer regions containing inverted Sox9 binding sites separated by 3 or 4 bp (30 -32). The Sox9 binding site identified in this study is also a Sox9 palindromic binding motif and again indicates that miR-140 is a critical chondrocyte-specific target. In fact, LacZ expression was not detected in the male gonads of any miR140 -3k-LacZ Tg embryonic mice (data not shown). We further showed that Sox9 A76E did not up-regulate miR-140 promoter activity even when L-Sox5 and Sox6 were co-expressed. Only one site mutation of the Sox9 palindromic motif completely abolished the ability of the Sox trio to up-regulate promoter activity despite a Sox9 motif mutation that decreased but retained Sox9 binding status. These findings suggest that DNA-dependent Sox9 homodimer formation is needed for L-Sox5 and Sox6 assembly to enhance its DNA binding and/or transactivation ability, but a more detailed mechanism remains undetermined. As previously mentioned, binding of the Sox9 homodimer to its palindromic motif is critical for cartilage-specific gene regulation, and the regulatory mechanism suggested in this study could apply to other cartilage targets. In fact, we observed that Sox9 A76E did not up-regulate luciferase activity from a Col2a1 enhancer reporter vector with L-Sox5 and Sox6 co-transfection (data not shown).
In conclusion, we report that a proximal region upstream of miR-140 has chondrogenic promoter activity in vivo, and its cartilage-specific expression is generated from its transcript within the host gene Wwp2. miR-140 promoter activity is upregulated by the critical transcription factors L-Sox5, Sox6, and Sox9, and their response elements and detailed binding sites were identified. Our findings suggest that the DNA binding and/or transactivation ability of Sox9 in its homodimer form is dependently boosted by L-Sox5 and Sox6.