Identification of novel SHOX target genes in the developing limb using a transgenic mouse model

Abstract

Deficiency of the human short stature homeobox-containing gene (SHOX) has been identified in several disorders characterized by reduced height and skeletal anomalies such as Turner syndrome, Léri-Weill dyschondrosteosis and Langer mesomelic dysplasia as well as isolated short stature. SHOX acts as a transcription factor during limb development and is expressed in chondrocytes of the growth plates. Although highly conserved in vertebrates, rodents lack a SHOX orthologue. This offers the unique opportunity to analyze the effects of human SHOX expression in transgenic mice. We have generated a mouse expressing the human SHOXa cDNA under the control of a murine Col2a1 promoter and enhancer (Tg(Col2a1-SHOX)). SHOX and marker gene expression as well as skeletal phenotypes were characterized in two transgenic lines. No significant skeletal anomalies were found in transgenic compared to wildtype mice. Quantitative and in situ hybridization analyses revealed that Tg(Col2a1-SHOX), however, affected extracellular matrix gene expression during early limb development, suggesting a role for SHOX in growth plate assembly and extracellular matrix composition during long bone development. For instance, we could show that the connective tissue growth factor gene Ctgf, a gene involved in chondrogenic and angiogenic differentiation, is transcriptionally regulated by SHOX in transgenic mice. This finding was confirmed in human NHDF and U2OS cells and chicken micromass culture, demonstrating the value of the SHOX-transgenic mouse for the characterization of SHOX-dependent genes and pathways in early limb development.

Figure 1. Generation and expression analysis of SHOX-transgenic mice.

(A): The SHOXa cDNA was tagged with a Lumio and SV40 Poly(A) sequence and cloned under the control of a murine Col2a1 promotor/enhancer expression cassette. (B): Genotyping was performed using specific primers spanning the first 409 nucleotides of the SHOXa cDNA. No PCR product was detected in wildtype animals. (C):-Southern Blot analysis of the two transgenic lines (1 and 2) used for our investigations. Genomic DNA was digested with BamHI, EcoRV and Hind III. BamHI digestion results in a 1.3 kb fragment that corresponds to the Lumio/SV40-tagged SHOX cDNA, which was flanked by BamHI sites. The presence of only one signal per lane indicates a single integration site of the transgene. (D): Relative quantitative expression of Col2a1 and SHOXa transcripts in limbs of wildtype and transgenic littermates (N = 5–8 per litter) at E12.5, E13.5 and E14.5. The expression of the transgene corresponds to the expression dynamics of Col2a1. SHOX levels are generally low with highest expression at E12.5. Values are variable among individual animals as indicated by the standard deviation (SD). (E): WISH of wildtype (Wt) and transgenic (Tg) embryonic limbs from E11.5-E14.5 (N = 20 for each stage). The transgene is weakly expressed in the developing limb at E11.5 and becomes defined around the cartilaginous anlagen at E12.5. From E13.5 onwards, the expression is mainly seen in the mesenchyme around the developing cartilage and in the perichondrium and decreases during later stages.

Figure 2. Analysis of postnatal bone parameters of Col2a1-SHOX-transgenic mice.

(A): Alcian Blue/Alizarin Red S staining at different developmental (E14.5, E18.5) and postnatal (P28) stages does not reveal apparent differences between transgenic and wildtype skeletal elements. (B): Postnatal in vivo time-course analysis of bone growth in 65 animals of two transgenic lines by μ-CT analysis. Tibiae and femora of wildtype and Tg(Col2a1-SHOX) littermates at the age of 4, 12 and 24 weeks were scanned, female and male individuals were evaluated separately. Total bone length, cortical bone thickness and bone volume do not show significant differences between wildtype and transgenic females or males. Some transgenic animals presented longer bones and weaker structures of the cortical bone in the subcartilaginous region (indicated in the μ-CT images). Other micromorphological parameters (bone mineral density (BMD), trabecular volume and thickness) showed no significant differences. Statistical analyses were performed using student's t-test. (C): hematoxilin and eosin (H&E) stainings of the growth plate in wildtype and transgenic tibiae. Consistent differences between wildtype and Tg(Col2a1-SHOX) adult growth plates (24 weeks of age) did not exist (N = 8), but some transgenic tibiae showed a buckling, and the columns of chondrocytes became shorter and were not strictly oriented in a parallel assembly compared to the wildtype (right image).

Figure 3. Regulated genes in transgenic mice and validation of Ctgf as a target.

(A): qRT-PCR using limb RNA (E12.5-E14.5) from wildtype (Wt) and transgenic littermates (Tg) (N = 8–10 for each stage). Measurements were carried out individually, in duplicates, and normalized to Adam9 and Sdha. Relative normalized values are presented on the y-axis. Significances are indicated in each diagram by asterisks (*: p≤0.05, **: p≤0.01, ***: p≤0.001). Variations are indicated by the standard deviation (SD). In 7/8 candidates an upregulation was confirmed as significant in at least one embryonic stage. (B): nCounter analysis of CTGF and SHOX expression in NHDF and U2OS cells after transient transfections of SHOX and p.Y141D. CTGF is significantly downregulated in NHDF cells, whereas it is significantly upregulated in U2OS cells. Values on y-axis represent absolute counts of mRNA, normalized to ADAM9, HPRT1 and SDHA. Significancies are indicated by asterisks. (C): In situ hybridization using a Ctgf antisense riboprobe on embryonic limbs from wildtype and SHOX-transgenic littermates (N = 8) at stage E12.5. In transgenic embryos, enhanced and distalized expression of Ctgf was detected in the middle part of the developing limbs.

Figure 4. Analysis of CTGF as a direct transcriptional target of SHOX.

(A): Genomic structure of the human CTGF region. ChIP-Seq analysis in ChMM cultures revealed an accumulation of Shox binding in the Ctgf promoter region (grey peaks), especially in a region 3–4 kb from the transcriptional start site (TSS) where an evolutionary conserved sequence (ECR) of 597 bp (human chr6:132317086-132318077) was identified (green bar). (B): Location of the pGL3 ECR and pGL3 ECR+ reporter constructs (grey bars) within the CTGF upstream region. The ECR+ construct encompasses the ECR and an upstream region including ATTA/TAAT motifs and palindromes. SHOX binding motifs (ATTA/TAAT sites and palindromes) in the CTGF 5′ region around the ECR are indicated by asterisks. Red bars represent the location of the generated oligonucleotides for EMSA. (C): Luciferase reporter gene assays in NHDF and U2OS cells. pcDNA4/TO SHOX was cotransfected with a luciferase reporter vector harbouring either the ECR or the ECR+ sequence. Transfections and measurements were carried out in triplicates. A significant activation in the luciferase activity was observed 24 h after SHOX transfection in NHDF cells using both reporter constructs (1.7-fold/2.5-fold with p = 0.02/0.007 for ECR/ECR+). In U2OS cells, an alteration was not observed for the ECR reporter, but a significant reduction was demonstrated for the ECR+ reporter construct (1.0-fold/2.8-fold with p = 0.1/0.003 for ECR/ECR+). (D): EMSA. The SHOX wildtype (Wt) and the mutant p.R153L proteins bind to oligonucleotides 1 and 2, whereas the defective proteins p.Y141D and p.A170P cannot. All fragments of oligonucleotides 1 and 2 containing an ATTA/TAAT site are sensitive to SHOX binding (1a–c, 2a–b). The fragment lacking this motif does not bind (oligonucleotide 2c). Using the SHOX-3 antibody (Ab), we demonstrate that the binding is SHOX-specific. (E): Immunohistochemistry performed on pubertal tibial growth plates. Staining was performed using preimmune serum as a negative control, SHOX antibody and a CTGF-specific antibody. Both the SHOX and CTGF proteins were detected in growth plate chondrocytes.