NPPB and ACAN, two novel SHOX2 transcription targets implicated in skeletal development

Abstract

SHOX and SHOX2 transcription factors are highly homologous, with even identical homeodomains. Genetic alterations in SHOX result in two skeletal dysplasias; Léri-Weill dyschondrosteosis (LWD) and Langer mesomelic dysplasia (LMD), while no human genetic disease has been linked to date with SHOX2. SHOX2 is, though, involved in skeletal development, as shown by different knockout mice models. Due to the high homology between SHOX and SHOX2, and their functional redundancy during heart development, we postulated that SHOX2 might have the same transcriptional targets and cofactors as SHOX in limb development. We selected two SHOX transcription targets regulated by different mechanisms: 1) the natriuretic peptide precursor B gene (NPPB) involved in the endochondral ossification signalling and directly activated by SHOX; and 2) Aggrecan (ACAN), a major component of cartilage extracellular matrix, regulated by the cooperation of SHOX with the SOX trio (SOX5, SOX6 and SOX9) via the protein interaction between SOX5/SOX6 and SHOX. Using the luciferase assay we have demonstrated that SHOX2, like SHOX, regulates NPPB directly whilst activates ACAN via its cooperation with the SOX trio. Subsequently, we have identified and characterized the protein domains implicated in the SHOX2 dimerization and also its protein interaction with SOX5/SOX6 and SHOX using the yeast-two hybrid and co-immunoprecipitation assays. Immunohistochemistry of human fetal growth plates from different time points demonstrated that SHOX2 is coexpressed with SHOX and the members of the SOX trio. Despite these findings, no mutation was identified in SHOX2 in a cohort of 83 LWD patients with no known molecular defect, suggesting that SHOX2 alterations do not cause LWD. In conclusion, our work has identified the first cofactors and two new transcription targets of SHOX2 in limb development, and we hypothesize a time- and tissue-specific functional redundancy between SHOX and SHOX2.

Figure 1. SHOX2 transactivates NPPB and ACAN.

Luciferase reporter activity of U2OS cells transfected with reporter plasmids containing the NPPB promoter (A) or the Acan enhancer (B), renilla luciferase control plasmid and different combinations of SHOX, SHOX2, SHOX2(p.L155V), SHOX2(p.Q234X), SOX6 and SOX9 expression plasmid as indicated. Fold-increase values were obtained by normalizing the relative luciferase units of each sample with the relative luciferase units of the sample transfected only with the reporter plasmid. All values represent the mean and standard deviation of three independent samples, with each sample assayed in triplicate. Significant p-values obtained comparing different independent samples are indicated with an asterisk (*p<0.05 and **p<0.001).

Figure 2. Identification and characterization of the SHOX2 dimerization.

A) SHOX2a and SHOX2b homo- and hetero-dimerize in the yeast-two hybrid system. The S. cerevisiae strain Y187 was cotransformed with pGBT9 (containing the GAL4 binding domain – BD) and pACT2 (containing the GAL4 activation domain – AD) vectors. Interactions were determined using a β-galactosidase liquid assay with CRPG as substrate. Empty vectors were employed as negative controls. B) Characterization of the SHOX2 domains involved in the SHOX2 dimerization. The SHOX2 protein structure showing the amino acid location of the homeodomain (HD) and the OAR domain is shown with the various analysed SHOX2 fragments, indicating their name and amino acids that they contain. To the right of each fragment the corresponding yeast two-hybrid results are shown. Y187 cells were cotransformed with SHOX2 in the pGBT9 vector and SHOX2 fragments in the pACT2 vector. Protein interaction percentages were obtained by normalizing the β-galactosidase units of the different SHOX2 fragments to that obtained with full-length SHOX2. Empty vectors were employed as negative controls. C) SHOX2 mutants impair the SHOX2 dimerization. SHOX2 protein structure scheme showing the localization of the seven analysed missense mutations. Yeast two-hybrid assay of Y187 cotransformed cells with SHOX2 in the pGBT9 vector and the different SHOX2 mutants in the pACT2 vector. Protein interaction percentages were obtained by normalizing the β-galactosidase units of the various SHOX2 mutants to the wildtype SHOX2. Empty vectors were employed as negative controls.

Figure 3. Identification and characterization of the SHOX2-SOX5 and SHOX2-SOX6 protein interactions.

A) SHOX2 interacts with SOX5 and SOX6 in the yeast-two hybrid system. The S. cerevisiae strain Y187 was cotransformed with the pGBT9 (BD) and pACT2 (AD) vectors. Interactions were determined by using a β-galactosidase liquid assay with CRPG as substrate. Empty vectors were employed as negative controls. B) SHOX2 interacts with SOX5 and SOX6 in human cells. Nuclear extracts of HEK293 cells overexpressing FLAG:SHOX2 and HA:SOX5 or HA:SOX6 were immunoprecipitated using anti-FLAG-agarose. Western blots (WB) of immunoprecipitates (IP) were probed with SHOX2, SOX5 and SOX6 antibodies. Nuclear extracts corresponding to 10% input were included as protein expression controls and nuclear extract immunoprecipitates of cells transfected only with HA:SOX5 or HA:SOX6 were included as negative controls. The western-blot images clearly show that SOX5 and SOX6 immunoprecipitate only in the presence of SHOX2. C) Characterization of the SHOX2 domains involved in the interaction with SOX6. An scheme of the SHOX2 protein structure showing the amino acid location of the homeodomain (HD) and the OAR domain is shown with the various analysed SHOX2 fragments, indicating their name and the amino acids that they contain. To the right of each fragment, the corresponding yeast two-hybrid results are shown. Y187 cells were cotransformed with SOX6 in the pGBT9 vector and the SHOX2 fragments in the pACT2 vector. Protein interaction percentages were obtained by normalizing the β-galactosidase units of the different SHOX2 fragments to that obtained with full-length SHOX2. Empty vectors were employed as negative controls. D) SHOX2 mutants impair the interaction with SOX6. SHOX2 protein structure scheme showing the localization of the seven missense mutations analyzed. Yeast two-hybrid assay of Y187 cotransformed cells with SOX6 in the pGBT9 vector and the different SHOX2 mutants in the pACT2 vector. Protein interaction percentages were obtained by normalizing the β-galactosidase units of the various SHOX2 mutants to the wildtype SHOX2. Empty vectors were employed as negative controls. E) Characterization of SOX6 domains involved in the interaction with SHOX2 using the yeast two-hybrid assay. SOX6 is schematically drawn showing the amino acid location of two dimerization domains, the first and the second coiled-coils (1st cc and 2nd cc, respectively), and the HMG DNA-binding domain. Depicted below are the SOX6 generated constructs, indicating the name of each fragment and the amino acids that they contain. Y187 cotransformed cells with the different SOX6 fragments in the pGBT9 vector and SHOX2 in the pACT2 vector. The protein interaction percentages were obtained by normalizing the β-galactosidase units of the different SOX6 fragments to that obtained with full-length SOX6. Empty vectors were employed as negative controls.

Figure 4. Identification and characterization of the SHOX2-SHOX interaction.

A) SHOX2 interacts with SHOX in the yeast-two hybrid system. The S. cerevisiae strain Y187 was cotransformed with the pGBT9 (BD) and pACT2 (AD) vectors. Interactions were determined by using a β-galactosidase liquid assay with CRPG as substrate. Empty vectors were employed as negative controls. B) SHOX2 interacts with SHOX in human cells. Nuclear extracts of HEK293 cells overexpressing FLAG:SHOX2 and HA:SHOX were immunoprecipitated using anti-FLAG-agarose. Western blots (WB) of immunoprecipitates (IP) were probed with SHOX2 and SHOX antibodies. Nuclear extracts corresponding to 10% input were included as protein expression controls and nuclear extract immunoprecipitates of cells transfected only with HA:SHOX were included as negative controls. The western-blot images clearly show that SHOX immunoprecipitate only in the presence of SHOX2. C) Characterization of the SHOX2 domains involved in the interaction with SHOX. A scheme of the SHOX2 protein structure showing the amino acid location of the homeodomain (HD) and the OAR domain is drawn together with the various SHOX2 fragments analysed, indicating the name of each fragment the amino acids that they contain. Yeast two-hybrid assay of Y187 cells cotransformed with SHOX in the pGBT9 vector and different SHOX2 fragments in the pACT2 vector. Protein interaction percentages were obtained by normalizing the β-galactosidase units of the different SHOX2 fragments to that obtained with full-length SHOX2. Empty vectors were employed as negative controls. D) Characterization of the SHOX domains involved in the interaction with SHOX2 using the yeast two-hybrid assay. An scheme of the SHOX protein structure showing the amino acid location of the homeodomain (HD) and the OAR domain is shown with the various SHOX fragments analysed, indicating their name and the amino acids that they contain. To the right of each fragment are the corresponding yeast two-hybrid results. Y187 cells were cotransformed with the different SHOX fragments in the pGBT9 vector and with SHOX2 in the pACT2 vector. Protein interaction percentages were obtained by normalizing the β-galactosidase units of the different SHOX fragments to that obtained with full-length SHOX. Empty vectors were employed as negative controls. E) SHOX mutants impair the SHOX2-SHOX interaction. Schematic structure of SHOX showing the homeodomain (HD), the OAR domain and the localization of the eight analyzed missense mutations. Yeast two-hybrid assay of Y187 cotransformed cells with the different SHOX mutants in the pGBT9 vector and SHOX2 in the pACT2 vector. Protein interaction percentages were obtained by normalizing the β-galactosidase units of the various SHOX mutants to the wildtype SHOX. Empty vectors were employed as negative controls.

Figure 5. SHOX2 is coexpressed with SHOX, SOX5, SOX6 and SOX9 in the 18-week human fetal growth plate.

Immunohistochemistry performed on 18-week human fetal tibia growth plates using antibodies against SHOX2, SHOX, SOX5, SOX6, SOX9 and the negative control (PBS). Specific staining can be observed in the reserve (R), proliferative (P) and hypertrophic (H) zones of the growth plate for all analysed proteins. Images performed at 20× magnification.