Differences in fat and muscle mass associated with a functional human polymorphism in a post-transcriptional BMP2 gene regulatory element

Abstract

A classic morphogen, bone morphogenetic protein 2 (BMP2) regulates the differentiation of pluripotent mesenchymal cells. High BMP2 levels promote osteogenesis or chondrogenesis and low levels promote adipogenesis. BMP2 inhibits myogenesis. Thus, BMP2 synthesis is tightly controlled. Several hundred nucleotides within the 3' untranslated regions of BMP2 genes are conserved from mammals to fishes indicating that the region is under stringent selective pressure. Our analyses indicate that this region controls BMP2 synthesis by post-transcriptional mechanisms. A common A to C single nucleotide polymorphism (SNP) in the BMP2 gene (rs15705, +A1123C) disrupts a putative post-transcriptional regulatory motif within the human ultra-conserved sequence. In vitro studies indicate that RNAs bearing the A or C alleles have different protein binding characteristics in extracts from mesenchymal cells. Reporter genes with the C allele of the ultra-conserved sequence were differentially expressed in mesenchymal cells. Finally, we analyzed MRI data from the upper arm of 517 healthy individuals aged 18-41 years. Individuals with the C/C genotype were associated with lower baseline subcutaneous fat volumes (P = 0.0030) and an increased gain in skeletal muscle volume (P = 0.0060) following resistance training in a cohort of young males. The rs15705 SNP explained 2-4% of inter-individual variability in the measured parameters. The rs15705 variant is one of the first genetic markers that may be exploited to facilitate early diagnosis, treatment, and/or prevention of diseases associated with poor fitness. Furthermore, understanding the mechanisms by which regulatory polymorphisms influence BMP2 synthesis will reveal novel pharmaceutical targets for these disabling conditions.

Fig. 1

The ultra-conserved sequence mediates repression of reporter genes driven by the BMP2 promoter or a heterologous viral promoter in C3H10T½ mesenchymal cells. Luciferase activity generated in C3H10T½ cells after transfection with reporter genes with the murine Bmp2 promoter (nt −1,237 to 471) or cytomegalovirus (CMV) promoter only (BmpLUC or CMVLUC) or with the mouse or human ultra-conserved sequence (mouse nt 9,574–10,204 or human nt 11,488–11,877) relative to the mouse or human distal promoters, respectively, inserted downstream of LUC (BmpLUCBmp or CMVLUCBmp). Relative reporter activity is shown ± SEM; for the murine Bmp2 promoter with the murine ultra-conserved sequence, n = 7; for the CMV promoter with the murine or human ultra-conserved sequences, n = 6 each.

Fig. 2

BMP2 RNAs with the A or C alleles differ in affinity for specific RNA binding proteins in extracts from mesenchymal cell types. A: Bars show the relative lengths and positions of BMP2 RNAs used for UV cross-linking. RNA W contains the entire murine ultra-conserved sequence (nt 9,574–9,938 relative to the mouse promoter [Abrams et al., 2004]). The human ultra-conserved sequence (nt 11,636–11,726 relative to the human promoter) differs at only 12 nt located away from rs15705. RNAs X (nt 9,736–9,938) and Y (9574–9782) contain the right or the left approximate halves of the ultra-conserved sequence. RNA Z contains the 34 nt flanking the rs15705 SNP. This region is identical between the mouse (nt 9,713–9,747) and human (nt 11,642–11,676) genes. The RNA Z sequences are AGAUUUAAAAUGUAUUU(A or C)GUUGUACAUUUUAUAUG. B: Capped and ³²P-labeled, in vitro synthesized, RNAs Z with the A or C allele were UV-cross-linked in extracts from MC3T3-E1 osteoblastic cells. The C RNA labeled nucleolin (Nuc) more efficiently; whereas the A RNA labeled HuR more efficiently (n = 2). C: RNAs containing the full-length murine ultra-conserved sequence (RNA W) were UV-cross-linked in extracts from ROS17/2.8 rodent osteosarcoma with excess polyA competitor. One, 5, 10, or 25 pmol of competitor RNA oligo with the A (lanes 2–5) or the C (lanes 6–9) sequence (human nt 11,642–11,676 as in RNA Z) were added to the reactions. The mouse and human sequences are identical in this region. The graph indicates the UV cross-linking to nucleolin in extracts relative to samples incubated without competitor RNA oligo. D: RNAs containing the full-length human ultra-conserved sequence (RNA W) were UV-cross-linked in extracts from HeLa cells with increasing amounts of purified human nucleolin (Invivogen, San Diego, CA). The labeling intensity of one band that comigrated with the purified nucleolin increased with the added protein. All other proteins that were labeled by the BMP2 RNA, including the protein marked with a “★” here, did not change in intensity. E: RNAs W, X, or Y were UV-cross-linked in extracts from retinoid-treated F9 embryonal carcinoma cells with excess polyA competitor. The reactions were subsequently immunoprecipitated with an HuR antibody (Santa Cruz, sc5261). The full-length and right half RNAs, but not the left half, immunoprecipitated efficiently.

Fig. 3

The rs15705 genotype alters the luciferase activity generated from reporter genes with the ultra-conserved 3′UTR. A: Nucleotides 11,649–11,740 relative to the human distal promoter are shown. The aligned sequences show the A or the C allele at the rs15705 SNP in the context of the CMVLUCBmp plasmid (Fig. 1). Unchanged nucleotides are indicated by “.”. The sequence of the miR-569 microRNA is indicated above the sequence with potential hybridization indicated by “|” for Watson-Crick pairing and “:” for non-Watson-Crick pairing. B: The relative reporter activity ± SEM of each plasmid in HeLa (n = 12), MC3T3-E1 osteoblastic (n = 3), and C3H10T½ mesenchymal (n = 6) cells. A Student's t-test was used to compare the luciferase activity of CMVLUCBmp with the A allele at the rs15705 SNP to the plasmid with the C allele. The P-value is shown in the bar for each cell type.