Retinoic acid and GM-CSF coordinately induce retinal dehydrogenase 2 (RALDH2) expression through cooperation between the RAR/RXR complex and Sp1 in dendritic cells

Abstract

Retinoic acid (RA)-producing dendritic cells (DCs) play critical roles in gut immunity. Retinal dehydrogenase 2 (RALDH2) encoded by Aldh1a2 is a key enzyme for generating RA in DCs. Granulocyte-macrophage colony-stimulating factor (GM-CSF) potently induces RALDH2 expression in DCs in an RA-dependent manner, and RA alone weakly induces the expression. However, how GM-CSF and RA induce RALDH2 expression remains unclear. Here, we show that GM-CSF-induced activation of the transcription factor Sp1 and RA-dependent signaling via the RA receptor (RAR)/retinoid X receptor (RXR) complex contribute to Aldh1a2 expression. The RAR antagonist LE540 and the Sp1 inhibitor mithramycin A inhibited GM-CSF-induced Aldh1a2 expression in fms-related tyrosine kinase 3 ligand-generated bone marrow-derived DCs (BM-DCs). ERK and p38 MAPK inhibitors suppressed GM-CSF-induced nuclear translocation of Sp1 and Aldh1a2 expression. Sp1 and the RARα/RXRα complex bound to GC-rich Sp1-binding sites and an RA response element (RARE) half-site, respectively, near the TATA box in the mouse Aldh1a2 promoter. The DNA sequences around these sites were highly conserved among different species. In the presence of RA, ectopic expression of RARα/RXRα and Sp1 synergistically enhanced Aldh1a2 promoter-reporter activity. GM-CSF did not significantly induce Aldh1a2 expression in plasmacytoid DCs, peritoneal macrophages, or T cells, and the Aldh1a2 promoter in these cells was mostly unmethylated. These results suggest that GM-CSF/RA-induced RALDH2 expression in DCs requires cooperative binding of Sp1 and the RAR/RXR complex to the Aldh1a2 promoter, and can be regulated by a DNA methylation-independent mechanism.

Figure 1. Sp1 participates in the Aldh1a2 expression.

(A) The genomic organization of the mouse Aldh1a2 gene and its 5′-flanking region is shown. A fragment containing exon 1 and its 5′-flanking region from −2,600 to +182 was inserted into reporter vectors. DNA binding sites (STAT-binding sites, NF-κB binding sites, a SREBP binding site, and putative RARE half-sites (RARE-h)), the TATA box, and the GC-rich region in the fragment are indicated. (B) Flt3L-generated BM-DCs were cultured with 10 ng/ml GM-CSF for 16 h in the presence or absence of 1 µM mithramycin A. After the culture, Aldh1a2 mRNA expression was assessed by real-time PCR (Left panel), and protein expression of RALDH2 (ALDH1A2) and α-tubulin was analyzed by Western blotting (Right panel). Relative mRNA expression levels were calculated by defining the Aldh1a2 mRNA expression level in the cells incubated with medium alone for 16 h was set to 1 (Left panel). Data are representative of three (Left panel) or two (Right panel) independent experiments. (C) Serial-deletion fragments derived from the 5′-flanking region of the mouse Aldh1a2 gene were inserted in the reporter vector, pGL3-basic. COS-7 cells were transfected in triplicate with one of the deletion constructs (1.25 µg) or the pGL3-RALDH2 (−2,600) reporter vector (1.25 µg) in combination with or without the 0.5 µg of pCMV-Myc-Sp1 expression vector. One day after the transfection, luciferase activity was measured. Relative promoter activities were calculated by arbitrarily defining the activity of pGL3-basic alone as 1. Data are presented as mean + SD of triplicate cultures. Statistical significance between two groups was determined by the Student's t test (**p<0.01, ***p<0.001). Data are representative of three independent experiments.

Figure 2. The MEK1-ERK-signaling pathway and the p38 MAPK-signaling pathway are required for the GM-CSF-induced Aldh1a2 expression and nuclear translocation of Sp1 in BM-DCs.

(A) Flt3L-generated BM-DCs were cultured with or without 10 ng/ml GM-CSF for 16 h in the presence or absence of 50 µM PD98059 (PD) or 25 µM SB203580 (SB). After the culture, Aldh1a2 gene expression was assessed by real-time PCR. The Aldh1a2 mRNA expression level in the cells incubated with medium alone for 16 h was set to 1. The results are shown as the mean + SD of triplicate cultures. Statistical significance was determined by the Student's t test (***p<0.001). (B) BM-DCs were cultured with or without 10 ng/ml GM-CSF in the presence or absence of 50 µM PD98059 or 25 µM SB203580 for 16 h. After the culture, cytosolic and nuclear proteins were analyzed for the presence of Sp1, α-tubulin, and lamin B1 by Western blotting. (C) Structures of the deletion constructs used in this figure are shown. Δ1, Δ2, and Δ3 constructs were prepared by deleting the region spanning from the −56 to −41, −99 to −78, and −139 to −123, respectively, from the pGL3-RALDH2 (−373) reporter vector. COS-7 cells were transfected in triplicate with the 1.25 µg of pGL3-RALDH2 (−373) reporter vector or the deletion mutants in combination with 0.5 µg of the pCMV-Myc-Sp1 expression vector or control empty vector. One day after the transfection, cells were stimulated with or without 5 ng/ml PMA, and luciferase activity was measured. Relative promoter activities were calculated by arbitrarily defining the activity of pGL3-basic alone as 1. Data are presented as mean + SD of triplicate cultures. Statistical significance between two groups was determined by the Student's t test (**p<0.01, ***p<0.001; NS, not significant). Data are representative of three independent experiments.

Figure 3. Sp1 binds to the Aldh1a2 promoter region.

(A) The locations and nucleotide sequences corresponding to Probe A, Probe B, and Probe C in the 5′-flanking region of the mouse Aldh1a2 gene are shown. COS-7 cells were transfected with the 0.5 µg of pCMV-Myc-Sp1 or control empty vector. One day after transfection, cell lysates were analyzed for DNA-binding activity by DNAP assay using the indicated biotinylated DNA probes and anti-Myc Ab. (B) Flt3L-generated BM-DCs were cultured in the presence or absence of 10 ng/ml GM-CSF. After 16 h, nuclear extracts were analyzed for the presence of lamin B1 and Sp1 by Western blotting using anti-lamin B1 and anti-Sp1 Abs (left panel), or assessed for DNA binding activity by DNAP assay using biotinylated DNA Probe C and anti-Sp1 Ab (right panel). Data are representative of at least three independent experiments.

Figure 4. RA enhances GM-CSF-induced Aldh1a2 expression via the RARα/RXRα heterodimer bound to the RARE half-site.

(A) Localization of the putative RARE half-sites (underlined) in Probe C and their mutants, Probe C(RARE-h mt1) and Probe C(RARE-h mt2), are shown. COS-7 cells were transfected with the 0.5 µg of pSG5-RARα and/or pSG5-RXRα. One day after transfection, cell lysates were subjected to DNAP assay using the biotinylated DNA probes. The precipitates were analyzed by Western blotting using anti-RARα(upper panel) and RXRα (lower panel) Abs. (B) Flt3L-generated BM-DCs were cultured with 10 ng/ml GM-CSF for 16 h in the presence or absence of 100 nM RA. LE540 (1 µM) was added to the indicated cultures. After the culture, Aldh1a2 mRNA expression was assessed by real-time PCR. The Aldh1a2 mRNA expression level in the cells incubated with medium alone for 16 h was set to 1. (C) BM-DCs were cultured with (closed circle) or without (open circle) 10 ng/ml GM-CSF for 16 h in the presence of graded concentrations of RA. After the culture, Aldh1a2 mRNA expression was assessed by real-time PCR. Relative expression levels were calculated by defining the Aldh1a2 mRNA expression level in the cells incubated with medium alone for 16 h was set to 1. Asterisks indicate a significant difference (***p<0.001) from the Aldh1a2 expression in BM-DCs stimulated with GM-CSF alone. (D) COS-7 cells were transfected in triplicate with 1.25 µg of the pGL3-RALDH2 (−873) reporter vector or that containing the mutated RARE half-site (mt2) with or without the 0.5 µg of expression vectors, pSG5-RARα and pSG5-RXRα. One day after transfection, cells were stimulated with or without 100 nM RA, and luciferase activities were measured. The relative promoter activities were calculated by arbitrarily defining the activity of pGL3-baic alone without RA as 1. Data in (B, C, and D) are presented as mean + SD (B and D) or mean ± SD (C) of triplicate cultures. Statistical significance between two groups was determined by the Student's t test (*p<0.05, **p<0.01, ***p<0.001; NS, not significant). Data are representative of at least three independent experiments.

Figure 5. Sp1 and RARα/RXRα enhance each other's binding to the Aldh1a2 promoter and cooperatively enhance its activity.

(A) COS-7 cells were transfected with the 0.5 µg of pCMV-Myc-Sp1, the combination of pSG5-RARα and pSG5-RXRα, or the three. One day after transfection, cell lysates were subjected to DNAP assay using anti-Myc Ab, anti-RARα Ab, or anti-RXRα Ab, and biotinylated DNA Probe C whose sequence is shown in Figure 3. (B) COS-7 cells were transfected in triplicate with the 1.25 µg of pGL4-RALDH2 (−873) reporter vector and the 0.5 µg of expression vectors, pCMV-Myc-Sp1, pCMV-Myc-Sp1db, pSG5-RARα, and pSG5-RXRα, or control empty vectors. One day after transfection, cells were stimulated with or without 100 nM RA for 16 h. Then luciferase activities were measured. Relative promoter activities were calculated by arbitrarily defining the activity of pGL4-RALDH2 (−873) alone without RA as 1. (C) Flt3L-generated BM-DCs were cultured with or without 10 ng/ml GM-CSF or 10 nM RA. These cells were subjected to ChIP assay with anti-Sp1 or anti-RARα Ab or control IgG1. Binding of Sp1 and RARα proteins to the Aldh1a2 promoter site was estimated by real-time PCR. Data in (B and C) are presented as mean + SD of triplicate cultures. Statistical significance between two groups was determined by the Student's t test (*p<0.05, **p<0.01, ***p<0.001). Data are representative of three independent experiments.

Figure 6. The short DNA regions containing the GC-rich region, the RARE half-site, and the TATA box in the 5′-flanking region of the Aldh1a2 genes are well conserved among various species.

The sequence of the 5′-flanking region of the mouse (Mus musculus) Aldh1a2 gene was compared with that of the human (Homo sapiens) ALDH1A2 gene, that of rat (Rattus norvegicus) Aldh1a2 gene, that of cattle (Bos taurus) ALDH1A2 gene, that of chicken (Gallus gallus) ALDH1A2 gene, and that of zebrafish (Danio rerio) aldh1a2 gene. The sequence data were obtained using the NCBI MapViewer. The shaded regions indicate homology with the mouse sequence. The locations of the conserved RARE half-sites (RARE-h) and TATA boxes are indicated by boxes.

Figure 7. Methylation of the CpG island in the Aldh1a2 promoter prohibits Sp1 to activate the promoter, whereas the Aldh1a2 promoter is largely unmethylated in BM-pDCs as well as in BM-cDCs.

(A) pCpGL-basic and pCpGL-RALDH2 (−873) reporter vectors were methylated with 1.25 µg of M.SssI. COS-7 cells were transfected with methylated or unmethylated pCpGL-basic or pCpGL-RALDH2 (−873) reporter vector in combination with or without the 0.5 µg of pCMV-Myc-Sp1 expression vector. One day after transfection, luciferase activity was measured. Relative promoter activities were calculated by arbitrarily defining the activity of pCpGL-basic alone as 1. (B) COS-7 cells were transfected with pCMV-Myc-Sp1. One day after transfection, cell lysates were analyzed for DNA binding activity by DNAP assay using DNA Probe B and Probe C methylated with M.SssI or left unmethylated. The bound proteins were analyzed by SDS-PAGE followed by Western blotting with anti-Myc Ab. (C) Flt3L-generated BM-DCs were stained with allophycocyanin-labeled anti-CD11c Ab and phycoerythrin-labeled anti-B220 Ab, and were sorted to cDC and pDC fractions with a FACSAria. (D) Sorted BM-pDCs and BM-cDCs were cultured for 16 h with or without 10 ng/ml GM-CSF. Expression of Aldh1a2 mRNA was analyzed by real-time PCR. Relative expression levels were calculated by defining the Aldh1a2 mRNA expression in the cells incubated with medium alone for 16 h was set to 1. Data in (A and D) are presented as mean + SD of triplicate cultures. Statistical significance between two groups was determined by the Student's t test (***p<0.001). Data in (A, B, and D) are representative of at least three independent experiments. (E) Genomic DNA was isolated from BM-pDCs and BM-cDCs, denatured, modified with sodium bisulfite, and used in nested PCR (−424 to −30) for bisulfite sequencing. Eleven and thirteen independent clones of pDCs and cDCs, respectively, were analyzed. The methylation patterns of 5 representative clones of each cell type are shown. Closed circles indicate methylated CpG and open circles indicate unmethylated CpG.