Identification of natural RORγ ligands that regulate the development of lymphoid cells

Abstract

Mice deficient in the nuclear hormone receptor RORγt have defective development of thymocytes, lymphoid organs, Th17 cells, and type 3 innate lymphoid cells. RORγt binds to oxysterols derived from cholesterol catabolism, but it is not clear whether these are its natural ligands. Here, we show that sterol lipids are necessary and sufficient to drive RORγt-dependent transcription. We combined overexpression, RNAi, and genetic deletion of metabolic enzymes to study RORγ-dependent transcription. Our results are consistent with the RORγt ligand(s) being a cholesterol biosynthetic intermediate (CBI) downstream of lanosterol and upstream of zymosterol. Analysis of lipids bound to RORγ identified molecules with molecular weights consistent with CBIs. Furthermore, CBIs stabilized the RORγ ligand-binding domain and induced coactivator recruitment. Genetic deletion of metabolic enzymes upstream of the RORγt-ligand(s) affected the development of lymph nodes and Th17 cells. Our data suggest that CBIs play a role in lymphocyte development potentially through regulation of RORγt.

Figure 1. RORγ reporter activity induced by sterols in insect cells

(A) Cholesterol is necessary and sufficient to promote RORγ reporter activity in insect cells. D.m. Kc167 cells were grown in continuous cultures in CDM with or without cholesterol and were transiently transfected with RORγ-gal4/UAS firefly luciferase reporter. Representative experiment of n=3. (B) RORγ reporter activity in insect cells grown in CDM is dependent on the coactivator taiman. Kc167 cells were transfected with RORγ reporter described in (A) plus dsRNA. Cells were incubated overnight with 10 μM 7DHC prior to measurement of luciferase activity. RNAi dsGal4 inhibits expression of the RORγ-gal4 fusion protein, RNAi EYFP is negative control and RNAi Tai is specific for the D.m. coactivator taiman. Representative experiment of n=10. (C) CBIs induce RORγ reporter activity in Kc cells maintained in sterol-free CDM. Kc cells were transiently transfected with RORγ-gal4 as described in (A). 25OH (25-hydroxycholesterol), 7DHC (7-dehydrocholesterol), and SQL (squalene). (D) Structure-activity relationships (SAR) of sterols that promote RORγ reporter activity in insect cells. Shown are one non-cyclic and 8 sterol backbones based on the position of the internal double bond. The color bar on the right depicts the range (nd to 0.04 μM) as EC₅₀ of RORγ reporter activity induced by each compound. We tested 78 sterols and 46 were bioactive. Compounds (n=120) used as negative controls included 101 non-sterol compounds such as PUFAs, fatty acids, phospholipids, glycerolipids, amino acid derivatives and retinoids. nd: no activity detected. 4C (4-cholest-4-en-3-one), 4C22OH (4-cholest-4-en-3-one-22-ol), 4,7-cholesten (cholest-4,7-dien-3-one), 4AC4MΔ⁰ (4α-carboxy, 4β-methyl-cholesta-3-ol), Δ⁷-daf (Δ⁷-dafachronic acid). Compound nomenclature: common sterols are labeled by abbreviation and common name i.e, 7DHC is 7-dehydrocholesterol, for rare compounds an abbreviation followed by standard nomenclature for sterol lipids (Nes, 2011) is provided, for example 4AC4MΔ⁰ (4α-carboxy, 4β-methyl-cholestan-3-ol).The results are a summary of n=33 experiments. Each compound was tested at least twice in technical triplicates. Error bars are standard deviations.

Figure 2. RORγ reporter activity maps to the cholesterol biosynthetic pathway

(A) Metabolic map of RORγ reporter activity. Top: enzymes in the pathway. Middle: common names of CBIs. The conventional nomenclature for these sterols is: Lanosterol (14α,4α,4β-trimethyl-cholesta-8,24-dien-3-ol), FF-MAS (4α,4β-dimethyl-cholesta-8,14,24-trien-3-ol), T-MAS (4α,4β-dimethyl-cholesta-8,24-dien-3-ol), 4ACD8 (4α–carboxy, 4β-methyl-cholesta-8,24-dien-3-ol), 4αM-Z-one (4α–methyl-cholesta-8,24-dien-3-one), 4αM-Z-ol (4α–methyl-cholesta-8,24-dien-3-ol). The intermediates in the conversion of lanosterol into FF-MAS and the formyl intermediates of 4ACD8 are not shown. (B) Heatmap of RORγ reporter activity after overexpression of cholesterol biosynthetic enzymes in HEK293T cells co-transfected with RORγ-gal4/UAS-firefly luciferase reporter. (C) RORγ reporter activity is blocked in Fdft1^−/− SXLT cells transfected as described in (B). Empty vector, A304F=inactive RORγ mutant, ROR=wt RORγ, Trb=thyroid hormone receptor, Trb_T3=Trb plus 50 μM Triiodothyronine. Representative experiment of n=8. (D) Transfection of hFDFT1 can rescue RORγ reporter activity in SXLT cells. Representative experiment of n=3. (E) Supplementation of media with squalene (SQL) can partially rescue the full length RORγ reporter activity from an endogenous Il-23r enhancer element in SXLT cells. Representative experiment of n=4. Statistical analysis: two-tailed unpaired Student’s t-test. **P<0.005 with a power of 0.87 for 4 aggregated experiments. (F) tTA-off inducible RORγ reporter activity in U937ROR cells grown in cholesterol supplemented CDM after removal of doxycycline. Representative experiment of n=4. (G) RORγ reporter activity in U937ROR cells maintained in media supplemented with exogenous cholesterol is inhibited in a dose dependent fashion by cholesterol biosynthetic inhibitors lovastatin and Ro 48-8071 fumarate. Representative experiment of n=3. Experiments are technical triplicates. Error bars are standard deviations.

Figure 3. RORγ reporter activity maps to 4α,4β-dimethylsterols in the cholesterol biosynthetic pathway

(A) Overexpression of Cyp51 results in increased RORγ activity in HEK293T cells. HEK293T cells were transfected as described in Figure 2B. Representative experiment of n=3. (B) FF-MAS is a weak agonist for RORγ. HEK293T Cells were transfected like A. Representative experiment of n=6 two-tailed unpaired Student’s t-test *P<0.05 power of 0.74 (6 aggregated experiments). (C) Human CYP51a1 specific shRNAs (shCyp) reduces RORγ reporter activity in HEK293T cells which was rescued by overexpression of shRNA-resistant Cyp51. Knockdown of mRNA was >80%. Representative experiment of n=3. (D). Residual RORγ reporter activity is detected in Cyp51^−/− fibroblasts. Fibroblasts from Cyp51^−/− embryos were immortalized and respective cell lines were transfected with the same plasmids as described in A. (E) Knockdown of SC4MOL reduced RORγ reporter activity in U937ROR cells. U937ROR cells were transfected with plasmid containing shRNA specific for SC4MOL. Cells were selected post-transfection with puromycin selected cells were cloned. Shown are the results of 4 clones compared to a scramble clone control. Representative of n=3 experiments. (F) Knockdown of NSDHL reduced RORγ reporter activity in U937ROR cells. U937ROR cells were transfected with plasmid containing shRNAs specific for NSDHL as described in D. Representative of n=2 experiments. (G). Knockdown of C14ORF1, the human homologue of 0610007P14rik results in increase of RORγ reporter activity. HEK293T cells were transfected with RORγ-gal4 driven luciferase reporter system plus plasmids containing C14ORF1 specific shRNAs from the Sigma shRNA mission library. The shRNA validation and average percentage of mRNA knockdown is shown at the bottom. Representative of n=3 experiments. Experiments are presented as averages of technical triplicates. Error bars are standard deviations.

Figure 4. Sterol lipids are RORγ ligands

(A) CBIs abundant in mammalian cells bind RORγ with high affinity. Binding affinity for recombinant RORγ protein was defined by ability of the compound to compete with fluoresceinated 25-hydroxycholesterol or by alphascreen technology (7DHC and lanosterol). Concentrations of different CBIs in total embryonic tissue measured by mass spectrometry are derived from (Keber et al., 2011) and thymic concentration of SC4MOL products 4ACD8 and 4α-hydroxymethyl-cholest-7-en-3β-ol (4AOHD7) were defined in supplemental material and methods. The concentration of 25-hydroxycholesterol are estimates from normal mouse serum and lung tissue (Bauman et al., 2009). ND=not detected NT=not tested. (B) Drawing of 4ACD8. (C) Sterols promote the recruitment of fluoresceinated NCOA2 coactivator peptide to RORγ. NCOA2 peptide recruitment was compared between holo and apo RORγ receptor produced in bacteria using fluorescence polarization assay. (D) Sterols increase stability of RORγ in thermal denaturation assays. The rate of protein denaturation with increasing temperature was measured comparing empty receptor (apo) with ligand-bound receptor (holo forms). For (C) and (D), Apo receptor (red), holo 25-hydroxycholesterol (blue), holo 4α-carboxy-cholest-7-en-3β-ol (4ACD7) (green) and 4ACD8 (lila). (E) Ribbon drawing showing the structure of RORγ bound to the CBI 4ACD8. The RORγ LBD is depicted in green with the ligand inside containing oxygen atoms (Red). An LXXLL motif-containing coactivator peptide is shown in red and Helix 12 in blue. (F) A 180 degree rotation of (E). Graphs with measurements are representatives of n=3 experiments. Cell culture experiments are in technical triplicates and error bars are standard deviations. (G) The 4α-carboxy group of 4ACD8 forms a salt-bridge with RORγ LBD residue Q286 that is homologous to the 4ACD8 contact residue of RORα (Q289).

Figure 5. Deficiencies in the cholesterol biosynthetic pathway affect lymph node development and Th17 cell differentiation

(A) Lti populations are reduced in cervical lymph node anlagen of day E14.5 Cyp51^−/− mouse embryos. (B) Brachial lymph node anlagen are absent of day E14.5 Cyp51^−/− mouse embryos. (C) Lti populations are reduced in axillary lymph node anlagen of day E14.5 Cyp51^−/− mouse embryos. Embryos were obtained from time-mated pregnancies and fixed with formaldehyde. Whole embryo serial sections of wt (n=3) and Cyp51^−/− (n=4) embryos were prepared and stained for IL7Rα chain (GREEN), CD4 (RED) and hematopoietic lineage marker CD45 (BLUE). (D) Quantitation of IL7Rα⁺ and CD4⁺ cells in lymph node anlagen of Cyp51^−/− embryos shown in (A) and (C). Left axillary, right cervical lymph node anlagen. Data are quantified as bar graphs of averages of per cent cells per embryo. (E) Impaired Th17 cell differentiation in RORγt-cre Sc4mol-deficient naïve CD4⁺ T cells. FACS analysis of Th17 cells stained for IL-17A and IFNγ 72 h after polarization (top and bottom panels) and 144 h after polarization (middle panel). Each experiment compares naïve CD4 T cells sorted from Sc4mol^+/+ and Sc4mol^−/− animals and polarized side by side as technical triplicates. Th17 polarization, representative experiment (n=6). **P<0.005 and power 0.96 for P<0.01. Th1 polarization, representative experiment (n=3). (F) Addition of 1 μM 4ACD8 rescues Th17 cell differentiation in RORγt-cre Sc4mol^−/− naïve CD4⁺ T cells. Cells were processed and analyzed as described in (E). Representative of (n=3) experiments with P<0.005 and power 0.99 for P<0.01. Data are quantified as bar graphs with averages (%) of differentiated cells. Error bars are standard deviations. Statistics: two tailed unpaired Student’s t-test with equal variance, power estimates are for aggregate experiments.