Differential modulation of nuclear receptor LRH-1 through targeting buried and surface regions of the binding pocket

Biochemical characterization.

Previous agonist development has demonstrated the utility of the R¹ sulfamide of 6N in improving potency,²⁷ while the R⁴ carboxylic acid tail of 10CA has recently been shown to make PL-like contacts at the mouth of the pocket, contributing to improved efficacy.^{28, 29} To capture the benefits of both agonists, we synthesized a hybrid compound (“6N-10CA”) (Scheme 1). We compared binding affinity and potency using fluorescence polarization (FP) competition³⁰ and luciferase reporter assays,^{25, 27, 28} respectively. Trends for 6N and 10CA mirrored reported data,^26–28 with 6N exhibiting greater affinity (K_i) and potency (EC₅₀) than 10CA, and 10CA exhibiting greater efficacy (fold-activation) than 6N (Fig. 2A, B, D). 6N-10CA displayed picomolar affinity, approximately 10-fold higher than that of 6N (Fig. 2A, D), representing the first picomolar binding affinity LRH-1 ligand to date. Luciferase reporter assays demonstrated a similar trend, with 6N-10CA exhibiting the highest potency and a fold activation similar to 10CA (Fig. 2B, D). Each compound increased the stability of the receptor relative to co-purified bacterial PLs, with 6N-10CA displaying the greatest degree of stabilization, improving thermal stability by approximately 12.2°C, followed by 6N (+9.4°C) and 10CA (+6.2°C) (Fig. 2C, D). Overall, this suggests that orthogonal contacts within the LBP (Fig. 1) enhance affinity, potency, and stabilizing effect on the receptor in an additive manner.

Gene expression.

To investigate how these 6HP-derived molecules compare in driving LRH-1 target gene expression, we performed RT-qPCR analyses of key LRH-1 target genes Cyp7a1^{9, 10, 31} and Shp⁸ in HepG2 cells, a liver-derived cell line commonly used to assess LRH-1 activity.^{23, 32, 33} 6N-10CA and activating PL DLPC drove Shp expression to the greatest extent. Alternatively, 6N-10CA and 6N each increased expression of Cyp7a1, while DLPC decreased expression (Fig. 3A). Interestingly, 10CA slightly decreased expression of both target genes, comparable to effects of LRH-1 antagonist Cpd3³⁴ (Fig. S1). These data highlight the importance of assessing LRH-1 activity at endogenous levels and monitoring gene expression to capture selective effects of ligands. Additionally, while the mRNA levels of certain genes such as Shp are consistently increased by agonizing ligands, others may be differentially responsive.

Given the differences observed in RT-qPCR, we performed RNA-seq analyses on HepG2 cells treated with 6N-10CA, 6N, and activating PL DLPC (Fig. 3; Fig. S2; Table S1–2), which revealed a number of shared gene expression patterns (Fig. S3). However, GO and KEGG pathway analyses revealed that DLPC- and 6N-treated cells exhibited divergent effects on metabolic pathways, with genes responsible for lipid and cholesterol homeostasis upregulated by DLPC and downregulated by 6N (Fig. 3B, Table S1–2). Closer inspection revealed 42 shared genes upregulated by DLPC and downregulated by 6N (Fig. 3C). The DLPC-induced expression pattern of most of these metabolic genes was mirrored in 6N-10CA-treated cells, with the exception of certain genes such as Acss3 and Snx33 (Fig. 3B, C; Table S1–2), which may be more responsive to deep pocket contacts. RT-qPCR analysis examining a subset of overlapping DLPC-upregulated and 6N-downregulated genes confirmed that 6N-10CA consistently drives an increase in mRNA levels of metabolic genes that are downregulated by 6N (Fig. 3A). This suggests that addition of the R⁴ carboxylic acid engages the allosteric networks required to drive DLPC-like signaling of genes regulating lipid and cholesterol homeostasis. The opposite pattern is seen in Cyp7a1 (Fig. 3A), where the R¹ sulfamide increases expression, while DLPC decreases expression. The minimal response of 10CA in RT-qPCR (Fig. S1) and RNA-seq (Fig. S2) may be the result of poor solubility, potency, or cell permeability, highlighting the importance of the 6N R¹ sulfamide for compound efficacy in cultured cells.

Coregulator binding and signaling.

A selective nuclear receptor modulator can increase or decrease transcriptional output by differentially recruiting coregulators. This is often visible by small molecules having tissue-selective effects that are governed by coregulator expression.^35–38 Therefore, the impact of a selective modulator will be highly dependent upon whether the ligand can enhance functional interactions between the nuclear receptor and those coactivators expressed in the cell.^{39, 40} We hypothesized that the divergence in ligand-mediated effects on metabolic gene expression in HepG2 cells is due to different effects on coregulator-LRH-1 signaling, as recent evidence suggests differential coregulator recruitment by 6N and 10CA.²⁹ We first sought to explore how synthetic agonists affect recruitment of LXXLL motif-containing peptides of coregulator proteins established to regulate LRH-1 activity: steroid receptor coactivator-3 (SRC3),⁴¹ peroxisome proliferator-activated receptor gamma coactivator 1-α (PGC-1α),^{42, 43} and small heterodimer partner (SHP).^{8, 44} While each peptide was sensitive to agonists, PGC-1α binding affinity was the most responsive (Fig. 4A). 6N and 10CA drove a 3-fold, 4-fold, and 2-fold increase in affinity for SRC3, PGC-1α, and SHP, respectively (relative to poorly activating, co-purified bacterial PLs^{18, 19, 22}). 6N-10CA promoted a similar affinity for SRC3 and an enhanced affinity for PGC-1α and SHP (10-fold and 4-fold over vehicle control, respectively) (Fig. 4A). Therefore, synthetic agonists drive increased interaction between the LRH-1 LBD and the binding interface of coregulator proteins, with 6N-10CA having the most pronounced effect on binding affinity. However, the ability of 6HP-derived small molecules to drive selective recruitment of coregulators is not seen between the LXXLL motif and the LRH-1 LBD.

To investigate how small molecules drive in-cell communication between full-length LRH-1 (FL-LRH-1) and coregulators, we used a luciferase reporter assay where we concomitantly overexpressed LRH-1 and coactivators SRC3, PGC-1α, p300, as well as corepressor nuclear receptor corepressor 1 (NCOR1). Overexpression of one coregulator enables us to observe how a small molecule modulates LRH-1 activity where that coregulator is abundant. Each coregulator modulated LRH-1 activity as expected, with coactivators SRC3, PGC-1α, and p300 increasing activity and NCOR1 decreasing activity, relative to the vector control (Fig. S4). Vector controls showed a consistent trend, with 6N and 6N-10CA displaying the greatest potency and 10CA and 6N-10CA exhibiting the greatest efficacy (Fig. 4B). However, small molecule-driven activity changed drastically depending upon which coregulator was overexpressed in the cell (Fig. 4B). Each small molecule is agonistic when NCOR1 is overexpressed, suggesting that NCOR1 expression has minimal effects on LRH-1 agonist activity. However, when PGC-1α is overexpressed, there is a dose-dependent decrease in receptor activity upon addition of RJW100 and 6N, and no response to carboxylic acid-containing 10CA and 6N-10CA. When coactivator p300 is overexpressed, LRH-1 is unresponsive or potentially antagonized by RJW100, though decreased activity was not statistically significant. This trend was mirrored in cells with SRC3 overexpressed (Fig. 4B). LRH-1-PGC-1α signaling is therefore the most sensitive to ligand identity, and our data suggest that synthetic ligands without LBP mouth contacts may act as inverse agonists for this pathway, providing useful tools for decreasing LRH-1 activity in a cell-type specific manner.

Structural studies.

To explore the molecular basis of 6N-10CA-mediated activation, we first solved the crystal structure of 6N-10CA bound to the LRH-1 LBD (PDB 7TT8, Fig. 5, Table S3). We observed clear electron density for each region of the small molecule, including the 6HP core, R¹ sulfamide moiety, and the R⁴ carboxylic acid tail (Fig. 5A). As seen with 10CA²⁹ and 6N²⁷ (Fig. 1), there was hydrogen bond contact with residues Y516, K520, G421, and A424 at the pocket mouth (10CA), as well as M345 and T352 deep within the pocket (6N) (Fig. 5B, C). Additionally, there is both π–π stacking and an indirect hydrogen bond contact with H390 through a conserved water molecule seen in previous structures (Fig. 1; Fig. 5C). RJW100, 6N, 10CA, and 6N-10CA have nearly identical binding poses, demonstrating that polar contacts anchor bicyclic compounds in a predictable orientation in the pocket (Fig. 5D). To examine whether there were notable differences between structures of the LRH-1 LBD bound to different synthetic ligands, we identified sites that may reflect ligand-binding status by performing an all-residue distance comparison (visualized as a distogram) between crystal structures of the active⁷ (DLPC-bound; PDB 4DOS¹⁹) and inactive^{18, 19} (apo; PDB 4PLD²²) LRH-1 LBD (Fig. 5E). Regions of high structural dissimilarity included H6 and H10, reflecting movement at the LBP mouth. We therefore compared the LBP mouth in crystal structures of LRH-1 in different ligand-bound states (Fig. 5F). The position of G421 (proxy for the H6/7 loop) was the most sensitive to the binding of DLPC and synthetic agonists. Each agonist promoted movement of H6 toward G421, with DLPC and carboxylic acid-containing compounds demonstrating the greatest effects (Fig. 5F). The same trend was seen in the distance between H3 and G421, though RJW100 and 6N narrowed this gap. Alternatively, each ligand-bound state drives H10 closer to G421 to an equivalent degree. Mobility of H3 relative to H6 and H10 was less informative, as DLPC-bound and apo states promote similar positioning. Therefore, orientation of the H6–H7 loop best reflects ligand activity, and ligands contacting the LBP mouth drive the greatest movement in this region.

Compound-mediated dynamics.

We then examined differences in predicted communication pathways between allosterically-linked sites on LRH-1 bound to different agonizing ligands using molecular dynamics (MD) simulations (1-μs) coupled with suboptimal paths analysis.⁴⁵ This analysis identifies chains of residues with the strongest patterns of correlated motion, as measured by cartesian covariance, identifying communication paths between two sites on the receptor. These strongest paths are thought to convey the most information between the sites, facilitating allosteric communication.^{46, 47} This analysis has previously detected allosteric patterns that are correlated with activity of different ligand-bound states of LRH-1.^{22, 27, 43} We obtained the strongest 1000 suboptimal paths connecting the Tif2 coactivator (as a sensor for the conformation and dynamics of the AF-2) and a secondary activation function surface (AF-B), which encompasses β1/2-H6-H7 and communicates ligand-binding status to the AF-2,^{19, 22, 43} in LRH-1 complexes with 6N, 10CA, and 6N-10CA (Fig. 5G). Paths in LRH-1-10CA traverse H4–H5 to reach the AF-2 surface, while 6N preferentially induces communication through H3. 6N-10CA induces communication between both H4–H5 and H3, indicating that correlated motions are induced along both pathways, in contrast to the more directionally restrictive pathways observed in 6N and 10CA. We also obtained the 1000 strongest paths connecting the ligand and Tif2 peptide. While 6N and 6N-10CA induce communication directly between H3 and Tif2, 10CA relies heavily on induced communication between H3-H12-peptide and H5-Tif2, with only a small fraction of paths directly linking H3 with Tif2 peptide (Fig. 5G). Overall, different ligands that contact the pocket mouth drive distinct positioning of H6–H7, and communication between this region (AF-B) and the AF-2 provides a predictive measure of ligand activity.

PGC-1α was the most responsive coactivator peptide to small molecule binding (Fig. 4A), promoted inverse agonist activity of RJW100 and 6N when overexpressed in reporter assays (Fig. 4B), and can discriminate between PL-bound and apo states in peptide binding assays.¹⁹ We therefore performed MD simulations (2-μs) on LRH-1 LBD-PGC-1α⁴³ bound to RJW100, 6N, DLPC, 10CA, and 6N-10CA, as well as apo LRH-1 LBD-PGC-1α, to more deeply explore the dynamics and allostery responsible for differential ligand responsiveness observed in RNA-seq and reporter assays. RMSF analysis and comparison of average structures revealed considerable variation in the AF-B (Fig. S5). This is consistent with mobility of, and communication from, H6–H7 in the LRH-1 LBD (Fig. 5E, F, G), as well as previous studies proposing an importance of this region in communicating ligand-binding status to the AF-2.^{19, 22}

To investigate the entire LBD for sites driving differential ligand responsiveness, we conducted suboptimal paths analysis between every pair of residues (excluding neighboring residues) on the protein and displayed these results as a matrix (Fig. 6A; Fig. S6). This analysis preferentially detects paths of communication between distant sites on the protein, enabling us to identify novel regions that are allosterically linked. There were many communication paths between the AF-B and H3-H4-H5 (residues 350–380; “site one”), H8-H9-H10 (residues 450–500; “site two”), as well as the PGC-1α peptide (Fig. 6A, B; Fig. S6). Site one includes the AF-2, while site two corresponds to the binding interface between LRH-1 and β-catenin, a coactivator of LRH-1.⁴⁸ Close inspection revealed G421, a residue at the mouth of the pocket between H6 and H7, as the “hotspot” of AF-B communication in 6N-10CA-bound LRH-1. Interestingly, this residue is both contacted by PLs and tail-containing ligands (Fig. 1; Fig. 5) and has previously been shown to decrease LRH-1 activity when mutated.¹⁹ We inspected how communication from G421 differed between ligand-bound states. DLPC, 10CA, and 6N-10CA promoted many paths with site one and site two (Fig. 6C). Alternatively, apo-LBD exhibited a similar number of paths with only site one, while RJW100 and 6N drove few paths of communication with both sites. Communication between G421 and PGC-1α was also highly ligand responsive. Apo-LBD, as well as RJW100- and 6N-bound states, communicated almost entirely through H3 and H12. However, DLPC, 10CA, and 6N-10CA promoted nearly identical paths of communication running through H7-H4-H5-PGC-1α (Fig. 6D). These paths strongly resemble previously reported LRH-1 LBD complexes bound to both activating ligand and coactivator.²² Overall, MD simulations reveal that communication with the AF-2 is strongly influenced by contact with G421, and that tail-containing compounds mimic PL-mediated allosteric communication from this residue. Because PGC-1α is known to be an important modulator of LRH-1 activity in the liver,⁴⁹ differential communication between AF-B and the AF-2 may be responsible for the distinct signaling patterns seen in metabolic gene expression in HepG2 cells (Fig. 3).

To gain additional mechanistic insight into how activating ligands drive differential signaling with FL-LRH-1, we performed hydrogen/deuterium exchange mass spectrometry (HDX-MS) on FL-LRH-1 bound to 6N, 10CA, 6N-10CA, and poorly activating (co-purified bacterial) PLs (Fig. 6E, Table S4, Fig. S7). Different ligands promoted deuterium exchange of distinct regions of the receptor. 6N-10CA and 6N stabilized H1–H2 on the LBD, while 10CA drove stabilization of the zinc-finger binding motif of the DBD. 6N-10CA stabilized part of the DBD (region proximal to the Ftz-F1 box), along with several other areas on the LBD (H7–H8, and H9–H10). One of these regions (H9–H10) intersects with site two identified in MD studies above (Fig. 6A, B, C). We mapped these regions onto a structure of LRH-1 with LBD and DBD interfaces proposed by recent modeling studies⁵⁰ (Fig. 6E). The model predicted three sites of inter-domain communication that primarily involve H1–H2 and H3 of the LBD. This corresponds to the region of the LBD stabilized by 6N-10CA and 6N (H1–H2), along with the region of the DBD stabilized by 6N-10CA and 10CA (proximal to H3). These HDX-MS data therefore support previous modeling studies and reveal novel insight into interdomain communication driven by distinct ligands. However, FL-LRH-1 complexed with different ligands (6N, 10CA, and 6N-10CA) revealed modest differences in DNA binding affinity (Fig. S8), suggesting that synthetic ligands play a limited role in DNA occupancy.

Chemical synthesis.

While the effects of 6N-10CA on LRH-1 signaling have not been explored, it was initially generated as an intermediate compound in the synthesis of a probe for an FP competition assay used to study small molecule binding to LRH-1.³⁰ The chemical synthesis of 6N-10CA has therefore been described previously.³⁰ Synthetic methods for 6N have also been previously published.²⁷ The carboxylic acid derivate (10CA) and RJW100 were synthesized as previously described.²⁸ All compounds in this study were isolated at greater than 95% purity with the exception of 10CA (91% exo and 9% endo). Representative HPLC traces for each compound used in the study are provided in the supplemental information.

Cell culture.

HeLa cells were cultured in phenol red-free MEMα + 10% fetal bovine serum (FBS) – charcoal/dextran treated, and HepG2 cells were cultured in phenol red-free DMEM + 10% fetal bovine serum. Cell lines were cultured under standard conditions (5% CO₂, 37°C). HepG2 cells used in RNA-seq studies were not tested prior to experimentation, but this cell line later tested positive for mycoplasma (LookOut^® Mycoplasma PCR Detection Kit). To verify that expression data was not affected by potential contamination, all RT-qPCR experiments were conducted in HepG2 cells ordered from ATCC and verified to be mycoplasma free. HeLa cells were verified to be mycoplasma free with the LookOut^® Mycoplasma PCR Detection Kit.

Protein expression and purification.

LRH-1 LBD was expressed and purified as described previously.^{26, 27} Briefly, BL21(DE3) E. coli cells were transformed with human LRH-1 LBD (residues 299–541) with an N-terminal 6xHis tag in a pMCSG7 vector. Bacteria were grown at 37°C in liquid broth until cells reached OD₆₀₀ 0.6. Protein expression was then induced with 1 mM IPTG for 4 hours at 30°C. Cells were centrifuged and stored at −80°C. After one freeze-thaw cycle, the pellet was resuspended in lysis buffer (20 mM Tris-HCl, 150 mM NaCl, 5% glycerol, 25 mM imidazole, 0.2 mM PMSF, DNase, lysozyme, pH 7.4) and lysed via sonication. Lysate was subjected to Ni²⁺ affinity chromatography. To remove co-purified bacterial lipids and to ensure a homogenous complex, LRH-1 LBD was incubated with four-fold molar excess of DLPC for 16 h at 4°C and then purified with size-exclusion chromatography (SEC) into assay buffer (150 mM NaCl, 20 mM Tris-HCl, 5 % glycerol, pH 7.4). Protein used for thermal stability and peptide recruitment assays was purified in a similar manner but was not complexed with DLPC prior to SEC purification. LRH-1 LBD used for crystallography was incubated with TEV protease to remove the 6xHis tag and subjected to a second round of nickel affinity chromatography. All protein was stored at −80°C until use.

Full-length human LRH-1 (FL-LRH-1; residues 1–495, iso-form 1, codon optimized for E. coli) was expressed and purified as described previously.²⁹ Briefly, FL-LRH-1 was cloned into a PET28b vector modified with a His-SUMO tag and transformed into BL21 (DE3) E. coli cells. Cells were grown in liquid broth media at 37°C to OD₆₀₀ 0.5. Protein expression was induced with 0.2 mM IPTG and grown overnight at 16 °C. Cells were harvested by centrifugation and resuspended into lysis buffer containing 20 mM Tris (pH 7.5), 500 mM NaCl, 5% glycerol, 25 mM imidazole, 0.5 mM tris(2-carboxyethyl) phosphine (TCEP), and 0.1 mM PMSF. Cells were lysed by sonication for 8 minutes and clarified by centrifugation. Protein was purified by Ni²⁺ affinity chromatography with elution buffer including 20 mM Tris HCl (pH 7.5), 500 mM NaCl, 5% glycerol, 0.5 mM TCEP, and 125 mM imidazole. Ion-exchange chromatography was used to remove DNA with tandem Hitrap Q-SP columns (GE Healthcare). Protein was flowed through the Q column onto the SP column, from which it eluted as a single peak using a linear gradient of NaCl from 0.35 M to 1 M.

Ligand binding assays.

FP competition assays were performed as previously described.^{26, 30} Briefly, experiments were conducted in black 384-well plates in assay buffer (150 mM NaCl, 20 mM Tris-HCl, 5% glycerol, pH 7.4). 6N conjugated to fluorescein amidite (FAM) (10 nM/well) was incubated with LRH-1 LBD (5 nM/well). Unlabeled compounds were added at concentrations indicated in figures. Each experiment was performed twice with four technical replicates each. Technical replicates were averaged and normalized independently prior to final data analysis. Using GraphPad Prism (version 9), data were fit to a one-site, fit K_i curve, assuming a final probe concentration of 10 nM and probe affinity of 1 nM.³⁰ Data were excluded from wells with 1.0 e-6 M 6N-10CA, as the resulting point dropped sharply after ligand saturation, distorting the curve fit.

Reporter assays.

LRH-1 reporter assays were conducted as described previously.^{25, 26} Briefly, HeLa cells were seeded at ~ 7,500 cells per well in 96-well plates (white-walled, clear bottom) in MEMα + 10% FBS. When cells reached 80–90% confluence, they were transfected with LRH-1 (5 ng/well), a reporter plasmid with an LRH-1 response element derived from the SHP promoter cloned upstream of firefly luciferase (50 ng/well), and a plasmid expressing Renilla luciferase constitutively from a CMV promoter (1 ng/well). Cells were transfected with FuGENE at a ratio of 5:1 (FuGENE:DNA). Twenty-four hours after transfection, compounds were diluted in Opti-MEM and introduced to cells at final concentrations indicated in figures (final DMSO concentration was 0.37%). After ~ 24 hours, luciferase signal was measured using the DualGlo kit (Promega). Luciferase reporter assays that overexpressed coregulators were conducted similarly but used a FuGENE:DNA of 4:1 and involved co-transfection of respective coregulators in a pDest26 (PGC-1α, SRC3, NCOR1) or pcDNA (p300) vector (10 ng/well). PGC-1α, SRC3, and NCOR1 vectors were generated via a Gateway^™ LR Clonase^™ (Invitrogen^™) reaction with an empty pDest26 vector and respective donor vectors. Compounds were added to reporter assays with overexpressed coregulators at low, medium, and high doses, corresponding to 1.0 e-7, 1.0 e-6, and 1.0 e-5 M for RJW100 and 10CA, and 5.0 e-8, 5.0 e-7, and 5.0 e-6 M for 6N and 6N-10CA (due to compound cytotoxicity at high doses). Each experiment was conducted with four biological replicates, each with three technical replicates averaged prior to data analysis. Firefly luciferase signal for each well was divided by the well’s Renilla signal intensity and then normalized relative to the DMSO control. Data were analyzed with GraphPad Prism (version 9) using a stimulating dose-response curve (Hill slope = 1). Data were excluded from analysis for cells treated with 3e-5 M of 6N-10CA, as the final signal showed a drastic decrease in overall signal, potentially indicating cell toxicity or compound insolubility. Data from reporter assays overexpressing coregulators were analyzed with a Two-way ANOVA and Dunnett Multiple Comparisons Test.

Protein stability.

Thermal stability of the LRH-1 LBD complexed with ligands was determined using a Tycho^™ NT.6 Nanotemper. LRH-1 LBD was incubated with 5-fold molar excess of ligand (final DMSO concentration was 1.4%) overnight at 4 °C in assay buffer (150 mM NaCl, 20 mM Tris-HCl, 5% glycerol, pH 7.4). Complexes were centrifuged at high speed for five minutes prior to measurement. Samples were then loaded into capillaries and tryptophan/tyrosine fluorescence was monitored at wavelengths 330 and 350 nm over a 30 °C/min gradient (35°C – 95°C). The melting point (inflection point) was determined with Tycho^™ NT.6 software. Three separate experiments were conducted in triplicate. Technical replicates were averaged, and data were plotted using GraphPad Prism (version 9).

RT-qPCR.

HepG2 cells were seeded at 400,000 cells per well in 24-well plates in DMEM + 10% FBS – charcoal/dextran treated. When cells reached ~ 90% confluence, media was exchanged with media containing compound or vehicle (DMSO for synthetic ligands and EtOH for DLPC) at the desired concentration (final vehicle concentration was 0.1%). Synthetic ligands were added at 10 μM, while DLPC was added at 100 μM to achieve a maximum response (concentrations were determined from luciferase reporter assays reported here and previously).^{7, 28} After 24 hours, media was decanted, cells were washed with PBS, and cells were collected in RLT (+ 1% 2-mercaptoethanol) lysis buffer. Cells were stored at −80°C. RNA was extracted from cells using the RNeasy^® Mini Kit (QIAGEN), with on-column DNAse digestion. RNA was reverse transcribed with the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems^™). cDNA was quantified using Power SYBR Green PCR Master Mix (Applied Biosystems), using TATA-Box Binding Protein as a housekeeping gene. C_t values were calculated by resident software on the StepOne Plus thermocycler. Data were normalized using the ΔΔC_t method.⁶⁶ Each experiment was conducted with four biological replicates, corresponding to distinct passage numbers or cells passaged independently for at least one week. Data were analyzed with GraphPad Prism, using a Brown-Forsythe and Welch One-Way ANOVA and Dunnett T3 Multiple Comparisons Test.

Primers used for RT-qPCR were as follows: hTBP (forward 5’-TGCACAGGAGCCAAGAGTGAA-3’, reverse 5’-CACATCACAGCTCCCCACCA-3’); hSHP²⁴ (forward 5’-GCTTAGCCCCAAGGAATATGC-3’, reverse 5’-GTTCCAGGACTTCACACAGC-3’); hCYP7A1⁶⁷ (forward 5’-GAGAAGGCAAACGGGTGAAC-3’, reverse 5’-GGATTGGCACCAAATTGCAGA-3’); hSCD1⁶⁸ (forward 5’-GCAGCCGAGCTTTGTAAGAG-3’, reverse 5’-GTTCTACACCTGGCTTTGGG-3’); hACSS2⁶⁸ (forward 5’-GGATTCCAGCTGCAGTCTTC-3’, reverse 5’-CATGCCACCACAAGTCAATC-3’); hDHCR7⁶⁹ (forward 5’-GGACTGGTTTTCACTGGCGAGCG-3’, reverse 5’-CCAGAGCAGGTGCGTGAGGAG-3’); hFDFT1⁷⁰ (forward 5’-GCAACGCAGTGTGCATATTTT-3’, reverse 5’-CGCCAGTCTGGTTGGTAAAGG -3’).

RNA-seq.

HepG2 cells were prepared and treated with compounds as described for RT-qPCR. The assay was run with three biological replicates (defined as cells cultured independently for at least one week) in parallel. Cells were then homogenized in RLT buffer and extracted using the RNeasy Mini kit (Qiagen) with on-column DNase digestion. RNA quality was assessed using a TapeStation 4200 (Agilent) and then ten nanograms of total RNA was used as input for cDNA synthesis using the Clontech SMART-Seq v4 Ultra Low Input RNA kit (Takara Bio), according to the manufacturer’s instructions. Amplified cDNA was fragmented and appended with dual-indexed bar codes using the NexteraXT DNA Library Preparation kit (Illumina). Libraries were validated by capillary electrophoresis on a TapeStation 4200 (Agilent), pooled at equimolar concentrations, and sequenced on an Illumina NovaSeq6000 at 100SR, yielding ~35 million reads per sample. Alignments were performed using BioJupies⁷¹ and subsequently analyzed using the R/Bioconductor package “DESeq2” to identify differentially expressed genes.⁷² Enrichr^73–75 was used for GO (GO Biological Process) and KEGG pathway analyses, using genes significantly up- and downregulated in cells treated with compounds. Venn diagrams were constructed using the following online tool: https://bioinformatics.psb.ugent.be/webtools/Venn/

The data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus⁷⁶ and are accessible through GEO Series accession number GSE199060.

Peptide binding.

Coregulator peptide binding assays were performed as described previously.⁴³ Briefly, LRH-1 LBD was incubated with compound (4-fold molar excess) or an equal volume of DMSO overnight at 4°C. Complexes were centrifuged at high speed for five minutes prior to assay setup. Complexes were serially diluted in assay buffer (150 mM NaCl, 20 mM Tris-HCl, 5% glycerol, pH 7.4) in black 384-well plates. FAM-labeled coregulator peptides were added to a final concentration of 50 nM. Sequences of peptides used were as follows: SRC3, ⁺H₃N-KKENNALLRYLLDRDDPSD-CO₂⁻; PGC-1α, ⁺H₃N-EAEEPSLLKKLLLAPANTQ-CO₂⁻; SHP, ⁺H₃NQGAASRPAILYALLSSSLK-CO₂⁻. After an overnight incubation at 4°C, fluorescence polarization was measured on a Bio-Tek Neo plate reader. Four independent experiments were performed in triplicate. Data were combined and fit to a one-site equilibrium binding equation with GraphPad Prism (version 9).

Crystallography and structure determination.

LRH-1 LBD-6N-10CA crystals were generated as described previously.^{26, 29} Briefly, cleaved (6xHis tag removed) LRH-1 LBD was incubated with 6N-10CA at four-fold molar excess overnight at 4°C. The complex was then purified via SEC into crystallization buffer (150 mM NaCl, 100 mM ammonium acetate, 1 mM EDTA, 2 mM CHAPS, 1 mM DTT, pH 7.4) and incubated with a peptide corresponding to human Tif2 NR box 3 (⁺H₃N-KE-NALLRYLLDKDD-CO₂⁻) at four-fold molar excess for two hours at room temperature. The complex was then concentrated to ~ 7 mg/mL and crystals were generated via hanging drop vapor diffusion in crystallant containing 0.1 M tri-Na cit-rate – pH 4.6, 2–14% tert-butanol, and 0–15% glycerol at 4°C. Crystals were then flash frozen in liquid N₂ using cryoprotectant consisting of crystallant supplemented with 30% glycerol. Data were collected remotely from the Southeast Regional Collaborative Access Team (SER-CAT) at the Advanced Photon Source (Argonne National Laboratories, Chicago, IL). Data were processed using HKL2000 and phased with molecular replacement, using PDB 5L11 as the search model. Structure refinement was performed with Phenix^{77, 78} and Coot⁷⁹ (data were processed in P3121 but refined in P3221). During refinement, residues 333–337 were removed due to poor electron density. Final figures were constructed with PyMOL (Schrödinger, LLC).⁸⁰ Bio3D⁸¹ was used to perform distogram analysis.

Molecular dynamics simulations.

MD simulations with LRH-1-Tif2 were performed on three complexes: i) LRH-1-Tif2-6N (PDB 6OQY), ii) LRH-1-Tif2-10CA (PDB 7JYD), iii) LRH-1-Tif2-6N-10CA (PDB 7TT8). Note that structures ii) and iii) used for MD underwent further refinement before being uploaded to PDB. The complexes were solvated in an octahedral box of TIP3P water with a 10 Å buffer around the protein complex. Na⁺ and Cl⁻ ions were added to neutralize the protein and achieve physiological buffer conditions (150 mM NaCl). Systems were set up using the xleap tool in Amber-Tools18 of Amber 2018⁸² with the ff14SB forcefield.⁸³ Parameters for 6N, 10CA, and 6N-10CA were obtained using Ante-chamber⁸⁴ in AmberTools18. Minimizations and simulations were performed with Amber18 (also part of Amber 2018). For minimization, 5000 steps of steepest descent were used, followed by 5000 steps of conjugate gradient minimization. Minimizations were first performed with 500 kcal/mol·Å2 restraints on all protein and ligand atoms. Restraints were then removed on all atoms except the ligand and Tif2 peptides, and the protocol was repeated. Restraints were subsequently lowered to 100 kcal/mol·Å2 and then finally removed from all atoms for two final rounds of minimization. Minimized systems were heated from 0 to 300 K with a 100-ps MD run, with constant volume periodic boundaries and 5 kcal/mol·Å2 restraints on all protein and ligand atoms. A 12-ns equilibration was performed for all complexes with 10 kcal/mol·Å2 restraints on the Tif2 peptide and ligand atoms using the NPT ensemble. Restraints were reduced to 1 kcal/mol·Å2 and 10-ns of additional equilibration was performed. Finally, production trajectories of 1000 ns were obtained for unrestrained complexes in the NPT ensemble. All bonds between heavy atoms and hydrogens were fixed with the SHAKE algorithm.⁸⁵ A cutoff distance of 10 Å was used to evaluate long-range electrostatics with particle mesh Ewald and for van der Waals forces. Structural averaging was performed using the CPPTRAJ⁸⁶ module of AmberTools.

Simulations of LRH-1 complexed with PGC-1α were conducted similarly. Ligand-bound (or apo) LRH-1 LBD was generated by modeling ligands into the crystal structure of LRH-1-PGC-1α-RJW100 (PDB 5UNJ).⁴³ Ligand position of RJW100 in LRH-1-PGC-1α-RJW100 was used to guide ligands into the pocket. Maestro (Schrödinger, LLC)⁸⁷ was used to optimize hydrogen bond assignments, add N- and C-terminal caps to LRH-1 and PGC-1α, and run initial minimization on the structure. Systems were set up as described above, using the xleap tool in AmberTools20 of Amber 2020,⁸⁸ with ff14SB⁸³ (protein), GAFF2⁸⁹ (ligand), and TIP3P⁹⁰ (water) forcefields. Minimization and equilibration steps were conducted as previously noted, though three 12-ns equilibration steps preceded the final run, with restraints applied to: i) all non-solvent atoms -- 10 kcal/mol·Å2, ii) peptide and ligand atoms -- 10 kcal/mol·Å2, iii) ligand atoms -- 1 kcal/mol·Å2). Water molecules critical for ligand engagement deep within the pocket (four for RJW100²⁵ and 10CA;²⁹ one for 6N²⁷ and 6N-10CA) were restrained along with ligand. Four-500 ns simulations were run and concatenated with CPPTRAJ,⁸⁶ with every fifth frame (total of 40,000 frames) used for data analysis. Water, Na⁺, and Cl⁻ molecules, along with N- and C-terminal caps, were removed for data analysis.

CPPTRAJ⁸⁶ was used to analyze RMSF. Dynamic networks were constructed from trajectories using the NetworkView plugin⁹¹ in VMD⁹² and the Carma program.⁹³ Networks are constructed by defining all protein Cα atoms as nodes, using Cartesian covariance (calculated in Carma) to measure communication within the network. Pairs of nodes that reside within a 4.5 Å cutoff for 75% of the simulation are connected via an edge. Suboptimal paths between nodes were identified using the Floyd−Warshall algorithm.⁹⁴ Matrices visualizing the number of paths between every residue on LRH-1-PGC-1α-ligand complexes was determined using a loop script: ./subopt contact.dat loop 50 $i $j (i=source; j=sink; 50=cutoff). Communication with residue K5 on the PGC-1α peptide (S₁L₂L₃K₄K₅L₆L₇L₈A₉) was used as the sink for pathway analysis. This residue is divergent among most LXXLL-containing coregulators and selectively displayed strong communication with DLPC, 10CA, and 6N-10CA. It was thus considered optimal for determining differences in communication driven by ligands with PGC-1α. Suboptimal paths were visualized with VMD.⁹²

HDX-MS.

To prepare LRH-1 ligand complex for HDX analysis, 20 μM of FL-LRH-1 protein was incubated with 30 μM of a 12bp LRH-1 binding element from the Oct4 promoter and 100 μM of each ligand, followed by overnight incubation at 4°C to ensure complete ligand exchange. The complex was further purified in buffer containing 20 mM HEPES (pH 7.5), 150 mM NaCl and 0.5 mM TCEP using SEC on S200 Increase 10/300 column. The final purified complex was further spiked with 2X ligand before HDX measurement.

Solution-phase amide HDX experiments were carried out with a fully automated system (CTC HTS PAL, LEAP Technologies, Carrboro, NC; housed inside a 4°C cabinet) as described,⁹⁵ with slight modifications. Peptides were identified using tandem MS (MS/MS) experiments performed on a QExactive (Thermo Fisher Scientific, San Jose, CA) over a 70 min gradient. Product ion spectra were acquired in a data-dependent mode and the five most abundant ions were selected for the product ion analysis per scan event. The MS/MS *.raw data files were converted to *.mgf files and then submitted to MASCOT (version 2.3 Matrix Science, London, UK) for peptide identification. The maximum number of missed cleavages was set at 4 with the mass tolerance for precursor ions ± 0.6 Da and for fragment ions ± 8ppm. Pepsin was used for digestion and no specific enzyme was selected in MASCOT during the search. Peptides included in the peptide set used for HDX detection had a MASCOT score of 20 or greater. The MS/MS MASCOT search was also performed against a decoy (reverse) sequence and false positives were ruled out if they did not pass a 1% false discovery rate.

For differential experiments, the LRH-1:12bpOct4 complexes with either phospholipid or compound bound were thawed and diluted to 10 μM. Next, the complexes (5 μl) were mixed with 20 μl of D₂O-containing HDX buffer (20 mM HEPES pD 7.9, 150 mM NaCl, 0.5 mM TCEP) and incubated at 4°C for 0, 10, 30, 60, 900 or 3,600 s. Following on-exchange, unwanted forward- or back-exchange was minimized, and the protein was denatured by the addition of 25 μl of a quench solution (1.2 M guanidine HCl, 1.8% FA pH 2). Samples were then immediately passed through an immobilized pepsin column (prepared in house) at 150 μl min⁻¹ (0.1% v/v TFA, 4°C) and the resulting peptides were trapped and desalted on a 2 mm × 1 cm C₈ trap column (Hypersil Gold, ThermoFisher). The bound peptides were then gradient-eluted (4–40% CH3CN v/v and 0.3% v/v formic acid) across a 2.1 mm × 5 cm C₁₈ separation column (Hypersil Gold, ThermoFisher) for 5 min. Sample handling and peptide separation were conducted at 4°C. The eluted peptides were then subjected to electrospray ionization directly coupled to an Orbitrap mass spectrometer (Exactive, ThermoFisher).

Each differential HDX experiment was performed with three technical replicates. The intensity weighted mean m/z centroid value of each peptide envelope was calculated and subsequently converted into a percentage of deuterium incorporation. This is accomplished by determining the observed averages of the undeuterated and fully deuterated spectra using the conventional formula described elsewhere⁹⁶. The fully deuterated control, 100% deuterium incorporation, was calculated theoretically, and corrections for back-exchange were made on the basis of an estimated 70% deuterium recovery and accounting for 80% final deuterium concentration in the sample (1:5 dilution in D₂O HDX buffer).

Deuterium uptake for each peptide is calculated as the average of %D for all on-exchange time points and the difference in average %D values between the unbound and bound samples is presented as a heat map with a color code given at the bottom of the figure (warm colors for deprotection and cool colors for protection). Statistical significance for the differential HDX data is determined by an unpaired t-test for each time point, a procedure that is integrated into the HDX Workbench software.⁹⁷ Peptides are colored by the software automatically to display significant differences, determined either by a >5% difference (less or more protection) in average deuterium uptake between the two states, or by using the results of unpaired t-tests at each time point (p-value < 0.05 for any two time points or a p-value < 0.01 for any single time point). Peptides with non-significant changes between the two states are colored grey. Each peptide bar in the heat map view displays the average Δ %D values, associated standard deviation, and the charge state. Additionally, overlapping peptides with a similar protection trend covering the same region are used to rule out data ambiguity.

The HDX data from all overlapping peptides were consolidated to individual amino acid values using a residue averaging approach. Briefly, for each residue, the deuterium incorporation values and peptide lengths from all overlapping peptides were assembled. A weighting function was applied in which shorter peptides were weighted more heavily and longer peptides were weighted less. Each of the weighted deuterium incorporation values were then averaged incorporating this weighting function to produce a single value for each amino acid. The initial two residues of each peptide, as well as prolines, were omitted from the calculations. This approach is similar to that previously described.⁹⁸ The consolidated HDX data is mapped onto the protein model in ChimeraX.⁹⁹

The data have been deposited to the ProteomeXchange Consortium via the PRIDE¹⁰⁰ partner repository with the data set identifier PXD031099.

DNA binding.

DNA binding was measured as described previously,²⁹ using fluorescence polarization with a 6-carboxyfluorescein amidite (FAM)-labeled 17bp double-stranded DNA probe containing the LRH-1 binding element from the Cyp7a1 promoter. FL-LRH-1 was titrated against 10 nM FAM-Cyp7a1 in buffer containing 20 mM Tris (pH 7.5), 150 mM NaCl, 5% glycerol and 0.5 mM TCEP, with a final volume of 50 μl, and incubated at room temperature for 30 min. Polarization was monitored on a Synergy 4 microplate reader (Biotek) at an excitation/emission wavelength of 485/528 nm. Curves were fit using a one-site specific binding equation with baseline-subtraction of wells containing protein but no probe. Experiment was conducted with two technical replicates.