The Oli-neuM line (Cellosaurus ExPASy CVCL_VL76) was grown in either growth medium (GM) or differentiation medium (DM) containing 500 μg/mL Geneticin (G418, Gibco^™, Thermo Fisher Scientific, Waltham, MA, United States, 10131027) and maintained at 37°C in 5% CO₂, as previously described (Porcu et al., 2015).

Budesonide (Selleckchem.com, S1286, Cologne Germany) was dissolved in 100% Dimethyl sulfoxide (DMSO; A3672, AppliChem, Damstad, Germany) and used at the final concentrations indicated in the text. Clobetasol (Prestw-781) was purchased from Prestwick Chemical Library^®, Illkirch-Graffenstaden, France and used at a final concentration of 10 μM as previously described (Porcu et al., 2015). Dorsomorphin (S7306, Selleckchem, 50,829 Cologne Germany), a selective inhibitor of AMPK (Meley et al., 2006), was purchased from Selleckchem.com, the stock concentration was 100 mM, and it was employed at a concentration of 3 μM, previously demonstrated as the working concentration in Oli-neuM cell line (Del Giovane et al., 2022). 0.5% DMSO max was added to the vehicle treatment in all experiments. Unless otherwise stated, drugs were administrated in DM for 48 h to cells left to grow in GM for 24 h. Culturing and time of drug treatments have been established previously based on the timing and concentration for optimal MBP expression in Oli-neuM (Meley et al., 2006; Porcu et al., 2015; Neumann et al., 2019; Del Giovane et al., 2022). Engagement tests were performed as previously described in growth chambers containing aligned polystyrene (PS) microfibers of 2–4 μm manufactured and used as previously described (Nocita et al., 2019). For microfiber engagement analysis, the cells were treated for at least 72 h, according to previously established protocols (Del Giovane et al., 2022; Dominicis et al., 2023).

Cells were seeded in 96-well plates (Grainer bio-one, Kremsmünster, Austria, 655,090 CHIMNEY WELL, μCLEAR^®, NERO, CELLSTAR^®, TC) pretreated with 10 μg/mL fibronectin (F0895; Sigma-Aldrich, Burlington, MA, United States) after 48 h with treatment in DM were fixed and processed for immunofluorescence (IF), as previously described (Nocita et al., 2019; Del Giovane et al., 2022). The automatized acquisition was performed using a Leica DMI6000 B epifluorescence inverted microscope (Leica Microsystems, Wetzlar, Germany), equipped with Leica Application Suite X and Matrix Screener software (version 3.0) at 20× magnification (HCX PL FLUOTAR 20 × NA 0.4). Images were then quantified and statistically analyzed with ScanR Analysis software (version 1.1.0.6 or 3.0; Olympus, Tokyo, Japan), as previously described (Porcu et al., 2015; Nocita et al., 2019). Hoechst 33342 (H3570, Thermo Fisher Scientific Inc., Waltham, MA, United States) and phalloidin (A12380; 1:40; Thermo Fisher Scientific Inc., Waltham, MA, United States) staining were performed to detect nuclei and actin cytoskeleton. Rat anti-MBP (MCA409S, 1:100; AbD Serotec, Hercules, CA, United States), was used as primary antibody (Ab) and Alexa Fluor 488 as secondary Ab (Thermo Fisher Scientific Inc., Waltham, MA, United States), as indicated in the text. Image analysis was performed as previously described (Porcu et al., 2015).

Cell culture chambers containing electro-spun PS microfibers were prepared as indicated in Nocita et al. (2019), UV-sterilized before use, and pre-treated with 10 μg/mL fibronectin (F0895; Sigma-Aldrich, Burlington, MA, United States). 60,000 Oli-neuM cells were seeded in GM. After 24 h, the medium was exchanged for either DM supplemented with 0.5% DMSO (vehicle) or the indicated drugs. After 72 h at 37°C in 5% CO₂, cells were processed for fixation and IF microscopy analyses. Acquisition and engagement analyses were performed as described in Nocita et al. (2019). Confocal images were acquired as previously described (Nocita et al., 2019) with a 40X objective, and 3D volume reconstruction was performed using Imaris software (Bitplane AG, Zúrich-Switzerland). Quantification of % of engagement was performed by analyzing 75 images per sample for each treatment in 3 biological replicates (n = 3). The percentage of engaged cells was estimated by counting the total number of nuclei within a range of 86 μm from the fiber. Cells were considered engaged if nuclei were touching the fiber. The mean data (± SEM) of three biological replicates (3n) were plotted on the graph using Graph Pad (version 7).

Hundred engaged cells for each sample of each biological replicate were randomly selected to measure the mean values of the length of the membrane extensions (processes) along the PS fibers. The length of the processes extending along the fibres (μm) was measured by using the Analyses/Measure tool of ImageJ (v.1.54d). The variability (SEM) observed among biological replicates indicated that the measurement of 100 cells for each biological replicate is sufficiently representative of the group variability for this parameter.

125,000 Oli-neuM cells/well were seeded in 12-well plates. After 48 h, GM was substituted with DM, to which treatment or vehicle was added. After 48 h, RNA extraction, quantification, and cDNA production were performed using RNA-Solv Reagent (R6830-01; VWR, Radnor, PA, United States). 2 μg of the RNA per sample was retro-transcribed following the High-Capacity cDNA Reverse Transcription Kit (4,368,814; Thermo Fisher Scientific Inc., Waltham, MA, United States) manufacturer’s instruction. Quantitative polymerase chain reaction (qPCR) was performed using SYBR Green Technology and the QuantStudio R 3 Real-Time PCR System (Applied Biosystems R, Thermo Fisher Scientific Inc., Waltham, MA, United States). Primer pairs used with StoS Quantitative Master Mix 2X SYBR Green-ROX (GeneSpin Srl, Milan, Italy) are indicated in Table 1. Specifically, Gapdh was used as an endogenous control to normalize data. 50 ng of cDNA per sample was used per reaction. qPCR was performed in triplicate in MicroAmp Fast Optical 96-Well Reaction Plate (Applied Biosystems R, Waltham, MA, USA). The 2^−ΔΔCT relative quantification method was used to determine fold change in expression. This consists of two main steps: normalization of threshold cycle (CT) values of the target mRNAs to the CT values of the endogenous control Gapdh, in the same samples (ΔCT = CT target–CT Gapdh) and further normalization to the control (ΔΔCT = ΔCT–ΔCT vehicle). The fold change in expression was then obtained as log₂(2^−ΔΔCT) and represented in the plots.

Typically, 2.75 × 10⁵ Oli-neuM cells were seeded in 6-well plates in GM media, until 70% confluence. Unless otherwise specified, cells were treated for 48 h in DM with the drugs indicated in the text. For immunoblot analyses the following Abs, diluted in TBS and 4% BSA, were used: anti-AMPKa1/2 (Santa Cruz Biotechnology, sc-74461), anti-p-AMPKa1/2 (Thr172) (Santa Cruz Biotechnology, sc-33524). Cell extract (CE) preparation and immunoblot analyses were performed as previously described (Porcu et al., 2015; Nocita et al., 2019). The band signal was detected by ChemiDoc™ Imaging System 12,003,153 (Bio-Rad) and intensity was estimated using the ImageJ 1.54d. The data were plotted on a graph using GraphPad Prism 7.0 (GrahPad Software, San Diego, CA, United States) as fold change versus vehicle, arbitrarily set to 1.

The Oli-neuM cells were typically seeded on Petri dishes and allowed to grow in GM for 24 h prior to media replacement with DM. The treatments are indicated in the text and were applied in DM for 48 h. The crude extract was obtained by scraping the plate with a Lysis Buffer (25 mM Tris–HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% NP-40, 5% glycerol, 0.1 μL/mL PMSF and 10 μL/mL PIM). Cells were lysed and total protein extraction was quantified using spectrophotometer Thermo Scientific NanoDrop™ 1,000 Spectrophotometer (Thermo Fisher Scientific Inc.; USA). To remove the non-specific binding 1 mg of total lysate was then incubated in a wheel with the Protein G PLUS-Agarose (Santa Cruz Biotechnology, sc-2002) beads for 90′ at 4°C. After centrifugation at 500 x g for 5′, the supernatant was removed and incubated at 4°C overnight in a wheel with 2 μg of anti-LKB1 Ab (Santa Cruz Biotechnology, sc-32245). Samples were further incubated with the Protein G PLUS-Agarose (Santa Cruz Biotechnology, sc-2002) beads for 2 h at 4°C and after centrifugation the immunoprecipitation (IP) pellet was resuspended in Washing Buffer (50 mM Tris–HCl pH 7.4, 150 mM NaCl, 5 mM EDTA, 1% NP-40, 5 mM MgCl₂, 0.1 μL/mL PMSF and 10 μL/mL PIM) three times before to SDS-loading buffer addition, meanwhile, the IP supernatant was kept at −20°C till SDS-PAGE analyses. IP pellets and IP supernatant were then mixed with SDS-loading buffer (6X) and incubated at 95°C for 5′. After SDS-PAGE electrophoresis western blotting and immunodetection were performed as previously described using Anti phosphorylated LKB1 Ab (Santa Cruz Biotechnology, sc-271924). Results were quantified on bands using ImageJ 1.54d.

In studies performed in multiwell plates (IF and qPCR), three experimental replicates are present in each plate, and the mean values obtained from data analysis were considered as one biological replicate. The effects of each drug treatment ratioed its internal control (vehicle) in IF experiments, WB, and qPCR data were analyzed to determine statistically significant differences among multiple single or combined treatments by paired two-tailed Student’s t-test and one-way analysis of variance (ANOVA) with Tukey’s tests, respectively. For IF quantitative analyses 25 images per well were acquired and the mean values of at least three wells (one biological replicate) were normalized toward an internal control, typically the vehicle (DM+ DMSO 0.5% max). The mean values ± SEM of three biological replicates were calculated and plotted on the graph using GraphPad Prism version 7.00 for Windows (GraphPad Software, Boston, Massachusetts USA, www.graphpad.com). Statistical analyses were performed using GraphPad Prism tools two tailed t test and one-way ANOVA multiple comparisons.

Docking calculations were performed using the 3D structure of human Smo (hSmo) in complex with cholesterol (PDB code: 5L7D, Byrne et al., 2016). This choice was driven by the chemical similarity between cholesterol and the investigated ligand, Budesonide. Notably, in this structure, Smo is in an inactive conformation, suitable for docking compounds with pharmacological activity like Budesonide, due to mutations which were reverted to the wild-type amino acids prior calculations. The receptor conformation was prepared using the Protein Preparation Wizard tool, implemented in the Maestro Suite 2021 (Madhavi Sastry et al., 2013). Any missing residues were added and conformationally optimized using the Prime toolkit (Jacobson et al., 2002, 2004). The missing intracellular loops (ICL) 2 and 3 were built for homology using the PDB structures 5L7I (Byrne et al., 2016) and 4JKV (Wang et al., 2013) as templates, respectively. Correct bond orders were assigned, missing hydrogen atoms were added, and all water molecules were deleted from the receptor structure. Protonation and tautomeric states at pH 7.4 were assigned to the side chains using Epik (Johnston et al., 2023). Finally, the positions of all the hydrogens were minimized. The docking search box (20 Å × 20 Å × 20 Å) was centered around the position of cholesterol in the experimental complex. Docking calculations were performed employing the Glide (version 9.3) SP protocol and the OPLS3A force field.

Prior to Molecular Dynamics (MD) simulation, the Smo N-terminus (P57) and C-terminus (W549) were capped with acetyl and N-methyl groups, respectively. For comparison, the apo-Smo system was obtained by removing the ligand from the same structure used for docking Budesonide (PDB code: 5L7D) and following the same preparation protocol described in the previous paragraph. For each system, the receptor was then embedded in a 110 Å x 110 Å (along the x and y axes) pre-equilibrated 1-palmitoyl-2-oleoylphosphatidylcholine (POPC)—cholesterol (7:3 molar ratio) bilayer and solvated using the TIP3P water model with the aid of the membrane-builder tool of CHARMM-GUI.org. The ff14SB (Maier et al., 2015) and lipid17 Amber force fields were used to parametrize the protein and the lipids, respectively. The bonded and van der Waals parameters for Budesonide were taken from the GAFF force field (Wang et al., 2004), whereas its atomic partial charges were computed using the two-stage restrained electrostatic potential (RESP) fitting procedure (Bayly et al., 1993) implemented in Antechamber (Wang et al., 2006). The required ESP was calculated using the Gaussian package (Frisch et al., 2009) at the Hartree-Fock level of theory with the 6-31G* basis set, following a geometry optimization performed with the B3LYP functional.

The GROMACS 2020.6 code (Berendsen et al., 1995) was used to perform the simulations. A cutoff of 12 Å was used for short-range interactions. The long-range electrostatic interactions were computed using the particle mesh Ewald method with a 1.0 Å grid spacing under periodic boundary conditions. The non-iterative LINCS (Hess et al., 1997) algorithm was applied to constrain bonds, allowing the use of a 2 fs integration time step. To resolve all steric clashes, each system underwent 30,000 steps of steepest descent energy minimization in three phases. In the first phase, the system’s heavy atoms were kept fixed to relax only the hydrogens and water molecules; during the second phase, the lipid bilayer was also released; and in the third phase, all atomic positions were minimized. Each complex was then equilibrated and heated to 300 K, alternating NPT and NVT cycles (for a total of 30 ns) with the Berendsen coupling bath and barostat (Berendsen et al., 1984), while gradually decreasing harmonic constraints on the heavy atoms of the membrane, protein, and ligands. During the production runs, the pressure of 1 atm and the temperature of 300 K were maintained with the stochastic velocity rescaling (Berendsen et al., 1984) and the Parrinello and Rahman (1981) algorithms, respectively. The stability of the MD simulated system was assessed by monitoring the protein RMSD with respect to the starting conformations over simulation time (Supplementary Figure S1).

Network parameters such as hubs, communities, and structural communication analyses were obtained by using the WebPSN 2.0 web-server (Felline et al., 2020).

The methodology builds the Protein Structure Graph (PSG) based on the interaction strength of two connected nodes, according to Equation 1:

where interaction percentage (I_ij) of nodes i and j represents the number of pairs of side-chain atoms within a given cut-off value (4.5 Å), while N_i and N_j are normalization factors. The interaction strength (represented as a percentage) between residues i and j (I_ij) is calculated for all node pairs. If I_ij is more than the minimum interaction strength cutoff (I_min) among the residue pairs, then it is considered to be interacting and hence represented as a connection in the PSG.

The O4 marker is found on the surface of oligodendrocyte progenitors and is commonly used as the earliest recognized marker specific for the oligodendroglial lineage. Previous research showed that Budesonide can promote O4 expression in the Oli-neu mouse oligodendroglia cell line (Joubert et al., 2010). However, further studies have not yet addressed whether Budesonide treatment can promote OPC differentiation till axon remyelination. This latter step is essential for drug efficacy in remyelination. To investigate the Budesonide potential for remyelination therapies, we used the Oli-neuM cell line. Unlike Oli-neu (De Vries and Boullerne, 2010), Oli-neuM cells, by constitutively expressing MyRF gene, can differentiate till they engage and myelinate synthetic axons upon promyelinating drug treatment (Porcu et al., 2015; Nocita et al., 2019).

We first established the effects of Budesonide on MBP expression in dose–response experiments, starting from the minimal effective concentration we have previously established in Oli-neuM cells (Porcu et al., 2015). Therefore, we tested Budesonide at concentrations of 0.1, 1 or 10 μM, using 10 μM Clobetasol as positive control or vehicle (max 0.5% DMSO) as negative control. Treatments were applied in differentiation medium (DM) for 48 h, after which the cells were processed for IF microscopy (Figures 1A,B) or qPCR analysis (Figures 1C–E). A clear dose–response relationship was observed in both IF and qPCR experiments for MBP (Figures 1B,C). To further validate these findings, we also analyzed the expression levels of the tetraspan myelin proteolipid protein (PLP), one of the most abundant proteins of compact myelin in the CNS, as well as of the myelin and lymphocyte protein MAL, a conserved tetraspan proteolipid component of both peripheral nervous system (PNS) and CNS myelin. Both Plp and Mal showed a significant increase in their expression compared to the vehicle upon 48 h of treatment with 10 μM Budesonide (Figure 1D).

To determine if Budesonide stimulates or inhibits Smo activity under our experimental condition we performed a dose–response experiment to measure Gli1 expression levels by qPCR (Figure 1E). As expected, Budesonide acts as an inhibitor of Smo activity: Gli1 is downregulated by Budesonide in a dose-dependent manner. Being 10 μM Budesonide the maximal Gli1 inhibition dose observed.

These results confirm that Budesonide treatment inhibits Smo activity while simultaneously promoting Oli-neuM differentiation and myelin gene expression.

To determine if Budesonide treatment could induce OPCs differentiation till axon engagement (Lee et al., 2012) we cultured Oli-neuM cells in chambers containing parallel electrospun aligned PS microfibers with a 4 μm diameter (here following named “synthetic axons”), as we previously described (Del Giovane et al., 2022). After 72 h of treatment with 10 μM Budesonide or vehicle in DM media, Oli-neuM cells were fixed and processed for IF confocal microscopy analyses (Figure 2). We observed that vehicle-treated cells poorly engaged the PS fibers compared to Budesonide-treated cells, although they were plated at the same concentration and treated in parallel experiments. Indeed, in this and previous experiments, we observed that a number of them, not engaging fibers, spontaneously undergo apoptosis. On the contrary, those cells engaging synthetic axons appear stable and extend the processes along the fibers expressing MBP (Figure 2A). To estimate the number of cells engaging the fibers (Figure 2B), 75 images were acquired randomly for each sample in each of the three biological replicates (3n). Cells were counted within 86 μm from the center of the nearby fiber. Those with the nuclei attached or on the PS fiber were considered engaged, the others not engaged (Figure 2B left panel). We also quantified the length of the membrane processes, expressing MBP, that extend onto fibers. The lengthening of the membrane over the fiber was expressed in micrometres and plotted on the graph as the mean value (±SEM) of three biological replicates (Figure 2B right panel). Both the percentage of engaging cells and the length of processes extending onto the PS fibers were significantly increased by Budesonide treatment.

Altogether these data conclusively show that Budesonide rapidly promotes Oli-neuM differentiation within 72 h of treatment and increases the capacity of Oli-neuM to contact axons and effectively wrap them as shown in the 3D reconstruction (Figure 2C) obtained using the surface tool of Imaris software, allowing to visualize the cell surface of a typical cell treated with vehicle or Budesonide, wrapping the PS fibers with their processes.

Since Budesonide binds to Smo CRD but also regulates glucocorticoid receptor (GR), we wished to clarify the contribution of each to MBP expression. Toward this aim, we investigated the impact of Smo on Budesonide-induced MBP expression by utilizing two isogenic cell lines, the Oli-neuM^shSmo and the Oli-neuM^shCTL, previously created by infection with Lentivirus particles carrying, respectively, the pLKO.1 vector expressing Smo shRNA (shSmo) and a pLKO.1 vector expressing a control shRNA (shCTL) (Del Giovane et al., 2022).

Oli-neuM^shSmo and the Oli-neuM^shCTL cells were treated with 10 μM Budesonide, 5 μM SAG, 10 μM Clobetasol or vehicle (DMSO<0.5%) in DM media for 48 h and processed in parallel for quantitative automated IF microscope analysis as described in Materials and Methods. 10 μM Clobetasol or SAG 5 μM were used as a positive or negative control, respectively. Clobetasol treatment is known to induce MBP by binding GR while the second stimulates Gli1 but not MBP in Oli-neuM (Del Giovane et al., 2022). Images were analyzed and the mean data of three biological replicates were plotted in the graph and statistically analyzed (Figure 3A and Supplementary Figure S2).

Results showed that both Clobetasol and Budesonide treatments upregulate MBP protein levels in Oli-neuM^shCTLand Oli-neuM^shSmo cells. However, Budesonide activity is significantly reduced in Oli-neuM^shSmo treated cells compared to Oli-neuM^shCTL indicating that Smo gene silencing markedly reduces MBP levels in Budesonide-treated cells. This effect is below significance in the presence of Clobetasol or vehicle treatment (Figure 3A). To corroborate these data, we have analyzed Mbp gene expression in Oli-neuM^shCTL and Oli-neuM^shSmo cells upon Budesonide and vehicle treatments (Figure 3B). We observed a striking reduction of Mbp mRNA in Oli-neuM^shSmo compared to Oli-neuM^shCTL Budesonide treated cells while the vehicle did not show significant differences (Figure 3B). These data suggest that MBP mRNA is more sensitive to treatment than MBP protein levels. This difference could result from the distinct turnover rates of the mRNA and protein.

To confirm the capacity of Budesonide to act as an inhibitor of Smo activity in our experimental conditions we analyzed the effect on Gli1 gene expression under Budesonide compared to Vismodegib treatments in Oli-neuM. Vismodegib is a well-characterized competitive antagonist of Smo activity and it binds to the receptor orthosteric site within the TMD bundle (Byrne et al., 2016; Huang et al., 2018). In preliminary experiments, we observed that the minimal effective Gli1 inhibitory dose in qPCR of Vismodegib is 0.1 μM (Figure 3C). Compared to 0.1 μM Vismodegib, 10 μM Budesonide has similar inhibitory activity on Gli1 gene expression (Figure 3C left panel).

We then wondered if the combined use of two Smo inhibitors, namely 0.1 μM Vismodegib and 10 μM Budesonide, could have combinatorial effects on MBP expression (Figure 3C right panel). Indeed, the combined use of 10 μM Budesonide and 0.1 μM Vismodegib reduces Mbp gene expression compared to 10 μM Budesonide alone indicating that they partially obstacle each other action. To confirm that this effect is reflected also on the MBP cytosolic levels we determined by IF analyses the effects on MBP levels of the combined use of 10 μM Budesonide with dose increase of Vismodegib (0.1, 1, or 10 μM; Figure 4). Clearly, at the increase of Vismodegib concentration, the efficacy of Budesonide to promote MBP increase diminishes. This latter result supports the view that Vismodegib by binding to the Smo TMD might contribute to reduce Budesonide access to the CRD, thereby reducing its efficacy in stimulating MBP.

We concluded that Budesonide acts as an inhibitor of Smo canonical pathway in Oli-neuM and through Smo promotes MBP gene expression (Figure 3) and consequently an increase in the MBP cytosolic levels (Figures 3, 4).

The molecular signaling through which Budesonide treatment stimulates MBP gene expression via Smo binding remained to be clarified.

In a previous study, we demonstrated that the quinolone GSA-10 activates MBP gene expression through AMPK activation (Del Giovane et al., 2022). Additionally, various ligands can activate non-canonical pathways by binding to Smo, resulting in AMPK activation in response to changes in nutrient intake or environmental conditions (Akhshi et al., 2022). Therefore, we hypothesized that also Budesonide might trigger this pathway to promote AMPK phosphorylation.

AMPK is a kinase involved in various intracellular processes and cytoskeleton rearrangement (Neumann et al., 2019). It is important to note that Gli1 is a known target of activated AMPK (Teperino et al., 2012). When activated, AMPK phosphorylates Gli1 at serine/threonine residues (Ser102, Ser408 and Thr1074), decreasing both its transcriptional activity and protein stability (Li et al., 2015; Di Magno et al., 2016). Conversely, it increases Gli1 cytoplasmic localization and its interaction with the E3 ubiquitin ligase β-TrCP, leading to Gli1 degradation by the proteasome (Zhang et al., 2017).

Therefore, we analyzed the effects of Budesonide treatment alone or in combination with the selective phospho-AMPK inhibitor Dorsomorphin (DRS) (Lo et al., 2019; Sun et al., 2021) on AMPK activation and MBP expression. 3 μM DRS either alone or in combination with 10 μM Budesonide was used to treat the cells for 48 h. After protein extraction, AMPK activity was determined by Immunoblotting (IB) analyses (Figure 5A). The results revealed that 48 h of Budesonide treatment induces AMPK phosphorylation, while the combined use of DRS substantially reduces phospho-AMPK levels to those observed in control conditions (Figure 5A) and MBP expression (Figure 5B). These data confirmed that 10 μM Budesonide treatment promotes AMPK phosphorylation after 48 h treatment and this signal is required for MBP expression.

In their 2012 study, Teperino et al. (2012) demonstrated that LKB1 can serve as an upstream regulator of AMPK phosphorylation via Smo. Therefore, we analyzed the effects of 10 μM Budesonide treatment on LKB1 activation in total lysates of Oli-neuM. Immunoblot quantification of the mean relative amounts of pLKB and LKB present in the total extracts (n = 3) reveals a significant increase of LKB phosphorylation in Budesonide-treated cells compared to vehicle (Figure 5C). To better visualize activated LKB1 we immunoprecipitated the LKB1 protein from total lysates of Oli-neuM cells treated as above (Figure 5D). Anti-LKB1 Ab (IP_αLKB1) or, as a mock reaction, with anti-FLAG Ab (IP_αFLAG) was used to IP the LKB1 protein. IP pellets and Supernatants were then processed for IB analyses using anti-PhoLKB1and anti-LKB1 Ab in parallel experiments (Figure 5D). Clearly, the IP_αLKB1 pellets of Budesonide-treated cells contained more phosphorylated LKB1 compared to vehicle-treated IP_αLKB1 pellets (Figure 5D). This result is specific as no phosphorylated LKB1 could be pelleted in the mock IP reaction (IP_αFLAG).

Altogether, the results show that Budesonide promotes MBP expression via a signaling that requires AMPK phosphorylation. Given the known role of LKB1 to act downstream of Smo and upstream of AMPK our data support the view that Budesonide treatment stimulates both kinase phosphorylation on its way to promote MBP expression.

Since Budesonide exerts its anti-inflammatory activity by binding to GRα subunit, we wondered if both GR and Smo receptors are involved in the stimulation of MBP gene expression. To address this, we determined the effects of Mifepristone alone or in combination with Budesonide in Oli-neuM after 48 h treatment. Mifepristone is an anti-progesterone drug that inhibits GR activity by binding to the GRα subunit (Pelaia et al., 2016; Caramori et al., 2022). In these experiments, Oli-neuM cells were plated for 24 h in GM, prior to being treated with 10 μM Budesonide, 0.1 μM Mifepristone, with both drugs or with vehicle control for 48 h in DM. Clobetasol was used as a positive control for MBP expression. Total lysates were processed for IB analysis as described in Material and Methods (Figure 6A). As expected, IB analyses showed a significant increase in MBP under Budesonide and Clobetasol treatment. No effect was observed under Mifepristone treatment compared to the vehicle. The co-treatment of Budesonide and Mifepristone (Bud+Mife) impaired the stimulatory effects of Budesonide on MBP levels (Figure 6A respective lanes). We then ascertain if Mifepristone had any effect on reducing Budesonide-mediated MBP gene transcription through qPCR (Figure 6B). We observed that while Clobetasol (Clob) and Budesonide (Bud) stimulate MBP gene expression, the combination treatment with Budesonide plus Mifepristone did not alter this pattern.

Overall, IB and qPCR data shows that Budesonide’s stimulation of MBP in Oli-neuM cells involves both GR and Smo receptors, however GR controls solely MBP protein levels but not MBP gene expression.

To further clarify how Budesonide can antagonize Smo canonical pathway, we performed extensive computational investigations. Given the importance of Smo as anticancer target, several crystallographic structures of the receptor bound to either agonists or antagonists are available for modeling studies (Rana et al., 2013; Byrne et al., 2016; Luchetti et al., 2016; Xiao et al., 2017). These structures allow identifying two main Smo domains to which receptor’s ligands can bind: the extracellular CRD and the seven-helix domain (TMD). The CRD is essential for the localization at the primary cilium (Rana et al., 2013; Byrne et al., 2016; Zhang et al., 2021), and it interacts with two major groups of small molecules: hydroxylsteroids and GCs. TMD can interact with both agonists such as SAG and antagonists, including Cyclopamine, Vismodegib and Smoothened antagonist (SANT-1) (Byrne et al., 2016; Luchetti et al., 2016). Notably, cholesterol and oxysterols have been shown to act as endogenous activators of Smo by engaging the extracellular CRD and, in turn, to positively regulate Smo/Gli1 signaling (Luchetti et al., 2016).

Despite the well-known capability of Budesonide and of other GCs to bind the CRD domain of hSmo, the precise structural details of this ligand-receptor interaction remain unclear (Rana et al., 2013). To date, only one structural study has been conducted by Huang et al. (2016), who used Nuclear Magnetic Resonance (NMR) experiments to outline the binding region of this corticosteroid in the CRD domain of both human and Drosophila Smo; however, the atomistic model of the binary complex was only released for the Drosophila variant. To get more insights on how GCs binding to Smo CRD can result in the inhibition of Smo/Gli1 signaling while promoting AMPK activation and MBP expression and to elucidate the molecular determinants of the Budesonide-hSmo interaction, we analyzed Budesonide’s binding mode within the hSmo CRD in silico.

Initially, molecular docking calculations were conducted using Glide 9.3 software and employing as a target the experimental 3D structure of hSmo in complex with cholesterol bound at CRD, due to its chemical similarity with Budesonide (see Materials and Methods for details). The lowest-energy and most recurring docking pose shows Budesonide occupying the same lipophilic cleft as cholesterol, delineated by residues L112, I156, V157, I496, and L489 (Supplementary Figure S5).

Then, to better evaluate the energetics of the predicted binding mode and account for solvation and protein flexibility effects, extensive MD simulations (3 μs long) were performed on the BUD-hSmo docking complex. The analysis of the ligand RMSD over time revealed a rearrangement during the initial microsecond of simulation, after which Budesonide reaches a conformation conserved for the rest of the simulated timescale (»2 μs) (Figure 7). Interestingly, the final binding mode of Budesonide closely resembles that of cholesterol in experimental complexes (Figures 8A,B). In detail, the pose is stabilized by a set of lipophilic contacts between the ligand’s steroid nucleus and the sidechains of L108, W109, L112, Y130, I156, V157, and P164, as well as by two hydrogen bonds detected between Budesonide’s 21-hydroxyl group and the backbone carbonyl moieties of I156 and V157 (Figures 7B, 8B). Notably, the predicted binding model aligns well with the available experimental data on the BUD-hSmo interaction. As shown in Figure 8, the ligand is indeed surrounded by the amino acids (L108, W109, G111, L112, G162 and W163) that exhibited the most significant chemical shift changes in the NMR experiments performed by Huang et al. (2016).

To investigate the relationship between the Budesonide’s binding mode and its pharmacological activity, a second MD simulation was performed on the apo-hSmo receptor to compare how Budesonide binding can affect the GPCR dynamics relative to its basal unliganded state. We particularly focused on the allosteric networks known to govern GPCR functioning by using a graph-based approach, namely protein structure network analysis (PSN) (see Materials and Methods for details), which identifies the strongest inter-residue interactions along an MD trajectory (hereafter referred to as to metapath). In both the BUD-hSmo system and the apo receptor, PSN detected long-range communications between the Budesonide/Cholesterol binding pocket in the CRD and the intracellular termini of TM5 and TM6, in line with the established regulatory role of the CRD in the activation process of Smo (Huang et al., 2018). However, interesting differences were observed in the receptor areas touched by the identified communication networks. Specifically, in the presence of Budesonide, the metapath shows a more direct involvement of the upper portions of TM5 and TM6, with a series of communication hubs (residues D473, Q477, W480, Y487, and Q491) detected exclusively under these conditions. Conversely, in the apo receptor, the interdomain edges primarily involve residues from the ECL1 and ECL2 loops.

In parallel, the distinct behavior of the receptor in the two systems was also highlighted by the analysis of CRD movements over the MD trajectories. As evidenced by the RMSD plots shown in Figure 9A, Budesonide appears to reduce the fluctuations of the CRD relative to the TMD compared to the apo receptor (Figure 9B). At an atomistic level, this stabilization is due to a more stable network of electrostatic interactions in the BUD-hSmo system, connecting the top part of the TMD and linker domains with the CRD. In line with previous computational studies (Bansal et al., 2023), we identified five salt bridges stabilizing this interdomains connection: R485-E160, R485-D209, R159-E208, R296-E305 and R512-E226 (Figure 9B). Indeed, while these contacts were maintained for most of the simulation in the BUD-hSmo trajectory, three of them (R485-E160, R485-D209, R159-E208) were gradually lost in the apo receptor, explaining the greater RMSD fluctuation observed in absence of the ligand (Figure 9B).

Altogether, our results suggest that Budesonide exerts its unique regulatory activity on Smo by promoting the compaction of the CRD domain onto the TMD and enhancing the allosteric communication between the CRD and helices TM5 and TM6. These findings, along with the detailed atomistic model of the Budesonide-hSmo interaction, may have significant implications in drug discovery, potentially guiding the design and development of new compounds with promyelinating properties like those of Budesonide.