A Structural Mechanism for Noncanonical GPCR Signal Transduction in the Hedgehog Pathway

Abstract

The Hedgehog (Hh) signaling pathway is fundamental to embryogenesis, tissue homeostasis, and cancer. Hh signals are transduced via an unusual mechanism: upon agonist-induced phosphorylation, the noncanonical G protein-coupled receptor SMOOTHENED (SMO) binds the catalytic subunit of protein kinase A (PKA-C) and physically blocks its enzymatic activity. By combining computational structural approaches with biochemical and functional studies, we show that SMO mimics strategies prevalent in canonical GPCR and PKA signaling complexes, despite little sequence or secondary structural homology. An intrinsically disordered region of SMO binds the PKA-C active site, resembling the PKA regulatory subunit (PKA-R) / PKA-C holoenzyme, while the SMO transmembrane domain binds a conserved PKA-C interaction hub, similar to other GPCR-effector complexes. In contrast with prevailing GPCR signal transduction models, phosphorylation of SMO promotes intramolecular electrostatic interactions that stabilize key structural elements within the SMO cytoplasmic domain, thereby remodeling it into a PKA-inhibiting conformation. Our work provides a structural mechanism for a central step in the Hh cascade and defines a paradigm for disordered GPCR domains to transmit signals intracellularly.

Extended Data Fig. 1:. SMO binds PKA-C via its intrinsically disordered C-terminus.

a, CD analysis of the purified SMO pCT, spanning residues 565-657 (40 [μM). b, AlphaFold 3 models of SMO from the phosphorylated SMO / PKA-C complex (blue) aligned with the AlphaFold 3 model of SMO alone, with phosphorylation sites included (green), or the AlphaFold 2.3.0 model of SMO alone, downloaded from the EBI protein structure database, in which phosphorylation sites were not explicitly specified (brown), c, Comparison of different SMO / PKA-C conformational states captured by AlphaFold 3 (blue) vs AlphaFold 2.3.0 (magenta or teal for conformations 1 or 2 (Conf1 or Conf2), respectively) (see Supplementary Discussion 2). Conf1 and Conf2 display near-identical positions for the PKI-like helix, RII-like helix, and reentrant loop, but with the PKA-C rotated such that it presents a new surface to the SMO 7TM domain. ipTM scores for Conf1 and Conf2 are 0.84 and 0.80, respectively. d, Left: empirical structures of agonist (SAG21k)-bound, active SMO (PDB: 6O3C) and PKA-C (PDB: 1ATP). Right: AlphaFold 3 model of the SMO / PKA-C complex. Note that in the AlphaFold model, the PKA-C N-terminus is pointing upwards the membrane.

Extended Data Fig. 2:. Evolutionary conservation of sequence and structural elements in SMO / PKA-C complexes.

a, Multiple sequence alignment of helix 8 and pCT domains from SMO orthologs. Conserved K/R residues (blue circles), GRK2/3 phosphorylation sites (orange), and other essential residues for SMO / PKA-C signaling (green, see main text for detailed explanation) are indicated above the alignment, while secondary structural elements defined here or in prior studies are indicated below, b, Disulfide trapping of PKA-Cα with soluble SMO pCT L637C (i.e., lacking the CRD and 7TM domains), performed as in Fig. 2b. The SMO pCT has no endogenous cysteines, so trapping must be with L637C (the engineered Cys at P+2). c, Comparable disulfide trapping occurs with a PKA-C C343S mutant. Because PKA-Cα has only two endogenous cysteines (C199 and C343), this result indicates that disulfide trapping involves PKA-C C199.

Extended Data Fig. 3:. SMO/PKA-C complex preparation and cryoEM data processing.

a, Size-exclusion chromatography and SDS-PAGE analysis of BS3-crosslinked SMO/PKA-C complex. b, Workflow of cryoEM data processing. c, Representative cryoEM micrograph (scale bar: 50 nm). d, Representative 2D class averages (scale bar: 10 nm). e, Gold-standard FSC curves of the EM map. f, Angular distribution plot of final particles.

Extended Data Fig. 4:. SMO and PKA-C sequence coverage in HDX-MS, and summary statistics of HDX-MS runs.

a, Left: HDX-MS sequence coverage of the SMO L637C construct presented as a “snake plot”. Residues covered by the MS measurements are colored blue, while residues that were not detected by MS are colored white. TM helices are numbered in navy. b, HDX-MS sequence coverage of PKA-C. Colors are as in (a). c, Summary of protein coverage, # of peptides, and redundancy for MS runs. d, list of MS runs presented in this study (see Methods for details on sample preparation). e, Back-exchange for SMO 541-546, the most extensively deuterated peptide in our data set, was measured at 19.8% (see Methods).

Extended Data Fig. 5:. HDX-MS studies of phosphorylated SMO / PKA-C complexes vs. unbound proteins.

Deuterium exchange difference (average number of deuterons) mapped for SMO (a) or PKA-C (b) in the phosphorylated SMO (pSMO) / PKA-C complex vs. the unbound protein, at the indicated timepoints following deuterium exchange. Key peptides are indicated on the plot, Negative differences denote decreased exchange (i.e., increased protection) in the complexed vs uncomplexed forms. Domain organization of SMO and PKA-C are indicated below each plot: CRD = cysteine rich domain, TM = transmembrane helix (numbered 1-7), pCT = membrane proximal C-tail. Standard deviations from replicate measurements are in gray. c, Mapping of HDX-MS results onto the AlphaFold3 model of the SMO / PKA-C complex. Peptides in SMO showing decreased or increased exchange (i.e, increased or decreased protection) are shown in blue and purple, respectively, while peptides in PKA-C showing decreased or increased exchange are shown in yellow and teal, respectively. d, Superposition of PKA-C complexes with wild-type SMO (blue) or SMO L637C (salmon), with key peripheral SMO structural elements indicated. P-site alanine is red. Although the PKI-like helix and RII-like helix are near-superimposable in both models, the reentrant loop is in slightly different positions within the SMO 7TM cavity; however, this is within the range of variability observed for this region even in the wild-type protein (see Fig. 6d, Extended Data Fig. 10c). Representative HDX-MS mass spectral envelope (t = 10 min) (e) or uptake vs. time plots (f) of the indicated peptides within the PKA-C active-site cleft or the SMO inhibitor sequence.

Extended Data Fig. 6:. Additional biochemical, cell biological, and functional studies of the SMO RII-like helix.

a, HDX-MS uptake vs. time plots for the indicated peptides in the SMO RII-like helix (top) and PKA-C hinge region (bottom). b, SDS-PAGE analysis of diamide-induced disulfide trapping between a minimal-cysteine PKA-C and SMO A574C, as in Fig. 3e. c, GLI transcriptional reporter assay was performed on the indicated SMO wild-type or mutant constructs following transfection into Smo^−/− MEFs. The assay was conducted as in Fig. 4a, except using the direct SMO agonist SAG21k (green) vs a vehicle control (black). d, Locations of SMO V626 and V630 (red), and the residues with which they interact, at the interface between the PKI-like helix and the RII-like helix. e, SAG21k-induced phosphorylation of the indicated myc-tagged SMO wild-type and mutant constructs (or an empty vector control) was assessed following stable transfection into HEK293 cells, isolation of SMO-myc from cell lysates on anti-myc nanobody resin, and blotting with anti-pSMO antibody (top) and anti-myc antibody (bottom). Treatment of wild-type SMO with the GRK2/3 inhibitor Compound 101 (Cmpd101) serves as a negative control. Note that although several of these SMO mutants exhibit stronger phosphorylation than the wild-type protein, none of them exhibit less phosphorylation, making it unlikely that their deficits in GLI transcriptional reporter assays arise from an inability to undergo GRK2/3-mediated phosphorylation. f, Ciliary localization of the indicated FLAG-tagged SMO constructs was assessed following transfection into NIH3T3 cells, followed by SAG21k treatment (to induce SMO ciliary accumulation) and antibody staining. Left: representative images of selected SMO wild-type or mutant constructs, with FLAG-SMO (anti-FLAG, green), cilia (acetylated tubulin (AcTub), magenta; Arl13b, yellow) and nuclei (DAPI, blue). SMO lacking its entire C-terminus (SMOΔCT), which cannot localize to primary cilia, serves as a negative control. Right: intensity-based quantification of FLAG-SMO localization to primary cilia. Note that although some of the mutants display a modest (<50%) decrease in ciliary localization compared to wild-type SMO, this is not sufficient to account for the near-complete loss of GLI activation in the transcriptional reporter assays. Scale bar = 10 microns.

Extended Data Fig. 7:. SPR studies of wild-type and mutant SMO / PKA-C complexes.

a, SPR sensorgram for binding of myristylated PKA-Cα, in concentrations ranging from 0.016 μM to 1 μM, to phosphorylated wild-type (WT) SMO or the indicated SMO mutants reconstituted into nanodiscs. b, Left: SPR sensorgram for 250 nM PKA-Cɑ binding to wild-type SMO (green) but displaying no binding to a negative control version of SMO lacking the entire C-terminus (SMOΔCT, magenta), consistent with prior findings^,. Right: Quantification of SMO WT or SMOΔCT SPR binding signal from n=3 replicates.

Extended Data Fig. 8:. Additional biochemical and functional studies of SMO phosphorylation.

a, Targeted MS-based quantification of phosphorylation of a phosphopeptide including pSMO 615 (left) and total SMO protein in each sample (middle), using FLAG-SMO protein isolated from HEK293 cells treated with vehicle, SMO agonist SAG21k, SMO inverse agonist KAADcyc, or GRK2/3 inhibitor Cmpd101. SMO sequence diagram indicating the location of S615 (yellow) along with the GRK2/3 phosphorylation sites previously mapped via MS, is shown at right. “Intensity” is a measurement of the abundance of phosphorylation sites (left) or total protein (right), derived from model-based estimation in Msstats which combines individual peptide intensities. Inset depicts SMO pCT sequence with new (yellow) and previously mapped (orange) GRK2/3 phosphorylation sites. b, Location of GRK2/3-phosphorylated S/T residues and K/R residues in the alternative conformation of the SMO / PKA-C complex (AlphaFold 2.3.0 Conf2, see Extended Data Fig. 1c). c, GLI transcriptional reporter assay on Smo^−/− MEFs transfected with the indicated SMO mutants and treated/analyzed as in Fig. 5c. SMO sequence from 561-581 is indicated, and residues mutated in this experiment are underlined.

Extended Data Fig. 9:. HDX-MS and disulfide trapping studies to monitor effects of phosphorylation on SMO / PKA-C complexes.

Deuterium exchange difference (average number of deuterons) mapped for SMO (a) or PKA-C (b) in the phosphorylated vs nonphosphorylated SMO / PKA-C complex. Data are presented as in Extended Data Fig. 5. c, HDX-MS uptake vs. time plots for the indicated peptide in SMO helix 8. d, Disulfide trapping of SAG21k-bound, phosphorylated vs nonphosphorylated SMO A574C / PKA-C complex, performed as in Fig. 5e. Quantification is below the gel and represents the mean +/− s.d. from 3 replicates. e, HDX-MS uptake vs. time plots for the indicated peptide in the SMO region C-terminal to the pseudosubstrate motif. GRK2 phosphorylation sites in SMO, and CaMKII phosphorylation sites in RyR, are highlighted in yellow. f, HDX-MS uptake vs. time plots for the indicated peptide C-terminal to the pseudosubstrate motif (see e). Bimodal deconvolution of the peptide revealed high-exchanging (blue) and low-exchanging (green) populations, suggesting that this region may interact with PKA-C via a conformational selection mechanism. g, Overlay of top 5 AlphaFold 3 models indicating a propensity of the region C-terminal to the pseudosubstrate motif to wrap around PKA-C in some cases.

Extended Data Fig. 10:. The SMO / PKA-C complex structurally mimics other GPCR-effector complexes.

a, Comparison of the SMO / PKA-C complex to SMO or other GPCRs bound to the indicated proteins. In each structure, the GPCR portion is colored cyan, and the structural elements that engage the intracellular 7TM cavity are colored maroon. PDB numbers: SMO/NbSmo8 (6O3C), m2AchR/Barr1 (6U1N), Rhodopsin/GRK1 (7MT9), SMO/Gi (6XBL). b, Left: AlphaFold model of complex between a C-terminally truncated SMO construct (“7TM”) and the reentrant loop in the pCT, modeled as separate sequences. Right: AlphaFold model of near-full-length SMO (including the entire pCT) from the SMO / PKA-C complex, for comparison. The reentrant loop sequence is colored maroon in both models. c, Left: overlay of the following GPCR / β-arrestin structures, aligned on the β-arrestin: β1-adrenergic receptor /β-arrestin1 (6TKO), M2AchR / β-arrestin1 (6U1N), neurotensin receptor / β-arrestin1 (6UP7). Right: overlay of top 24 AlphaFold models of SMO / PKA-C complex (light blue), along with conf1 (pink) and conf2 (cyan) as shown in Extended Data Fig. 1c.

Extended Data Fig. 11:. Influence of SMO ICLs and membrane lipids on SMO / PKA-C interactions.

a, Left: representative HDX-MS mass spectral envelope (at t_ex = 5 min) for PKA-C 247-261 in the αG-αH loop. Right: αG-αH sequence protected in the SMO / PKA-C complex mapped onto conf2 of the SMO / PKA-C complex. b, Uptake vs. time plots of PKA-C 313-327. c, Electrostatic surface potential for each conformation (conf1, conf2) of the SMO / PKA-C complex. Scale is shown at bottom.

Extended Data Fig. 12:. Allosteric changes in SMO induced by PKA-C binding.

HDX-MS analysis of the indicated peptide in the TM5-TM6 region (a) or CRD (b) in phosphorylated SMO (pSMO) alone (top spectrum in each pair), or in complex with PKA-C (bottom spectrum in each pair). In each panel, representative HDX-MS mass spectral envelope and uptake vs. time plots are indicated at left, and regions of SMO that have undergone increased (purple) or decreased (dark blue) deuterium exchange upon SMO / PKA-C interaction are mapped onto the SMO-SAG21k-NbSmo8 structure (PDB: 6O3C).

Figure 1:. AlphaFold modeling of the SMO / PKA-C complex.

a, AlphaFold 3 model of phosphorylated mouse SMO in complex with mouse PKA-Cɑ (rendered as a surface with small N-lobe in white and large C-lobe in olive). Positions of the SMO extracellular cysteine rich domain (CRD), transmembrane spanning domain (7TM), and proximal cytoplasmic tail (pCT) are indicated, and approximate location of membrane is shown in gray. The PKI-like helix, inhibitor sequence (seq), RII-like helix, and reentrant loop are labeled (see main text). The AlphaFold 3 predicted local confidence (pLDDT) score (0-100, with higher values indicating greater confidence) is represented by a color scale (upper left). The turquoise sphere in SMO’s inhibitor sequence represents the P-site. b, Sequence of SMO helix 8 and pCT. Secondary structural elements are underlined, GRK2/3 phosphorylation sites are highlighted in orange, and key amino acids in the pseudosubstrate motif are labeled in red. c, Structural snapshots of the complex between phosphorylated SMO (blue) and PKA-C (olive) as assessed by molecular dynamics (MD) simulations (one snapshot every 300 ns, 3 μs of simulation time). Each frame is aligned on the PKA-C subunit. Left: overall view of the complex, Right: zoomed-in view of the interface between PKA-C and SMO, highlighting the stability of the PKI-like helix, inhibitor sequence, and RII-like helix. d, AlphaFold3 models of SMO / PKA-C complexes from the indicated species, aligned on PKA-C from each complex. e, Docking of two AlphaFold 2.3.0 models (see Supplementary Discussion 2) into a low-resolution cryoEM map obtained from a BS3-crosslinked SMO / PKA-C sample in a GDN detergent micelle. Conformation 1 and conformation 2 are colored blue and dark magenta, respectively. CryoEM density is shown in gray and displayed at a contour level of 0.1.

Figure 2:. A central role for the SMO pseudosubstrate motif in the SMO / PKA-C complex.

a, Left: empirical structure of PKIɑ(5-24) (PDB: 1ATP, red) in complex with PKA-C (olive/white as in Fig. 1a). Right: AlphaFold model of phosphorylated SMO (blue) in complex with PKA-C. In this and all subsequent figures, SMO cartoon (top) represents the orientation of the SMO / PKA-C complex (in this case with SMO at the bottom and PKA-C at the top, with the PKA-C N-lobe pointing upwards and the C-lobe pointing downwards). Locations of residues discussed in the text are indicated. Note salt bridge between the P-6 arginine in PKIɑ and PKA-C E203 (left); this interaction is absent in SMO (right), b, SDS-PAGE analysis of purified wild type (WT) or L637C mutant SMO incubated with PKA-C treated with or without the oxidizing agent diamide (to induce disulfide bond formation). Sequence alignments above the gel image show the mutation present in SMO and the location of the disulfide bond between SMO L637C and PKA-C C199. The endogenous cysteine in the SMO intracellular domain (C554) in wild-type SMO provides a negative control for nonspecific SMO / PKA-C disulfide bond formation. Location of SMO, PKA-C, and the disulfide-trapped SMO / PKA-C complex are indicated at right. PKA-C runs as a doublet following diamide treatment, due to formation of an intramolecular disulfide bond between C199 and C343 as shown previously.

Figure 3:. Structural mimicry of the PKA-RIIβ holoenzyme by a SMO amphipathic helix.

a, Top: cartoon of tetrameric PKA-RIIβ holoenzyme (RIIβ₂-C₂, left cartoon) in which each PKA-C subunit (C, dotted outline) contacts the inhibitor sequence from one PKA-R subunit (R, red) and the β4/β5 loop from the other PKA-R subunit (R’, green), and vice versa. Bottom: overlay between PKA-RIIβ holoenzyme empirical structure (PDB: 3TNP) and phosphorylated SMO / PKA-C AlphaFold 3 model aligned on the PKA-C in each complex. SMO is shown in blue, with the RII-like helix in the foreground (dark blue) and additional SMO sequences, including the inhibitor sequence and PKI-like helix, in the background (light blue). The SMO and PKA-RIIβ complexes are shown separately at right, with key residues and motifs indicated as described in the main text, b, MD structural snapshots of SMO F577 (light blue, one snapshot every 150 ns, 3 μs of simulation time) within the SMO C-tail (dark blue), when aligning the simulation frames with PKA-C (brown). c, MD structural snapshots (one snapshot every 30 ns, 1.5 μs of simulation time) show the positions of the PKI-like helix (615-630, green) and RII-like helix (570-581, red) during simulations of wild-type SMO (WT) or the indicated SMO mutants. Overall root-mean square fluctuations (in Å) for the residues indicated in the model at left are shown below each simulation. d, Representative HDX-MS mass spectral envelope (deuterium exchange time t_ex=10 min) of peptides in the indicated region in phosphorylated SMO (pSMO) or PKA-C alone (top spectrum in each pair), or in complex with PKA-C (bottom spectrum in each pair). The deuterium uptake is shown for each spectrum on the top right. Centroids are indicated by red dashed lines. SMO peptide = 577-590, and PKA-C peptide = 121-128. e, SDS-PAGE analysis of diamide-induced disulfide trapping between pSMO S570C and PKA-C G136C in a minimal-cysteine ((-)Cys) background (see Methods). Wild-type (WT) SMO (harboring an endogenous cysteine in helix 8, see Fig. 1) or (-)Cys PKA-C, serve as negative controls. Locations of SMO S570, A574, and PKA-C G136 are indicated in the cartoon at left. See Fig. 5e for quantification.

Figure 4:. The SMO RII-like helix is essential for Hh signal transduction.

a, GLI transcriptional reporter assay in Smo^−/− murine embryonic fibroblasts (MEFs) transfected with GFP (negative control), WT SMO, or the indicated SMO mutant constructs. Cells were treated with conditioned medium containing the N-terminal signaling domain of Sonic hedgehog (ShhN) alone (green), or in the presence of the SMO inverse agonist vismodegib (ShhN + vismo, blue), or control conditioned medium lacking ShhN (black). Data represents the mean +/− s.d., n=3 independent biological samples. Locations of SMO residues are indicated in the cartoon at left, b, A second panel of SMO mutants were analyzed as in (a). Data represents the mean +/− s.d., n=3 independent biological samples, c, Left: Steady-state SPR analysis of binding interactions between PKA-C and wild-type SMO or the indicated SMO mutants reconstituted into nanodiscs (see Figure 5 for definition and functional characterization of the 5KE mutant); Right: K_D determinations for the experiments shown at left (n=4 measurements per condition).

Figure 5:. SMO phosphorylation generates new structural elements that stabilize the PKA-C complex.

a, AlphaFold model of the SMO / PKA-C complex (SMO in blue, PKA-C in olive/white), with GRK2/3-phosphorylated S/T residues (red) and conserved K/R residues highlighted in the SMO pCT. SMO helix 8 and the RII-like helix are indicated for orientation. b, Stability of the SMO pCT in its phosphorylated vs nonphosphoryated states was assessed in MD simulations by applying a force vector (indicated by the direction of the blue arrow) to unfold the pCT, then quantifying the force required to achieve a constant velocity of unfolding^,, monitored as the distance between the red and green spheres. Top: Beginning (1) and ending (2) states for the MD simulation (states are marked on the bottom right graph) with a vector indicating the direction of applied force. Bottom: Graphs showing "extension " (distance between the red and green spheres at the ends of SMO) on the X-axes and either "polar contacts" or "pulling force" on the Y-axis for phosphorylated (blue) vs nonphosphorylated (black) SMO, showing that SMO phosphorylation increases the number of intramolecular polar contacts leading to an increased force required to unfold the SMO pCT. c, Wild-type SMO or the 5KE mutant (underlined K/R residues mutated to E) were analyzed via a GLI transcriptional reporter assay in Smo^−/− MEFs, performed as in Fig. 4a, b. d, Representative HDX-MS mass spectral envelope (t_ex= 5 min) of the indicated peptide in SMO helix 8 (539-546) or the RII-like helix (577-590) in the nonphosphorylated vs phosphorylated SMO / PKA-C complex (SMO / PKA-C (3^rd row) vs pSMO / PKA-C (4^th row), respectively). The deuterium uptake is shown for each spectrum on the top right. Centroids are indicated by red dashed lines. e, Disulfide trapping of PKA-C wild-type or G136C (in a minimal cysteine ((-)Cys construct) with phosphorylated vs nonphosphorylated, SAG21k-bound SMO (pSMO vs SMO, respectively), as in Fig. 3e. Band intensity is quantified in the graph at right, and represents mean +/− standard deviation from 3 independent trials. Representative gel image corresponding to disulfide trapping of phosphorylated wild-type vs S570C pSMO is shown in Fig. 3e.

Figure 6:. Structural mimicry of canonical GPCR-effector signaling complexes.

a, Comparison of the SMO / PKA-C model (left) with the m2AchR / β-arrestin1 structure (right). Locations of interaction sites in the C-terminus (including the reentrant loop), 7TM domain (including ICLs), and membrane are indicated in each complex, b, The SMO 7TM interface. Residues in the αD-αE and αG-αH loops of PKA-C interacting with SMO ICL domains are indicated. The reentrant loop region of the SMO pCT (residues 602-609) is shown in red. c, GLI reporter assay for the indicated deletions or alanine substitutions in the SMO reentrant loop (see (b)). d, Structural snapshots of the complex between phosphorylated SMO (blue) and PKA-C (olive) (one snapshot every 300 ns, 3 μs of simulation time). Each frame is aligned on the PKA-C subunit. Reentrant loop region of SMO pCT is highlighted in green. e, HDX-MS data for region of PKA-C C-terminal to the αI helix (residues 313-327) in its apo (top), SMO-bound (middle), or pSMO-bound (bottom) forms, determined at t_ex = 10 min and plotted as in Fig. 3d. Bimodal deconvolution of the PKA-C 313-327 spectra revealed high-exchanging (blue) and low-exchanging (green) populations, suggesting two non-interconverting states of the SMO / PKA-C complex.

Fig. 7:. A structural mechanism for Hh signal transduction by the SMO / PKA-C complex.

(Left) In the Hh pathway "off" state, SMO (blue) is inactive, with its pCT largely disordered (indicated by wavy lines) and unable to effectively engage PKA-C (orange) due to insufficient phosphorylation by GRK2/3. (Middle) In the Hh pathway "on" state, inhibition of PTCH1 (not shown) enables sterols (gray) to bind the SMO extracellular and 7TM domains, leading to SMO activation and phosphorylation by GRK2/3. This stabilizes secondary structures (PKI-like and RII-like helices) in the pCT and enables SMO to form a complex with PKA-C. PKA-C binding reinforces the SMO pCT secondary structures, further enhancing complex formation. (Right) The SMO inhibitor sequence (blue wedge) enters the PKA-C active site to interrupt PKA-C’s catalytic cycle. Consequently, GLI is released from phosphorylation-induced inhibition, leading to transcription of Hh pathway target genes. See main text for additional details.