Structural transitions enable interleukin-18 maturation and signaling

Abstract

Several interleukin-1 (IL-1) family members, including IL-1β and IL-18, require processing by inflammasome-associated caspases to unleash their activities. Here, we unveil, by cryoelectron microscopy (cryo-EM), two major conformations of the complex between caspase-1 and pro-IL-18. One conformation is similar to the complex of caspase-4 and pro-IL-18, with interactions at both the active site and an exosite (closed conformation), and the other only contains interactions at the active site (open conformation). Thus, pro-IL-18 recruitment and processing by caspase-1 is less dependent on the exosite than the active site, unlike caspase-4. Structure determination by nuclear magnetic resonance uncovers a compact fold of apo pro-IL-18, which is similar to caspase-1-bound pro-IL-18 but distinct from cleaved IL-18. Binding sites for IL-18 receptor and IL-18 binding protein are only formed upon conformational changes after pro-IL-18 cleavage. These studies show how pro-IL-18 is selected as a caspase-1 substrate, and why cleavage is necessary for its inflammatory activity.

Keywords: IL-18; NMR; caspase-1; conformational change; cryo-EM; cytokine cleavage; inflammatory activity; pro-IL-18.

Figure 1.. Biochemical characterization of caspase-1 cleavage and binding activity

(A-C) The non-target control and two clones (shown as clones 1 and 2) of CASP1^−/− THP-1 cells primed with LPS before treatment with nigericin for 2 h. IL-1β (A), IL-18 (B), and LDH (C) released into the cell culture were quantified. (D) Quantification of caspase-1 catalytic efficiency in vitro (k_cat/K_m) on its main substrates, pro-IL-18, pro-IL-1β, pro-IL-37, and GSDMD. (E-F) Gel filtration chromatography analysis of the binding between caspase-1 p20/p10 (C285A catalytic mutant) and pro-IL-18 (E) or pro-IL-1β (F). The elution volume of each complex is indicated. The SDS-PAGE gel containing individual fractions from the gel filtration run of the complex is shown, highlighting comigration and complex formation. The 6xHis-SUMO tag cleaved from pro-IL-1β (F) is indicated by an asterisk. The data shown are representative of at least three independent experiments. Bars and error bars represent the mean ± SEM of at least three independent experiments. Statistical significance was determined by one-way ANOVA with Tukey’s multiple comparisons test: ns = not significant; **p < 0.01; ***p < 0.001; ****p < 0.0001. See also Figure S1.

Figure 2.. Complete auto-processing of caspase-1 is essential for cleavage activity

(A) The domain organization of different auto-processed forms of caspase-1 used in this study. The sequence of the linker region between p20 and p10 is shown with acidic residues in red. (B, C) Quantification of catalytic efficiency in vitro of different auto-processed forms of caspase-1 on pro-IL-18 (B) or pro-IL-1β (C). (D) Cleavage of the chromogenic peptide substrate WEHD-pNA by indicated caspase-1 forms. (E-G) Gel filtration analysis of the binding of caspase-1 p33/p10 (E), p22/p10 (F) or p20/p12. The SDS-PAGE gel containing individual fractions from the gel filtration run of the complex indicates different degrees of complex formation. Comparison with the gel filtration analysis on the interaction between caspase-1 (p20/p10) and pro-IL-18 should refer to Figure 1E. The data shown are representative of three independent experiments. Bars, graphs and error bars represent the mean ± SEM of at least three independent experiments. Statistical significance was determined by one-way ANOVA with Tukey’s multiple comparisons test: **p < 0.01; ***p < 0.001; ****p < 0.0001. See also Figure S2

Figure 3.. Cryo-EM map and model of the caspase-1/pro-IL-18 complex

(A) Domain organization of CARD-deleted caspase-1 and pro-IL-18, with p20 in light purple or coral, p10 in blue or light pink, and the removed linker between p20 and p10 in gray. The approximate location of the catalysis-inactivating C285A mutation is indicated by a red arrow. The prodomain and the mature domain of pro-IL-18 are in green and orange, respectively. (B) Representative 2D classes of the caspase-1/pro-IL-18 complex. (C) Homogenous refinement of the caspase-1/pro-IL-18 complex of all 3D classes without applying symmetry, showing the heterogeneity in conformations. The percentage of total particles is shown for each class. (D, E) The cryo-EM map of the major class (III) of the caspase-1/pro-IL-18 complex (contoured at 4.0 σ), colored by local resolution (D) and by domains (E). The global resolution of the complex is 3.5 Å. (F) The caspase-1/pro-IL-18 interaction regions, with the active site region and the exosite region blocked by red and blue rectangles, respectively. The pro-IL-18 molecule is shown as a ribbon diagram, whereas the dimer of the caspase-1 p20/p10 complex is shown as a ribbon diagram for the left monomer and a ribbon diagram superimposed with the cryo-EM map for the right monomer. Both the model and map are colored by domains. (G) The cryo-EM map of the caspase-1/pro-IL-18 complex with one molecule of pro-IL-18 bound to caspase-1 without applied symmetry. The map is colored by local resolution (contoured at 8.0 σ). The global resolution of the complex is about 5.4 Å. Active site and exosite are highlighted in red and blue circles, respectively. (H) The fitted atomic model of caspase-1 and pro-IL-18 binding via active site only (class IV, open conformation), and via both the active site and exosite (class III, closed conformation). Locations for the active site and exosite are in red and blue circles, respectively. (I) Comparison of two modes of pro-IL-18 recognition by caspase-1. The arrow indicates the rotation needed to change from an open to a closed conformation by engaging the exosite binding. See also Figures S3–S5 and Table S1

Figure 4.. The interface between caspase-1 and pro-IL-18 in the complex

(A, B) Zoom-in views of the interactions at the active site region (A) and the exosite region (B). Cryo-EM densities for the prodomain of pro-IL-18 involved in the interactions are shown in blue mesh. Selective important residues are shown with side chains and labeled. (C) Sequence alignment among caspase1/4/5 in humans, caspase-1/11 in mice, and the hybrid of caspase-1/4 in dogs. Identical and similar amino acids are highlighted and colored in red, respectively. Sequence alignment was performed by using ClustalOmega online tool and plotted in ESPript 3.0. Residues with large buried surface areas in the complex are mapped at the aligned sequences, with orange rectangles for > 60 Ų buried surface areas and green rectangles for 30 – 60 Ų buried surface areas. Certain residues at the interface are indicated by red or blue arrows for the active site and the exosite, respectively. The sequence number of human caspase-1 and caspase-4 are indicated above and below the sequence alignment, respectively. *indicate the two residues with differential contributions in pro-IL-18 binding and processing by caspase-1 and caspase-4. (D) The potential effect of the IDL linker residues in p22 and p12 for the interaction with pro-IL-18. At the left, the ribbon diagram of the complex is shown together with the electrostatic surface of pro-IL-18 in which red is for negatively charged surface, blue is for positive charged surface and white is for neutral surface. At the right, a zoom-in view of the exosite and the IDL sequence are shown. The caspase-1 structure shown is the p20/p10 complex, and the additional IDL residues if they were included as in p22 or p12 are shown as dotted lines with red as acidic residues and black as other residues. The acidic region of the IDL included in p12 would have been closer to the negatively charged surface of pro-IL-18, and thus could affect the interaction. See also Figure S6

Figure 5.. Mutations of key residues at the two interfaces impair the binding and cleavage of pro-IL-18 by caspase-1

(A) Mature IL-18 by ELISA in combined supernatant and cell lysate for each WT or mutant caspase-1-reconstituted CASP1^−/−THP-1 cells. White bars and filled circles: primed with LPS for 4 h; gray bars and filled rectangles: primed with LPS for 4 h followed by stimulation with nigericin for 3 h. (B) IL-18-deficient THP-1 macrophages reconstituted with indicated pro-IL-18 variants primed with LPS for 4 h before treatment with nigericin for 2–3 h. IL-18 was immunoprecipitated from cell culture supernatants and analyzed by immunoblot. (C-D) Quantification of in vitro catalytic efficiency of WT and mutant caspase-1 p20/p10 on WT pro-IL-18 (C), and of WT caspase-1 on WT and mutant pro-IL-18 (D). The pro-IL-18 WT control was the same experiment as the caspase-1 WT control. (E) Quantification of in vitro catalytic efficiency of WT and mutant caspase-1 p20/p10 on pro-IL-1β. (F-L) Gel filtration analysis of the binding between mutant caspase-1 and WT pro-IL-18 (F-I) and between WT caspase-1 and mutant pro-IL-18 (J-L). The SDS-PAGE gel containing individual fractions from the gel filtration run of the complex indicates different degrees of complex formation. The residues involved in the active site and exosite interaction are in red and blue, respectively. The data shown are representative of three independent experiments. Bars and error bars represent the mean ± SEM of at least three independent experiments. Statistical significance was determined by one-way ANOVA with Tukey’s multiple comparisons test: **p < 0.01; ***p < 0.001; ****p < 0.0001. See also Figure S6

Figure 6.. Solution NMR structure of apo pro-IL-18

(A) Ca trace of the NMR structure ensemble of best (lowest Rosetta scores) 10 structures resulting from CYANA structure calculation and subsequent refinement using Rosetta (see Materials & Methods) (left) and a single cartoon representation of pro-IL-18 (right). Prodomain and mature domain are shown in green and orange, respectively, except that the tetrapeptide is displayed with dark blue sticks. β-strands are labeled, and “N” and “C” denote the N- and C-termini, respectively. (B) The best structure of the bundle (lowest Rosetta score), shown from two different views and in domain colors except for the 33-LESD-36 tetrapeptide which is displayed with dark blue sticks. (C) A plot of the ¹⁵N transverse relaxation rate, R₂, of the backbone amide of each residue. As described in the text, residues with lower R₂ are more flexible on the ps-ns timescale. The data points shown in green and orange are residues within the β-strands of pro and mature domains, while the points shown in dark blue are residues of the tetrapeptide. Both the tetrapeptide and the extra β-strand, β*, not seen in the complex, are highlighted by dashed blue and orange boxes, respectively. (D) Overlay of cryo-EM structure of pro-IL-18 complexed with caspase-1 (yellow) with the solution NMR structure of apo pro-IL-18 (prodomain in green; mature domain in orange). While the structures are quite similar overall, with a backbone heavy atom RMSD of 1.6 Å over ordered residues (11–28, 45–60, 80–88, 97–104, 107–112, 114–120, 133–145, 147–165, 170–179, 185–190 as determined by ¹⁵N R₂ values), several differences are highlighted. (i) Residues 53–80 are missing from the complex structure due to a lack of density. In the apo structure, this region is mostly comprised of a flexible loop; however, a small β-strand involving residues 55–58 (β*) is observed. (ii) The average orientation of the tetrapeptide-containing loop in the apo structure is distinct from that in the complex. The tetrapeptide is shown in blue sticks (apo) and yellow sticks (complex). (iii) The side chain of I48, which forms a critical exosite interaction with W294 of caspase-1 in the complex, points towards the protein core in the apo structure. The I48 side chain is shown in orange sticks (apo) and yellow sticks (complex), while the W294 side chain is shown in gray sticks. See also Table S2

Figure 7.. The topology of β-trefoil fold differs in pro-IL-18 and mature IL-18

(A) The structures of apo pro-IL-18 (prodomain in green; mature domain in orange) and mature IL-18 (cyan) are overlaid. The N-terminus of pro-IL-18 is indicated by the green “N”, while that of mature IL-18 is marked by the cyan “N”. While the overall alignment of the structures is quite poor (backbone heavy atom RMSD = 6.5 Å over ordered residues, see residue ranges given in Figure 6 legend), the ordered residues from β4 through β13 align well, with a substantially lower backbone heavy atom RMSD of 1.9 Å. (B) Topology diagrams of pro-IL-18 apo (prodomain in green; mature domain in orange) and pro-IL-18 in complex with caspase-1. The loop that is absent in the pro-IL-18 in complex with caspase-1 is indicated by dash lines. The tetrapeptide of pro-IL-18 in apo or bound form is indicated by blue. (C) Topology diagram of mature (cyan) IL-18 (PDB:3WO2). (D) The conformational difference between pro-IL-18 (left panel) and mature IL-18 (right panel), showing the structural rearrangement. For pro-IL-18, the prodomain is in green, and the mature domain is in orange except for the region immediately after the cleavage site (purple) and the following disordered loop (magenta). Mature IL-18 is in cyan except for β1 (purple) and the β2-β3-α1 region (magenta). The short β1 in pro-IL-18 rearranges into the long β1 in mature IL-18 (purple). The β* and the following region in pro-IL-18 rearrange into β2-β3-α1 in mature IL-18 (magenta). (E) The crystal structure of the IL-18/IL-18Rα complex (PDB: 4R6U) (left) and a zoom-in view (right), indicating the binding interface between mature IL-18 and IL-18Rα. The β2-β3-α1 region (magenta) in mature IL-18 participates in receptor binding. See also Figure S7