3D structures of IL-1 ligands show a similar structural fold, i.e., they are constituted by a 12-stranded beta-trefoil domain with the linkers between them  . These ligands share only 22% sequence identity, but they are structural homologs . This information can prove the flexibility of IL-1 family ligands for sequence alteration and their ability to preserve overall conformation. This idea encouraged us to design a series of truncated-IL-1Ra that preserve parental structure and contact regions to the receptor (IL-1R1). Structural superimposition of IL-1β/Ra/EBI-005 reveals several similarities that may account for the ligand-receptor binding sites (Fig. 1b,c).
Here we utilized Molegro Virtual Docker for in silico screening of ligands-non-conserved sequences, which they do not implicate in ligand-receptor interaction. This information led to the recognition of the identical non-binding site at all three ligands (IL-1Ra/β/EBI-005). Subsequently, we mapped the deletion cluster of IL-1Ra, around residue 50-100 (β5-β6-β7) which is not implicated in ligands-receptor (IL-1R1) binding interface and is far from the core of protein structure (Fig. 1a). Altogether, the results demonstrated that most of the deletions in this site would not cause a considerable alteration in ligand conformation. Among the truncated IL-1Ra proteins, the 3D structure of T-IL-1Ra (57-98 residues) had a much greater construction similarity to the native protein than other truncations in the targeted site (β5-β6-β7), therefore, we selected this truncated form as our model (Fig. 2)
T-IL-1Ra/IL-1R1 binding interface in comparison with IL-1Ra/IL-1R1
As mentioned, the deletion clusters in the β5-β6-β7 strands, suggesting that this area is favorable for truncation. It is evident in the structure that this area is located far from the core of protein structure and it does not participate in the interaction with IL-1R1. Crystallography data analysis of IL-1Ra/IL-1R1complex (Fig. 3a, b) revealed that the interface between antagonist and IL-1R1 contains strong contacts between D1D2 domain of receptor and β1–β2, β2–β3, β3–β4, β10–β11 loops of ligand and a few interactions between IL-1R1-D3 domain and ligand. As it is demonstrated in figure 2, the binding interface of the engineered ligand is significantly similar to the native protein, besides few extra interactions made in the interface of the T-IL-1Ra/IL-1R1 complex (Fig. 3c). The Z-score for T-IL-1Ra and IL-1Ra were predicted to be -6.77 and -4.54, respectively by the ProsaWeb server (Fig. 3d, e). The higher negative score of T-IL-1Ra than the wild-type protein ensures the maximum quality of the modeled truncated protein.
T-IL-1Ra antagonistic feature survey via comparing of IL-1/IL-1R1/IL-1RAcP complexes
The crystallography structure studies of ligands in the complex with not only recognition receptor (IL-1RI (A001241)), but also co-receptor (IL-1RAcP (A003536)) suggested low affinity of IL-1Ra to the IL-1R1-D3 domain in the conformational basis of the antagonism. IL-1RAcP is a co-receptor that only can bind to the binary complex of the IL-1R1/IL-1α-β complex . For the stabilization of ternary complex interactions (IL-1/IL-1R1/IL-1RAcP), the D3 domain of IL-1RAcP has to turn around the binary complex to bind the D3 domain of IL-1R1. The D3 domain of IL-1R1 in the complex with IL-1Ra stays far away from the D3 domain of IL-1RAcP, which is anticipated to disrupt the D3-D3 binding interface. As it is obvious in our protein model, the binding interface of T-IL-1Ra is similar to the parental protein which is important for antagonistic features, therefore the D3 domain of IL-1R1 stays far away from the complex conduces decreased affinity between D3-D3 domains of IL-1R1-IL-1RAcP (Fig. 4).
Protein structure validation
Molecular dynamic simulation
To investigate the structural changes in the protein-protein complex induced by ligand binding, several conformational properties were analyzed, such as root RMSD, root mean square fluctuations (RMSF), the radius of gyration (Rg), number of hydrogen bonds (NHBs), electrostatic interactions and salt bridges. RMSD (nm) vs. time (ns) for all the backbone atoms of IL-1Ra/IL-1R1 and T-IL-1Ra/IL-1R1 complex simulations were calculated to survey the stability of complexes. As shown in figure 5a, early in the simulation of complexes, IL-1R1 domains turn around the ligands because of the flexibility of the linker between the D1D2 and D3 domain, causing an immediate ascent in the overall RMSD value. From 8.5 ns onwards truncated complex attained the approximate equilibrium phase with the RMSD value averaged around 4.6 Å, whereas, the native complex trajectory experienced an ascending trend, which suggested relatively higher stability of T-IL-1Ra complex than native complex. Both systems gradually tended to converge in the last 8 ns (Fig. 5a).
RMSF of T-IL-1Ra/IL-1R1 and IL-1Ra/IL-1R1 complexes were computed to investigate changes in protein flexibility of the complex upon ligand binding. RMSF fluctuation plot of Cα ca
rbon atoms vs time (50 ns) separately for two complexes is shown in figure 5b Residues in T-IL-1Ra/IL-1R1 complex experienced minor fluctuation and the overall RMSF of the truncated complex was lower than the native complex, which indicated this complex was relatively more stable during the simulation process. New interactions involved in stabilizing the truncated complex could play an important role in minimizing the fluctuations and maintaining the proteins in a rigid structure to simplify the formation of the complex (Fig. 5b).
The radius of gyration is a significant parameter to survey the compactness of protein. The radius of gyration for T-IL-1Ra/IL-1R1 and IL-1Ra/IL-1R1 complexes showed fluctuation in Rg value until 10 ns, afterward attained virtually stable Rg value. It was indicated that the Rg average values for truncated and native systems were around 3.11 nm and 3.13 nm, respectively. The lower Rg value of truncated ligand bond to the receptor than the native complex can be attributed to the elimination of ligand-space-barrier which lets the IL-1R1 domains encompass the ligand more tightly. Therefore, T-IL-1Ra/IL-1R1 complex showed higher compact than the native ligand (fig. 5c).
Interaction energetic feature
MM-GBSA method was used to calculate the binding free energy of systems. The average binding free energies and detailed energetic contribution components of 50 ns were shown in table 1. Interestingly, the free energy of the truncated system (-1087.037 KJ/mol) is significantly lower than the native system (-836.819 KJ/mol) which could demonstrate a higher binding affinity of the truncated ligand with its receptor than the native protein. This result conforms to the outcomes obtained from RMSF analysis. Furthermore, dissecting the binding free energy into contributing components showed that the electrostatic interaction in truncated complex (-2214.386 KJ/mol) had a major role in the low free energy of the truncated system and the system stability.