Molecular analysis of the type III interferon complex and its applications in protein engineering

Abstract

Type III interferons (IFNλs) are cytokines with critical roles in the immune system and are attractive therapeutic candidates due to their tissue-specific activity. Despite entering several clinical trials, results have demonstrated limited efficacy and potency, partially attributed to low-affinity protein-protein interactions (PPIs) responsible for receptor complex formation. Subsequently, structural studies of the native IFNλ signaling complexes remain inaccessible. While protein engineering can overcome affinity limitations, tools to investigate low-affinity systems like these remain limited. To provide insights into previous efforts to strengthen the PPIs within this complex, we perform a molecular analysis of the extracellular ternary complexes of IFNλ3 using both computational and experimental approaches. We first use molecular simulations and modeling to quantify differences in PPIs and residue strain fluctuations, generate detailed free energy landscapes, and reveal structural differences between an engineered, high-affinity complex, and a model of the wild-type, low-affinity complex. This analysis illuminates distinct behaviors of these ligands, yielding mechanistic insights into IFNλ complex formation. We then apply these computational techniques in protein engineering and design by utilizing simulation data to identify hotspots of interaction to rationally engineer the native cytokine-receptor complex for increased stability. These simulations are then validated by experimental techniques, showing that a single mutation at a computationally predicted site of interaction between the two receptors increases PPIs and improves complex formation for all IFNλs. This study highlights the power of molecular dynamics simulations for protein engineering and design as applied to the IFNλ family but also presents a potential tool for analysis and engineering of other systems with low-affinity PPIs.

Figure 1

Molecular modeling and simulation of the IFNλ3 ternary complex reveals differences in PPIs between native and engineered ligands. (A) An engineered IFNλ3, H11 (blue), was used to solve the structure of the type III IFN in complex with its receptors, IFNλR1 (silver) and IL10Rβ (gold) (PDB: 5T5W). Residues on the ligand mutated in the engineered protein are shown in red. Five amino acid mutations separate the wild-type IFNλ3 from the engineered H11. (B) Residue contact patterns differ greatly between wild-type and engineered IFNλ3, leading to significant differences in PPIs throughout simulations. Δτ > 0.25. (C) Hotspots of interactions (residue pairs with the highest values of Δτ) between each set of protein chains are shown in red. Visualized interactions within protein chains can be seen in Fig. S6.

Figure 2

Residue strain fluctuations reveal sites of interaction known to drive protein complex formation. Residues experiencing high differences in strain (Δε) $\Delta ϵ$between the engineered and wild-type simulations align with known regions of interaction in the 5T5W structure (site 1, IFNλR1-IFNλ3 interface; site 2, IL10Rβ-IFNλ3 interface; site 3, IFNλR1-IL10Rβ interface).

Figure 3

Mutations in H11 increase ligand stability by reducing chain fluctuations. (A) Differences in root mean-square fluctuation (ΔRMSF) for the IFNλ3 protein between the WT and engineered simulations are shown. The five mutations which separate the proteins are called out with gray lines, with a red bar indicating the region experiencing the largest differences. (B) The primary modes of the first principal component, when superimposed, reveal a highly flexible region (red backbone) for the WT IFNλ3. This same region remains stable within H11. Star, location of H120 (WT) or R120(H11).

Figure 4

Gaussian-accelerated molecular dynamics simulations reveal that mutations in H11 help achieve complex stability through the formation of a secondary structure. Relative free energy minima generated from Gaussian-accelerated molecular dynamics (GaMD) simulations are plotted using the root mean-square deviation (RMSD) of IFNλ3 and IFNλR1. The engineered complex shows two distinct free energy minima, whereas the wild-type complex only shows one. As the engineered complex progresses through the simulation, the formation of a secondary structure is observed (in red).

Figure 5

Mutation on IL10Rβ identified by molecular simulation leads to an increase in predicted PPIs between IL10Rβ and wild-type IFNλ3. (A) Residue contact patterns differ greatly between wild-type and engineered IL10Rβ, biasing the system toward receptor-ligand contacts. Δτ > 0.25. (B) The top 5 residue-residue interactions occurring between proteins are shown in red, demonstrating a shift in PPIs occurring at the IL10Rβ-IFNλ3 interface.

Figure 6

Computationally designed IL10Rβ shows improved binding for all type III IFNs and enables purification of the wild-type IFNλ3 ternary complex. (A) Fold increase in binding data for yeast displaying IFNλ1, IFNλ2, or IFNλ3 in complex with IFNλR1 and either wild-type IL10Rβ (WT) or IL10Rβ-N147D (N147D). n = 3, plotted with SEM, statistical significance determined by two-way ANOVA. (∗p = 0.0206, ∗∗∗∗p < 0.0001) (B) Size-exclusion chromatogram (Superdex S75 column) of the WT IFNλ3/IFNλR1/IL10Rβ-N147D ternary complex (top). Volume fractions are shown on the SDS-PAGE protein gel (bottom).