Structural Evaluation of Interleukin-19 Cytokine and Interleukin-19-Bound Receptor Complex Using Computational Immuno-Engineering Approach

Abstract

Interleukin 19 (IL-19) is an anti-inflammatory cytokine that belongs to the IL-10 family, where IL-20 and IL-24 also exist. While IL-19 and IL-20 share some comparable structural folds, there are certain structural divergences in their N-terminal ends. To date, there are no reported IL-19 receptors; although, it has been suggested in the literature that IL-19 would bind to lL-20 receptor (IL-20R) and trigger the JAK-STAT signaling pathways. The present report examines the structure of the IL-19 cytokine and its receptor complex using a computational approach. Specifically, the postulated modes of interactions for IL-20R as an IL-19 receptor are examined on the basis of a set of computational findings. The author has used molecular docking and molecular dynamics simulation to generate a 3D model for the IL-19 complex with IL-20R. When a protein’s crystal structure is not available in the literature, predictive modeling is often employed to determine the protein’s 3D structure. The model assessment can be based on various factors, which include stability analysis using RMSD calculations, tracking changes in time-based secondary structures and the associated Gibbs energies, ΔG. Since one model complex (referred to as model A throughout this paper) can be used as a working hypothesis for future experiments, this structure has been explored here in detail to check its stability, subunit interfaces, and binding residues. The information gathered in this approach can potentially help to design specific experiments to test the validity of the model protein structure. Additionally, the results of this research should be relevant for understanding anti-inflammatory mechanisms and, eventually, could contribute to the efforts for therapeutic developments and targeted therapy.

1. Introduction

Immuno-engineering is a newly discovered field that has generated considerable interest in the area of computational biochemistry, especially since the outbreak of COVID-19. Immuno-engineering utilizes molecular-engineering-based methods to develop new proteins and peptides. Computational results of immuno-engineering and immuno-informatics can aid the development of neo-antigens/neo-adjuvants as well as targeted therapeutics, which in turn could contribute to the broader efforts of preventing and combating virus-borne pathogens/diseases [1,2,3,4,5,6].

Cytokines are broadly categorized in three main groups, namely lymphokines, interleukins and chemokines. The interleukins (IL) play an important role in immune-modulatory and immune-regulatory processes. Interleukin receptors (IL-R) are transmembrane glycoproteins that bind to specific IL molecules and promote signal transduction. Currently, there are nearly forty reported IL molecules, but all the receptors for these ILs have not been identified yet. The current work focuses on this latter topic using a computational structural biology approach to develop a tentative model of the receptor–cytokine pairs [7,8,9,10,11].

Aside from the IL-19, the IL-10 family includes other interleukin members such as IL-10, IL-20, IL-22, IL-24, IL-26, IL-28, Il-29, and certain interferons like interferon-lambda1 (IFN-λ1) and -4. IL-10 family members also share several receptors. IL-19 and IL-20 share a comparable structural fold; although, there are certain structural divergences in their N-terminal ends as noted by Logsdon et al. [12]. It has been suggested that IL-20 receptor (IL-20R) might work as IL-19 receptor [12,13,14,15,16,17]. However, this function of IL-20 is only poorly understood, and experimental knowledge about this subject is also currently limited. Computational approaches can be particularly useful to explore this topic in detail.

It has also been suggested that the IL-19 signals activate the JAK/STAT signaling pathways through the IL-20 receptor [12,13,14,15,16,17,18,19]. IL-20 ligand and IL-20αR and IL-20βB receptors form a ternary complex. Type-1 IL-20R consists of IL-20αR and IL-20βB receptors, whereas type-II IL-20R consists of IL-22R and IL-20RβB subunit. The ternary signaling complex of IL-20 ligand and IL-20αR and IL-20βB receptors have three interfaces that exist between IL-20 ligand/IL-20αR, IL-20 ligand/IL-20βB, and IL-20αR/IL-20βB. Among these, the majority of the residue interactions are identified between IL-20 ligand and IL-20βB interfaces [12]. IL-20αR and IL-20βB receptor chains are often described as IL-20R1 and IL-20R2.

In addition to the IL-20 and IL-19 ligands, IL-20R is also a receptor for the other IL-10 family members, including IL-24, and this ligand–receptor complex is capable of triggering the JAK/STAT signaling pathway. Incidentally, some of these IL-10 family members (IL-19, IL-20, IL-22, IL-29) function as monomers [20]. However, some IL-10 family members operate as oligomers, like dimer (IL-10, and IL-22) and tetramer (IL-22). IL-24 and IL-26 can exist as functional monomers or dimers. But to this author’s knowledge, no higher-order multimers have been observed in the IL-10 family. Though the structural and/or functional relevance of this monomer or multimeric interleukin ligand is not known, the dimeric IL-10 is stabilized by a subunit interface, whereas IL-8 (does not belong to IL-10 family) monomer is a high-affinity ligand with coupled interactions [21,22].

It was observed that IL-20/IL-20αR binding was weak, while the interactions between IL-19 and IL-20αR have not been experimentally detected. However, the IL-19 ligand/IL-20βB complex formation has been experimentally confirmed by surface plasmon resonance (SPR) experiments, which suggested that IL-19 would more strongly bind to IL-20βB than IL-20 [12]. However, it is still not clear from the available data if IL-19 sequentially interacts with IL-20βB and then with IL-20αR to form a signaling complex. Several authors have proposed that IL-20βB and IL-20αR could couple to generate the receptor for the IL-19 ligand [12,15]. The IL-19 or IL-20 and the melanoma differentiation associated with the gene-7 (mda7) ligand may also interact with the IL-20αβ complex. The association of IL-20βB with IL-20αR has been suggested as a mechanism to form the heterodimeric receptor for IL-19 [16].

Our previous reports have examined various immuno-regulatory proteins and their impact on cellular signaling cascades [23,24,25,26]. These earlier studies have also explored the structural basis of the ligand–receptor interactions in immunologically relevant proteins [27,28]. The primary goal of the current study is to predict a structural model for the IL-19 bound IL-20R complex. Here, the IL-20-bound IL-20R complex is used as a benchmark for the hypothesized IL-19/IL-20R ligand–receptor complex. The model incorporates previously reported as well as some unrestricted combinations of ligand and receptors that form a relatively stable and reliable IL-19/IL-20R complex. While some of these IL-19 and IL-20R bindings are non-specific, the computationally designed model structures may exhibit functions that are relevant for preventing and treating various immune disorders.

2. Materials and Methods

2.1. Selection of Protein Models

For the receptor structure, the IL-20 hetero-dimeric 4DOH receptor chains αR and βB were selected as proposed IL-19 receptors [12]. For the IL-19 ligand protein, 1N1F was chosen [14]. Both 4DOH and 1NIF structures have been determined by X-ray crystallography at 2.80Å and 1.95Å resolution, respectively. Cytoscape was employed for the protein–protein interaction analyses [29] and the ligand (IL-19)-receptor (IL-20αR/IL-20βB) model protein complexes were predicted using GrammX and Patchdock/Firedock docking server, respectively [30,31,32]. A GrammX server uses a Fast Fourier Transformation method to predict the docked structure, whereas PatchDock/FireDock uses shape complementarity criteria for docking.

2.2. Modeling and Simulation

NAMD 2.14 and QuikMD of VMD 1.9.3 were used for time-based simulation studies for IL-19 ligand and IL-19/IL-20R complex, model A, and model B [33,34,35]. These models are discussed in a later part of this report. “Structure manipulation” helped to prepare and generate the protein structure files (.psf). An implicit solvation method was selected where a Generalized Born implicit solvent (GBIS) was utilized [36]. In all cases, a CHARMM36 force field was used. Energy minimization (2000 steps), annealing, and equilibration of the system were performed before the final simulation run, and annealing was carried out for 0.03 ns. By default, the system temperature was gradually increased from 60 to 300 K for 0.24 ns. Equilibration was performed for 0.04 ns at 300 K. The final production run was performed for 20 ns at 300K using Langevin dynamics, and r-RESPA was used. For all simulations, the time step was selected to 2 femtosecond (fs), applying the NVT ensemble. The backbones were restrained for the minimization and the equilibration processes, whereas for the final simulation, no atoms were constrained. Most of these parameters were preset as default. Proteins’ structural illustrations were generated by Biovia’s Discovery Studio Visualizer [37]. Origin graphical program was used to construct root mean square deviation (RMSD) and root mean square fluctuation (RMSF) plots. Additional experimental details of the simulation used in this work have been described elsewhere [27].

3. Results and Discussion

Figure 1 shows the structure of the IL-19 ligand, 1N1F.PDB (A), and the IL-20 ligand-bound IL-20 receptor system, 4DOH.PDB (B). From Figure 1A, it is evident that IL-19 is predominantly α helical in structure, and Figure 1B indicates that the ligand IL-20 binds mostly between the D1 domains of the IL-20αR and βB receptor. Figure 1C displays the cavities within the 4DOH receptor chains αR and βB. An interaction scheme of the IL-19/IL-20R pathway is displayed in Supplementary Materials S1 (SM 1) Figure S1.

The protein (IL-19 ligand)–protein (IL-20 receptor) interaction networks were analyzed using Cytoscape, where the interactions between the IL-19 and IL-20βB receptor (Figure 2A) were established by various experimental confirmations like cross-linking study, surface plasmon resonance spectroscopy, and molecular sieving. Few in vivo interactions are possible between the IL-19 and IL-20αR. The UniProt ids for IL-20βB and -αR are Q6UXL0 and Q9UHF4, respectively [38]. The protein–protein, bipartite expansion network of IL-20αR and -βB based on iRefindex are displayed in Figure 2B,C [39]. It is observed from Figure 2A that IL-19 interacts with both IL-20βB and -αR. Additionally, Figure 2B demonstrates that both IL-19 and IL-20 ligands are interacting with IL-20αR. While some biological interactors can be identified in Figure 2, in many cases, the data from large-scale interactome projects (mostly remaining unexplored for experimental validations) can be treated as hypotheses.

Though the interactions between IL-19 and IL-20R heterodimeric structure are documented in the published literature [12,17], to date, there is no available crystal structure for IL-19-bound receptor complex. This lack of information on this topic could be linked to certain practical difficulties in reconstructing the complex. Within the IL-20 family, the IL-19 is known to be a weak cytokine that provokes downstream epidermal keratinocytes responses [40].

A sequence alignment of IL-19 and IL-20 is reported in SM 1 Table S1. The identity matrix reveals a 40.96% degree of homology between these two proteins. Both IL-19 and IL-20 cytokines activate STAT3 transcription factor, and the relevant pathway common to both IL-19 and IL-20 is reported as the JAK/STAT signaling pathway.

The IL-19-ligand-bound IL-20 receptors-docking schemes, as obtained using the GrammX server, are designated in Figure 3. The top ten docked model structures of the ligand–receptor protein were chosen on the basis of smoothed Lennard–Jones potentials, refinement, and scoring. Determining the binding between IL-19 and IL-20R was difficult since in most of these trial cases, the binding process resulted in non-specific bindings and/or weak binding. While the exact reasons behind these non-specific bindings are not clear at this time, sometimes they are related to the absence of specific signature structures or sequences and/or protein’s instability. According to earlier studies, the IL-19 binds to IL-20Rβ more tightly than IL-20 (and it is postulated that the IL-19 binds to the IL-20R heterodimeric structure).

The schemes for three possible models emerge from the above considerations as follows: (1) IL-19 binds to IL-20αR first, (2) IL-19 binds to IL-20βB first, and (3) IL-19 binds to the heterodimeric receptor IL-20αβ. These postulated schemes are displayed in Figure 3. In scheme 1, the IL-19 has been docked to the IL-20αR first, and then the IL-19/IL-20αR complex has been docked to the -βB receptor. In the case of scheme 1, there are no suitable trimeric complexes as most of them undergo non-specific binding. Therefore, in our modeling experiment, an IL-19/IL-20αRβB complex formation was not identified. Surface plasmon resonance (SPR) analyses by previous authors also did not identify the IL-19/IL-20αR complex formation, though IL20/IL-20αR interactions were considered as a possibility [12]. The details of this scheme 1 are displayed in the SM 1 Figures S2 and S3.

In the second scheme of Figure 3, we have assumed that the IL-20βB receptor-bound IL-19 was docked to the –αR receptor (Figures S4 and S5). This is a proposed scheme, while experimental proof for the IL-19 and IL-20βB receptor interaction is documented in the literature [12]. Here also, in most cases, the βB bound IL-19 either did not bind to -αR receptor or bound to the wrong domain (D2) of the -αR (non-targeted binding) and thus, formed a poorly coordinated complex. Among these relatively poorly assembled structures, the one labeled as model 6 (of Figure S5) visually resembled the reported structure of IL-19-bound IL-20R. A detail scrutiny of this structure suggests that the ligand in this case binds to the D2 domain of the alpha R receptor. Therefore, this specific complex formation cannot be considered reliable. Additional details of this scheme and the associated models are described in SM 1 Figures S4 and S5.

For the third scheme in Figure 3, the IL-20αβ receptor dimer and the IL-19 ligand were considered. The GrammX server successfully confirmed this scheme, and hence, we used its corresponding model 1 in Figure 4. The results in Figure 4 suggest that, except for two cases, the ligand-binding pocket in most of these models is the same, which is the groove between the two receptors. Models 3, 4, 9, and 10 are visually similar to model 1. In model 5, the ligand binds to beta receptor D2. Model 1 in Figure 4 will be denoted as model A in the remainder of this report. The 3D structure of model A is included and can be obtained from Supplementary Materials S2 (SM 2).

Since this current paper is completely theoretical and based on simulations, potential experimental validation of the results cannot be covered explicitly. Nevertheless, the neighbor residues of ligand IL-19 and receptors IL-20αR and IL-20βB are displayed in SM 1 Tables S2 and S3. The fine structural details of model A are displayed in SM 1 Table S4. This displayed information can be useful to experimentalists for designing mutations with one of the subunits having stronger binding for further studies. Since model A can be used as a working hypothesis for future experiments, the author has examined this model structure at the molecular level to check its surface area, binding interfaces, and surface complementarity.

The binding interface of model A is calculated using PDBePISA [41]. The interfacial region between subunits αR:A, βB:αR, and βB:A are 859 Ų, 697Ų, and 525.2 Ų, respectively. The solvation-free energy gain (kcal/mol) due to the formation of interfaces between αR:A, βB:A, and βB:αR are −4.5, −5.1, and −5.4, respectively. This value is the change in total solvation energies of individual and interfacial structures. Negative ∆G denotes better and stronger complex formation.

According to earlier results, the formation of the IL-20 receptor dimer seems to occur after ligand binding [13]. Therefore, it is unclear at this point how the dimeric receptor forms without the presence of the IL-19 ligand and how the IL-19 binds to hetero-dimeric IL-20R1/R2. However, it should be noted that the formation of a heterodimeric receptor (IL-20αR/IL-20βB) for IL-19 was reported in a previous publication [16]. It is expected that further aspects of the associated problems will be studied in additional computational work and that new results will be presented in future reports.

A further docking and scoring experiment was carried out to predict a justifiable, correct model of the IL-19-bound receptor. For this second experiment, Patch Dock and Fire Dock server were used (SM 1 Figure S6). In the case of docking with the Patchdock–Firedock server, the very first docking step of scheme 1 was unsuccessful (Figure S7). Thus, the second docking step of scheme 1 was not necessary. As we did not obtain any suitable model for this system, we did not proceed further using scheme 1 of Figure S6. For the second scheme of Figure S6, one model (model 3 of Figure S9) was tentatively selected for further analyses. A detailed scrutiny reveals that most of the neighboring residues of the ligand IL-19 mostly reside within the D2 domain of αR1 receptor with respect to the αR1:D2 of 4DOH. Therefore, this complex formation is not fully reliable. This model is considered as model B and included in SM 1 Figures S8 and S9.

Visual assessments also reveal that most models from this scheme 2 form weak or non-specific protein–protein binding. In the case of scheme 3 of Figure S6, the binding pockets are mutually similar for all models and are rather close to the description of Figure 1B. However, no specific models were selected here due to their positive binding energies, as illustrated in Figure S10.

To date, no established crystal structures for the IL-19/receptor complex can be found in the readily available literature. In such cases, predictive modeling is often employed to set up the 3D structures of such exploratory proteins. In the present work, Ramachandran plots were generated to assess the required protein structures, and the results are shown in Figure S11 of the SM 1. As seen there, these plots for the 4DOH.PDB and model A are mutually comparable (Figure S11). Additionally, model “suitability” can be evaluated by various factors, such as stability analysis using RMSD calculations and an examination of the secondary structure. The negative-solvation-free energy gain upon complex formation is also an indicator of a stable complex formation.

Approximate results of ΔG and the dissociation constant (Kd) for the 4DOH system and model A were calculated using PRODIGY [42]. The ΔG values of model A (−17.7 kcal mol⁻¹) are comparable to those of 4DOH (−20.1 kcal mol⁻¹). Additionally, the Kd value (M) of model A at 0 °C is 1 × 10⁻¹³, which is a value indicative of high binding affinity between IL-19 ligand (chain A) and IL-20 receptor chains (IL-20αR and IL-20βB). The Kd (M at °C) value of 1.9 × 10 ⁻¹⁵ for 4DOH indicates very high affinity between the IL-20 ligand (chain A) and IL-20 receptor (chains αR, βB).

Figure 5 describes the RMSD plots of wt IL-20/IL-20R and model A (Figure 4: model1). From this figure, it is evident that model A is quite stable. An RMSD comparison of wt IL-20/IL-20R and model A demonstrate that they are comparable (Figure 5). The RMSD plot of model B is included in the SM 1 Figure S12. From the RMSF plot (Figure S13), it is evident that, except for few residues, the fluctuations in receptor chains βB and αR are mutually comparable between the wt and model A structures. Some residues of the IL-19 ligand show relatively higher flexibility, particularly those near amino acids 80. The Pro80 residue within IL-19 is close to the receptor chain IL-20βB. Residues 79–80 within IL-19 form a loop-like structure and, thus, exhibit high flexibility. At the same time, the residues around Cys80 of the IL-20 ligand within 4DOH form an alpha-helical structure and hence are stable in nature.

The secondary structure of model A does not manifest major structural variations with time (Figure 6). Therefore, the proposed model of the IL-19 ligand attached to the dimeric IL-20 receptor (α, β) seems kinetically stable and invariant with time. The color codes of secondary structure analysis are displayed in Figure S14. The secondary structural analyses of model B are also included in Figure S14.

Several experimental conformational preferences and proofs of direct IL-19/IL-20R interactions, including SPR signal measurement, crosslinking, and molecular sieving, are plotted in Figure 2. Some of these experimental bases for this claim of IL-20R as IL-19 receptors have been identified by previous authors [12,17]. Thus, in addition to the direct interactions, the proteins’ meta-analyses were based on biological, largely functional, or correlative (e.g., co-expression) interactions. The structural quality of the IL-19/IL-20R model (model A) is evaluated by RMSD calculations and secondary structure analyses; based on the results, this model is found to be comparable to the IL-20/IL-20R wt protein complex.

The present author has previously reported computational methodologies similar to those used here mostly to study the wild vs. mutant structures and the impact of mutation on proteins structures and functions. Considering the absence of readily available crystal structure data for the IL-19 receptor and IL-19-ligand-bound receptor, 3D computational modeling has been used in the present investigation to predict the structures of these systems. This predictive approach to modeling is somewhat underexplored and, thus, represents a relatively novel element of this study. The results presented here can be further tested experimentally and corroborated in future studies.

4. Conclusions

Based on the computational data discussed in this report, it is possible to conclude that model A can be taken as a working model for IL-19 and the IL-20αR/IL-20βB complex. The subunit-interacting area and the interaction residues within model A are tabulated in SM 1 Tables S2–S4. Adding to the absence of experimentally confirmed IL-19 receptors, the reliance on computational predictions seems to be a limitation of the findings of the specific binding residues as potential candidates for mutagenesis studies. In this regard, the proposal presented here for complex formation should remain open for confirmation by experiments.