IL-2 Flexible Loops Might Play a Role in IL-2 Interaction with the High-Affinity IL-2 Receptor: A Molecular Dynamics (MD) Study

Te clinical use of high-dose IL-2 in cancer immunotherapy faces several drawbacks such as toxicity and unfavorable pharmacokinetic profle. Tese drawbacks can be avoided by inhibiting IL-2 interaction with the CD25 subunit, which is a component of the high-afnity IL-2 receptor (IL-2R αβ c ). Several studies showed mutations of potential IL-2 residues such as R38, F42, Y45, and Y72 would produce IL-2 that is CD25-independent. In essence, structural comparison between wild-type (WT) IL-2 and CD25-independent IL-2 can be very insightful to assess the role of IL-2 fexibility and conformation in the IL-2 receptor in-teractions. Here, we investigated the fexibility loops and conformation of IL-2m (F24A, Y45A, and L72G), which is known to be CD25-independent, and IL-2m2 (F42Y and L72R) mutants along with WT IL-2 using MD simulations. Despite residue mutations, both IL-2m and IL-2m2 showed comparable conformational compactness and better stability than WT IL-2. Interestingly, IL-2m and IL-2m2 mutants showed rigid BC and CD loops in comparison to WT IL-2 . Also, the AB loop conformation of IL-2m was a bent structure compared to the WT IL-2 and IL-2m2. Principal component analysis (PCA) and free-energy landscape results suggested IL-2m and IL-2m2 have stable conformations compared to the WT IL-2. Terefore, these mutation sites of IL-2 produced stable and rigid loops that might prevent IL-2 from binding to the CD25 subunit. Our results can help to assess IL-2 fexibility loops to design new CD25-independent IL-2 mutants without compromising the IL-2 structure.


Introduction
Interleukin-2 (IL-2) is an immunoregulatory cytokine with a molecular weight of 15.5 kDa and four antiparallel α-helices structure that plays a crucial role in the immune response [1][2][3]. IL-2 exerts its stimulation function by binding to monomeric, dimeric, or trimeric IL-2 receptors (IL-2Rs). Te high-afnity trimeric receptor (IL-2Rαβc) of IL-2 consists of α (CD25), β (CD122), and c (CD132) subunits, while the moderate-afnity dimeric receptor (IL-2Rβc) consists of β and c subunits [4]. IL-2 acts as a lymphocyte growth and stimulating factor that promotes the expansion, diferentiation, and survival of antigen-activated T cells and B cells as well as the cytolytic activity of natural killer (NK) cells and regulated T (Treg) cells [5,6]. Terefore, the antitumor activity of IL-2 is attributed to its ability to expand and activate innate and adaptive efector cells. Te high dose of IL-2 was used as an immunotherapy agent to eradicate cancer cells by expanding efector T cells and NK cells population in patients [7][8][9]. Recombinant IL-2 (aldesleukin) was the frst immunotherapy agent to gain the FDA approval in 1992 to treat metastatic renal cancer and later for metastatic melanoma [10].
However, using IL-2 as a therapeutic agent is limited due to severe toxic efects such as capillary leak syndrome, hypotension, hypoxia, and neurological efects, as well as the induction of immunosuppressive responses and bearing a short half-life [2]. Also, IL-2 might induce pulmonary edema as a result of IL-2 binding to CD25 (IL-2Rα) on lung endothelial cells [11]. Te CD25 subunit is highly expressed in Treg cells, thus binding of IL-2 to the high-afnity IL-2 receptor (i.e., IL-2Rαβc) deviates the IL-2 activity toward expansion of Treg cells. Tis limits the bioavailability of IL-2 to stimulate antitumor efector T cells and NK cells [12]. Terefore, preventing IL-2 binding to CD25 (α subunit) can minimize its toxicity profle as a cancer immunotherapeutic agent. On the other hand, expansion of Treg cells is required in the treatment of other diseases such as autoimmune diseases (AD); therefore, low dose of IL-2 treatment has been evaluated in clinical trials to treat AD [13][14][15][16].
In cancer immunotherapy, modifed IL-2-based agents have been developed to avoid the abovementioned toxic efects and improve the pharmacokinetics (e.g., short halflife) and pharmacodynamic profles. Tis can be achieved by maintaining or improving IL-2 binding to IL-2Rβc receptor while abolishing or minimizing binding to IL-2Rαβc receptor. For instance, site-directed mutagenesis of IL-2 has been developed to lower its afnity to the high-afnity IL-2 receptor (CD25-independent IL-2). In previous work, F42K or R38A mutations in IL-2 decreased its afnity toward IL-2Rαβc receptor while maintaining the afnity toward IL-2Rβc receptor [17,18]. Another study showed that mutation of F24A, Y45A, and L72G residues abolished the IL-2 mutant (IL-2m) interaction with the CD25 subunit [19]. IL-2m (CD25-independent) was conjugated to a specifc cergutuzumab amunaleukin (CEA) antibody to improve therapeutic and pharmacokinetic profles of IL-2. Moreover, IL-2 mutation strategies can also strengthen small molecule inhibitors to block IL-2-CD25 interaction [20]. Other strategies such as adding polyethylene glycol (PEG), Fc domain of immunoglobulin (Ig) moieties, or specifc antibodies to IL-2 would enhance the IL-2 half-life.
Protein conformations play a crucial role in their biological activities and binding interactions. Te intrinsic fexibility of IL-2 plays a signifcant role in the IL-2 receptor interaction by modulating the strength of the binding interaction [21,22]. Here, we assess IL-2 loop fexibility and conformations of IL-2m (i.e., CD25-independent IL-2) along with wild-type (WT) IL-2 using MD simulations. Also, we designed IL-2 mutant 2 (IL-2m2) by mutation of F42Y, and L72R residues to investigate the relationship between potential residues sites and IL-2 fexibility loops. Te mutation of Y42 and R72 residues in IL-2m2 was based on their similarity to F42 and L72 residues in size and chemical structure as well as their position between AB loop and helix B. Comparative investigation of conformation and fexibility between WT and CD25-independent interleukins might be insightful to assess their loops roles in IL-2 structures. Tus, it allows researchers to design IL-2 mutants that do not bind to the CD25 subunit while maintaining the required conformation and stability along with the ability to interact with moderated-afnity IL-2 receptor.

Result and Discussion
2.1. IL-2 Flexibility. C-α root mean square deviation (RMSD) was used to assess the stability of MD simulations systems of WT IL-2, IL-2m, and IL-2m2 mutants. All systems reached convergence during the simulations and thereby stable systems were obtained (Figure 1(a)). Te fexibility results of interleukin showed that IL-2m and IL-2m2 structures signifcantly have higher rigidity than the WT IL-2 in helix C, BC, and CD loops as well as slightly in the AB loop ( Figure 1(b)). Te dynamic of the AB loop is essential toward IL-2Rαβc binding; therefore, binding of antibodies or small molecule inhibitors to this loop prevents IL-2 interaction with IL-2Rαβc receptor [23]. Because of similar observed rigidity of IL-2m and IL-2m2 mutants, sites 42 and 72 might play an important role in the IL-2 fexibility despite residue types ( Figure 2). Taking the intrinsic dynamics of IL-2 into account, our results suggest IL-2 might need the high fexibility loops to interact with IL-2Rαβc. Te fexibility of these loops might not be required to interact with IL-2Rβc since IL-2m maintained its afnity toward IL-2Rβc. Although the absent interaction of IL-2m with IL-2Rαβc was attributed to the loss of potential residues in IL-2R interaction, our results suggest mutation of the sites 42 and 72 induced the loops rigidity of IL-2m.

IL-2 Conformation and Radius of Gyration (Rg).
Te intrinsic dynamic of IL-2 is well-known to play a role in their biological function; particularly during interaction with IL-2R [22,24]. A single or multiple mutations in proteins can change proteins conformation and thereby their biological activity [25][26][27]. Te BC and CD loops were rigid structures in the IL-2m mutant (CD25-indpendent) possibly because of removing F42 and L72 residues, which could act as hinders between the AB loop and helix B in the WT IL-2. Terefore, the AB loop showed bent structure in IL-2m compared to WT IL-2 and IL-2m2 ( Figure 3). However, Y45 residue might play less role in the AB, BC, and CD loop structures since Y45 residue was not mutated in IL-2m2 mutant that showed similar loops rigidity compared to IL-2m mutant. Surprisingly, F42A, Y45A, L72G or F42Y, and L72R mutations did not change the Rg values in IL-2m or IL-2m2, respectively (Figure 4(a)). Rg values for WT IL-2, IL-2m, and IL-2m2 have similar values (around 1.52 nm). Tis indicated that the conformational compactness of the WT and its mutants were similar. Terefore, mutation sites (42, 45, and 72) might not infuence the whole IL-2 conformation while infuencing the IL-2 loops fexibility. Tis fnding could help to design further CD25-independent IL-2 mutants without compromising IL-2 conformation or its stability.

Solvent Accessible Surface Area (SASA).
Te solvent accessible surface area (SASA) was assessed for IL-2m and IL-2m2 in comparison to WT IL-2 in this study [28,29]. Te SASA can estimate exposed or buried residues of IL-2 during the MD simulations. In consistent with Rg results, all-residue SASA values of WT IL-2, IL-2m, and IL-2m2 were similar, especially after 60 ns of simulations (Figure 4(b)). Terefore, similar conformation was maintained for WT IL-2, IL-2m, and IL-2m2 even though there were three or two mutated residues. We can suggest that the mutations at sites 42 and 72 might decrease the fexibility of IL-2 but not compromise their solvent exposure. Moreover, we elaborated further on SASA values in the AB loop residues in comparison to WT IL-2. Again, SASA of the AB loops were similar for both systems, especially after 90 ns of simulations (data not shown). Terefore, there were no conformational changes due the F42A, Y45A, and L72G mutation.

Principal Component Analysis (PCA).
Due to the importance of IL-2 dynamics in its biological activities, we investigated the essential dynamics space on C-α of IL-2m (CD25-independent IL-2) and IL-2m2 along with WT IL-2 using the principal component analysis (PCA). Clearly, WT IL-2 covered a wider phase space than IL-2m by trajectories projection of the frst principal components PC1 and PC2 ( Figure 5). Te PCA results suggest mutation of F42A, Y45A, and L72G in IL-2m might increase the stability by less fexibility observed in comparison to WT IL-2. Likewise, IL-2m2 showed similar PCA results of IL-2m; thus F42Y and L72R increased the structure stability. From a conformational dynamic perspective, IL-2m might not interact with IL-2Rαβc due to the observed dynamic and bent structure of the AB loop along with absence of potential bindinginteracting residues (i.e., F42, Y45, and L72). Although IL-2m2 has not been tested experimentally yet, we assume IL-2m2 might be CD25-independent IL-2 based on our in silico     Figure 4: Te radius of gyration (Rg) was plotted against simulation time (ps). Similar Rg values around 1.52 nm were obtained for WT IL-2 (black), IL-2m (red), and IL-2m2 (green) during the MD simulations (a). Tis indicates similar conformational compactness of WT IL-2 and IL-2m that reported as CD25-independent (i.e., not interacting with IL-2Rαβc). Also, IL-2m2 showed similar conformational compactness. Te surface area solvent accessibility (SASA) for WT IL-2 (black), IL-2m (red), and IL-2m2 (green) as a function of simulation time (b). Tere was no diference between WT IL-2 and its mutants. results. Tis is due to similar efects of alanine compared to tyrosine and glycine compared to arginine in IL-2m and IL-2m2, respectively. Further, the free-energy changes were assessed using free-energy landscape (FEL) analysis for WT IL-2, IL-2m, and IL-2m2 through principal component plots. In consistent with previous results, IL-2m showed minima within a single basin while WT IL-2 showed wider basins (Figures 6(a)-6(b)). Terefore, these mutations in IL-2m might stabilize interleukin conformation. In previous work, IL-2m was prevented from interacting with IL-2Rαβc (high-afnity IL-2 receptor) while interacting with IL-2Rβc [19]. For IL-2m2, we observed slightly similar basins in comparison to WT IL-2 ( Figure 6(c)). Tis might be explained due to the chemical similarity of phenylalanine to tyrosine and lysine to arginine. Terefore, our fnding suggests that mutation of IL-2m prevents IL-2Rαβc binding due to rigid and stable IL-2 structure especially in the potential loops.

Conclusion
Te WT IL-2 binds to IL-2Rαβc and IL-2Rβc receptors while IL-2m only binds to IL-2Rβc receptor. Preventing IL-2 binding to IL-2Rαβc (α subunit known as CD25) receptor has clinical benefts due the known toxicity profle of IL-2. In our study, we performed a comparability study between conformation and fexibility of the WT IL-2 and its mutants IL-2m and IL-2m2 using MD simulations. Our results showed F42A and L72G mutation in IL-2m rigidifed the BC, CD, and AB loops and maintained similar conformational compactness and solvent accessibility compared to WT IL-2. Moreover, IL-2m showed a better dynamic stability according to the PCA and FEL results. Terefore, we suggest rending IL-2 loops fexibility might play a role in the IL-2 interaction to IL-2Rαβc. IL-2m2, F42Y, and L72R mutation produced similar fexibility and conformational results of IL-2m with less efect on the AB conformation. However, fndings of our study can be confrmed using conformational and dynamic experimental studies such as fuorescence and H/D exchange instruments, respectively.

Structure Preparation and MD Simulations.
Te structure of WT IL-2 was obtained from Protein Data Bank (PDB ID 1M47) [30]. Mutations of WT IL-2 to produce IL-2m (F42A, Y45A, and L72G) and IL-2m2 (F42Y and L72R) mutants were performed using PyMOL [31]. For molecular dynamic (MD) simulations, the X-ray structure and modeled structures were used as starting trajectories for WT IL and IL-2 mutants, respectively. Afterward, MD simulations was performed using GROMACS 5.1.4 program with CHARMM27 force feld similar to our previous work with slight changes [26,27,32,33]. Particle mesh Ewald (PME) [34] was used instead of the cutof scheme, and a short-range nonbonded cutof distance of 1.3 nm was used to compute the long-range electrostatic interactions. Ten, solvation of all structures was done with TIP3P water [35] with the minimal distance of 0.407 nm between the solute and the wall of the cubic box the cubic box of length 7.47827 nm. Protonation states were assigned for titratable residues based on pH 7.0 condition using GROMACS pdb2gmx program. To mimic an ionic strength of 0.15 M, an equivalent amount of Na and Cl ions instead of water molecules was added. Ten, a brief energy minimization (20 ps) followed by unconstrained equilibration MD simulation (100 ps) was done at a constant temperature (300 K) and pressure using Berendsen and Parrinello-Rahman coupling methods, respectively. Finally, for WT IL-2, IL-2m, and IL-2m2 solution structures production, a 100 nslong production MD simulation at a constant temperature of 300 K was performed for each system; the temperature was controlled using a velocity-rescale thermostat and a time step of 2 fs.  (PCA). PCA (known as essential dynamics analysis) has been utilized to flter large motions in macromolecules by reducing the number of dimensions that describe protein motions [36,37]. To obtain PCA, frst a covariance matrix was produced by using Cα atomic coordinates of WT, IL-2, and IL-2m2 trajectories. Ten covariance matrix was diagonalized to produce eigenvalues and eigenvectors by gmx covar in GROMACS utilities as 85% of motion is cumulated in frst 2 vectors. Finally, the frst eigenvector, refect WT, IL-2, and IL-2m2 motion behavior was visualized and used to plot the PCA.

Visualization and
AnalysisF. MD simulation trajectories for each system were analyzed using GROMACS 5.6 tools. Root mean square of deviation (RMSD) and root mean square of fuctuation (RMSF) were calculated for WT IL-2, IL-2m, and IL-2m2 using gmx rmsd and gmx rmsf. For radius of gyration (Rg) and solvent accessible surface areaSFF (SASA), gmx gyrate and gmx sasa commands were used. All 2D plots were depicted using xmgrace in GROMACS utilities. PyMol was used to visualize and represent all IL-2 structures and to depict Porcupine plot (Sean M. Law et al.). Graphic visualization for WT IL-2 and its mutants were done using PyMOL 2.1.

Data Availability
All data generated or analyzed during this study are included in this published article.

Conflicts of Interest
Te author declares that he has no conficts of interest.