A comprehensive analysis of Protein Data Bank reveals low desolvation 1 penalty in π-cation system

28 Cation-π interactions widely exist between ligand-protein interfaces, attracting much attention in 29 molecular recognition in recent years. Interactions named cation-π and π-cation (cationic vs 30 arene small molecular ligands) shall be separately considered in drug and pesticide design 31 process. The two interactions involved in ligands and protein pockets may differ in energy 32 features and therefore offers significant inspiration for drug and pesticide design. However, an 33 in-depth study on differences between cation-π and π-cation systems from an energy perspective 34 is still lacking. In this study, we calculated and compared cation-π and π-cation systems in terms 35 of physicochemical properties of ligand/protein and solvation effect. It seems that the desolvation 36 penalty of the cation-π systems was relatively higher than the π-cation pairs, even though these 37 interactions both can improve the ligand activity. This is the reason for evolution converged on 38 π-cation interactions in the cation-π-mediated proteins. The π-cation interaction facilitating the 39 inhalation of ligand to the pocket may provide a new sight for the molecular design of 40 pharmaceuticals and pesticides. 41 42 47 Surface Area; PDB: Protein Data Bank; IC 50 : the half-maximum inhibition concentration; K i : the inhibitory constant; MAPK: the mitogen-activated protein kinase; LTA4H: the leukotriene 4


3
Introduction 49 The cation-π interaction, dominated by the electrostatic attraction between electron-rich aromatic 50 rings and positively charged groups, is an essential topic in structure-based drug and pesticide  interactions into cation-π and π-cation pairs in drug or pesticide and target interactions (Fig. 1) 66 [13]. The π element in a cation-π pair is typically provided by the side-chains of aromatic 67 residues, Trp, Tyr, and Phe, while the positively charged ligands act as "cation". For π-cation 68 pairs, two protonated amino acids (Arg and Lys) are "cation", with arene ligands providing the π  approach can be precise in drug and pesticide design. However, the energetic difference between 80 cation-π and π-cation in drug and protein interaction systems remains unclear. While 81 understanding the difference between the two systems makes it easier to use cation-π and 82 π-cation systems for rational drug and pesticide design. Therefore, there is a strong need to 83 perform a systematic analysis of the energetic difference between these two interactions in all 84 available protein structures.

86
In this study, we get 1334 cation-π systems and 2174 π-cation systems from 141,706 crystal and promote the coevolution of proteins and ligands. The reduced desolvation penalty may be 94 induced by the physiochemical property difference between π-supplier and cation-supplier. 95 Therefore, taking a proper π-supplier of the ligand scaffold into account will guide how to design 96 potent drugs and pesticides.

99
The definition of cation-π and π-cation interactions 100 The PLIP was used to geometrically identify cation-π interactions with default parameters [21].

101
Furthermore, the cation-π and π-cation interactions were classified by the ligand groups.

102
Tertamine, quartamine, guanidine, and sulfonium ligands provided cations for cation-π 103 interactions. Aromatic ligands were π components in π-cation interactions.   The Sander module of Amber16 was used to minimize the energy of complex structures with 113 implicit solvent [28]. The AMBER ff14SB force field was used for residues, and the general 114 6 AMBER force field (gaff ) was used for ligands [29,30]. The cutoff distance for the long-range 115 electrostatic interaction was set at 10.0Å. The minimization procedure consisted of the following 116 three steps [31]. First, only hydrogens, ions, and water molecules were allowed to move, and the 117 solute remained fixed with a constraint of 500 kcal mol -1 Å -2 . Then, the backbone atoms of the 118 protein were fixed, and other atoms were relaxed. Finally, all the atoms of the system were free 119 to move. In each step, the steepest descent method was used for the first 2000 cycles and the 120 conjugate gradient method was used for the next 1000 cycles to perform energy minimization  The ΔG solv can be divided into two partsthe electrostatic desolvation penalty (ΔG PB/ GB ) and the   Further, the data analysis for cation-π and π-cation residue pairs was performed. 67.42%, which was relatively higher than that of Lys (32.58%, Fig. 2b). These results implicated 159 that Trp and Arg were essential in the cation-π and π-cation interactions, which was consistent  (Fig. 3a). However, there was a slight difference between cation-π and π-cation: the 171 hydrophobic scores of π-cation pairs clustered in the interval of -5 to 45, while the most scores of 172 cation-π moved to the interval of 5 to 45. From violin plots, we found that most score (cation-π: 173 79.60%, π-cation: 77.60%) lied in the range of 5-35 (Fig. 3b). Nevertheless, the shape of π-cation 174 moved down compared with that of cation-π, revealing that the binding pockets of cation-π were 175 much more hydrophobic. ligands was studied to make the difference between cation-π and π-cation systems. Just like the 181 hydrophobic score, the ligands in π-cation pairs were the lower pKa biased (Fig. 4a). When the 182 pKa was in the range 0 to 25, the possession percentage of π-cation systems was lower than the 183 cation-π. For the range -15 to 0, the possession percentage of π-cation systems was much larger 184 than that of cation-π. Fig. 4b showed the different shapes of two violin plots. In the π-cation 185 system, there were two wide shapes in the interval of -10 to 0 and the interval of 0 to 10. For the 186 cation-π systems, three wider sections were shown. Meanwhile, the first wide part of cation-π 187 was relatively higher than that of π-cation. On the other hand, the last wide shape of π-cation was 188 lower than the cation-π. The median of cation-π (~10) was higher than π-cation (~5). These 189 results indicated that the basic ligands preferred to act as cations in cation-π systems. is a difference in the desolvation penalty between cation-π and π-cation systems is still unknown.

206
In this study, the SOL_value, avoiding the system interference, was defined to determine the there was more cation-π interaction (45.50%) than π-cation interaction (27.37%). When the 216 SOL_value is ranged from 0.45 to 0.47, the percentage of cation-π (26.76%) was much larger 217 than that of π-cation (13.01%). In contrast, the percentage of π-cation was larger than the SOL_value range of 0.00~0.30, in which the percentage of π-cation was 10.64% higher than that 221 of cation-π. The bottom of the violin plot of cation-π was much thinner than that of π-cation 222 systems, which demonstrates that there were fewer values in comparison to π-cation systems 223 11 when tending to 0.00 (Fig. 5b). In summary, the SOL_values of over 72% of π-cation systems 224 are much closer to 0.00. This means that the π-cation interactions in the complexes may result in 225 a lower desolvation penalty which is conducive to the binding affinity. It is maybe the reason for 226 the enrichment of π-cation in the natural systems.

228
Case study 229 The above results revealed that the hydrophobicity of cation-π pockets was relatively higher,  (Table S1 and S2). Further, we 235 used two cases to explain the difference between cation-π and π-cation in the protein systems 236 ( Fig. 6a and b). Serine/threonine kinase acts as an essential component of the mitogen-activated interaction with Tyr267 and Tyr378. It was common that the hydrophobicity score of the 245 MAPK1 pocket for π-cation pair was lower than that of LTA4H for cation-π. Meanwhile, the 246 12 SOL_value was improved after the introduction of cation-π and π-cation interactions. However, 247 these interactions both can strengthen the interaction of the ligand with receptors, even though 248 the SOL_value of 27P-LTA4H (0.41) was much higher than 33A-MAPK1 (0.18). Meanwhile, it 249 should be noted that the activity fold changed of 33A (26.74) was much higher than 27P (3.35).

250
These results are consistent with our conclusion that the π-cation interaction has a lower 251 desolvation penalty than cation-π systems, which would contribute to the ligand binding. The cation and π interactions play an essential role in drug(pesticide)-target interaction.

255
Differentiating cation-π and π-cation systems can facilitate the rational molecular design of 256 pharmaceuticals and pesticides. We comprehensively compared the physicochemical properties 257 of protein pockets and ligands and solvation energy for cation-π and π-cation systems in this 258 study. Compared with a cation-π system, a π-cation system probably results in a lower  Competing interests 286 The authors declare that they have no known competing financial interests or personal 287 relationships that could have appeared to influence the work reported in this paper.

Figure legends
398 Fig. 1 Definition of cation-π and π-cation interactions. For the cation-π pair, the π system is 399 typically provided by the aromatic side-chains of Trp, Tyr, and Phe, while the protonated ligand 400 exists as the cation (PDB code: 1ax9). In the π-cation system, protonated Arg and Lys always act 401 as cations with the arene ligands acting as a π partner (PDB code: 1bkm).   π-cation interactions for IC 50 /K i , hydrophobic score (Hyd score), and SOL_Value.