According to our algorithm, 14 hydration sites were found in the Rhodopsin-G protein system. Among them, 10 sites were identified as functional hydration sites (Table 2, Fig. 3a/4). Hydration sites Wr45, Wr3 and Wr38 formed a hydrogen-bond network with corresponding residues N2.40, Q7.59 and N7.57. One thing to be noted is that the reference number Wr is equal with the initial abbreviation of “Water-rhodopsin”, which means Water molecules in Rhodopsin system. In addition, GPCR residue number, such as N2.40, Q7.59 and N7.57, was encoded based on Ballesteros/Weinstein numbering rules. Hydration sites Wr48, Wr42, Wr43, Wr44 and Wr36 formed a hydrogen-bond network with corresponding residues M7.56, K7.58 and E6.23, while hydration sites Wr40 and Wr47 were hydrogen-bonded to residue T2.37 and R7.61, respectively (Fig. 3a). The hydration site Wr36 was found with the lowest free binding energy, -6.6 kcal/mol, and formed polar interactions with residue E6.23 and K7.58 (Table 2). In the Rhodopsin-Arrestin structure, 14 hydration sites were found, while 9 sites were identified as functional hydration sites (Table 2, Fig. 3e/4). Hydration sites Wr22, Wr17, Wr16, Wr19, Wr18, Wr12 and Wr13, together with corresponding residue K7.58, N7.57, Q7.59, N2.40 and T2.37, formed a hydrogen-bond network (Fig. 3e). The hydration site Wr21 exhibited the lowest free binding energy, -2 kcal/mol, and formed polar interactions with residue R3.50 (Table 2). From these analyses, we can see the interaction energy of several functional hydration sites in Rhodopsin-G protein system was obviously lower than those in Rhodopsin-Arrestin system, in addition to the different hydrogen binding sites caused by the G protein/Arrestin binding interface surface.
In the M2R-G protein system, 10 hydration sites were captured and 7 of them were identified as functional hydration sites (Table 2, Fig. 3b/4). Water molecules Wm21, Wm26, Wm22, Wm19, Wm20, Wm27 and Wm7 were hydrogen-bonded to corresponding residue N2.40, N7.57, Y7.53, D3.49, T6.36 and R3.50 (Fig. 3b). The hydration site Wm22 had the lowest binding free energy, -3.3 kcal/mol, and formed polar interaction with residue N7.57 (Table 2). However, only two functional hydration sites, Wm2 and Wm3, were identified in the M2R-Arrestin structure, and formed hydrogen bonds with residue R3.50 and N2.40, respectively (Fig. 3f). Comparison of the structures of M2R-G protein and M2R-Arrestin systems revealed that the open cytoplasmic part of TM6 accommodated α5-helix of Go and “Finger Loop” of Arrestin [60]. Likewise, the slanted TM6 accommodating the “Finger Loop” of the β-Arrestin 1, which occupied a similar position to the α5-helix, involved a smaller contact surface with a slightly lower depth (Figure S2b). In the Rhodopsin-G protein/Arrestin structures, the insertion depth of the α5-helix and “Finger Loop” was basically same (Figure S2a), which might explain why the number of hydration sites were almost the same in these two structures.
The hydration sites located at the intracellular crevice region of NTSR1 showed obvious diversities in the two activation modes. Firstly, the quantity of functional hydration sites in the G protein-related system was larger than that in Arrestin-associated system (Fig. 3c/3g). 14 hydration sites were caught in the G protein-related system, and 10 functional hydration sites were identified (Table 2, Fig. 3c/4). The functional hydration sites included Wn23, Wn29, Wn5, Wn24, Wn37, Wn30, Wn38, Wn34, Wn8 and Wn32, formed hydrogen bonds with corresponding residue A3.53, L2.35, S2.34, Q2.36, R3.50, Y7.53, L7.52, A6.29, V7.56 and S7.57. However, in NTSR1-Arrestin system, only one functional hydration site, Wn35, was found to have hydrogen bond with residue S7.57 (Fig. 3g). Secondly, the minimum of free binding energy of hydration sites in G protein-related system is smaller than that in Arrestin-associated system. In the G protein-related system, hydration site Wn37 formed polar interaction with residue R3.50 (Fig. 3c) and showed the lowest free binding energy of -3.6 kcal/mol (Fig. 4). While in Arrestin-associated system, the hydration site Wn35 had the lowest binding energy of -3 kcal/mol. Similar with M2R system, changes of the receptor structure might be responsible for the above-mentioned diversities. Alignment analyses showed that the insertion of α5-helix from G protein into the intracellular cleft region was about 6 Å deeper than that of “Finger Loop” from NTSR1-Arrestin complex (Figure S2c), which also provided a larger contact surface.
Different from the above-mentioned three types of receptor systems, the quantity of hydration sites was more abundant in the β1AR-Arrestin system than that in G protein-related structure (Fig. 3), which was probably caused by the specific structure of β1AR complexes. The “Finger Loop” of Arrestin was deeply inserted into the intracellular cleft region, about ~ 3 Å deeper than that induced by the α5-helix of G protein (Figure S2d). In the β1AR-G protein system, 10 hydration sites were found, and 6 sites were identified as functional water molecules (Table 2, Fig. 3d/4). However, in the Arrestin-associated system, 18 hydration sites were found, and 14 sites were identified as functional hydration sites (Table 2, Fig. 3h/4). Moreover, hydration sites Wb8, Wb28, Wb9, Wb3, Wb22 and Wb24 were found to form hydrogen bonds with corresponding residue Y3.60, D3.49, R3.50, V5.61 and K6.32 in the G protein-related system, while the hydration site Wb28 had the lowest binding free energy of -6.2 kcal/mol and formed polar interaction with residue D3.49 (Table 2). In the Arrestin-associated system, hydration sites Wb37, Wb7, Wb33, Wb32 and Wb36, together with residue R3.50, Y3.60, Q3.61 and M3.64, formed a hydrogen-bond network. At the same time, hydration sites Wb42, Wb41 and Wb10 formed a hydrogen-bond network with residue T2.39, T2.37 and N2.40. The hydration sites Wb29, Wb51, Wb46, Wb52, Wb50 and Wb39 interacted with residue A3.53, Y5.58, T6.36, I7.52 and D7.58 (Fig. 3h). The hydration sites Wb32 was found with the lowest binding free energy, -3.7 kcal/mol, and formed polar interaction with residue Q3.16 (Table 2). In general, the functional water molecules with the minimum of free binding energy could be found in the interface surface of every GPCR-G protein-related system, instead of Arrestin-associated structure.
Studies have confirmed that water molecule is rich in protein-protein interface. Francis et al performed analyses on the water molecules fixed at the protein-protein interface of 115 homomeric proteins and 46 protein-protein complexes [61]. They found that there were 15 water molecules on average in per 1000 Å of each protein interface. Moreover, previous studies have indicated that water-mediated contacts might carry momentous information complementary to direct contact information content through the method of the energy landscape theory of protein folding/binding [62]. Our research showed that hydration sites differed significantly in terms of existing number and binding energy between G protein-related and Arrestin-associated structures, which might contribute to some roles for the functional selectivity of the receptor. Water-mediated polar interactions were more abundant at the protein-protein interfaces with a deeper insertion of G protein or Arrestin than that with shorter embedding, but not superficially the interface area size (Figure S3). To note that, the functional water molecules with the lowest free binding energy could be found in the interface surface of every GPCR-G protein-related system, instead of corresponding Arrestin-associated structure (Fig. 4/S4, Table S5). In addition, in the interface surface area of Arrestin-associated systems, a small fraction of water molecules with the energy < 0 formed stable hydrogen bonds with residues, while the rest of water molecules acted to fill the space.