2.3.1. Interaction Networks
Interaction network analyses were performed on the crystallographic structures of KirBac3.1 WT (PDB 2WLJ) [14
] and on the KirBac3.1 W46R mutant model (see Materials and Methods) to elucidate the impact of the mutation W46R on the structural and dynamical properties of the protein. The residue W46 in KirBac3.1 WT can adopt two alternative conformations, called flipped-in, in the majority of cases (side chain of W46 oriented towards the channel) and flipped-out (side chain of W46 oriented towards the outside of the channel) (Figure 3
A). Figure 3
C presents details on the flipped-in structure. The atomic structure of the W46R mutant shows that the four R46 side chains point away from the channel pore in the crystal structure of the mutant W46R (Figure 3
B). This is also observed in the computational structures of KirBac3.1 W46R (98.68% of the 25415 relaxed MDeNM structures).
The first analysis focused on the KirBac3.1 WT: we compared the interaction networks (hydrogen bonds, van der Waals interaction, π-π and π-cation interactions) at the W46 residue level in flipped-in and flipped-out configurations. Interaction networks around W46 for these two configurations are shown in Figure 3
D–G respectively for chains A, B, C, and D, separately (for more details, see Figure S2
). W46 is tightly packed against the C-terminal end of TM2 of the same subunit and interacts with R134 and F135 at the bottom of this inner helix in the flipped-in conformation. Conversely, in the flipped-out conformation, these interactions are not possible, confirming the movement of the W46 residue away from the bottom of the inner helix and the center of the channel.
The second analysis focused on comparing the interaction network around the W46 residue, in its flipped-in conformation (the most frequent conformation) in KirBac3.1 WT and the interaction network around R46 in the W46R mutant (Figure 3
H–K and Figure S2
for more details). Some similarities are evident, e.g., in all the four subunits, corresponding to chains A, B, C, and D, both W46 and the mutated R46 residues interact with the residues I50 and F49 on the same external helix TM1. However, there are some clear differences. The interactions between the flipped-in W46 residue and the two R134 and F135 residues at the bottom of the TM2 are lost in the case of the W46R due to the orientation of the mutated residue R46 toward the outside of the channel. Along with this loss of interactions, a new network around R46 is established and interestingly new interactions can take place with D36 and/or W38 residues in the slide helix of the adjacent subunit (n − 1). It is important to highlight that only R46 residues from two opposite chains (A and C) interact with the slide helix. Such contact is not evident for the other two opposite chains (B and D). These observations suggest that the mutation R46 alters the interaction of the outer and inner helices at the level of the bundle crossing, disrupts a hydrophobic cluster (Figure S3
), and promotes the interaction of the bottom end of the outer external helix with the slide helix of the adjacent (n − 1) subunit. The R46 side chain points towards the outside of the channel and shows similarities with the flipped-out configuration of W46 residues in the wild type (WT).
2.3.2. Contacts between Residues
We calculated the accessible surface areas (ASAs) of residue 46 (W in WT and R in the W46R mutant) on all the 25,415 relaxed MDeNM structures. This value measures the residue’s interactions in this position and its degree of exposure to the solvent. The histograms of the ASA values given in Figure 4
A show that R46 in the mutant is more exposed than W46 in WT. The corresponding average value is 87.42 Å² with a standard deviation (SD) of 24.87 Å² for the WT and 158.7 Å² (SD = 30.81 Å²) for the mutant. These values are in accordance with our crystallographic data showing that the residue R tends to point out of the channel. We should also consider the different hydrophobicity of the two residues. R is more hydrophilic than W and therefore has more favor interactions with the water environment or the polar heads of the phospholipids.
In the closed flipped-in state, the side chain of W68 in Kir6.2 (PDB 6JB1) is directed towards the C-terminal end of TM2 and holds hydrophobic contacts with residues I167, K170, and T171 [24
]. In the homologous KirBac3.1 channel, W46, corresponding to W68 in Kir6.2, is also directed to the C-terminal end of TM2 (Figure 1
B and Figure 2
C) and is close to the I131 (corresponding to I167 in Kir6.2), R134 (corresponding to K170 in Kir6.2), and F135 (corresponding to T171 in Kir6.2).
In KirBac3.1, the proximal residues R134 and F135 are more exposed to the solvent in the mutant W46R than in the WT (Figure 4
B,C), supporting the suggested loss of interaction with R46 in the mutant protein. It is worth noting the significant increase in ASA value for R134 compared to F135. R134 shows an ASA average value of 21.77 Å² (SD = 10.96 Å²) in the KirBac3.1 WT and 41.05 Å² (SD = 18.19 Å²) in the mutant W46R. F135 shows an average value of 38.90 Å² (SD = 13.87 Å) and 44.27 Å² (SD = 17.59 Å²) in KirBac3.1 WT and W46R mutant, respectively. These results are explained by the fact that R134 is closer to the bottom end of TM2 and, therefore, more accessible to the solvent. It is also interesting to note that R46 has a higher ASA value than R134, indicating the significant exposure and flexibility of the latter.
The distribution of shortest distances between residues W46/R46 and I131 calculated on the MDeNM set of conformations shows that the shortest distance is higher (for all the chains) in the mutant (Figure 4
D). The average values are 2.45 Å (SD = 0.37 Å) and 6.20 Å (SD = 0.9 Å) for the WT and the W46R mutant, respectively.
All the evidence gathered with this study points to a clear disruption of these crucial interactions between TM1 and TM2 in the case of the W46R mutant, compared to the WT protein.
2.3.3. Impact of the Mutation on the Gating
In order to investigate the effects of the mutation W46R on various dynamical components of the gating mechanism from the selected normal modes described in Materials and Methods, we carried out MD simulations using MDeNM including the entire environment (water, lipids, ions). Structural and dynamical features taken into account are (i) the movement of the slide helix of each chain, (ii) the movement of the whole cytoplasmic domain, and (iii) the coordinated movement of the transmembrane helices.
The W46R mutation’s impact on the motion of the slide helix was analyzed considering the interaction of the R46 residue with residues of the slide helix of the adjacent protein monomer (n − 1). To this end, we calculated the correlations between the upward movement of the slide helices (definition of the angle in Material and Methods and Figure 5
B, in pink) and the shortest distances between R46 at the extremity of TM2 and D36 located at the extremity of the slide helix. This movement has been previously highlighted as important in the gating of the channel [18
]. Figure 5
A shows the positive correlation between helix movements and inner-residues distances in blue circles, while red circles indicate negative correlations. The diameter and color intensity of each circle are proportional to the strength of the correlation. The shortest distances between R46 and D36 of neighboring chains (n − 1) are mainly correlated with the upward movement of slide helices. Therefore, it can be inferred that a smaller distance between these two residues induces a larger upward movement of slide helices.
More precisely, (i) the shortest distance d46a36d, d46b36a,
(see Figure 6
’s caption for definitions) are correlated with the upward movement of the chain A (upa
); (ii) distances d46c36b,
are correlated with the upward movement of the chain B (upb
); (iii) the distances d46d36c,
are correlated with the upward movement of the chain C (upc
). Altogether, this evidence points towards a clear correlation in the mutated protein between the upward movement of the slide helices and shortened distances between R46 and D36 residues belonging to the adjacent chain (n + 1 or n − 1). In the WT protein, which is mostly in flipped-in conformation, (98.68% of the relaxed MDeNM structures), such correlations are still visible but involve the adjacent chain (n + 1) (Figure 5
We analyze here the movement of the cytoplasmic domain since it has been considered an essential factor in gating [18
]. To investigate its relation with the upward movement of the slide helices, the shortest distance between D35, the first residue of the slide helix, and R167, located at the top of the cytoplasmic domain on the CD-loop, was taken as a measure of the proximity of these two structural regions. Also, inter-chain distances were taken into account to corroborate the findings. Significant correlations were observed between these distances and the upward movement of the slide helices, as shown in Figure 5
A. Furthermore, strong negative correlations are present between inter-chain distances and the upward movement of the slide helices to the membrane. When these distances decrease, the slide helices come closer to the membrane. More precisely, d35a167b
are inversely correlated with the upward movement of the chain A (upa), and d35b167c
are inversely correlated with a similar upward movement of the chain C (upc); similarly, d35d167a
has an inverse correlation with the upward movement of chain D (upd). Interestingly, the WT protein structure displays the same type of correlations (Figure 5
For each monomer of the protein, the comparison of correlations involving the distance between residue 46 and residue 36 (d46n36n − 1) on the neighboring chain (n − 1) and the distance between residue 35 and residue 167 (d35n167n + 1) on the neighboring monomer (n + 1) for KirBac3.1 WT and KirBac W46R reveals clear differences. Indeed, these distances are correlated in the case of KirBac3.1 W46R but not in the case of the KirBac3.1 WT. We hypothesize that this is the consequence of a new interaction present in the W46R mutant but not in the WT protein between R46 and D36 on the slide helix. While in the WT, the residue D35 interacts with R167 on the same polypeptide chain, the mutation at position 46 creates an inter-chain interaction between D36 and R46. The slide helix is, therefore, held in place by two interactions in different directions. In the case of the WT, the interaction D35-R167 promotes the orientation of the slide helix towards the cytoplasmic domain, significantly reducing its upward movement. Conversely, the formation of the new R46-D36 interaction in the W46R favors the movement of the slide helices toward the transmembrane domain.
The correlation between the upward movement of the slide helix and the tilting of the inner and outer transmembrane helices may affect the channel’s gating at the constriction points L124 and Y132. Definition of the inner and outer tilts is shown in Figure 6
B in red and blue, respectively, and Materials and Methods. The MDeNM sets for the mutant protein were analyzed to highlight correlations between structural determinants involved in the slide-helix movement and the tilting of the transmembrane helices (Figure 6
A). High correlation values indicate the interplay between the inner and outer helices’ tilt angles with the movement of the slide helices. Similar correlations were obtained for the WT (Figure 6
In the center of the channel, two constriction points corresponding to residues L124 and Y132 are important for the gating as they control the passage of the potassium ions through the channel. In more detail, the gating at L124, represented by inter-chain distances g124ac and g124bd is correlated with the upward movement of the slide helices (upa,b,c,d). In addition, the g124ac distance is correlated with the tilt of the transmembrane helices (tiltin, tilton). Conversely, the opening of the gate at the level of the residue Y132 represented by the inter-chain distances g132ac and g132bd has an inverse correlation with the upward movement of the slide helices (upa,b,c,d) and the tilt of the transmembrane helices (tiltin, tilton). In the case of KirBac3.1 WT, data from the MDeNM set show no clear correlation between the gating at the level of residue L124, while the opening of the gate at the level of the residue Y132, specifically for chains B and D, is mainly correlated with the upward movement of the slide helices and the tilting of the transmembrane helices (conversely to KirBac3.1 W46R).
The upward motion of the slide helix causes the tilt of both transmembrane helices in such a way that the central portions of the transmembrane helices move away from the center of the channel (L124) while their cytoplasmic ends get closer to the bottom of the channel (Y132). This causes the increase of the opening of the channel at the level of L124 but a decrease in the channel’s opening at the level of Y132.
2.3.4. Population of Open and Closed States Corresponding to the MDeNM Relaxed Structures
Note that in this work, a “closed” state is defined by a conduction pathway that is sterically occluded and an “open” state in which the pathway is sufficiently wide to accommodate at least a non-hydrated potassium ion. Four channel-gating states can be defined based on the open or closed conformation of the two main constriction points (L124 and Y132) as observed in the relaxed structures (for more details, see [18
]): (1) Fully open state when the shortest distances at the two constriction points are larger than the ionic diameter of K+
(diameter of K+
ion considered 3.53 Å); (2) Fully closed state when both distances are smaller than the ionic diameter of K+
; (3) Half-open state 1 when the gate at residue L124 is open and the gate at residue Y132 is closed; (4) Half-open state 2 when the gate at residue Y132 is open and the gate at residue L124 is closed. Interestingly, the fully open state structures constitute about 7.4% of the whole population for the mutant protein, a slightly higher percentage (0.5%) than KirBac3.1 WT (6.8% Table 2
). Furthermore, we can notice that the gate at residue Y132, closer to the cytoplasmic domain, is more often closed than the gate at residue L124 (32.5% versus 10.8%, respectively). It follows that the residue Y132 in both protein WT and the W46R mutant protein is the most restrictive point of the ion pathway. A more significant observation is that the “gate” at L124 is more open in the case of the mutant W46R compared with WT (33% and 29%, respectively). Conversely, the gate at Y132 is less open in the mutant W46R than WT (11% and 14%, respectively).
More details about the amplitude of the channel opening at the gating points on the scatter plots of shortest distances between the opposite chains A and C and those between chains B and D at both constriction points over all the MDeNM relaxed structures are shown in Figure 7
A,B for the WT and the W46R mutant proteins, respectively. The shortest distance between opposite chains at the level of the two constriction points is always lower than 15 Å for WT protein (Figure 7
A) but higher than 15 Å for a small population of the simulated structures of the W46R mutant (Figure 7
B). Interestingly, for this population, the opening is more prominent at residue Y132 compared with residue L124.