Dynamics and Interactions of GPI-Linked Lynx1 Protein with/without Nicotinic Acetylcholine Receptor in Membrane Bilayers

Abstract

Nicotinic acetylcholine receptor (nAChR) participates in diverse biological processes, e.g., mood, learning, and addiction. Glycosylphosphatidylinositol (GPI)-linked lynx1 is an allosteric modulator of nAChR function, including shifts in agonist sensitivity, reduced desensitization, and slower recovery from desensitization. This modulation is thought to be achieved by lynx1’s interaction with nAChR subunits, particularly at the α4:α4 interface. In this study, we used molecular modeling and simulation to study the structure, dynamics, and interactions of lynx1 when bound to nAChRs, as well as unbound, monomeric lynx1, when embedded in membranes. Though lynx1 structures are similar in both states, lynx1 dynamics are more restricted in the bound state than in the unbound one. When bound, interactions between lynx1 and nAChR are observed to be maintained throughout the simulations. Of particular note, lynx1 demonstrates prolonged interactions with the receptor C-loop in one of the nAChR α4 subunits, a region important for agonist binding and possibly the transition between open/close states. During interactions with lynx1, an α4 C-loop tends to be restricted in either close or open state, whereas the C-loop state transitions are more evident in when the nAChR is unbound by lynx1. Interestingly, the conformational change of the C-loop is stochastic, suggesting that lynx1 can influence nAChR (critical for its multimodal action), for instance by shifting its agonist sensitivity and recovery from desensitization.

Graphical Abstract

Introduction

The cholinergic system is involved in diverse functions including mood, anxiety, learning and memory, and addiction ^1,2. The nicotinic acetylcholine receptors (nAChRs) of the cholinergic system are comprised of 15 different subunits (α1-10 and β1-4) that assemble into both homopentameric receptors such as (α7)₅ ^3-5 or (α9)₅ ⁶ and heteropentameric receptors such as α4β2 ^7-11. Subunit composition and even the position of specific subunits within pentamers can influence the biophysical, pharmacological, and cell biological properties of the receptor, with profound effects on information processing in neurons that express these nAChRs.

The basic conformational states of nAChRs are the closed state, the open state, and the desensitized state. There is considerable evidence that the movement of the C-loop is important in the transition from the closed (agonist unbound) state to the open (agonist bound) state ^12,13. Agonist binding is thought to accompany movement of the C-loop in closer approximation to the agonist binding site (i.e., the so-called capping of the agonist binding site) along with subsequent changes in conformation of the pore lining transmembrane domain ¹⁴. Multiple factors can influence these transition states, including subunit composition, nicotine exposure, calcium, etc.

Subunit composition influences the properties of the two possible pentameric stoichiometries of α4β2 nAChRs, the most abundant heteromeric nAChR subtype in the brain. The lower sensitivity subtype (LS) receptors adopt the (α4)₃(β2)₂ stoichiometry, while the higher sensitivity (HS) receptors adopt the (α4)₂(β2)₃ stoichiometry. These two stoichiometries exhibit different desensitization kinetics and respond differently to chronic nicotine treatment. HS receptors are upregulated in response to nicotine, while LS receptors are unaffected. This is thought to play an important role in mechanisms of nicotine addiction. Additionally, subunit composition could possibly influence sub-cellular localization of receptors on the plasma membrane ¹⁵.

The lynx family proteins are a subset of the Ly6/uPAR superfamily, coding for cysteine-rich proteins adopting a three-fingered toxin fold ^16-21. A superfamily is related to the elapid snake venom toxin genes, such as α-bungarotoxin, consisting of multiple internal disulfide bonds between the conserved cysteine residues ^22-25. α-bungarotoxin and other such toxins exemplify a highly-conserved receptor binding motif that apparently evolved from lynx proteins ²². α-neurotoxins, such as α-bungarotoxin, have become widely used probes for the investigation of the properties of nAChRs.

Studies have indicated preferential interactions of lynx1 to the LS stoichiometry, particularly the α4:α4 nAChR subunit interface, which only exists with the (α4)₃(β2)₂ stoichiometry of α4β2 nAChRs ²⁴. Specifically, hydrogen bonds and cation-π interactions were observed between Arg38 of lynx1 and Trp156 / Tyr204 on the α4:α4 interface ²⁶. Computational modeling of lynx1 complexes with (α4)₃(β2)₂ suggests steric hindrance of lynx1 at the α4:β2 interface that would not exist at α4:α4 interfaces, proving a possible explanation for lynx1’s preferential binding to the LS stoichiometry over the HS one ²⁶.

A glycosylphosphatidylinositol (GPI)-linked prototoxin molecule such as lynx1 has been demonstrated to alter the stoichiometry and assembly of nAChRs ²⁴. Further, the presence or absence of a GPI-anchor has been shown in some cases to influence neuronal signaling ²⁷. Therefore, the existence of the GPI-anchor on lynx1 proteins could contribute to the structure, composition, or function of nascent nAChRs as well. GPI-anchored neurotoxin-like receptor binding proteins are confined to a volume above the extracellular face of the membrane. This topology positions then exert effects by binding to the extracellular portion of receptors.

Lyukmanova et al. ¹⁹ determined the first three-dimensional (3D) structure of the water-soluble domain lynx1 (i.e., without the GPI-anchor) by NMR. This structure becomes an important model for the lynx1-mediated nAChR structure-function studies. For example, based on this water-soluble model, a series of lynx1 mutations were also studied ²⁸ for understanding the binding and functional properties of lynx1 protein. However, there is little detailed biophysical information on how lynx1 interacts with nAChR or how these interactions lead to changes in nAChR function. The most abundant nAChRs in central neurons are α4β2 heteropentamers and α7 homopentamers ^16,29-31. Due to the lack of 3D structures for α4β2 and α7 nAChRs, most studies used acetylcholine-binding proteins (AChBPs) and Torpedo nAChR to elucidate the nAChR functions. Recently, Walsh et al. ³² determined the 3D structure of human α4β2 nAChR and Nissen et al. ²⁶ modeled a first complex structure of lynx1 with α4β2 nAChR. Preferable binding of lynx1 at the α4:α4 interface of nAChR was obtained through all-atom molecular modeling and molecular dynamics (MD) simulations of lynx1 at the α4:α4 interface (i.e., without the GPI-anchor and with only two α4 interface).

A GPI-anchor is presumably important in the function of lynx1 on nAChR function, but none of the previously published simulations have been conducted with GPI-anchored lynx1. Here, we performed all-atom molecular modeling and MD simulations of GPI-anchored lynx1 with the α4:α4 nAChR interface in the context of a membrane-embedded (α4)₃(β2)₂ nAChR. We also performed all-atom MD simulations of GPI-anchored lynx1 in a bilayer without nAChR. These simulations provide insights into the structure, orientation, and dynamics of GPI-anchored lynx1 on a bilayer surface in the presence or absence of nAChR, as well as detailed interactions between GPI-anchored lynx1 and nAChR in a bilayer.

Methods

To gain insights into how the GPI-anchored lynx1 protein behaves in a membrane bilayer and whether the lynx1-nAChR interactions influence lynx1’s structure, orientation, dynamics, and nAChR C-loop motions, we performed all-atom MD simulations of GPI-anchored lynx1 (lynx1-only system), a GPI-anchored lynx1-nAChR complex system, as well as an apo-nAChR system with three independent replicas of each system. All simulations were performed with NAMD ³³. We used the CHARMM36(m) force field for protein ³⁴, carbohydrates ^35-37, and lipids ^38,39. The NMR structure of lynx1 (PDB ID: 2L03) ¹⁹ and structure of nAChR (PDB ID: 6CNK) ³² were taken from the Protein Data Bank ⁴⁰. We used the TIP3P water model ⁴¹ and counter ions of K+ and Cl⁻ with a concentration of 0.15 M to neutralize the system.

Cholesterol (Chol), phosphatidyl-ethanolamine (PE), and phosphatidyl-choline (PC) are reported to be enriched in synaptic plasma membranes ⁴². Chol and sphingomyelin (SM) are postulated as important lipid types in membrane trafficking ^43-45. We chose to use the same lipid composition as in Sunshine et at. ⁴⁶ (DOPE:DOPC:PSM:Chol = 2:1:5:2) for all systems (Figs. 1A, C) to represent a neuronal plasma membrane using dioleoyl-phosphatidyl-ethanolamine (DOPE), dioleoyl-phosphatidyl-choline (DOPC), and palmitoyl-sphingomyelin (PSM). Lynx1 was anchored to the membrane with a GPI-anchor linked to the C-terminus of the protein. The GPI structure is shown in Figure 1B. The membrane and GPI-anchor structure were generated through CHARMM-GUIMembrane Builder ^47,48 and Glycolipid Modeler ^49,50. The initial system size was 100 Å × 100 Å × 115 Å for the lynx1-only and apo-nAChR system and 150 Å × 150 Å × 160 Å for the lynx1-nAChR system. All simulation systems and simulation inputs were generated through CHARMM-GUI ^51,52. Visualization was done via VMD and PyMOL ^53,54, and all analyses were done by CHARMM ⁵⁵.

Figure 1.

(A) Snapshot of the lynx1-only system: lynx1 in yellow, GPI-anchor in magenta sticks, and membrane bilayers colored with DOPE in brown, DOPC in cyan, PSM in red, and Chol in green. Lipid head groups are shown as van der Waals spheres and tails as sticks with each lipid color. (B) A GPI-anchored structure used in this study: mannose (green circles), glucosamine (blue cross square), phosphoethanolamine (EtNP), and myo-inositol 1-phosphate (white hexagon). (C and D) Snapshots of the lynx1-nAChR system: (α4)₃(β2)₂ structure of nAChR was employed here with α4:α4 facing lynx1. β2 colored in light gray and α4 in pink, green, and blue, respectively. The C-loop of α4₁ in dark blue as critical interaction site with lynx1 protein. The lipid composition is the same in the two systems: DOPE:DOPC:PSM:Chol = 2:1:5:2. In (A) and (C), water molecules, ions, and lipids located in front of proteins are not shown for clarity. (E) Schematic representation of lynx1 tilt (α) and rotation (β) angles with respect to the membrane whose normal is the Z axis. The three selected residues used to define α and β are shown as van der Waals spheres colored red (Arg38), pink (Thr30), and green (Pro47).

In this work, the van der Waals interactions were smoothly switched off over 10-12 Å by a force-based switching function ⁵⁶ and the long-range electrostatic interactions were calculated by the particle-mesh Ewald method ⁵⁷ with a mesh size of ~1 Å. The timestep was set to 2 fs and we constrained bonds containing hydrogen atoms with the SHAKE algorithm ⁵⁸. Temperature was held at 310 K, the constant temperature was controlled by Langevin dynamics with a damping frequency of 50 fs^{−1 59}. We first relaxed the system in a canonical ensemble (NVT) with all the solute atoms subjected to harmonic restraints. A 100-120 ps isothermal–isobaric ensemble (NPT) was then applied to adjust the solvent density. The Langevin piston method was used to control the constant pressure ^60,61. A dihedral restraint force constant was set to 1 kcal/(mol·rad²) to maintain the carbohydrate chair conformation during these equilibration steps. Simulations were performed using NAMD ³³ (lynx1-only and lynx1-nAChR) and OpenMM ⁶² (apo-nAChR). For the production run, each lynx1-only replica was simulated for 1.2 μs, each lynx1-nAChR replica for 500 ns, and each apo-nAChR replica for 300 ns. Trajectories were saved every 25 ps. All the results shown in this work were based on the analysis for production simulations.

The (α4)₃(β2)₂ nAChR stoichiometry selected for this work was based on the most abundant subtypes expressed in the brain and previously reported interactions of lynx1 with the α4:α4 interface ²⁶ (Figs. 1C, D). It was demonstrated previously that the interaction between lynx1 and nAChR occurs preferentially within the α4₁:α4₂ interface and mostly with Arg38 of lynx1 protein (Figure 1E). Thus, the initial lynx1-nAChR structure was generated with the α4₁:α4₂ interface towards lynx1. After manually opening the C-loop of α4₁, Arg38 was placed into the α4₁:α4₂ interface to generate a starting conformation. This conformation of nAChR was used for the generation of apo-nAChR systems. And this initial conformation was also set as the reference calculating the C-loop motion of nAChR α4₁ subunit. For the lynx1-nAChR system, distance restraints of 3 Å were applied between Arg38 of lynx1 and Trp156 and Tyr204 on nAChR α4₁ subunit during the first 100-ns of the MD simulations to relax the system, then all the restrains were released for 500-ns production. Trp156 and Tyr204 are the residues of the α4₁ subunit located in the α4₁:α4₂ interface area that are known to form hydrogen bonds with Arg38 in our previous work ²⁶.

Tilt (α) and rotation (β) angles defined in this study are schematically shown in Figure 1E. Three residues located near the central section of lynx1 were selected for α and β calculation: Arg38, Thr30, and Pro47. A vector from Arg38 to Thr30 (${\overrightarrow{r}}_1$) and a vector from Arg38 to Pro 47 (${\overrightarrow{r}}_2$) were used to define α as the angle between ${\overrightarrow{r}}_2$ and $\overrightarrow{z}$ (unit vector in the Z direction) and β as the angle between the normal vector formed by ${\overrightarrow{r}}_1\otimes {\overrightarrow{r}}_2$ and $\overrightarrow{z}$.

The Z-coordinate distribution range of the C-loop in each nAChR subunit was regarded as the potential interaction range for the lynx1-only system in this work. This criterion was made because we observed multi-mode interactions between lynx1 and nAChR C-loop in all complex system replicas (see Results and Discussion). Thus, we regard the C-loop as a critical interaction site for lynx1 protein in this study.

Results and Discussion

Lynx1-nAChR Interactions are Multi-modes

To obtain an insight into which specific regions of the nAChR are involved in lynx1-nAChR interactions, we have produced time series of interaction patterns between lynx1 and its environment in three replicas (Figure 2). Trp156 and Tyr204 are two sites on nAChR α4₁ subunit that are located within the interaction cut-off distance (4 Å) from lynx1 Arg38 in the initial structure. And, the “α4₁” label represents residues on nAChR α4₁ subunit excluding Trp156 and Tyr204. Considering interactions with nAChR α4₁ / α4₂ subunits all together (all colored region in Figure 2), Arg38 spent the majority of the time located within the interaction region of nAChR, which provides support of these interactions throughout our simulation time. Even though these interactions can be observed in either replica, the specific residues that Arg38 interacts with are different in each situation.

Figure 2.

Interaction patterns of lynx1 Arg38 with its environment in the lynx1-nAChR complex system. The interaction pattern graph shows the probability of occurrence within 4 Å from each of α4₂ (green), α4₁ (pink), α4₁ Trp156 (red), and α4₁ Tyr204 (orange) from three individual systems. Note that α4₁ Trp156 and α4₁ Tyr204 are initial interaction sites in the lynx1-nAChR complex system. α4₁ represents residues on α41 excluding Trp156 and Tyr204.

In Figure 3, snapshots of the lynx1 protein and nAChR α4₁ / α4₂ subunits were taken at 0-ns, 50-ns, 100-ns, 250-ns production times, respectively. Arg38 is represented in yellow sphere, and Trp156 and Tyr204 in red and orange ones. The C-loop of nAChR α4₁, colored in dark blue, was initially opened (compared to the structure in PDB 6cnk) and part of lynx1 (including Arg38) was placed within the α4₁:α4₂ interface. There are three regions in the C-loop: Ile203-Pro205 (inside facing region), Tyr197-Glu202 (front turn region), and Thr194-Lys196 (outside facing region). As Figure 3A series (A-i to A-iv) represent, Arg38 gradually detached from the α4₁:α4₂ interface and the C-loop closed. Interestingly, even though Arg38 left the initial interaction sites, it still kept interacting with other residues on α4₁ subunit, specifically, with the C-loop front turn residues, preventing the C-loop from opening again. This interaction is clear in Figure 4A that represents the averaged distance between each α4₁ C-loop residue and Arg38 over the entire simulation. A small distance value indicates a high frequency of Arg38’s presence in close to a specific residue. In replica-1, Cys199-Ala201 located in the front turn region show the closest contacts to Arg38 (Figure 4A), due to the outward movement of Arg38 from initial position as discussed in Figure 3A. Figure 3B series show the positions of α4₁, α4₂, and lynx1 when Arg38 maintaining partially detached from the initial interaction interface. In this system, Arg38 did not move far enough for the C-loop to close. Instead, Arg38 actually kept interacting with the front turn on the C-loop, which to some degree stabilizes the movement of the C-loop and lynx1 (Figure 3B-iv). Accordingly, these front turn residues have closer contact to Arg38 than other residues (see replica-2 in Figure 4A). This relative position between Arg38 and the C-loop persists for the rest of the simulation. Though Arg38 interacting mostly with front turn residues, Arg38 in replica-1 detached more complete from initial sites than replica-2 (Figure 2) that differs the C-loop motion dramatically. Unlike the events in Figures 3A and 3B, in Figure 3C, Arg38 was observed to fluctuate only within the initial interaction interface, making Arg38 interact with the inside facing residues in close contact (see replica-3 in Figure 4A and Figure 2). Fluctuating within the α4₁:α4₂ interface, Arg38 infrequently moves close to α4₂ subunit performing some non-specific interactions. In this case, with lynx1 protein kept inside the α4₁:α4₂ interface region, the C-loop was stabilized in the open state for the entire simulation time.

Figure 3.

Snapshots of α4₁, α4₂, and lynx1 in the lynx1-nAChR complex system: α4₁ in pink, α4₂ in green, and lynx1 protein in yellow. Arg38 (yellow), α4₁ Trp156 (red), and α4₁ Tyr204 (orange) are shown in van der Waals spheres and the C-loop of α4₁ in dark blue. (i-iv) represents the relative position of the proteins at 0-ns, 50-ns, 100-ns, and 250-ns production times, respectively.

Figure 4.

(A) Averaged distance between each α4₁ C-loop residue and Arg38 over the entire simulation in each replica. (B) Schematic representation of α4₁ C-loop internal angle (γ) for the representation of C-loop motion. The three selected residues used to define γ are shown as van der Waals spheres colored in cyan (α4₁ Pro205), blue (α4₁ Cys199), and green (α4₂ Phe40). (C) Density distributions of α4₁ C-loop γ angle in lynx1-nAChR complex and (D) apo-nAChR system. Each replica is colored in same as in (A). The value of initial model structure is represented in orange dashed line

The nAChR C-loop has been reported to be crucial for maintaining functional properties of nAChRs. Lynx1 Arg38 was initially placed inside the α4₁:α4₂ interface with the α4₁ C-loop opened. This is a configuration that is associated with the closed functional state of the nAChR ^63,64. As suggested in Figure 3A series, once closed, the C-loop would be hard to reopen due to the interaction between its front turn residues and Arg38 until lynx1 disassociates from the nAChR. Note that the opposite phenomenon was observed in Figure 3C series with maintaining the C-loop in the open state. Figures 4C-D show the α4₁ C-loop motions in lynx1-nAChR complex and apo-nAChR systems. The internal angle (γ) was employed to represent the open and closed states of C-loop; the smaller the angle is, the C-loop more likely in the closed state. Note that there is no hard threshold to distinguish the open and closed states, so we use the value of the initial structure as a reference. As shown in Figure 3A series in which the C-loop was prevented from re-opening due to the interaction with lynx1, γ values (black curve) are mostly small in Figure 4C. Expectedly, for Figure 3C series, another tight distribution around larger-than-initial γ is shown in Figure 4C (green curve). Opposite to the relatively narrow distribution observed in lynx1-nAChR complex systems, much wider two-peak distributions are shown for apo-nAChR systems (Figure 4D). Without the presence of lynx1, α4₁ C-loop shows a larger γ fluctuation range mainly with two peaks around 50° and 100°, representing close and open states, respectively. It is clear that due to the interaction with lynx1, α4₁ C-loop tends to be restricted in either close or open state, and a state transition is harder to be observed than in apo-nAChR. Movement of the C-loop over the ligand to the closed position is thought to transmit important information to the pore lining region of the channel, resulting in channel widening and allowing ion flow. MD simulations of the a7 nAChR alone indicates that a relatively low energetic barrier separates the two C-loop conformations ⁶⁵. Interactions of lynx1 with an nAChR subdomain (e.g. the C-loop) could provide substantial restrictions on these transitions. This has been supported by our MD results as less transition between the open and closed states are observed in lynx1-nAChR complex than apo-nAChR systems (Figure 4C,D). Although our models involve the C-loop at α4:α4 interfaces, a presumptive non-agonist binding site, presumably interactions with any of the C-loops of the pentamer can impede the tilt required of the pore-lining barrels needed for the enlargement of the pore diameter ⁶⁶.

Interactions with nAChR Reduce Lynx1’s Activity Dynamics

To gain a better understanding of the influence of lynx1-nAChR interactions on lynx1 structure, dynamics, and orientation (with respect to the membrane), we also performed the simulations of the lynx1-only systems. Figure 5A represents the density distributions of the center of mass (COM) of lynx1 Arg38 along the Z axis in either the lynx1-only (black curve) or lynx1-nAChR complex (green curve) system. In lynx1-only, Arg38 fluctuates within a wide Z-coordinate range (25-60 Å) with the possibility of falling into the nAChR interaction range (47-60 Å) and the lipid region (<30 Å). Therefore, some intermittent lynx1-lipids interactions were observed in the lynx1-only system when Arg38’s Z-coordinate is less than 30 Å, which did not happen in the lynx1-nAChR system. It is important to note that because the lynx1-only system mimics the unbound state, the estimation of lynx1-nAChR interactions was extrapolated from the Z-coordinate distribution of the nAChR C-loop of each subunit. Thus, the possibility discussed here is a necessary condition for the actual interaction. On the other hand, Arg38 shows a relatively small Z-coordinate fluctuation (50-60 Å) in the lynx1-nAChR complex system, which manipulates the multi-mode interactions between lynx1 and nAChR C-loop. As for the COM of the entire lynx1 protein backbone (Figure 5B), there is no significant difference observed in these two systems because lynx1 is GPI-anchored to the membrane bilayers. In other words, lynx1 can change its orientation relative to the membrane, but its Z-COM limits are bound.

Figure 5.

(A) Density distributions of lynx1 Arg38 and (B) lynx1 protein center of mass in the Z axis. The background shading is based on the averaged regions of lipids (red), water (white), and tentative nAChR interaction sites (blue) as the nAChR C-loop range in Figure 1. (C and D) Density distributions of (C) tilt and (D) rotation angles of lynx1 (defined in Figure 1). Note that green curves represent the lynx1-nAChR system and black for the lynx1-only system. Each distribution was calculated using all production trajectories of three replicas.

In terms of the tilt/rotation angle, as shown in Figs. 5C and 5D, there is a similar trend, i.e., the tilt/rotation angles show a high fluctuation range in the lynx1-only system and a relatively small one in the lynx1-nAChR complex system. From the schematic shown in Figure 1E, the greater the upward movement of Arg38, the larger tilt angle was observed. Thus, Figure 5C shows a very similar profile as those in Figure 5A. In the lynx1-only system, the tilt angle fluctuates all the way between 0-160° range, with 0° being when Arg38 lies within the lipid area, and 160° referring to the widest possible nAChR interaction range from the lipid. In contrast, in the lynx1-nAChR complex system, the tilt angle exhibits smaller fluctuations (120°-l50°). This is mainly due to the lynx1-nAChR interactions that were elaborated previously. With respect to the rotation angle, lynx1 fully rotates around the protein’s principal axis. Even with Arg38 at the α4₁:α4₂ interface and the C-terminal position fixed by the GPI-anchor, there is little limitation for rotating lynx1 as compared to the relatively restricted freedom of tilting. Thus, there is still a relatively large rotation angle change of lynx1 even in the lynx1-nAChR complex system.

We also employed the root mean square deviation (RMSD) and root mean square fluctuation (RMSF) to investigate the structure and dynamics changes of lynx1 in unbound/bound states (Figure S1). In terms of RMSD, both unbound/bound states show similar lyxn1 structure behaviors. In terms of RMSF, residues of β-strands (yellow shaded regions) show a lower fluctuation than other sections of the protein. In bound state, lynx1 has lower fluctuations especially for residues of turn II (residue number: 35-38) and the disordered region (residue number: 53-61). These two regions are located near nAChR and contain the reported interaction residue (Arg38) that is highly involved in the lynx1-nAChR interactions. However, in the back-side regions that are far away from nAChR, there is not much difference observed.

Generally speaking, interactions with lynx1 appear to restrict the conformational change of the C-loop that influences the nAChR function, and on the other hand, these interactions also lock the lynx1 protein (especially the residues facing nAChR) within the interaction region. These reciprocal interactions can only be observed when lynx1 and nAChRs are physically located closely, though it remains to be determined how frequently this happens on the neuron in situ. It is also reasonable to believe that once the interactions form, the influence on either nAChR or lynx1 might last for a significantly long time. We expect the GPI anchor on lynx1 to decrease the search time for nAChR on the membrane surface, by reducing the degree of freedom of lynx1 to the membrane plane. Similarly, this may also prolong the lynx1-nAChR interaction. Note that we only show the influence of lynx1 protein on (α4)₃(β2)₂ stoichiometry of nAChR in this work, it is possible that lynx1 protein may also have influences on other subtypes of nAChRs (i.e., α7, α3*) that need further investigation.

Conclusions

In this study, we elucidate the interactions between nAChR and lynx1 that either modulate the dynamics fluctuation of lynx1 protein or influence the open/close state change of the C-loop in one of the nAChR subunits that could influence its desensitization related to functional expression. The lynx1-only system was built to represent the unbounded state of lynx1, while the lynx1-nAChR complex systems mimic the bound state. The restrictive influence on lynx1 was inferred from dynamics analysis of systems when it is in bound vs unbound state. When unbound, lynx1 protein (especially Arg38) showed a high dynamic activity. The center of mass along the Z direction of Arg38 show 3 times wider range of fluctuation than in the bound state (30 Å vs. 10 Å). Unbounded lynx1 protein tilt and rotation angles show 5 (160° vs. 30°) and 1.3 (180° vs. 140°) times larger fluctuations, respectively, compared to the bound one. Arg38 intermittently interacts with different lipids when unbound while there are no lipid interactions in the bound one. Lynx1 belongs to the three-finger protein family that has relatively flexible loops. When bound, this loop flexibility would easily be influenced by interactions. The turn and disordered sections in the loops of lynx1 show decreased RMSF values (2 Å) when facing the nAChR. The flexibility of these loose sections is restricted by association with nAChR and which would be extended during long-term interactions.

The simulations described here also provide some insights into how these interactions could influence the activity of the nAChR. In the three replicas performed in this study of the bound state, there were significant and stable interactions between lynx1 (Arg38) and the receptor C-loop. The biological hypothesis of the lynx1 protein as an allosteric modulator on nAChR function with multimodal actions on nAChR function by reducing agonist sensitivity, accelerating the rate of desensitization, and prolonging the recovery from desensitization of nAChRs ^23,67. Results presented in this work support these hypotheses. The interactions observed between lynx1 and nAChR show some non-deterministic influences on the C-loop behavior of α4₁ subunit – preventing a closed C-loop from reopening, maintaining the intermediate state between open and closed, or maintaining the C-loop open state – depending on the position of lynx1 protein. Due to these interactions, nAChR would have a higher barrier between the open to closed transition, thus potentially inhibiting agonist sensitivity or inhibiting the recovery from desensitization.