Exploring the inhibition mechanism of adenylyl cyclase type 5 by n-terminal myristoylated Gαi1

Adenylyl cyclase (AC) is an important messenger involved in G-protein-coupled-receptor signal transduction pathways, which is a well-known target for drug development. AC is regulated by activated stimulatory (Gαs) and inhibitory (Gαi) G proteins in the cytosol. Although experimental studies have shown that these Gα subunits can stimulate or inhibit AC’s function in a non-competitive way, it is not well understood what the difference is in their mode of action as both Gα subunits appear structurally very similar in a non-lipidated state. However, a significant difference between Gαs and Gαi is that while Gαs does not require any lipidation in order to stimulate AC, N-terminal myristoylation is crucial for Gαi’s inhibitory function as AC is not inhibited by non-myristoylated Gαi. At present, only the conformation of the complex including Gαs and AC has been resolved via X-ray crystallography. Therefore, understanding the interaction between Gαi and AC is important as it will provide more insight into the unknown mechanism of AC regulation. This study demonstrates via classical molecular dynamics simulations that the myristoylated Gαi1 structure is able to interact with apo adenylyl cyclase type 5 in a way that causes inhibition of the catalytic function of the enzyme, suggesting that Gα lipidation could play a crucial role in AC regulation and in regulating G protein function by affecting Gαi’s active conformation.


Introduction
Many proteins are involved in cell communication of which one type is the G-protein-coupled receptor (GPCR), embedded in the membrane. GPCRs are part of a major signalling pathway, the GPCR signal transduction pathway, which enables the transfer of a signal from the extracellular region to the intracellular side and is a key target for drug development. A large diversity of GPCRs can be found in nature as about 800 human genes are involved in storing different types of GPCRs that can interact with neurotransmitters, hormones or exogenous ligands, for example [1].
In the cytosol, G proteins, composed of an α, β and γ subunit, are the first interaction partner of activated GPCRs. When a heterotrimeric G protein is activated by a GPCR, the trimer dissociates, resulting in an α subunit and a βγ dimer [2]. Activated Gα subunits transport the signal from the membrane to other regions of the cell by stimulating or inhibiting reactions via protein-protein interactions. Besides direct activation by GPCRs, the function of G proteins can also be influenced by other environmental factors, such as lipidation. Permanent N-myristoylation, for instance, is known to change the structure and function of the inhibitory G-protein subunit Gα i1 in its active GTP-bound state [3][4][5][6].
While a wide range of GPCRs exists, a relatively low diversity is present in the G protein family, e.g. in the human body. The human body includes only a relatively small variety of 21 α, 6 β and 12 γ subunits [1]. The Gα subunits are divided into four major subfamilies based on their sequence homology and function [7]: stimulatory Gα s , inhibitory Gα i , Gα q and Gα 12 [8,9]. Overall the structures of the Gα subfamilies are similar (S1 Fig, Fig 1), including a Ras domain and an alpha helical (AH) domain. The Ras domain is present in all members of the G-protein superfamily and can perform hydrolysis of GTP to GDP during deactivation of the Gα subunit [10]. In addition, the domain includes an interaction site for GPCRs as well as regions that can interact with the βγ dimer. Moreover, the Ras domain can also undergo lipidation. Except for Gα t , all Gα proteins are able to reversibly bind a palmitate to their N-terminal helix. Besides palmitoylation, Gα i can also permanently bind a myristoyl moiety to the N-terminus that appears to be crucial for the function of the subunit (Fig 1C) [4,5,9,11].
The AH domain is unique to the Gα subfamilies, which is composed of six α helices and interacts with the Ras domain when GTP or GDP is present (Fig 1C). However, this interaction between the AH and Ras domain is weakened when a nucleotide is absent in Gα's active site [12][13][14]. The high structural similarity among members of the Gα subfamilies is illustrated by aligning the X-ray structures of stimulatory Gα s and inhibitory Gα i1 , resulting in a root mean square deviation (RMSD) of only 1.07 Å between the Cα atoms of the two structures ( Fig 1A) [15,16]. Hence, from a comparison of the structures it is difficult to conclude what the origin is of their inverse action, i.e., how the structure can be related to a stimulatory respectively an inhibitory effect.
An example of a protein in which both Gα i and Gα s are important for regulation is adenylyl cyclase (AC). Ten isoforms of AC are known of which nine are membrane-bound (AC1-9) and one is soluble (sAC) [17]. These different types of AC are found throughout the body in different concentrations. AC5, for instance, is present in high quantities in the brain, the spinal cord and the heart, and is associated to congestive heart failure and pain perception [18,19]. G proteins have the ability to either stimulate (G s ) or inhibit (G i ) adenylyl cyclases' conversion of adenosine triphosphate (ATP) to cyclic adenosine monophosphate (cAMP) and pyrophosphate [20,21].
ACs consist of two membrane-bound regions, each built from six trans-membrane domains, and a catalytic region in the cytosol that includes two pseudo-symmetric domains, C1 and C2 (Fig 1) [22]. GTP-bound Gα s is known to bind to the C2 domain for which the interaction site is known from X-ray structures of Gα s interacting with AC ( Fig 1B) [23]. Such data is absent for the case of Gα i . In the absence of direct experimental information, a putative interaction site of GTP-bound Gα i has been suggested in analogy to the known structure of the complex of Gα s and AC (Gα s :AC) as the pseudo-symmetric site on the C1 domain ( Fig 1B). However, how the interaction of Gα i on the C1 domain should induce inhibition is not obvious [5]. Furthermore, since with this hypothesis the interaction sites of Gα i1 and Gα s are highly similar in addition to their structures, it is unclear how the α subunits can differentiate the two binding sites on AC and what the cause is of the stimulatory versus the inhibitory effect induced by the subunits.
A factor that could play an important role in differentiating the action of Gα i and Gα s is the difference in lipidation of both subunits. Although the X-ray structures in the Protein Data Base (PDB) [24] of the active inhibitory and stimulatory Gα subunits tightly align, the N-terminus, which is not resolved for Gα i or Gα s , is not myristoylated during the expression process of Gα i as lipidation can hinder crystallisation [4]. Hence, it is not clear to what extent the missing N-terminal myristoyl moiety affects the Gα i structure of the remaining protein while the bound myristoyl group has been known to be crucial for Gα i 's conformation and function as the ability to interact with AC5 is abolished upon removal of the myristate [4][5][6]. Classical molecular dynamics (MD) simulations of myristoylated GTP-bound Gα i1 , Gα i1 myr , demonstrate the stability of the myristoyl moiety on the Ras domain due to a hydrophobic pocket formed by β2-β3, α1 and the C-terminus α5 ( Fig 1C) and show that myristoylation can have a significant effect on the conformation of the subunit [25]. The findings suggest the possibility of an alternative novel interaction mode and open up new possibilities for selective interactions with AC. This is because the found structural changes in the classical MD simulations of Gα i1 myr [25] suggest that the subunit will not be able to interact with C1 as Gα s interacts with C2.
Here, we investigate the interaction between Gα i1 myr and AC, using classical MD simulations. To this end, the initial structure of Gα i1 myr was taken from reference [25] in which a 2 μs classical MD simulation of Gα i1 myr is described. Gα i1 myr can inhibit only particular isoforms of AC: AC1, AC5, AC6 [26]. In this study AC5 is used because X-ray structures of AC's catalytic domains are composed of isoforms AC2 and AC5. Ca. 16 AC structures can be found in the PDB with different resolutions and/or crystallisation conditions. All available structures have been co-crystallised with a Gα s subunit and correspond therefore to stimulated conformations at various levels of activation, depending on the nature of bound cofactors (e.g. cofactor-free complex of AC, substrate-bound AC complex). When AC5 becomes active, roughly three conformational options are possible: a complex of Gα s and AC5, Gα s :AC5, Gα s in complex with ATP-bound AC5, Gα s :AC5(ATP), or a complex of Gα s and AC5 bound to the reaction products cAMP and pyrophosphate, Gα s :AC5(cAMP). Currently, it is not known which one of these forms is most likely to interact with Gα i1 myr , or if Gα i1 myr can inhibit all of them. In this study, the structure of the AC5 protein was taken from a crystal structure of the cofactor-free Gα s :AC5 complex. This apo AC5 structure was used as it could provide insight into Gα i1 myr 's inhibitory effect on a stimulated Exploring inhibition of adenylyl cyclase type 5 by n-terminal myristoylated Gα i1 conformation of AC5 in the absence of ATP. The selected AC5 structure was employed to build a Gα i1 myr :AC5 complex ( Fig 1C) and to explore if the binding of Gα i1 myr is able to affect the active conformation initially induced by Gα s . The absence of ATP in the active site provides the opportunity to investigate Gα i1 myr 's ability to prevent the formation of AC's fully activated form by altering AC's conformation unfavourably prior to substrate association. In order to verify which changes are due to the interaction of AC5 with Gα i1 myr and which alter- :AC5 complex and the free AC5 system suggest two possible ways of AC inhibition in its apo form. First, Gα i1 myr seems to inhibit AC's conversion of ATP to cAMP by preventing active-site formation as the Gα i1 myr subunit perturbs the conformation of the active site at the C1/C2 interface. Second, the effect of Gα i1 myr on the AC structure leads to a closed conformation of the Gα s binding site on C2, decreasing the probability of Gα s association and thus of a counter-balancing re-stimulation of the AC5 activity. Taken together, the observed events lead to a suggestion for a putative Gα i1 myr inhibition mechanism of apo AC5 in which lipidation is crucial for Gα i1 myr 's function and its protein-protein interactions. Hence, the results of this study provide a possible indication that lipidation could play a significant role in regulating G protein function and therefore could impact signal transduction in G protein mediated pathways [4][5][6].

Initial structures
The PDB structure 1AZS, including the Gα s :AC complex with AC in the apo form, was used as a template, including 1AZS's C1 and C2 domain, for the initial AC5 structure of Rattus norvegicus (UniprotKB Q04400) [27][28][29]. The structure of the Rattus norvegicus Gα i1 myr subunit (UniprotKB P10824) interacting with GTP and Mg 2+ was taken from reference [25] (S2 and S3 Figs).

Docking of Gα i1 myr on AC5
The HADDOCK web server [30] was used for docking ten conformations of Gα i1 myr on the catalytic domains of AC5 of Rattus norvegicus. The Gα i1 myr snapshots were extracted at the end of the Gα i1 myr classical MD trajectory (around 1.9 μs) discussed in reference [25], with a time interval of 0.5 ns. The active region of Gα i1 was defined in HADDOCK as a large part of the AH domain (112-167), the switch I region (175-189) and the switch II region (200-220), allowing for a large unbiased area on the Gα i1 myr protein surface to be taken into account during docking. The active region of AC5's C1 domain was defined as the α1 helix (479-490) and the C-terminal region of the α3 helix (554-561) because experimentally it has been found that Gα i1 myr is unable to interact with C2 and its main interactions with AC are with the C1 domain [5]. Passive residues, residues that could take part in protein-protein interaction, were defined as residues around the active residues that are on the protein surface and within a radius of 6.5 Å of any active residue [30].
The initial Gα i1 myr :AC5 complex for the classical MD simulations was selected based on (1) the absence of overlap between the C2 domain and Gα i1 myr , (2) no overlap with Gα i1 myr 's GTP binding region and the interaction site of Gα i1 myr with C1 and (3) presence of similar complexes in the top-ten docking results of the docking calculations performed for all ten Gα i1 myr snapshots. The first property of the selection criteria is important since Gα i1 myr is unable to interact with C2 [5]. The second criterium has been defined since GTP is located in the active site of Gα i1 myr in the classical MD simulations, but was not incorporated in the docking procedure because this is not possible in HADDOCK. Therefore, no overlap between the GTP binding site and the C1 domain should be present in the docking result as otherwise the GTP molecule will not be able to fit in Gα i1 myr 's active site. The last criterium is the presence of similar Gα i1 myr :AC5 complexes of the selected complex in all top-ten docking results which increases the probability that complexation of the two proteins is not conformation specific, but is robust as similar complexes can be obtained using different conformations of Gα i1 myr .

Classical molecular dynamics simulations
The Gα i1 myr :AC5 complex was used to simulate the protein complex for 2.5 μs at 310 K and 1 bar using a Nosé-Hoover thermostat and an isotropic Parrinello-Rahman barostat. In the active site of Gα i1 myr one Mg 2+ ion and a GTP molecule are present. In order to closer mimic an AC5 system with which ATP or a product such as pyrophosphate would be able to interact, a Mg 2+ ion was added to the active site of AC5 (see S1 Appendix). Additionally, about 68 000 water molecules and 150 mM KCl are present in the simulated system. The force fields used for the protein and the water molecules are AMBER99SB [31] and TIP3P [32], which were employed by Gromacs 4.6.6 [33,34] to perform the runs. For GTP, the force field generated by Meagher et al. was used [35]. The adjusted force field parameters for the K + ions and the Clions were taken from Joung et al. [36]. The Mg 2+ ion parameters originated from Allnér et al. [37] and the parameter set for the myristoyl group was taken from reference [25]. The charges for the myristoyl group were obtained with Gaussian 09 [38] based on Hartree Fock calculations in combination with a 6-31G Ã basis set and using the AMBER RESP procedure [39]. Appropriate atom types from the AMBER99SB force field were selected to complete the myristoyl description.
Electrostatic interactions were calculated with the Ewald particle mesh method with a real space cutoff of 12 Å. Bonds involving hydrogen atoms were constrained using the LINCS algorithm [40]. The time integration step was set to 2 fs. The free AC5 system was simulated with the same setup as the Gα i1 myr :AC5 complex. The system was solvated in 30 000 water molecules and a 150 mM KCl concentration. The initial location of the Mg 2+ ion in the active site of the enzyme was the same as in the Gα i1 myr :AC5 complex system.

Structure superpositions and images
Multiprot [41] and VMD [42] were used to align protein structures. Uniprot [43] was used to align protein sequences. Images were prepared with VMD [42].

Results and discussion
The stability of the docked Gα i1 myr :AC5 complex and the effect of Gα i1 myr association was verified via investigating the conformational changes in the complex through classical MD simulations on the μs time scale. The X-ray structure used for the initial conformation of AC5 in the simulated complexes, was co-crystallised with Gα s but is not interacting with substrate or products from the ATP conversion reaction (see Introduction). Consequently, this conformation of AC5 could be viewed as a structure that is present before ATP association, but is already in an active form, due to its interactions with Gα s . Because of the crystallisation circumstances used to obtain the selected AC5 structure, the found conformational changes in the catalytic region of AC5 in the Gα i1 myr :AC5 complex are compared to the conformation found in the classical MD trajectory of the free AC5 system and the Gα s :AC X-ray structure in order to verify which structural alterations are due to the presence of Gα i1 myr and which changes are the result of the absence of Gα s .
Protein-protein interface of the Gα i1 myr :AC5 complex The initial conformation of the Gα i1 myr :AC5 complex suggests that Gα i1 myr 's proposed interaction site (see S1 Appendix) affects the conformation of C1 in a different way than Gα s stabilises the C2 domain (Figs 1, 2 and S3 Fig). Unlike Gα s , Gα i1 myr is not located between the helices of AC5's catalytic domain, but appears to clamp the C1 domain into its inactive conformation. Gα i1 myr is positioned around AC5's α3, interacting with α1, α2, and α3 via its switch I, II and III region together with the C-terminal domain of αB (Fig 2 and S3 Fig). Since C1's α1 helix appears to decrease its distance with respect to the C2 domain when an ATP analog, adenosine 5-(α-thio)-triphosphate (ATPαS), is present in the active site (S2 Fig), the interactions between Gα i1 myr and C1's α1 in the Gα i1 myr :AC5 complex could suggest that one way by which Gα i1 myr is able to inhibit ATP's conversion is by preventing C1's α1 to rearrange upon ATP binding. The initial Gα i1 myr conformation is very stable over the entire course of the classical MD trajectory. Since the starting structure of the Gα i1 myr :AC5 complex is unrelaxed, the conformation of the Gα i1 myr subunit and the Gα i1 myr /AC5 interface have been investigated during the MD trajectory to study the stability of the Gα i1 myr :AC5 complex and the effect of the AC5 interaction on the Gα i1 myr structure. In fact, the structure of Gα i1 myr only changes minimally by a slight adjustment in the orientation of the alpha helical domain (Fig 2B and  S6 Fig). A striking feature of the interface between Gα i1 myr and C1 is the fact that the switch regions, I, II and III, remain involved in AC5 binding, as well as the region on the alpha helical domain that rearranged upon myristoyl binding (Fig 2A and 2B) [25]. Gα i1 myr is stabilised on AC5 via the C1 domain without major interactions with C2 ( Fig 2C). The relative orientation of Gα i1 myr with respect to the C1 domain stabilises after * 400 ns (S6 Fig). This slight orientational repositioning is probably a consequence of the relocation of C2's β7-β8 loop, which could be due to the removal of the Gα s subunit from the C2 domain in the initial complex because this β7-β8 loop displacement is present in Gα i1 myr :AC5 as well as in free AC5 (Fig 3). Besides the orientation of Gα i1 myr , also minor alterations can be observed in Gα i1 myr 's active site. The interactions of the Mg 2+ ion near the GTP binding site (S7 Fig) are moderately altered compared to the free Gα i1 myr system [25]. In Gα i1 myr :AC5 both of Asp200's oxygens, OD1 and OD2, are able to interact with Mg 2+ , while in free Gα i1 myr only one of Asp200's oxygens is interacting with the Mg 2+ ion [25]. The interaction between Mg 2+ and Asp200's second oxygen in Gα i1 myr :AC5, OD2, can be temporarily perturbed by a water molecule, which leads to sudden jumps in the Mg 2+ and Asp200OD2 distance (S7 Fig). Structural effect on AC5's active site Gα i1 myr 's interactions with C1 impact the entrance of the ATP binding site. The active site of AC5 is located at the interface between C1 and C2. Structural changes of both domains upon Gα i1 myr binding can potentially influence the protein's activity. When comparing the root-mean-square deviation of the two domains with respect to AC5's initial conformation, it is clear that in the free AC5 as well as in the Gα i1 myr :AC5 complex the C2 domain experiences more changes than C1 (Fig 3C). In the X-ray structure (PDB code 1AZS) used as a template to construct the initial AC5 structure, the C2 domain is interacting with a Gα s subunit. The removal of Gα s from the C2 domain could affect the RMSD of the domain in the Gα i1 myr :AC5 and free AC5 systems as the initial structure of AC5 is influenced by the presence of Gα s .
During the classical MD runs, the first alteration to C2's initial structure that can be observed in both Gα i1 myr :AC5 and free AC5 is the relocation of the β7-β8 loop, positioned on the cytosolic side (Fig 3). The location of C2's β7-β8 loop is important for the active conformation of the active site at the C1/C2 interface (Fig 3 and S2 Fig). Loop relocation and the accompanying movement of the two domains appear to have a deactivating effect on the active site   (Fig 3 and S2 Fig). For instance, a residue that is known to be part of the active site, Lys1245, located in the β7-β8 loop, is unable to maintain its orientation towards the ATP binding site (S8 Fig). A reason for this conformational change in the Gα i1 myr :AC5 and free AC5 simulations could be the absence of Gα s at the C2 domain that destabilises β7-β8's location. The presence of Gα i1 myr appears to increase the stability of the β7-β8 loop relocation compared to the free AC5 system (Fig 3). An alteration that also occurs around AC5's active site in the Gα i1 myr :AC5 system is the decrease in distance between C2's α4 helix and C1's β2-β3 loop (Fig 4, S3 Fig). This change in the C1/C2 interface appears to make the positioning of ATP in the active site less favourable because ATP's adenine moiety, which is positioned between C2's α4 helix and C1's β2-β3 loop when ATP interacts with AC5 (S2 Fig), is unlikely to fit between the C1 and the C2 domain due to the diminished distance between C1 and C2 compared to the initial AC5 structure. Although the free AC5 system also undergoes a change in this region, the α4 helix of C2 remains closer to the X-ray location than in the Gα i1 myr :AC5 complex (Fig 4). In summary, the conformational changes at the C1/C2 interface of Gα i1 myr :AC5 seem to affect AC5's active site by interfering with ATP binding. In fact, the new position of the β7-β8 loop can even block the active site entrance, which could prevent ATP entry. The relocation of C2's β7-β8 loop, present in the Gα i1 myr :AC5 and the free AC5 systems, could be due to the removal of Gα s , which interacts with the C2 domain in the template X-ray structure. The interaction between Gα s and AC5 seems to stabilise the position of β7-β8 at the C1/C2 interface, close to C1's α1. As C2's β7-β8 is part of the active site, the position of β7-β8 could be viewed as a stimulation feature that can be switched on by Gα s or switched off by the absence of Gα s (free AC5) or by the presence of Gα i1 myr , which stabilises the relocation of the β7-β8 loop even more (Fig 3). Gα i1 myr 's interactions with C1 impact AC5's active site around ATP's adenine binding site. While the RMSD of free AC5 and Gα i1 myr :AC5 compared to the initial AC5 structure (obtained via the Gα s :AC X-ray structure) appear similar for C1, the C2 domain of Gα i1 myr : AC5 diverges more from the initial structure than free AC5 (Fig 3C). One of the major differences between the C2 domain of the free AC5 system and the Gα i1 myr :AC5 complex lies on the membrane side of the proteins. In the case of free AC5, C2's β4-β5 loop is interacting with the C1 domain, while in the Gα i1 myr :AC5 complex, the loop does no longer interact with the C1 Exploring inhibition of adenylyl cyclase type 5 by n-terminal myristoylated Gα i1 domain, leading to an unfavourable ATP binding site at the C1/C2 interface (Fig 5A, 5B and 5C). A weakening of the ATP binding site at the C1/C2 interface is also apparent from C1's interactions with the C2 domain near the active site region (S8 Fig). Asp1198, for example, which is important for stabilising ATP's adenine moiety in the Gα s :AC complex, reorients as the residue is part of C2's β4-β5 loop. Due to Asp1198's change in location, Lys1124, which is interacting with Asp1198, alters its orientation as well. Lys1124 also influences the stability of the active site in Gα s :AC as the residue stabilises ATP in a similar fashion as Asp1198 (S8 Fig). Although C1's RMSD is low (Fig 3C and S6 Fig), a significant change in conformation can be observed close to the Gα i1 myr /C1 interface where a kink in α2 occurs, which is less pronounced in free AC5 (Fig 5D). However, overall, the C2 domain seems to be affected most by the absence of Gα s and the presence of the inhibitory Gα subunit. This observation is in line Exploring inhibition of adenylyl cyclase type 5 by n-terminal myristoylated Gα i1 with the hypothesis that Gα i1 myr is able to constrain C1's conformation via its tight interactions with this domain, leading to a perturbation and destabilisation of the active site at the C1/C2 interface (Fig 5 and S7 Fig). This change of the C1/C2 interface prevents the catalytic domains to sample the conformation in which ATP could be positioned in the active site due to a decrease in distance between C2's α4 helix and C1's β2-β3 loop and a relocation of C2's β4-β5 and β7-β8 loops (Figs 3, 4 and 5), which play important roles in the construction of the active site.

Presence of Gα i1 myr induces closure of Gα s 's interaction site on AC5's C2 domain.
Besides the direct conformational changes in AC5's active site, another mechanism that could Exploring inhibition of adenylyl cyclase type 5 by n-terminal myristoylated Gα i1 induce inhibition is decreasing the probability of Gα s binding to the C2 domain. In free AC5 as well as in Gα i1 myr :AC5, the Gα s site seems to become less favourable for Gα s binding since the distance between C2's α2 and α3, α2-α3, is significantly decreased with respect to the Xray structure of the Gα s :AC complex (Fig 6A and 6B). In the free AC5 system α2-α3 changes from around 16 Å (initial distance between the Cα carbons of Asn221 and Phe1171) to an average distance of * 11 Å (Fig 6A and 6B). In the Gα i1 myr :AC5 simulations, the distance between the two residues can decrease even more severely to a distance of * 7 Å (Fig 6A and  6B), leading to a closed Gα s binding site conformation. Important residues that stabilise this closed conformation of the Gα s binding site in Gα i1 myr :AC5 are AC5's: Glu1083, Leu1088, Ala1090, Phe1171, Asn1172 and Asn1173 (Fig 6D). Hence, due to the interaction with Gα i1 myr , the catalytic domains of AC5 appear to deactivate AC5's catalytic ability via deformation of the active site as well as by sampling closed Gα s binding site conformations.

Possible mechanism of Gα i1 myr inhibition
The simulation of Gα i1 myr :AC5 in comparison with the free AC5 trajectory and the Gα s :AC Xray structure demonstrate that the first step in decreasing AC5's activity in the apo form is the relocation of the β7-β8 loop (Fig 7, step one). In fact, the β7-β8 loop seems to have an important role for the stimulatory response since the presence of Gα s leads to the stabilisation of the Exploring inhibition of adenylyl cyclase type 5 by n-terminal myristoylated Gα i1 loop, forming ATP's binding site (Fig 7, starting conformation of AC in left panel) [23]. This loop conformation is lost as soon as Gα s is absent, as observed for both free AC5 and Gα i1 myr : AC5. In step two of Fig 7 the Gα i1 myr :AC5 complex undergoes a rearrangement in the C2 domain (absent in free AC5), which leads to a further perturbation of AC5's active site. The classical molecular dynamics simulations also show that in the presence of Gα i1 myr , there appears to be a decrease in probability for Gα s association (Fig 7, step 3 and Fig 6). Hence, through these rearrangements Gα i1 myr could deactivate apo AC5 as well as decrease the probability of reactivation via Gα s .

Conclusion
The results of this study suggest that Gα i1 myr deactivates the apo form of adenylyl cyclase type 5 via constraining C1's active site region. Inhibition and stimulation of AC5 appear to follow different pathways. While Gα s binds between the helices of C2, increasing the stability of the C1:C2 dimer, Gα i1 myr is able to clamp the helices of the C1 domain, promoting an inactive conformation of AC5's catalytic domains and a possible decrease in affinity for Gα s on the C2 domain. Structurally, Gα s and non-myristoylated Gα i1 are very similar, however, when myristoylation has taken place on the N-terminus of Gα i1 , the conformation of the subunit changes drastically, leading to a structure that differentiates itself from the active Gα s subunit and enables the protein to function in an inhibitory fashion as is shown via the presented classical MD simulations. Hence, in line with experimental studies, myristoylation appears to be crucial for G i 's function and demonstrates how important even relatively small changes to a protein structure can be for its function. The difference in active AC conformation is shown, depending on AC's interaction partners (e.g. analog of the substrate ATP, the inhibitor Ca 2+ , substrate free). The alignment of the initial Gα i1 myr :AC5 complex with three AC:Gα s complexes demonstrates that the initial apo AC5 structure used in this study is different from the fully active AC conformation (yellow) to which ATPαS is bound. The yellow structure is a complex of Gα s :AC with ATPαS and forskolin (PDB code 1CJK). ATPαS is depicted in transparent yellow in which oxygen atoms are red, phosphor is tan and carbon is cyan. This structure is an active conformation of the AC catalytic domain. The blue (PDB code 3MAA) [44] is suggested to be an inactive Gα s :AC complex and interacts with methylpiperazinoforskolin (FKP) together with ATPαS and a Ca 2+ ion. The red structure (PDB code 1AZS), used as main template for the AC5 conformation in this study, is an Gα s :AC complex that only interacts with FKP in the catalytic domain and is more similar to the 3MAA structure than the fully active 1CJK structure around AC's active site. (E) Alignment of the Gα i1 non :RGS4 complex (PDB code 1AGR) and Gα i1 myr . In case of the non-myristoylated Gα i1 :RGS4 structure (Gα i1 non :RGS4), Gα i1 non is shown in red and RGS4 is shown in orange. The myristoylated Gα i1 is depicted in cyan. The location of Thr182, Glu207 and Lys210 are shown for both complexes as these residues are important for the interaction between RGS4 and Gα i1 [45]. The Gα i1 non :RGS4 residues are labeled in red and the Gα i1 myr residues are labeled in blue.