Modeling protein-protein interactions in axon initial segment to understand their potential impact on action potential initiation

Piyush Bhardwaj , Don Kulasiri , , Sandhya Samarasinghe

Abstract The axon initial segment (AIS) region is crucial for action potential initiation due to the presence of high-density AIS protein voltage-gated sodium channels (Nav). Nav channels comprise several serine residues responsible for the recruitment of Nav channels into the structure of AIS through interactions with ankyrin-G (AnkG). In this study, a series of computational experiments are performed to understand the role of AIS proteins casein kinase 2 and AnkG on Nav channel recruitment into the AIS. The computational simulation results using Virtual cell software indicate that Nav channels with all serine sites available for phosphorylation bind to AnkG with strong affinity. At the low initial concentration of AnkG and casein kinase 2, the concentration of Nav channels reduces significantly,suggesting the importance of casein kinase 2 and AnkG in the recruitment of Nav channels.

Key Words: Alzheimer’s disease; ankyrin-G; axon initial segment; casein kinase-2;microtubules; voltage-gated potassium channel; voltage-gated sodium channel

Introduction

Communication between neurons through action potentials(AP) is necessary for the nervous system to perform its functions. An AP is characterized by the changes in membrane potential or membrane voltage due to the efflux and influx of cations. Several studies have confirmed that AP is initiated at the axon initial segment (AIS) (Lemaillet et al., 2003; Rasband,2009; Kole and Stuart, 2012; Gulledge and Bravo, 2016), which is a non-myelinated region (Figure 1), with a length of 10 to 60μm, at the beginning of an axon (Kole et al., 2008; Buffington and Rasband, 2011; Kole and Stuart, 2012; Leterrier et al.,2015; Jones and Svitkina, 2016; Nelson and Jenkins, 2017;Fan and Markram, 2019). The main proteins in the AIS are voltage-gated sodium channels (Nav), voltage-gated potassium channels (Kv), microtubules (MTs), casein kinase 2 (CK2), and ankyrin-G (AnkG). AnkG is an essential protein in the AIS because the recruitment of other proteins (Kv, Nav, neurofascin 186 [NF186], and neuronal cell adhesion molecule [NrCAM])into the AIS depends on it (Hedstrom et al., 2008; Jones and Svitkina, 2016). To evaluate the importance of AIS in neurological processes and to show its relationships with various neurological diseases, it is necessary to understand the structural organization of the AIS region (Figure 1).

CK2, a serine/threonine-specific kinase, is acidophilic in nature and mediates phosphorylation of various proteins in the AIS(Meggio and Pinna, 2003). CK2 is constitutively active and does not require any second messenger or phosphorylation event to be activated (Meggio and Pinna, 2003; Bian et al., 2013).During the developmental stage, the early presence of CK2 in neurons has been observed, even before axon formation,and the experiments have shown the shortening of axons by 30% after CK2 inhibition (Ponce et al., 2011). CK2 plays a very critical role in AnkG-Navbinding to recruit Navchannels into the AIS facilitating the phosphorylation of Navchannels to increase their affinity towards AnkG (Rasband, 2008;Yamada and Kuba, 2016). Navchannels comprise a special type of amino acid sequence called the AIS motif, where the binding with AnkG takes place. The AIS motif contains four serine residues at different positions, S1112, S1123, S1124,and S1126 (Figure 2; Schafer et al., 2009). In addition to that,the AIS motif also contains the acidic residues: glutamate(E) and aspartate (D) residues are present at 1111 and 1113 positions respectively (Figure 2). All these residues have different functions in the AIS motif: the serine residues are responsible for the phosphorylation of Navchannels through CK2, and acidic residues (glutamate and aspartate) increase the affinity of phosphorylation process (Meggio and Pinna,2003). All these serine sites are phosphorylated by CK2 and this increases the binding affinity of the Navchannels towards AnkG to restrict them at the AIS (Meggio and Pinna, 2003;Brechet et al., 2008; Bian et al., 2013).

Any alteration in neuronal transmission can cause various neurological conditions, such as Alzheimer’s disease (AD),epilepsy, schizophrenia and bipolar disorder (Buffington and Rasband, 2011; Kaphzan et al., 2011; Bi et al., 2012; Harty et al., 2013; Rueckert et al., 2013; Peltola et al., 2016). AD is the most common neurodegenerative disease in humans; thus, it is the most common cause of dementia and, as yet, there is no known cure for AD (Hardy and Higgins, 1992; Maccioni et al., 2010; Craig et al., 2011; Karran et al., 2011; Checler and Turner, 2012; Alzheimer’s Association, 2013; Kametani and Hasegawa, 2018). Several studies suggest the correlation of AD with AIS proteins and have shown the loss of AnkG and Navchannels in AD brain samples (Kim et al., 2007; Sun et al.,2014). Cleavage of Navchannel is associated with increase in Amyloid β peptides (Aβ) production, a protein responsible for the amyloid plaques in AD patients (Kim et al., 2007; Kovacs et al., 2010). Moreover, loss of AnkG from AIS structure could play a critical role in AD pathogenesis by disturbing the protein trafficking within the neuron by altering the function of molecular motors (Kinesin and Dynein) (Kim et al., 2007;Sun et al., 2014).

Biological systems are complex as their features can be explained using different methods, and this makes them very difficult to study and to predict their behaviors (Bailey et al., 2002). However, computational models can be used to understand the plausible molecular mechanisms in an organism at the molecular level if we carefully develop the questions to be investigated. As we mentioned before, AP initiation is associated with the high number of Navchannels in the AIS, and these channels are phosphorylated by CK2 at their serine sites before their accumulation in the AIS.Further, AnkG recruits the phosphorylated Navchannels in this region. The dynamics of the phosphorylated channels, CK2,and AnkG would have crucial effects on AP initiation, and any impairment of these interactions would perhaps explain the experimentally observed link of AnkG to AD pathogenesis.Hence, we develop a mathematical model to investigate the role of serine-specific phosphorylation of Navchannels by CK2 before the accumulation of Navchannels into the AIS. Also,we test the effect of changes in the initial concentration of CK2 and AnkG on the accumulation of Navchannels at the AIS. Here, we suggest the AIS association with Alzheimer’s disease (AD) pathogenesis by reporting CK2 mediated serine specific phosphorylation is necessary for the recruitment of Navchannels into the AIS through AnkG. CK2 mediated phosphorylation increases the binding affinity with AnkG.Moreover, initial concentration of both AnkG and CK2 is also important and can disturb the final Navchannel population in AIS. According to the literature, alteration in Navchannels is associated with amyloid β peptides main component of amyloid plaques in AD patients (Kim et al., 2007; Kovacs et al.,2010).

Materials and Methods

Development of AIS models

It is necessary to make simplifying assumptions in the development of the model so that we can understand the interactions between the proteins within AIS. By using Virtual cell (version 7.2.0, Uconn Health, USA) software with currently available experimental data and insights (see References), the following assumptions are made:

1) The AIS region is considered as a cylinder with a length of 40 μm and a diameter of 1.5 μm (Gulledge and Bravo, 2016).A total of 12 species are present in our model which includes four different Navchannels (NaA, NaB, NaC, and N1111), four AnkG species (G1, G2, G3, and G4) and four CK2 species (C1,C2, C3, and C4).

2) In this model, four types of Navchannels are considered based on different phosphorylation events within the channels. We assume five phosphorylation conditions: no site phosphorylation, single-site phosphorylation, doublesite phosphorylation, triple-site phosphorylation and foursite phosphorylation (Figure 3). The no site phosphorylation means that all serine sites are in the non-phosphorylated state. We do not consider any Navspecies in this model as not being phosphorylated because without any serine phosphorylation, Nav-AnkG interactions are impossible(Brechet et al., 2008). Similarly, none of the Navspecies are assumed to be in the single-site phosphorylation condition as well, because AnkG does not bind to the Navchannels with single-site phosphorylation (Brachet et al., 2008). Four Navspecies in the model are assumed based on doublesite phosphorylation, triple-site phosphorylation, and foursite phosphorylation. Due to the lack of literature on the concentration of Navchannels before its restriction into the AIS, we estimate that the concentration of each of NaA, NaB,NaC, and N1111as 3.32 × 10–12μM.

3) The four phosphorylation sites in the AIS motifs (S1112,S1123, S1124 and S1126) are denoted by S1 = S1112, S2 =S1123, S3 = S1124, and S4 = S1126. All these sites within the Navchannels are responsible for equally strong interactions between the Navchannels and AnkG (Cantrell et al., 2018).

Here are the phosphorylation sequences for all four sites in the Navspecies:

· The phosphorylation of S1 is the first preference for CK2 because the regions near S1 (E, S1, D, F, and E) fulfill the criteria for the minimum consensus sequence (S/T, X, X, and D/E/pS/pT) for CK2 phosphorylation (Figure 4). The presence of acidic residues at then+ 1 andn+ 3 positions makes S1 the most suitable site for phosphorylation by CK2 (Meggio and Pinna, 2003).

· S3 also fulfills the minimum criterion for the consensus sequence (S/T, X, X, and D/E/pS/pT). S3 also has acidic residues at then+ 1 andn+ 3 positions, similar to S1 (Figure 4).However, S3 is assumed to be the second preference after S1 because S1 is located near the N-terminal.

· S4 is considered the third phosphorylation site because of the presence of an acidic residue (D) at then+ 1 position(Figure 4). The fourth, and final, phosphorylation site is S2 (S2, pS, E, pS, and D) because the region near S2 fulfills the minimum consensus sequence due to the presence of phosphorylated S or T at then+ 1 andn+ 3 positions (Figure 4).

4) Phosphorylation of the Navspecies (NaA, NaB, NaC, and N1111) by CK2 (C1, C2, C3, and C4) and its interaction with AnkG (G1, G2, G3, and G4) in this model are based on the mass action law.

5) In general, the phosphorylation is modeled using Michael-Menten kinetics (Wang and Wu, 2002) but due to rapid binding and unbinding of CK2 to the sites, we assume that the bindingunbinding can be modeled as elementary reactions and follows the mass action kinetics within the first few minutes;i.e., the serine related phosphorylation of Navchannels by CK2 is considered as a reversible elementary reaction. Because of the importance of the serine residues, the main aim is to understand the binding of CK2 with the Navchannels. Protein kinases have significant effects on the target protein they bind to. Protein kinases go on and off by auto-phosphorylation by binding with inhibitors or activator proteins or other small molecules. Their activity is highly regulated. However, this is not the case with CK2 due to its pleiotropic nature: CK2 does not require any mediators to activate because CK2 is always active. There is no evidence in the literature for any regulation of CK2 activity. CK2 not only phosphorylates the Navand Kv channels but also AnkG. Another reason for considering the mass action law to model the phosphorylation of Navspecies(NaA, NaB, NaC, and N1111) is the abundance of CK2 in the brain; therefore, we assume negligible competition of other species for CK2. In the absence of data, the phosphorylation of all four Navspecies is assumed to occur at the same association(Kon) and dissociation rates (Koff).

6) In the brain, more than 300 substrates are available for CK2-mediated phosphorylation (Meggio and Pinna, 2003;Nishi et al., 2014). We have no data on the exact number of CK2 molecules required for phosphorylation of the Navchannels. As CK2 present in abundance, we assume that the total CK2 concentration is 5.28 × 10–8(μM). Specifically,for this model, total CK2 concentration is divided into four equal concentrations designated as C1, C2, C3, and C4 to phosphorylate NaA, NaB, NaC, and N1111, respectively.

Figure 1 ｜A schematic diagram of the AIS structure.

Figure 2 ｜A schematic representation of the AIS motif within Nav channels.

Figure 3 ｜Classification of the Nav species in the AIS model.

Figure 4 ｜Consensus sequence for CK2 mediated phosphorylation.

7) According to the literature, the total number of Navchannels present in the AIS after binding with AnkG is 100–300 molecules/μm (Kole and Stuart, 2012). On the other hand, according to Srinivasan et al. (1988), AnkG numbers are ten times higher than the Navchannels (Srinivasan et al., 1988). Based on this observation, in this model initial AnkG concentration is assumed as 1.32 × 10–13μM. Similar to CK2 species, for this model, we divide AnkG species into four species named as G1, G2, G3 and G4 with the equal concentration of 3.32 × 10–14μM; G1, G2, G3, and G4 bind with the phosphorylated form of NaA, NaB, NaC, and N1111,respectively.

Equations in the AIS model

The schematic of the model is shown in Figure 5. All parameters in the model, including their initial concentrations and rate constants, were estimated in accordance with the literature (Srinivasan et al., 1988; Meggio and Pinna, 2003;Ubersax and Ferrell, 2007; Brechet et al., 2008; Kole et al.,2008; Tables 1 and 2).

Phosphorylation of Nav species by CK2

Equations 1 to 20 models the dynamic changes in all Navspecies (NaA, NaB, NaC and N1111) after phosphorylation mediated by CK2 (C1, C2, C3, and C4) based on ordinary differential equations. The phosphorylated forms of the Navspecies are denoted aspNaA,pNaB,pNaCandpN1111.[NaA], [NaB], [NaC], [N1111], [pNaA], [pNaB], [pNaC], [pN1111],[C1], [C2], [C3] and [C4]represent the concentration of the species. The phosphorylation of all Navspecies by CK2 species in Equation (1), (6), (11), and (16) are achieved at specific association rates (K01,K02,K03andK04) and dissociation rates(K10,K20,K30andK40).

Binding of phosphorylated Nav species with AnkG

Equations 21 to 40 describe the accumulation of Navspecies at the AIS after binding with AnkG. The ordinary differential equations for binding between the phosphorylated form of Navspecies (pNaA,pNaB,pNaCandpN1111) and AnkG (G1, G2,G3, G4) are written using the mass action law. [pNaA], [pNaB],[pNaC], [pN1111], [pNaA×G1], [pNaB×G2], [pNaC×G3], [pN1111×G4], [G1], [G2], [G3] and [G4] represent the concentrations of the species. The binding of phosphorylated Navspecies with AnkG species in equations 21, 26, 31 and 36 is accomplished through the association rates (K05,K06,K07andK08) and the dissociation rates (K50,K60,K70andK80).

Results

Using the computational model, we tested the role of serine site in the recruitment of Navchannels into the AIS to understand the nature of binding of all phosphorylated Navspecies (NaA, NaB, NaC, and N1111) with AnkG species (G1,G2, G3, and G4) (Figure 6). The binding of pN1111with G4 and achieves the highest concentration of the pN1111recruited by G4 (2.64 × 10–15μM or equivalent to 1565 molecules within the AIS volume) among all Navspecies. These results are supported by the study in which 100–300 Navmolecules/μm of the AIS are documented (Kole et al., 2008). The higher concentration of pN1111after binding to G4 indicates that in the presence of all serine sites, the binding with AnkG takes place with a strong binding affinity, which corroborated with the experimental evidence (Fache et al., 2004; Brechet et al., 2008). This strong binding affinity is responsible for the high density of Navchannels at the AIS. Navspecies other than pN1111shows a dramatic decrease in their protein concentrations after binding with AnkG. After binding with G1, pNaA species has the second-highest concentration (7.2× 10–16μM or equivalent to 433 molecules within the AIS volume). As mentioned in the assumptions, the fact that pNaA consists of three serine sites including S1 could be the reason for this significant reduction of pNaA as compared to pN1111.Remaining Navspecies (pNaB and pNaC) after their respective binding with G2 and G3 show almost similar concentration but lesser than pN1111and pNaA (3.8 × 10–16μM or equivalent to 205 molecules in the AIS volume, and 3.8 × 10–16μM or equivalent to 228 molecules in the AIS volume) respectively.Similar concentrations of pNaB and pNaC after binding with AnkG species demonstrate the importance of the S1 site in the AIS motif. Similar results are given by Brechet et al. (2008),where the concentration of the Navchannels after S1 mutation is the same as the Navchannels with a double site mutation in the presence of S1 (Brechet et al., 2008). The S1 site within the AIS motif of a Navchannel is surrounded by acidic residues, such as glutamate and aspartate. The presence of these acidic residues increases the importance of the S1 site by increasing its probability of being phosphorylated by CK2 as the first preference. This could be the reason for pNaB to show a similar behavior compared to the pNaC species (S1 absence).

Therefore, the availability of all serine sites for CK2-mediated phosphorylation is necessary for the regulation of the Navchannel population at the AIS to maintain the conditions suitable for AP initiation. Besides, any alteration in the phosphorylation conditions can cause a disturbance in the recruitment of Navinto the AIS by preventing Nav-AnkG binding. According to the literature, any alteration in the Navchannel densities at the AIS can increase the threshold voltage required for AP initiation. This disturbance in the AIS system can shift the AP initiation zone from AIS to a region such as the nodes of Ranvier with a high number of Navchannels after AIS (Lemaillet et al., 2003; Kole et al., 2008; Rasband, 2009;Gulledge and Bravo, 2016; Satake et al., 2017). However, NOR,as an AP initiator zone, cannot hold the voltage stress during AP due to its position away from the soma (Kole et al., 2008,2012).

Table 1 ｜Initial concentration of all species considered in the axon initial segment model

Table 2 ｜Rate constants used in the axon initial segment model with their biological meaning

Effect of change in the initial concentrations of CK2 on Nav channel recruitment at AIS

To test the significance of the initial concentration of CK2 on the Navchannel recruitment at AIS, we phosphorylated N1111species at ten different CK2 (C2) initial concentrations and observed its impact on the binding of pN1111with AnkG (G4)(Figure 7). The initial concentration of C2 was taken within a range of 6.6 × 10–9μM to 1.98 × 10–8μM. According to the results, at low initial C2 concentration (6.6 × 10–9μM), lesser pN1111molecules bound to G4 (1.38 × 10–15μM or equivalent to 833 molecules in the AIS volume). As we increased the initial C2 concentration, pN1111molecules were also increased after binding with G4. These results indicate the dependency of binding affinity between Navchannels and AnkG on the CK2 concentration. Similar results were reported by various studies that supported the importance of CK2 (Brechet et al., 2008;Hsu et al., 2017). Hsu et al. (2017) showed that the inhibition of CK2 using 4,5,6,7-tetrabromobenzotriazole significantly altered the binding of Navchannels with AnkG by eliminating their CK2-mediated phosphorylation. In addition to that, CK2 inhibition shortened the axons by 30% (Ponce et al., 2011).Our model and the experimental studies so far show that CK2 plays a critical role in the AIS by enhancing the binding affinity between Navchannels and AnkG (Brechet et al., 2008; Ponce et al., 2011; Hsu et al., 2017).

Effects of the initial concentrations of AnkG on Nav channel recruitment at AIS

To test the role of AnkG, we introduced ten new initial concentrations of G4 in the present model. The concentration range of G4 was set to from 1.66 × 10–14to 4.98 × 10–14μM.Binding of G4 with pN1111was taken into account to test the impact of AnkG on the binding of Navchannels into the AIS. According to the results (Figure 8), at the low initial concentration of G4, pN1111levels were lower after binding with G4 (1.36 × 10–15μM ～806 molecules of pN1111in the AIS).However, as we increased the G4 initial concentration, the concentration of pN1111after binding with G4 increased. At the highest initial concentration of G4, pN1111achieved the highest value after binding with G4 (3.96 × 1014μM /2394 molecules).The population of the Navchannels in the AIS region is higher than those in the soma region (Rasband, 2010). The high number of Navchannels in the AIS region is reported by Kole et al. (2008); their experimental results suggest that the density of Navchannels in this region is higher by 50-fold than that in the soma (Kole et al., 2008). The high number of Navchannels in the structure of the AIS supports the AIS skeleton in overcoming the load-induced during AP initiation. Moreover, due to their high number, Navchannels in the AIS region activate and deactivate much faster than the Navchannels present in the soma. Also, the voltage required to change the membrane potential in AIS is low compared with the soma (Ogawa and Rasband 2008; Yoshimura and Rasband, 2014). However, the functions of AnkG are not limited by Navchannel recruitment because, according to previous studies, AnkG is responsible for the recruitment of ion channels (Navand Kv), molecular motors(kinesin and dynein) and CAMs (NF186 and NrCAM) into the AIS (Zhang and Bennet, 1998; Zhou et al., 1998; Hedstrom et al., 2008; Cunha and Mohler, 2009; Nelson and Jenkins,2017). Moreover, the function of AnkG is also related to protein trafficking within the neurons because AnkG is connected with microtubules through the EB3 protein. While the kinesin and dynein motors travel through microtubules to facilitate the proper trafficking of the proteins, AnkG related mutations can cause disturbances in the transportation of the proteins within neurons (Sun et al., 2014). Moreover, dendritic proteins, such as integrin-β1 and microtubule associated protein-2, have been found in the axons after AnkG mutation indicating an alteration in protein trafficking (Hedstorm et al., 2008).

Possible role of AIS in AD pathogenesis

The AIS mutations are not only restricted to the reduced ability of a neuron to generate action potential but also are associated with the development of seizures and epilepsy (Hauser et al.,1986). The role of AIS in AD pathogenesis has been studied.Experimental evidence showed that patients with sporadic AD have the risk of developing seizures especially after the onset of dementia (Hauser et al., 1986). On the other hand, the studies using AD brain samples found the presence of disturbed Navchannels (Kim et al., 2007). Disturbed level of Navchannels has potentially been associated with elevated BACE-1 (β-site amyloid precursor protein (APP) cleavage enzyme-1). Increased BACE-1 possibly inhibits the trafficking of channels to the AIS by preventing the molecular motors to perform their functions(Kim et al., 2007). The increased level of BACE-1 precipitated in AD pathogenesis by increasing pathogenic Aβ production and is associated with the cleavage of Navchannel subunits (Kim et al., 2007; Kovacs et al., 2010). In mouse models of AD, the area around Aβ plaques showed synaptic losses, axonal swelling,and mutations in the neuronal network (Marin et al., 2016). Aβ plaques could cause damage near or around the region of AIS by targeting AnkG and βIV spectrin and decreasing the density and length of the AIS (Marin et al., 2016; Sohn et al., 2016).Disturbance in the AIS functions could be a consequence of the calpain-mediated proteolysis of AnkG and βIV spectrin due to the induction of Aβ (Buffington and Rasband, 2011; Marin et al., 2016). The low AnkG concentration can dismantle the structure of AIS by loosening the proteins anchored by AnkG.Dismantling the structure results in the disruption of neuronal polarity by abolishing the functions of the important proteins present in this region (Buffington and Rasband, 2011). The results are shown by Sun et al. (2014) support the relationship between AIS and AD by observing the low level of AnkG in AD transgenic mice. The lower level of AnkG could results in the alteration of Navchannel at the AIS.

Conclusion

The importance of the serine sites (S1, S2, S3, and S4) in the accumulation of Navchannels into the AIS to create suitable conditions for AP initiation has been shown through this modeling study which corroborates with the experimental findings. The results suggest that, in the presence of all the serine sites in Navchannels for CK2-mediated phosphorylation,the binding between the Navchannels and AnkG takes place with a strong binding affinity. We showed the significance of the initial concentrations of CK2 and AnkG on the recruitment of Navchannels. From the simulation results, we observed that the low initial levels of CK2 and AnkG reduced the population of the Navchannels at the AIS. At low levels of CK2, the Navchannels are not fully phosphorylated, which weakens the binding between the Navchannels and AnkG. Moreover, at low AnkG concentration, the AIS structural assembly dismantles which could create difficulties in the proper trafficking of proteins into the AIS such as Navchannels. Low levels of Navchannels increase the voltage required to initiate AP. Overall,these results indicate the potentially strong association of AIS proteins in AD pathogenesis. Future experimental investigation into the activity levels of AIS related CK2 in AD brains may shed light into the more confirmed role of CK2 in AD pathogenesis.

Author contributions:DK conceptualized and led the project. PB developed the model with DK. DK and SS supervised the project. PB and DK wrote the paper. SS independently critiqued the first draft. All authors approved the final version of the paper.

Figure 8 ｜Effect of different concentrations of AnkG on the Nav channel recruitment at AIS.

Conflicts of interest:There are no competing interests.

Financial support:None.

Institutional review board statement:Ethical issues are not needed to be considered in undertaking this study.

Copyright license agreement:The Copyright License Agreement has been signed by all authors before publication.

Data sharing statement:Datasets analyzed during the current study are available from the corresponding author on reasonable request.

Plagiarism check:Checked twice by iThenticate.

Peer review:Externally peer reviewed.

Open access statement:This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-Non-Commercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms.