Basal Glutathionylation of Na,K-ATPase α-Subunit Depends on Redox Status of Cells during the Enzyme Biosynthesis

Many viruses induce oxidative stress and cause S-glutathionylation of Cys residues of the host and viral proteins. Changes in cell functioning during viral infection may be associated with glutathionylation of a number of key proteins including Na,K-ATPase which creates a gradient of sodium and potassium ions. It was found that Na,K-ATPase α-subunit has a basal glutathionylation which is not abrogated by reducing agent. We have shown that acute hypoxia leads to increase of total glutathionylation level of Na,K-ATPase α-subunit; however, basal glutathionylation of α-subunit increases under prolonged hypoxia only. The role of basal glutathionylation in Na,K-ATPase function remains unclear. Understanding significance of basal glutathionylation is complicated by the fact that there are no X-ray structures of Na,K-ATPase with the identified glutathione molecules. We have analyzed all X-ray structures of the Na,K-ATPase α-subunit from pig kidney and found that there are a number of isolated cavities with unresolved electron density close to the relevant cysteine residues. Analysis of the structures showed that this unresolved density in the structure can be occupied by glutathione associated with cysteine residues. Here, we discuss the role of basal glutathionylation of Na,K-ATPase α-subunit and provide evidence supporting the view that this modification is cotranslational.


Introduction
Viral infections lead to disruption of the redox status of mammalian cells. As a rule, the infections cause oxidative stress [1] and change the thiol redox status, which depends on the ratio of reduced (GSH) and oxidized (GSSG) glutathione [2,3]. Under normal conditions, the reduced glutathione level in cells (1-5 mM) is 100-fold higher than the oxidized glutathione level. Under oxidative stress, this ratio can be reduced to 1 [4]. The shift of intracellular redox conditions to the oxidized state induces protein glutathionylation that protects the thiol groups of proteins from irreversible oxidation and changes their function [5,6]. Many viruses that cause oxidative stress induce glutathionylation not only in host proteins but also in their own viral proteins. For example, the HIV-1-protease critical for viral maturation is activated by glutathionylation [7,8]. Changes in functioning of cells during viral infection may be associated with glutathionylation of a number of key proteins [5,6]. Notably, the acute (Flu) and chronic (hepatitis) viral infections can lead to the activation of factor Nrf/ARE [9,10], which causes activation of glutathione transferase participating in glutathionylation of proteins [11,12].
Na,K-ATPase creates a gradient of sodium and potassium ions necessary for all living mammalian cells. This protein is sensitive to changes in the redox status of cells [13][14][15][16][17]. Earlier we have shown that glutathionylation of Na,K-ATPase catalytic -subunit is a determinant of the enzyme redox sensitivity [13]. Four cysteines of Na,K-ATPasesubunit can undergo regulatory glutathionylation in case of GSSG increase, which leads to reversible inactivation of the enzyme, preventing exhaustion of ATP in the cells under oxidative stress. In addition to the regulatory glutathionylation of Na,K-ATPase -subunit, its basal glutathionylation was found which is not removed by DTT [13]. This leads to the suggestion that the basally glutathionylated cysteine residues are located in a region of protein structure inaccessible to solvent. Basal glutathionylation was identified for the Na,K-ATPase -subunit from various organisms, duck, mole rats, rabbit, and rat, in both cell lysates and purified enzyme preparations [13,17,18]. Basal glutathionylation is described for a variety of proteins: inhibitory kappa kinase beta IKK , actin, and aldose reductase [5,6]. Although basal glutathionylation of proteins is a widespread phenomenon, its role remains unclear. Despite the fact that the number of proteins with experimentally observed glutathionylation is growing rapidly [6], there is very small number of crystallographic structures with the identified bounded glutathione; that is, glutathione molecules are not reported in the existing X-ray structures of Na,K-ATPase -subunit. Identification of glutathione in the structures of glutathionylated proteins is a common problem and a number of approaches are used to achieve this, as in the structure of mitochondrial ABC transporter Atm1 where authors compared the free and glutathione-bound protein structures to prove the presence of bonded glutathione [19]. They showed that there is unresolved electron density in the glutathione-bound protein, and it corresponds to the glutathione molecule. Based on this work, we have assumed that associated glutathiones in the Xray structures of -subunit of Na,K-ATPase should be shown as regions with unresolved density. We have analyzed all Xray structures of the Na,K-ATPase -subunit from pig kidney and found that there are a number of isolated cavities with unresolved electron density close to the relevant cysteine residues. Analysis of the structures showed that the unresolved density in the structure can be occupied by glutathione associated with cysteine residues. We have shown that acute hypoxia leads to increase of total glutathionylation level of Na,K-ATPase -subunit; however, basal glutathionylation of -subunit increases under prolonged hypoxia only. We have found that basal glutathionylation could be removed from fully denatured enzyme only. In this paper, we discuss the role of the basal glutathionylation and provide evidence supporting the view that this modification is cotranslational.

2.1.
Modelling. Models of the glutathionylated Na,K-ATPase were constructed by COOT program using 3B8E, 3KDP, 3WGU, 3WGV, and 4HYT (PDB ids) structures as templates [20][21][22]. Two model structures of glutathione, the "linear" and "compact," were built and minimized ( Figure 1). In order to create models of the glutathionylated Na,K-ATPase, a corresponding model structure of glutathione was manually inserted into the cavity with unresolved electronic density (negative density and blob) using COOT. We assumed that the cavity was isolated when the distance between atoms of the residues forming the cavity was less than the sum of two van der Waals radii. Initially, GSH molecule has been placed in the cavity to verify conformity of the size of GSH molecule and the cavity. Subsequently, the GSH molecule was positioned in the cavity so as to form disulfide bond with a cysteine residue inside the cavity, and there were no structure overlaps. The resulting models of Na,K-ATPase with the S-S bonded glutathione molecule were locally minimized in the MMFF94x force field with the MOE program, version 2009.10. This force field accounts for atomic charges and hydrogen and ionic bonds between neighboring atoms. RMSD values of templates and glutathionylated models were calculated with MOE program (all atom types and parameters of symmetry were taken into account). Analysis of the presence of disulfide bonds in the Na,K-ATPase -subunit structures was performed using DCCP program [23].

2.3.
Immunoblotting. The level of S-glutathionylation of Na,K-ATPase 1-subunit was estimated using immunoblotting. Proteins of cell lysates were separated on SDS-PAGE and transferred to a PVDF membrane. After the blocking procedure, mouse monoclonal anti-glutathione antibody (Chemicon Millipore, MAB5310) was added. Mouse monoclonal anti-Na,K-ATPase 1 antibody clone C464-6 (Upstate Millipore) was applied to detect total amount of 1-subunit, followed by horseradish peroxidase-conjugated secondary antibodies. Membrane was stained using a commercial kit SuperSignal6 West Femto Maximum Sensitivity Substrate (Thermo Scientific) and chemiluminescence was detected using Bio-Rad ChemiDoc MP instrument. Densitometric analysis was performed by Image Lab (Bio-Rad) program and the results were represented as ratio of glutathionylatedsubunit to total -subunit band intensity ((GSS-1)/total 1). The comparison was made between samples applied on the same membrane. The ratio of the bands (GSS-1)/total 1 in control was taken as 1.

Immunoprecipitation.
Immunoprecipitation was performed on lysates of SC-1 cells. Anti-Na,K-ATPase 1 antibody (3 g) was added to the cell lysate and the sample was incubated for 15 minutes at 4 ∘ b and constant agitation. The resulting immune complex was added to a tube containing protein A agarose and incubated for 2 h at 4 ∘ b and constant agitation. The sample was then centrifuged for 1 min at 15000 g and the supernatant was removed. The precipitate was washed with PBS for three times and then heated with 4x Laemmli buffer containing 8 M urea and 8% SDS at 80 ∘ b for 5 min to elute the protein. The sample was centrifuged and the supernatant was collected. The supernatant containing Na,K-ATPase 1-subunit was divided into two parts. One part was incubated without and another with 25 mM TCEP dissolved in Tris buffer, pH 7.4, during 30 min at 37 ∘ C. Glutathionylation of Na,K-ATPase 1-subunit in samples was detected by immunoblotting.

Structures of Na,K-ATPase from Pig Kidney Have Isolated Cavities with Unresolved Density near Several Cysteine
Residues. There are eight structures of Na,K-ATPase from pig kidney, but only five of them have resolution better than 4Å (Table 1), which allow for identifying the isolated cavities containing unresolved electron density near cysteine residues. In these structures, the isolated cavities were identified close to cysteine residues Cys 204, Cys 242, Cys 336, Cys 349, Cys 367, Cys 421, Cys 452, Cys 456, Cys 457, Cys 511, Cys 549, Cys 577, Cys 656, Cys 599, and Cys 698 of Na,K-ATPase -subunit (Table 2). It has been found that the SH groups of pairs of cysteine residues were oriented into the same cavities: bys 204, Cys 242; Cys 367, Cys 698; Cys 452, Cys 456 (457); Cys 511, Cys 549. However, using the DSSP program [23], we have shown that there are no disulfide bonds in all structures of Na,K-ATPase -subunit.
To identify glutathione molecules bonded with cysteine residues, we searched the cavities with unresolved electron density. Unresolved density is a region of relatively high residual electron density that cannot be explained by presence of water [25]. Large areas of unresolved density were found in closed cavities near cysteines by COOT program [25]. The greatest number of areas with unresolved density, similar in shape and size to glutathione molecule, was found near the residues Cys 204, Cys 242; Cys 367, Cys 698; Cys 452, Cys 456, Cys 457; and Cys 599. It should be noted that there is no unresolved density near Cys 421 in all these structures, which coincides with the mass spectrometry data indicating that cysteine residue 421 (Cys 423 of duck 1) has never been found S-glutathionylated [13].
We have analyzed which of the cysteine residues in the pair may be more accessible for glutathionylation. According to [4,26,27], cysteine residues located close to positively charged residues (arginine, lysine, and histidine) are more accessible for glutathionylation because they attract the electron density, increasing the probability of glutathione binding. On the contrary, negatively charged residues close to cysteine reduce the probability of glutathionylation. Analysis of the amino acid composition near Cys residues showed that in the pairs of Cys residues pointing into the same cavity one of the cysteines had a positively charged amino acid in its close vicinity, whereas the other Cys had a negatively charged residue (Table 3). Considering these data, cysteine residues 204, 452, 599, and 698 are preferable for glutathionylation. The mass spectrometry data obtained earlier confirmed basal glutathionylation of residues 204 and 698 [13], but it was not possible to assess glutathionylation of residues 452 and 599.

Glutathione Fits into the Identified Isolated Cavities and
Areas with Unresolved Electron Density. Models of the Na,K-ATPase -subunit with glutathione in isolated cavities were built manually using COOT to verify fitting of glutathione molecules to the cavities. For this purpose, two models of glutathione molecule were used (Figure 1), which were placed in isolated cavities avoiding overlap with the resolved parts of structure. Glutathione molecule was placed in isolated cavities to fit unresolved densities, and thiol group of glutathione has been directed towards the thiol group of a cysteine in the cavity. Areas of unresolved electron density, large enough to insert a glutathione molecule, were found in a number of isolated cavities in Na,K-ATPase near the cysteine residues listed in Table 4. It was found that the electron density capable of accommodating glutathione molecule by shape and size is located in cavities near the cysteine residues Cys 204, Cys 242 ( Figure 2 Table 4). The role of these Cys residues in the functioning of the enzyme was evaluated previously with point mutagenesis [24]. Replacing one of Cys 367 and Cys 698 residues or all three Cys residues (452, 456, and 457) with alanine or serine resulted in reduced activity by more than 75%, while replacing Cys 204 and Cys 599 residues resulted in reduced activity by more than 50% (Table 4). This indicates that these Cys residues are not critical for the enzyme activity but their glutathionylation could play an important role in regulating the protein function. Substitution of Cys 242 to Ala or Ser residues was lethal to the cells, suggesting the central role of this residue for the functioning of Na,K-ATPase. According to our data, Cys 242 is one of the residues undergoing regulatory glutathionylation in oxidative stress [13].
The mutagenesis data allowed for concluding that the presence of disulfide bridges is not required for folding of the Na,K-ATPase -subunit and its subsequent activity [24]. Indeed, analysis of the available structures shows that there are no disulfide bonds between the cysteine residues. Moreover, according to our data, closely located cysteine residues Oxidative Medicine and Cellular Longevity 5 a According to MALDI-TOF mass spectrometry data in [13]. b "Density": unresolved density that can fit by glutathione. c "Small size of density": unresolved density but too small to fit glutathione. d "Distant density": unresolved electron density at a distance of more than 5Å from Cys (glutathione fits density, but the density is far from the residue). e "Fragmented": no intact unresolved density. f "nd": no unresolved density close to the residue.

Energy Minimization of the Models.
To verify the obtained data, an SS bond has been formed between the glutathione molecule and the nearest cysteine residue and energy minimization of the obtained models carried out using the MOE program. After minimization, in 8 out of 9 models, glutathione not only stayed within the unresolved density but has moved to a more consistent location in the given density. Only in a single case out of nine, the glutathione molecule after minimization was partially located outside the unresolved density (3WGU, Cys 452). Changes in the Na,K-ATPase structure after minimization are given in Table 5. RMSD between the initial and minimized Na,K-ATPase structures is in the range of 0.05-0.09Å, indicating that that minimization with the bound glutathione does not result in significant changes of the Na,K-ATPase structure.
We have compared the distances between sulfur atoms in the pairs of the cysteines in the initial structure of the protein (3WGU) and in the model structures (obtained by us) with glutathione attached to one of the cysteine residues (Table 6). This comparison revealed that glutathione binding has little effect on the distance between sulfur atoms in the Cys pairs. After local minimization, RMSD value between the initial and glutathionylated Na,K-ATPase structures was smaller then RMSD value between the initial structure of Na,K-ATPase and Na,K-ATPase with Cys-S-S-Cys bridges (Table 5). Thus, shifting cysteine residues towards each other until the distance between them reached SS bridge length (2.04Å) and subsequent formation of disulfide bridges make greater changes in the structure of the molecule than incorporation of glutathione. Introduction of glutathione in the regions of protein with unresolved density has almost no effect on protein structure and location of cysteine residues determined by crystallography.

Prolonged Incubation of Cells under Hypoxic Conditions
Alters the Basal Level of Na,K-ATPase Glutathionylation. It can be hypothesized that the basal glutathionylation depends on the redox status of cells during protein folding. To test this, we incubated SC-1 cells at 0.2% and 20% pO 2 within 3.5 h and 96 h. Then, cell lysates were treated with the reducing agent TCEP (25 mM) for 30 min. Acute hypoxia (3.5 h) leads to increase of total glutathionylation level of Na,K-ATPase -subunit ( Figure 6(a)) that corresponds to our data, obtained earlier [13,28]. However, basal glutathionylation of -subunit does not change at these conditions ( Figure 6(b)). In contrast, at prolonged hypoxia (96 h), basal level of Na,K-ATPase -subunit glutathionylation was significantly higher than that under 20% pO 2 (Figure 6(c)). Since the basal glutathionylation is not removed by reducing agents and the glutathionylated cysteines are located in isolated cavities, we assume that the reaction of glutathionylation occurs during protein folding. If so, then the basal glutathionylation can be removed only from the unfolded protein. We    : Schematic representation of Na,K-ATPase glutathionylation depending on intracellular redox status. At normal redox status, the level of GSH is about 100 times higher than GSSG. Under these conditions, during biosynthesis, the proteins are slightly glutathionylated. Redox stress leads to the shift in GSH/GSSG ratio that induces protein glutathionylation. At normal redox status and at redox stress, the basal levels of glutathionylation (glutathionylation of solvent-inaccessible cysteine residues) are different. Subsequent change in the redox status does not affect basal glutathionylation, which demonstrates that the protein "memorizes" a cellular redox state during its biosynthesis. In contrast, glutathionylation of the solvent-accessible cysteine residues (regulatory glutathionylation) depends on the current redox status of cell. performed immunoprecipitation of Na,K-ATPase -subunit from lysates of SC-1 cells incubated at 0.2% and 20% pO 2 within 3.5 h and 96 h. Analysis of immunoprecipitatedsubunit revealed increasing of its glutathionylation under hypoxic conditions (Figure 6(d)). Then, we treated Na,K-ATPase -subunit obtained by immunoprecipitation with TCEP under denaturing conditions (8 M urea and 8% SDS). In this case, the glutathionylation was fully removed (Figures 6(e) and 6(f)). Based on these data, we can conclude that the basal glutathionylation is a cotranslational modification which, for example, is necessary to prevent the formation of disulphide bridges between the neighboring cysteine residues during protein folding. Formation of disulfide bridges can increase rigidity of the structure and prevent conformational lability of the molecule. In particular, residues Cys 204 and Cys 242 are located in the actuator domain of Na,K-ATPase that performs large amplitude transitions during the catalytic cycle, which would be impossible if SS bridges were formed between these residues. In addition, the formation of SS bridge between Cys 204 and Cys 242 residues will prevent the exposition of SH group of Cys 242 to the solvent from the cavity and its regulatory glutathionylation. It is also possible that glutathionylation is necessary for correct protein folding. Only glutathione and oxidized glutathione are required for glutathionylation of -subunit of Na,K-ATPase [13] without any additional enzymes and cofactors. Thus, the ability of SH group to undergo glutathionylation directly depends on the redox state of cells [5]. Reduced SH groups interact with GSSG, the level of which increases in oxidative stress. In the case of oxidation of thiol groups to SOX, they interact with GSH, the concentration of which in cytosol is 1-5 mM. Unlike the genetic code and histone code that underlie information storage and utilization, the epigenetic code and redox code modulate operation of the genetic and histone codes in the organizational structure, differentiation, and adaptation of an organism to the environment [29]. Redox signaling and redox control in multicellular organisms evolved and diversified with the increase in atmospheric O 2 about 600 million years ago. We assume that cotranslational glutathionylation allows for "remembering" a cellular redox state during protein biosynthesis, and as a result proteins