Cation currents in human airway epithelial cells induced by infection with influenza A virus

Cell culture

Standard, serial culture techniques were used to maintain H441 cells in RPMI medium supplemented with 8.5% fetal bovine serum (FBS), 8.5% newborn calf serum, 2 mm glutamine, 5 μg ml⁻¹ transferrin, 5 ng ml⁻¹ selenium, 20 nm insulin and an antibiotic/antimycotic mixture (Sigma Chemical Co., Poole, Dorset). All cells used in the present study were therefore maintained in medium supplemented with insulin. Whilst this medium is identical to that used in our first electrophysiological studies of these cells (Clunes et al. 2004), it differs slightly from the medium used in our more recent study of cells that had been strictly deprived of hormones/growth factors (Brown et al. 2008). For experiments, cells removed from culture flasks using trysin/EDTA were plated onto glass cover slips. After ∼24 h incubation in standard medium, the cells were washed and incubated (24–48 h) in medium identical to that described above except that the serum components were replaced with FBS (8.5%) that had been dialysed to remove hormones/growth factors. Whilst some experiments were undertaken using cells that had simply been cultured (24–48 h) either in this nominally hormone-free medium or in medium supplemented with 0.2 μm dexamethasone, most data are derived from glucocorticoid-deprived cells that were either transfected (Lipofectamine Transfection Reagent, Invitrogen, UK) with pGL3 plasmids (Invitrogen) incorporating appropriate cDNA sequences (see below), or infected with influenza A virus (see below).

Electrophysiology

Membrane currents (I_m) were recorded from cells plated onto glass cover slips using the perforated patch recording technique as detailed elsewhere (Horn & Marty, 1988; Clunes et al. 2004; Brown et al. 2008). The standard bath- and pipette-filling solutions (see Clunes et al. 2004) were designed to maintain quasi-physiological ionic gradients and so, under standard recording conditions, the equilibrium potentials for Na⁺, Cl⁻, and K⁺ (E_Na, E_Cl and E_K) were 68 mV, −42 mV and −82 mV, respectively. Since the cells were plated at relatively low density, it was possible to record currents both from cells growing singly and from cells growing as small groups of 2–5 cells. The input capacitance (C_m) and access resistance (R_a) of each preparation was carefully noted at the onset of each experiment and appropriate corrections applied to the voltage clamp circuitry. We typically achieved 50% compensation of the voltage drop across R_a, and the residual error in V_Hold (< 5%) was considered to be negligible. C_m and R_a were monitored throughout all experiments and all data included in the present study are derived from cells/groups of cells in which these parameters remained stable. Experiments were undertaken by recording the mean current evoked by driving the holding potential (V_Hold) through a series of four ramps (−113 to 87 mV in 2 s), firstly during exposure to the standard bath solution ([Na⁺]_o= 144.4 mm) and 20–30 s after [Na⁺]_o had been lowered to 10 mm by isosmotically substituting N-methyl-d-glucammonium (NMDG⁺). All reported potentials are corrected for the liquid junction potential between the pipette and bath solutions and, since the bath was earthed via a salt bridge filled with 3 m KCl, the solution changes imposed during the present study had negligible effects upon this potential. Data were analysed by constructing plots showing the relationship between I_m and V_Hold and the [Na⁺]_o-dependent component of the total membrane current (I_Na) was quantified by digitally subtracting the current persisting at low [Na⁺]_o from the corresponding control current. A directly analogous method was used to quantify the amiloride-sensitive (I_Amil) and La³⁺-sensitive (I_La) components of I_m. Cited values of membrane potential (V_m) were inferred from the observed values of the reversal potential (V_Rev) and values of total membrane conductance (G_Tot) and membrane Na⁺ conductance were estimated by regression analysis of data recorded at negative values of V_Hold. All reported currents and conductances have been normalized to the value of C_m which we associated with recordings from single cells (35 pF). This manipulation was undertaken to ensure that differences between the sizes of individual cells/cell clumps did not contribute to the variability in the experimental data, and such data are presented as picoamps per average-sized cell (pA cell⁻¹) and picosiemens per average-sized cell (pS cell⁻¹). The magnitude of the membrane Na⁺ current was quantified by measuring the Na⁺-dependent current flowing at −82 mV (I_{Na(−82 mV)}) a potential chosen since it equates to E_K. In some experiments [Na⁺]_o was lowered to 10 mm by isosmotically substituting K⁺. Bupivicaine (3 mm) was present in all solutions used in these experiments since this substance blocks the channels that underlie K⁺ conductance in resting cells (Inglis et al. 2007). The inclusion of bupivicaine therefore ensures that K⁺ currents flowing through these channels do not contribute to the current recorded during exposure to the K⁺-rich solution.

Cell transfection

The sequence encoding wild-type NS1 protein was derived from the PR8 strain influenza A virus (PR8/NS1) whilst the mutant form of this protein (PR8/NS1-Y89F) included an amino-acid substitution at Tyr⁸⁹ that prevents the binding/activation of PI3K (Hale et al. 2006). Heterologous expression of these proteins was achieved by transfecting cells with pGL3 plasmid encoding the appropriate gene sequences (Brown et al. 2008). The role of PI3K was explored by transiently expressing a chimeric protein (CD2-P110α) consisting of the catalytic P110α domain of PI3K (PI3K-P110α) attached to the extracellular and transmembrane domains of the rat CD2 surface antigen. This anchors the PI3K-P110α subunit to the inner surface of the plasma membrane thus increasing cellular PI3K activity (Reif et al. 1996). All cells were co-transfected with a second plasmid (pEGFP, 0.1 μg) encoding green fluorescent protein (GFP) so that successfully transfected cells could be identified (GFP fluorescence) visually.

Infection with influenza A

Influenza A virus (strain A/Udorn/72 [H3N2], referred to as rUd) was propagated in Madin–Darby canine kidney (MDCK) cells, titrated by plaque assay and stored as frozen stocks (see e.g. Takeda et al. 2002). For experiments, H441 cells on 10 mm diameter glass cover slips were placed in 35 mm Petri dishes (3 cover slips per dish) and bathed with ∼2 ml of hormone-deprived culture medium for 12–24 h before live virus was added to the medium. Preliminary experiments showed that H441 cells could support the replication of this virus and defined a viral titre (2 × 10⁷ plaque-forming units ml⁻¹) giving infection of ∼90% of cells. In some experiments cells were infected with a genetically engineered virus that incorporated the NS1-Y89F mutation (rUd/NS1-Y89F). Mock-infected cells were not exposed to virus but were otherwise treated identically.

Analysis of extracted protein

Mock-infected and virally infected cells grown on multi-well plates were lysed and aliquots of extracted protein were then denatured, fractionated on polyacrylamide gels and blotted onto nitrocellulose membranes which were then probed using antibodies against β-actin, NS1 or the Ser⁴⁷³-phosphorylated form of protein kinase B (PKB), which provides a read-out of cellular PI3K activity (Bayascas & Alessi, 2005). These methods are detailed elsewhere (Hale et al. 2006).

Untransfected cells

Analysis of currents recorded from untransfected cells showed that C_m was 25–40 pF in cells growing singly (see Methods) whilst cells growing in small groups (2–5 cells) displayed larger capacitances (50–100 pF). Since C_m provides an indication of membrane surface area (see Hamill et al. 1981), this finding confirms (Clunes et al. 2004; Brown et al. 2008) that H441 cells become electrically coupled to each other. However, irrespective of the extent of cell–cell coupling, glucocorticoid-deprived cells (C_m= 22–78 pF) displayed hyperpolarized values of V_m that were unaffected by lowering [Na⁺]_o (control: −62.4 ± 5.3 mV; Low Na⁺: −62.3 ± 4.8 mV) whilst dexamethasone-treated cells were relatively depolarized but became hyperpolarized when [Na⁺]_o was lowered. Whilst clear Na⁺ currents could thus be recorded from glucocorticoid-treated cells (see also Clunes et al. 2004; Brown et al. 2008), amiloride (10 μm) had no effect upon the Na⁺ currents recorded from cells growing singly but hyperpolarized electrically coupled cells by blocking the glucocorticoid-induced Na⁺ current (Table 1). These data therefore confirm (i) that glucocorticoids depolarize H441 cells by activating a Na⁺-permeable conductance (see also Clunes et al. 2004; Brown et al. 2008; Inglis et al. 2009) and (ii) that the properties of this glucocorticoid-induced conductance are influenced by the extent of cell–cell coupling (Brown et al. 2008). Our recently published data (Brown et al. 2008) therefore show that glucocorticoids depolarize electrically isolated cells by inducing a non-selective (P_Na/P_K≈ 0.6) cation conductance that is insensitive to amiloride but blocked by La³⁺ (see also Shlyonsky et al. 2005). In electrically coupled cells, however, these hormones induce a selective (P_Na/P_K≈ 16), amiloride-sensitive conductance similar to that co-expression of α-, β- and γ-ENaC co-expression (Clunes et al. 2004; Shlyonsky et al. 2005; Brown et al. 2008).

GFP-expressing cells

To ensure that exposure to transfection reagents and/or expression of heterologous protein did not alter the electrical properties of the cells, we explored the effects of transiently expressing GFP the fluorescent protein used as a marker of transfection (see Methods). Only 20–40% of cells expressed sufficient GFP to be visible using epifluorescence optics and, since we almost never saw groups of 2–5 GFP-expressing cells in contact with each other, our studies of transfected cells were undertaken by recording currents from cells growing singly to ensure that our data were derived exclusively from cells expressing heterologous protein (see also Brown et al. 2008). GFP-expressing cells maintained in glucocorticoid-free medium were hyperpolarized and insensitive to low [Na⁺]_o (Fig. 1A), and I_Na was therefore negligible (Fig. 1B) and G_Na close to zero. However, GFP-expressing cells that had been exposed to dexamethasone were depolarized and responded to low [Na⁺]_o with a hyperpolarizing shift in V_m (Fig. 1C; ΔV_m=−27.5 ± 3.5 mV; P < 0.001). These cells thus displayed clear Na⁺ currents (Fig. 1D) and regression analysis indicated that the magnitude of the underlying conductance was 766 ± 133 pS cell⁻¹, and it is therefore clear that the currents recorded from untransfected and GFP-expressing cells growing singly are indistinguishable. Further studies of GFP-expressing cells that had been exposed to dexamethasone showed that LY294002 (50 μm, 2–3 h), an inhibitor of PI3K, essentially abolished this glucocorticoid-induced Na⁺ current (Fig. 1D) and these data confirm (see also Thomas et al. 2004; Inglis et al. 2009) that this response was dependent upon PI3K.

Effects of PR8/NS1

Figure 2A shows currents recorded from electrically isolated PR8/NS1-expressing cells and, despite the fact that these cells had been maintained in glucocorticoid-free medium, V_m was more depolarized (P < 0.0001) than the control value measured in GFP-expressing cells (Fig. 2A). Moreover, lowering [Na⁺]_o hyperpolarized the PR8/NS1-expressing cells (Fig. 2A, ΔV_m=−24.6 ± 2.9 mV, P < 0.001) which displayed clear Na⁺ currents (Fig. 2B), and regression analysis indicated that the magnitude of the underlying conductance (866 ± 216 pS cell⁻¹) was similar to that measured in dexamethasone-stimulated cells. Exposing PR8/NS1-expressing cells to LY294002 (50 μm, 2–3 h) hyperpolarized V_m to −52.5 ± 5.9 mV, a value which did not differ from control. This inhibitor of PI3K also blocked the hyperpolarizing response to low [Na⁺]_o and abolished the PR8/NS1-induced Na⁺ current (Fig. 2B). Transient expression of PR8/NS1 therefore mimics the effects of glucocorticoid stimulation by evoking PI3K-dependent activation of a Na⁺-permeable conductance.

The data in Fig. 3A confirm that PR8/NS1 expression causes depolarization but, despite this clear finding, amiloride (1 mm) had no effect upon the currents recorded from these cells whilst the subsequent application of La³⁺ (1 mm) inhibited the currents flowing at positive and negative values of V_Hold and hyperpolarized V_m. Lowering [Na⁺]_o caused further hyperpolarization by inhibiting the currents flowing at negative values of V_Hold (Fig. 3A). Further analysis of these data confirmed that amiloride had no effect upon I_{Na(−82 mV)} (see Methods) whilst La³⁺ caused 70.5 ± 9.6% inhibition of this current (P < 0.02, Student's paired t test). Moreover, the La³⁺-sensitive current in these cells reversed at −10.5 ± 2.6 mV and, since this potential differed significantly (P < 0.0001, one-sample t test) from E_Na, this PR8-NS1-induced current cannot be Na⁺ selective. This situation was further analysed by assigning the K⁺ permeability (P_K) of the underlying conductance a value of unity and re-iteratively adjusting the value of P_Na to identify a solution to the Goldman–Hodgkin–Katz (GHK) equation consistent with the observed value of V_Rev. This analysis indicated that P_Na/P_K was 0.53, and it is therefore clear that expression of PR8/NS1 depolarizes hormone-deprived H441 cells by activating a poorly selective cation conductance essentially identical to the glucocorticoid-induced conductance seen in electrically isolated cells (Brown et al. 2008).

Physiological basis of the PR8/NS1-induced Na⁺ current

Transient expression of PR8/NS1 mimics the effects of glucocorticoid stimulation by activating a cation conductance and to study the signalling pathways underlying this response subsequent experiments explored the effects of expressing a mutant form of NS1 that is unable to activate PI3K (Hale et al. 2006). This mutant protein (PR8/NS1-Y89F) depolarized V_m to ∼−30 mV (Fig. 4A, P < 0.005 vs. data from cells expressing GFP alone, Student's t test) and, although this potential was more negative than that measured in cells expressing wild-type PR8/NS1 (see above), this apparent difference was not statistically significant. However, whilst PR8/NS1-Y89F-expressing cells displayed hyperpolarizing responses to low [Na⁺]_o, this response (ΔV_m=−11.1 ± 1.6 mV; P= 0.002, Student's paired t test) was smaller (P < 0.01, Student's t test) than that seen in cells expressing wild-type protein. Moreover, although PR8/NS1-Y89F expression did evoke Na⁺ current (Fig. 4C), the magnitude of the underlying conductance (∼280 pS cell⁻¹) was only ∼40% (P < 0.05, Student's t test) of that seen in cells expressing the wild-type protein (Fig. 4D).

PR8/NS1-Y89F therefore displays a mutant phenotype and the hypothesis that this abnormally small response reflects this protein's inability to activate PI3K (Hale et al. 2006) was tested by exploring the effects of expressing CD2-P110α, a chimeric protein that activates PI3K (Reif et al. 1996). Analysis of currents recorded from CD2-P110α-expressing cells showed that V_m was −46.0 ± 5.0 mV (Fig. 4B) and, although this potential is relatively hyperpolarized, it did differ (P < 0.02, Student's t test) from control, and CD2-P110α can therefore evoke a small depolarization (∼15 mV). However, although CD2-P110α-expressing cells did not hyperpolarize significantly when [Na⁺]_o was lowered (Fig. 4B), expression of this chimeric protein did seem to induce some Na⁺ current (Fig. 4C) since ionic substitution studies showed that these cells did express a Na⁺-permeable conductance whose magnitude (89.0 ± 26 pS cell⁻¹) exceeded the value measured in cells expressing GFP alone (P < 0.05, Student's t test). Parallel studies of cells co-transfected with both CD2-P110α and PR8/NS1-Y89F revealed a value of V_m (−16.1 ± 3.6 mV) similar to that measured in cells expressing wild-type PR8/NS1 (see Fig. 2). Moreover, the hyperpolarizing response to low [Na⁺]_o seen in the co-transfected cells (ΔV_m=−20.3 ± 4.9 mV) was also essentially identical to the response associated with wild-type PR8/NS1, whilst the Na⁺ currents recorded from these cells were also similar in magnitude to the currents induced by wild-type PR8/NS1 (Fig. 4C). Although CD2-P110α evokes only a barely perceptible increase in conductance in its own right, this PI3K-activating protein augments this PR8/NS1-Y89F-induced conductance to the wild-type level (Fig. 4D). It is therefore clear that co-expression of CD2-P110α can rescue the mutant phenotype associated with PR8/NS1-Y89F. However, the fact that this mutant protein can activate a Na⁺-permeable conductance in its own right indicates that the response to the wild-type PR8/NS1 must involve both PI3K-dependent and PI3K-independent signalling pathways.

Virus-induced expression of NS1

The fact that the transient expression of PR8/NS1 can activate a Na⁺-permeable conductance in glucocorticoid-deprived cells raised the possibility that such responses to NS1 may also occur during virus infection. We therefore explored the physiological consequences of infecting glucocorticoid-deprived cells with influenza A viruses. Analysis of protein extracted from cells infected with the wild-type rUd virus (see Methods) clearly revealed NS1 expression within the infected H441 cell (Fig. 5A) and, since this virus-encoded protein does not form part of the viral particle itself, this result shows that these cells are susceptible to infection. The presence of NS1 within the infected cells became apparent after ∼8 h and, as anticipated (Hale et al. 2006), the presence of this virus-encoded protein was associated with increased activity of PI3K (Fig. 5A). Parallel studies of glucocorticoid-deprived cells infected with a genetically engineered virus incorporating the NS1-Y89F mutation (rUd-NS1/Y89F) showed that this mutant virus also evoked NS1 expression within the infected cell, but did not activate PI3K (see also Hale et al. 2006). The images in Fig. 5B show cells plated onto glass cover slips under the conditions needed for electrophysiological analysis (see Methods) that had been infected with rUd or rUd/NS1-Y89F. The cells were prepared for immunocytochemical analysis after 12 h, and staining with an antibody against NS1 (Hale et al. 2006) showed that this virus-encoded protein was present in ∼90% of rUd- or rUd/NS1-Y89F-infected cells.

Physiological effects of virus infection

Figure 6A shows currents recorded from mock-infected cells maintained in dexamethasone-free medium. Since these currents were recorded both from single cells and from small groups of cells, C_m ranged from 19 to 77 pF (41.4 ± 3.3 pF, n= 34) but these cells all displayed hyperpolarized values of V_m that were insensitive to low [Na⁺]_o and these data confirm (present study; Clunes et al. 2004; Brown et al. 2008; Inglis et al. 2009) that G_Na is negligible in glucocorticoid-deprived cells. Figure 6B shows currents recorded from rUd-infected cells at identical passage (see Methods). During these experiments currents were recorded both from cells growing singly and from cells growing in groups of 2–5 but, irrespective of the pattern of growth, C_m (38.0 ± 3.4 pF, n= 22) was almost invariably close to ∼35 pF, the value which we associate with electrically isolated cells (Clunes et al. 2004). Values of C_m exceeding 50 pF were thus seen in only five instances and we never observed values of C_m > 70 pF, the value which we associate with cells growing in groups (see Table 1). Infection with rUd therefore seems to disrupt the normal electrical coupling between cells growing in contact with each other. Analysis of the currents recorded during exposure to the standard bath solution showed that rUd-infected cells were depolarized (V_m=−31.6 ± 2.2 mV, n= 22; P < 0.0001, Student's t test) when compared to mock-infected cells (Fig. 6B). Moreover, lowering [Na⁺] consistently hyperpolarized these cells (ΔV_m=−18.0 ± 2.8 mV; P < 0.0005, Student's t test) and further analysis revealed Na⁺ currents essentially identical to those seen in cells transiently expressing PR8/NS1 (Fig. 6C). The magnitude of the conductance underlying this virus-induced Na⁺ current was ∼500 pS cell⁻¹ (Fig. 6D).

Irrespective of the pattern of cell growth, cells infected with rUd/NS1-Y89F, which does not activate PI3K (Hale et al. 2006), displayed values of C_m close to ∼35 pF indicating that this mutant virus disrupts the electrical interconnections between neighbouring cells. Moreover, the value of V_m measured during exposure to the standard bath solution (−33.2 ± 3.3 mV, n= 14) was similar to the potential measured in cells infected with rUd and it is therefore clear that this virus can also cause depolarization. The rUd/NS1-Y89F-infected cells displayed hyperpolarizing responses to low Na⁺ and, although the measured fall in V_m (−10.5 ± 2.5 mV) appeared smaller than that seen in rUd-infected cells, this apparent difference was not statistically significant. However, analysis of the Na⁺ currents recorded from these cells (Fig. 6C) showed that the magnitude of the underlying conductance was only ∼30% of that measured in cells infected with the wild-type virus (Fig. 6D). The viral genotype therefore influences the magnitude of the influenza A virus-induced Na⁺ current. Although influenza A viruses encode their own ion channel (M2), this protein cannot explain the response described here since these channels are essentially impermeable to Na⁺ (Chizhmakov et al. 1996). Furthermore, both viruses used in the present study encode identical M2 proteins and yet there are clear differences in the rUd- and rUd/NS1-Y89F-induced currents.

Biophysical properties of the virus-induced conductance

Figure 7A shows data from experiments in which currents were again recorded from rUd-infected cells. These cells displayed values of C_m consistent with recording from electrically isolated cells (∼35 pF) and, although they had been maintained in glucocorticoid-free medium, all were relatively depolarized (V_m= 36.5 ± 6.0 mV, P < 0.0005 vs. data from mock-infected cells, Student's t test). Amiloride had no discernible effect upon the recorded current (Fig. 7A) but La³⁺ (1 mm) caused hyperpolarization (ΔV_m=−23.7 ± 6.0 mV, P < 0.02, Student's paired t test) by inhibiting the currents flowing at both negative and positive values of V_Hold. Subsequently lowering [Na⁺]_o to 10 mm had no further effect (Fig. 7A). Examination of the relationship between I_La and V_Hold (Fig. 7B) showed that this current reversed at −8.3 ± 4.3 mV, a potential which differed significantly (P < 0.0001, one-sample t test) from E_Na and it is therefore clear that this virus-induced current cannot be Na⁺ selective. Further analysis (see above) indicated that P_Na/P_K was 0.61.

In a further series of experiments I_La was quantified in virus-infected cells during exposure to a Na⁺-rich bath solution (n= 4) or to low Na⁺ solution (10 mm) containing K⁺ as the principal cation (n= 5). The values of C_m measured in these experiments (Na⁺-rich saline: 34.0 ± 4.0 pF; K⁺-rich saline: 24.8 ± 2.6 pF) were consistent with recordings from single cells and analysis of data recorded at high [Na⁺]_o showed that V_m (−15.7 ± 1.7 mV) was less negative than the value reported above (P < 0.05, Student's t test). It thus appears that bupivicaine, a K⁺ channel blocker present in all solutions used in this series of experiments (see Methods), depolarizes virus-infected cells. However, La³⁺ (1 mm) consistently hyperpolarized these cells (ΔV_m=−10.4 ± 1.2 mV; P < 0.005, Student's paired t test) and examination of the I_La–V_Hold relationship showed that this current reversed at −3.7 ± 2.6 mV. This potential does not differ significantly from that presented above, and indicates that P_Na/P_K must be 0.57. Moreover, the value of V_Rev measured during exposure to the K⁺-rich bath solution (−0.4 ± 5.5 mV) did not differ (one-sample t test) from the value predicted (GHK equation) for a conductance displaying this degree of Na⁺ selectivity. The rUd-induced depolarization therefore reflects the activation of a non-selective cation conductance essentially identical to the glucocorticoid- and the PR8/NS1-induced conductances seen in electrically isolated cells (Brown et al. 2008).

The response to dexamethasone

Although dexamethasone consistently depolarized H441 cells by activating a Na⁺-permeable conductance (see also Clunes et al. 2004; Brown et al. 2008; Inglis et al. 2009), amiloride had no effect upon the glucocorticoid-induced Na⁺ currents recorded from cells growing singly, but caused ∼90% block of this current in electrically coupled cells. This result accords with recently published data which show that glucocorticoids depolarize electrically isolated H441 cells by activating a cation conductance that is insensitive to amiloride but blocked by La³⁺ whereas, in electrically coupled cells, these hormones induce a Na⁺-selective, amiloride-sensitive ‘ENaC-like’ conductance (Brown et al. 2008). Such differential expression of poorly selective/selective Na⁺ conductances in single cells/cell clumps was first documented in renal epithelia (Korbmacher et al. 1993) and it is also clear that isolated alveolar type II (ATII) cells express both selective and poorly selective Na⁺ channels (Jain et al. 2001). Interestingly, the relative abundance of these two channel types in ATII cells is influenced by the culture conditions, with poorly selective channels being dominant in cells cultured under standard conditions on impermeable supports and ENaC-like channels dominating in cells grown in steroid-supplemented media on permeable supports at an air–liquid interface (Jain et al. 2001). Moreover, assays of surface protein abundance and the use of anti-sense oligonucleotides to suppress expression of specific ENaC subunits showed that both channel types were dependent upon α-ENaC, whilst the selective channels seemed to require the additional expression of β- and γ-ENaC (Jain et al. 1999, 2001). The conditions used to culture absorptive epithelial cells therefore influence the biophysical properties of the endogenous, Na⁺-permeable conductance (Korbmacher et al. 1993; Jain et al. 2001; Shlyonsky et al. 2005; Brown et al. 2008) and this seems to reflect an effect upon the relative abundance of ENaC subunits at the cell surface (Jain et al. 1999, 2001).

Transient expression of PR8/NS1

Transiently expressing PR8/NS1 in electrically isolated cells mimicked the effects of glucocorticoid stimulation by activating a Na⁺-permeable conductance essentially identical to the glucocorticoid-induced conductance seen in single cells (Brown et al. 2008). As far we are aware, this observation provides the first indication that any virus-encoded protein can act in this way and, since PR8/NS1 activates PI3K (Hale et al. 2006), it is tempting to attribute this response to the recently-reported activation of this kinase, particularly since PI3K is clearly important to the control of epithelial Na⁺ absorption (Record et al. 1998; Blazer-Yost et al. 1999; Snyder, 2002; Thomas et al. 2004; Tong et al. 2004; Pochynyuk et al. 2007; Brown et al. 2008). However, earlier experiments showed that activating PI3K, either by insulin stimulation or CD2-P110α expression (see Methods), did not evoke Na⁺ current in H441 cells that had been strictly deprived of hormones/growth factors (Brown et al. 2008). Although the cells in the present study were maintained under slightly different conditions (see Methods), the present data show that CD2-P110α evokes only a barely detectible increase in conductance, and it is therefore clear that activating PI3K does not provide a stimulus sufficient to mimic the effects of glucocorticoid stimulation (present study; Brown et al. 2008). However, PI3K inhibitors do abolish this response to glucocorticoid stimulation in these cells (present study; Thomas et al. 2004; Inglis et al. 2009) and we have also shown that PI3K enhances the glucocorticoid-evoked Na⁺ current (Brown et al. 2008) and so, whilst PI3K is clearly important to the induction/maintenance of epithelial Na⁺ absorption (Record et al. 1998; Blazer-Yost et al. 1999; Snyder, 2002; Thomas et al. 2004; Tong et al. 2004; Pochynyuk et al. 2007; Brown et al. 2008), the role of this kinase appears to be largely permissive in H441 cells.

The fact that PR8/NS1 evokes Na⁺ current as effectively as dexamethasone therefore suggests that the response to this virus-encoded protein cannot reflect a simple increase in PI3K activity and this, in turn, suggests that PR8/NS1 must be able to activate signalling pathways other than those dependent upon this kinase. To test this hypothesis we explored the effects of a mutant form of PR8/NS1 (PR8/NS1-Y89F) that is unable to activate PI3K (Hale et al. 2006). Whilst the expression of this protein did activate a Na⁺-permeable conductance, this response was smaller than that induced by the wild-type protein and PR8/NS1-Y89F therefore displays a mutant phenotype. Moreover, this phenotype could be rescued by co-expression of the PI3K-activating CD2-P110α protein, which shows that the abnormally small response to PR8/NS1-Y89F reflects a failure to activate PI3K (Hale et al. 2006). However, the most important result to emerge from these studies was that PR8/NS1-Y89F did evoke a small response in its own right, and this clearly shows that the PR8/NS1-induced Na⁺ current does not occur as a simple response to increased PI3K activity.

The full response to wild-type PR8/NS1 must therefore involve at least two signalling pathways, and it is therefore interesting that recently published data have shown that NS1 is phosphorylated by the extracellular signal-regulated kinases (ERK1/2) (Hale et al. 2009). Although the physiological significance of this is unclear, one possibility is that the presence of NS1 within the infected cells could suppress the ERK1/2-mediated phosphorylation of endogenous proteins by providing an alternative substrate for this kinase (Hale et al. 2009). Such competitive inhibition might well be significant since glucocorticoids inhibit ERK1/2 by evoking expression of a protein known as GILZ (glucocorticoid-induced leucine zipper), and this response appears to be an important part of the mechanism that allows these hormones to evoke Na⁺ transport (Soundararajan et al. 2005, 2007). Glucocorticoid-deprived H441 cells therefore display high levels of ERK1/2 activity (Inglis et al. 2008) and these kinases inhibit α-ENaC gene expression via multiple mechanisms (Zentner et al. 1998; Lin et al. 1999; Wang et al. 2000). Moreover, ERK1/2 also phosphorylates residues within β- and γ-ENaC, and this phosphorylation facilitates a binding interaction between ENaC and Nedd-4/2 (Yang et al. 2006). Since Nedd-4/2 is a ubiquitin ligase that targets ENaC for ubiquitilation, internalization and degradation (see for example Snyder, 2002, 2004), ERK1/2-mediated phosphorylation of these subunits leads to a loss of ENaC function (Yang et al. 2006). However, the mechanisms that allow glucocorticoid hormones to evoke Na⁺ transport are complex and incompletely understood and there are several alternative mechanisms that might underlie the PI3K-independent response to PR8/NS1 report here. Identifying the signalling pathways that allow PR8/NS1 to evoke Na⁺ currents in these cells may therefore clarify the mechanisms that allow glucocorticoids to regulate Na⁺ transport.

Infection with influenza A virus

As far as we are aware, the present data provide the first indication that a virus-encoded protein can activate an endogenous conductance in an epithelial cell and this raised the possibility that such responses to NS1 might also occur in virus-infected cells. We therefore explored the effects of infecting glucocorticoid-deprived H441 cells with influenza A viruses, and initial analyses of protein extracted from cells infected with a wild-type strain of influenza A (rUd) confirmed the presence of NS1 within virus-infected cells and showed that the expression of this virus-encoded protein was associated with increased activity of PI3K (Hale et al. 2006). Analysis of currents recorded from rUD-infected cells showed that this virus induced depolarization by activating a Na⁺-permeable conductance, and evidence that this response was dependent upon NS1 came from studies of cells infected with a virus incorporating the NS1-Y89F mutation. Although this mutant virus induced NS1 expression within the infected cell, it did not activate PI3K and, although this mutant virus did evoke Na⁺ current, the magnitude of the underlying conductance was only ∼30% of that induced by the wild-type virus. It is therefore clear that mutation of the NS1 gene modifies the physiological response to this virus and the virus-induced increase in conductance is therefore dependent, at least in part, upon NS1. However, the fact that rUd/NS1-Y89F did evoke some Na⁺ current indicates that the response to the wild-type virus must involve multiple signalling pathways, and this conclusion accords with the data derived from cells transiently expressing wild-type and mutant PR8/NS1 (see above).

Whilst uninfected cells growing in contact with each other normally become electrically interconnected, the values of C_m measured in virus-infected cells were almost invariably close to 35 pF, the value which we associated with recordings from single cells. The influenza A virus therefore appears to disrupt the electrical coupling between neighbouring cells, and it is therefore interesting that influenza A has also been shown to cause a loss of transepithelial resistance in confluent MDCK cells by opening the tight junctions between cells (López-Vancell et al. 1984). Taken together, these observations suggest that this virus may cause a loss of epithelial barrier function and, if this occurred in vivo, then this response to infection would tend to increase the volume of liquid in the lungs/airways. However, in the present context, the loss of cell–cell coupling had a broader significance since work from this laboratory (present study; Brown et al. 2008) and elsewhere (Shlyonsky et al. 2005), has shown that the biophysical properties of the hormone-induced conductance in H441 cells are influenced by the extent to which the cells have formed electrical contacts with each other. Whilst our original intention was to record currents from virus-infected cells growing singly and in small groups in order to explore the effects of influenza A upon the poorly selective and the selective Na⁺ conductances in these cells (Shlyonsky et al. 2005; Brown et al. 2008), the fact that this virus caused a loss of electrical coupling made this goal impossible. All virus-induced Na⁺ currents reported here therefore reflect the activation of an amiloride-resistant cation conductance that can be blocked by La³⁺. This conductance is identical to the dexamethasone-induced conductance seen in electrically isolated cells (Shlyonsky et al. 2005; Brown et al. 2008), and the present data show that infection with influenza A virus mimics the effects of glucocorticoid stimulation.

Significance of present findings

The present data show that the virus-encoded NS1 protein provides a mechanism that allows influenza A virus to activate an endogenous conductance in hormone-deprived H441 cells and, as far as we are aware, these data provide the first indication that a virus can act in this way. Whilst NS1-mediated activation of PI3K (Hale et al. 2006) will clearly have many downstream effects, the present response was not simply a side effect of PI3K activation since the full response to this virus-encoded protein involved at least two signalling pathways. The involvement of multiple signalling pathways suggests that the ability of influenza A to alter the biophysical properties of the infected cell may be a conserved aspect of virus biology that confers evolutionary benefit. However, the present data present an apparent paradox since apical cation channels provide a route for Na⁺ influx in absorptive epithelia (see for example Junor et al. 1999; Marunaka et al. 1999; Kemp et al. 2001) suggesting that the NS1-induced increase in G_Na which we now report may provide a mechanism that would tend to reduce the volume of liquid in the infected lung, whereas functional studies have now shown clearly that these viruses have the opposite effect (Chen et al. 2004; Wolk et al. 2008). It is possible that the NS1-induced conductance which we now report may be inserted into the basolateral membrane where it would tend to oppose the action of the Na⁺ pump by providing a route for Na⁺ influx/K⁺ efflux, and such an effect may well lead to a reduction in the overall rate of Na⁺ transport. However, the severity of influenza A-induced lung disease is determined by multiple interactions between the infective virus and its host (Hale et al. 2008; Wolk et al. 2008), and so the overall effect of the influenza A virus on the volume of liquid in the lung may well reflect a balance between the previously documented inhibitory effects (Kunzelmann et al. 2000; Chen et al. 2004; Wolk et al. 2008), effects on epithelial barrier function (see above) and the novel responses to NS1 which we now report. Interestingly, NS1 proteins encoded by different influenza A strains vary in their ability to activate PI3K (Hale et al. 2008) and a better understanding of the ways in which these virus-encoded proteins modify the physiology of the infected cell may therefore clarify the mechanisms underlying strain-specific variations in disease severity. Future work must therefore expand upon the present observation by defining the effects of virus-encoded NS1 proteins upon integrated experimental systems (see for example Wolk et al. 2008) that display a polarized epithelial phenotype which more closely models the situation in the human lung.