Dr Stuart M Wilson, Centre for Cardiovascular and Lung Biology, Division of Medical Sciences, Ninewells Hospital and Medical School, University of Dundee, Dundee DD1 9SY, UK. E-mail: firstname.lastname@example.org
BACKGROUND AND PURPOSE Insulin-induced Na+ retention in the distal nephron may contribute to the development of oedema/hypertension in patients with type 2 diabetes. This response to insulin is usually attributed to phosphatidylinositol-3-kinase (PI3K)/serum and glucocorticoid-inducible kinase 1 (SGK1) but a role for protein kinase B (PKB) has been proposed. The present study therefore aimed to clarify the way in which insulin can evoke Na+ retention.
EXPERIMENTAL APPROACH We examined the effects of nominally selective inhibitors of PI3K (wortmannin, PI103, GDC-0941), SGK1 (GSK650394A) and PKB (Akti-1/2) on Na+ transport in hormone-deprived and insulin-stimulated cortical collecting duct (mpkCCD) cells, while PI3K, SGK1 and PKB activities were assayed by monitoring the phosphorylation of endogenous proteins.
KEY RESULTS Wortmannin substantially inhibited basal Na+ transport whereas PI103 and GDC-0941 had only very small effects. However, these PI3K inhibitors all abolished insulin-induced Na+ absorption and inactivated PI3K, SGK1 and PKB fully. GSK650394A and Akti-1/2 also inhibited insulin-evoked Na+ absorption and while GSK650394A inhibited SGK1 without affecting PKB, Akti-1/2 inactivated both kinases.
CONCLUSION AND IMPLICATIONS While studies undertaken using PI103 and GDC-0941 show that hormone-deprived cells can absorb Na+ independently of PI3K, PI3K seems to be essential for insulin induced Na+ transport. Akti-1/2 does not act as a selective inhibitor of PKB and data obtained using this compound must therefore be treated with caution. GSK650394A, on the other hand, selectively inhibits SGK1 and the finding that GSK650394A suppressed insulin-induced Na+ absorption suggests that this response is dependent upon signalling via PI3K/SGK1.
protein encoded by n-myc downstream regulated gene 1 and gene 2
phosphoinositide-dependent protein kinase 1
plekstrin homology domain
provirus integration site for Moloney murine leukaemia virus kinase
protein kinase B (also known as Akt)
polo-like kinase 1
40 kDa proline-rich substrate of Akt (PKB)
70 kDa ribosomal S6 kinase
sodium dodecyl sulphate
serum and glucocorticoid-inducible kinase 1
smooth-muscle myosin light-chain kinase
TORC1 and 2
target of rapamycin complex 1 and 2
The regulated re-absorption of Na+ within the aldosterone-sensitive epithelia of the distal nephron (ASDN; late distal convoluted tubule and the cortical collecting duct) is a physiologically important process that determines the amount of Na+ lost in urine, and this absorptive mechanism is therefore of central importance to Na+ homeostasis, whole body fluid balance and the control of blood pressure (reviewed by Loffing and Korbmacher, 2009). Na+ absorption within the ASDN is dependent upon epithelial Na+ channels (ENaC), transport proteins composed of 3 subunits (α-, β- and γ-ENaC) that form the highly selective Na+ channels, which allow apical Na+ entry in absorptive epithelia (Canessa et al., 1993; 1994). The activity and/or surface expression of these channels is controlled by the hormones involved in the control of Na+ and water balance (aldosterone and vasopressin) but, in addition to its role in carbohydrate metabolism, insulin is also known to stimulate Na+ absorption within the ASDN (Atchley et al., 1936; Defronzo et al., 1975; 1976; Blazer-Yost et al., 1998; Tiwari et al., 2007; Loffing and Korbmacher, 2009). While this effect may normally have little physiological significance (Loffing and Korbmacher, 2009), it can become relevant in patients with type 2 diabetes, as this condition is almost invariably treated using drugs that exert a generalized insulin-sensitizing action. At least some such drugs can cause tissue oedema, high blood pressure and increased risk of congestive heart failure, particularly if co-administered with insulin itself. While the mechanism underlying this oedema is not well understood (see review Buckingham and Hanna, 2007), it does seem to involve inappropriate stimulation of renal Na+ retention (Hong et al., 2003; Guan et al., 2005). Ideally, type 2 diabetes should therefore be treated using drugs that promote glucose uptake without altering renal Na+ handling, but this goal will only become achievable once the signalling pathways that allow insulin to control Na+ transport are fully understood. The present study therefore explores the mechanisms that allow insulin to stimulate Na+ transport in a cell line (mpkCCD) derived from the mouse cortical collecting duct (Bens et al., 1999).
Standard techniques were used to maintain the mpkCCD cell line, which is derived from the mouse cortical collecting duct (Bens et al., 1999), in serial culture. The medium used was phenol red-free DMEM/Ham's F12 supplemented with glutamine (2 mM), foetal bovine serum (FBS, 2%, v/v), penicillin/streptomycin mixture (1%; Sigma, Poole, Dorset, UK), sodium selenate (60 nM), transferrin (5 µg·mL−1); dexamethasone (50 nM), triiodothyronine (1 nM), epidermal growth factor (10 ng·mL−1) and insulin (5 µg·mL−1). For experiments, cells were removed from culture flasks using trypsin/EDTA and plated onto Snapwell membranes (5 × 105 cells·cm−2, electrometric studies) or Transwell membranes (2.5 × 106 cells·cm−2, Western analysis). Cells were then cultured in complete medium that was replaced every 48 h, and after 6 days, this standard medium was replaced with hormone-free medium (phenol red-free DMEM/Ham's F12 medium containing antibiotics and 2% charcoal-stripped FBS). After a further 24 h, serum was withdrawn and the cells used in experiments after a further 18–24 h.
Quantification of Na+ transport
Snapwell membranes bearing confluent cells (see above) were mounted in Ussing chambers and bathed with bicarbonate-buffered physiological salt solution (composition in mM: NaCl, 112; NaHCO3, 25; KCl, 4.7; MgSO4, 1.2; KH2PO4, 1.2; CaCl2, 2.5 and D-glucose, 11.6, pH 7.3–7.5 when bubbled with 5% CO2). All cells were maintained under open circuit conditions and transepithelial potential difference (Vt) was monitored (DVC-1000 Voltage/Current Clamp, World Precision Instruments, Stevenage, UK) and recorded (4 Hz) directly to computer disk (ADI Powerlab Interface and associated software; AD Instruments, Chalgrove, Oxfordshire, UK). Experiments were initiated once Vt had stabilized (40–50 min) and, during each recording, standard pulses of transepithelial current (20 s, −10 µA·cm−2) were injected every 40 s. The magnitudes of the resultant deflections in Vt (ΔVt) were then used to calculate transepithelial resistance (Rt) which allowed the equivalent short circuit current (IEq; i.e. the current that would be required to hold Vt at 0 mV) to be calculated using the expression IEq=Vt/Rt. Such calculations were undertaken using spreadsheet software (Microsoft Excel) and, because all experimental manoeuvres were carefully timed, we were able to align the individual data sets, which allowed us to calculate mean values showing the pooled data for each series of experiments undertaken.
All values of Vt are shown relative to an earth electrode in the basaolateral bath, which implies that the current carried by cations moving from the lumen to the interstitium will be negative. Such currents are therefore shown as downward deflections of the traces. While this convention differs from that used in many previous electrometric studies of cultured epithelia (see e.g. Wilson et al., 2010), it does accord with more general conventions that are invariably used in electrophysiological studies of single cells. Moreover, the experimental approach used in this study differs from that followed in our earlier studies because, until now, we have always measured short circuit current (ISC) directly from cultures held under voltage clamp (see e.g. Wilson et al., 2010). The values of IEq reported here are, however, very similar to our recently reported values (Wilson et al., 2010) and it is therefore clear that the two approaches do provide essentially identical data. We believe that the present method is preferable because, even in hormone-deprived cells, Vt is −20 to −40 mV and this potential can hyperpolarize to −70 mV during insulin stimulation. Holding Vt at 0 mV in order to measure ISC directly would therefore hyperpolarize the apical membrane potential and establish electrochemical driving forces for ionic movements that may not occur in vivo. This has the potential to alter the physiological properties of the cells.
Western analysis of extracted protein
Cells on Transwell membranes were washed with ice-cold phosphate buffered saline and scraped into ice-cold lysis buffer containing protease and phosphatase inhibitors (1% Triton; 50 mM Tris – HCl, pH 7.5; 1 mM EGTA; 1 mM EDTA; 1 mM Na orthovanadate; 10 mM glycerol phosphate; 50 mM NaF; 5 mM Na pyrophosphate; 270 mM sucrose; 0.1% β-mercaptoethanol; 1 Roche Mini Protease Inhibitor tablet per 10 mL). Lysates were then transferred to Eppendorf tubes, ultrasonicated to ensure complete cellular disruption and their protein contents determined using Bradford reagent. Aliquots of extracted protein were then reduced and denatured by heating (95°C, 5 min) in the presence of 5% β-mercapto-ethanol and 10% sodium dodecyl sulphate (SDS) and fractionated on 10% SDS-polyacrylamide gels. The fractionated proteins were then transferred to Hybond-P membranes (Amersham, Bucks., UK) that were probed using the antibodies described below. Immunoreactive proteins were visualized by enhanced chemiluminesence.
Activity of endogenous kinase activity
PI3K activation was monitored using Western analysis (see above) to monitor changes in the cellular abundance of Ser473-phosphorylated PKB, as it is now clear that the phosphorylation status of this residue provides a reliable read out of cellular PI3K activity (Bayascas and Alessi, 2005). These analyses were undertaken using antibodies against Ser473-phosphorylated and total PKB. Serum and glucocorticoid-inducible kinase 1 (SGK1) activity was monitored by assaying the phosphorylation status of residues (Thr346/356/366) within NDRG1 that are physiological substrates for SGK1 and not for other, closely related kinases including PKB (Murray et al., 2005a; García-Martínez and Alessi, 2008; Inglis et al., 2009). Similarly, the phosphorylation of PRAS40-Ser246 was monitored as an indicator of PKB activity (Kovacina et al., 2003), while P70-S6K-Thr389 phosphorylation was assumed to indicate kinase activity of the target of rapamycin signalling complex 1 (TORC1). The results of all such experiments were quantified by densitometry using the Syngene Genegenius image capturing system, GeneSnap image capturing program and GeneTools densitometry analysis program (Synoptics Ltd., Cambridge, UK).
Experimental design and data analysis
All experiments were undertaken using strictly paired protocols in which control and experimental cells were age-matched and at identical passage. Care was taken to ensure that the control and experimental cells were handled identically and electrometric studies were therefore undertaken using parallel Ussing chamber systems, so that we could simultaneously record currents from control and experimental cells. Analyses of extracted proteins were similarly undertaken using paired experimental designs in which the phosphorylation of endogenous proteins was assayed both in hormone-deprived and insulin-stimulated cells. All experiments involving putative inhibitors of protein kinases involved the analysis of protein extracted from pairs of hormone-deprived/insulin-stimulated cells that had either been exposed to the test compound or to the solvent vehicle (0.1% DMSO). Initial experiments showed that this concentration of DMSO had no effect on any of the parameters studied (not shown). All data are presented as mean ± SEM, and values of n refer to the number of times a protocol was repeated using cells at different passage. The statistical significance of differences between data derived from hormone-deprived and insulin-stimulated cells were assessed using Student's paired t-test, whereas the results of experiments undertaken using more complex protocols were analysed by one- or two-way analysis of variance/Bonferroni post hoc test.
Amiloride, 5-(N-Ethyl-N-isopropyl) amiloride (EIPA), benzamil, insulin, culture reagents and all general laboratory reagents were from Sigma (Poole, UK) while PI103, rapamycin and inhibitor of Akt (PKB) 1/2 (Akti-1/2, Barnett et al., 2005; Lindsley et al., 2005) were from Merck (Beeston, UK). GSK650394A and GDC-0941 were a generous gift from Prof D.R. Alessi, who had arranged for these compounds to be synthesized within the MRC Protein Phosphorylation Unit (MRC-PPU) at the University of Dundee. Antibodies against Ser473-phosphorylated and total protein kinase B (PKB), and Thr389-phosphorylated and total 70 kDa ribosomal S6 kinase (P70-S6K) were from Upstate (Watford, UK) while the antibodies against Thr346/356/366-phosphorylated and full length forms of the protein encoded by the n-myc downstream regulated gene 1 (NDRG1), and the Ser246-phosphorylated and total forms of the proline-rich 40 kDa substrate of Akt (PKB) (PRAS40) were prepared within the antibody production unit within the MRC-PPU. We are grateful to Prof Sir Phillip Cohen (MRC-PPU) for allowing us access to these antibodies.
Bioelectric properties of hormone-deprived cells
Initial studies of confluent cells (n= 8) showed that Vt, Rt and IEq were normally ∼−43.8 ± 1.5 mV, ∼2.5 ± 0.2 kΩ·cm2 and ∼−16.2 ± 1.7 µA·cm−2, respectively, and, as anticipated (see Bens et al., 1999), amiloride (10 µM) caused a rapid (5–10 s) and almost complete depolarization of Vt (−3.9 ± 0.4 mV, P < 0.001). As this response was accompanied by an increase in Rt (4.4 ± 0.4 kΩ·cm2, P < 0.001), this ENaC blocker essentially abolished IEq (−0.8 ± 0.1 µA·cm−2, P < 0.001). Further experiments (n= 9) in which the apical concentration of amiloride was increased progressively showed that these effects were concentration-dependent and established that concentrations >10 µM were maximally effective. The concentration needed for half maximal inhibition (IC50) of IEq was 0.74 ± 0.01 µM. Benzamil reproduced these effects of amiloride fully but was ∼35-fold more potent (EC50= 22.7 ± 1 nM, n= 9) and, while EIPA also depolarized Vt and increased Rt, the highest concentration tested (3 × 10−4 M) caused only ∼75% inhibition of IEq which made it difficult to estimate IC50 accurately. EIPA was, however, ∼100-fold less potent than amiloride (IC50= 90 ± 20 µM, n= 7). The rank order of potency among these compounds is therefore benzamil > amiloride > EIPA. This observation confirms (see Bens et al., 1999) that hormone-deprived mpkCCD cells spontaneously absorb Na+ from the apical bath via an ENaC-dependent mechanism.
Bioelectric response to insulin
Figure 1 shows the results of experiments that explored the effects of insulin on the bioelectric properties of these cells. Vt was ∼−50 mV at the onset of these experiments and, because Rt was ∼2 kΩ·cm2, these data confirm (see above) that IEq is normally ∼−20 µA·cm−2 (Figure 1). The control data show that Vt tended to depolarize slightly with time and, as Rt was stable, this effect is reflected by a slight decline in IEq (Figure 1). Insulin (20 nM, basolateral) hyperpolarized Vt to ∼−60 mV and this response became apparent after 3–5 min latency and reached a plateau after ∼45 min (Figure 1). This hyperpolarization was accompanied by only a small fall in Rt (Figure 1) and further analysis showed that insulin evoked an augmentation of IEq that reached a plateau after ∼30 min (Figure 1). Apical amiloride (10 µM) abolished Vt and increased Rt in unstimulated and insulin-stimulated cells and, even after stimulation with insulin, only negligible currents persisted in the presence of amiloride (Figure 1). The insulin-induced augmentation of IEq must therefore reflect stimulation of ENaC-mediated Na+ absorption. Interestingly, insulin also increased the value of Rt measured in the presence of amiloride (Figure 1), indicating that this hormone must have other effects on these cells (e.g. increasing paracellular resistance). The physiological basis of this action was not investigated.
Insulin-induced phosphorylation of endogenous proteins
Insulin (20 nM, basolateral) increased the cellular abundance of the Ser473-phosphorylated PKB without altering the overall abundance of this protein (Figure 2A,B) and this result shows that insulin evokes PKB-Ser473 phosphorylation. This response reached a maximum after 15–30 min and, although there was some decline from this peak value, increased phosphorylation of PKB-Ser473 persisted for at least 6 h (Figure 2A,B). As the phosphorylation of this residue is dependent upon PI3K (Bayascas and Alessi, 2005), this result shows that insulin normally activates this phospholipid kinase (see review, Cohen, 2006). Insulin (20 nM, basolateral) also increased the abundance of Thr346/356/366-phosphorylated NDRG1 and this response, in common with the phosphorylation of PKB-Ser473, occurred with no change in the overall expression of this protein. The insulin-induced phosphorylation of these residues followed a very similar time course to the phosphorylation of PKB-Ser473 (Figure 2C,D) and, because NDRG1-Thr346/356/366 phosphorylation is catalyzed by SGK1 and not by other related kinases, including PKB (Murray et al., 2005a,b), this result shows that insulin also activates SGK1 (Cohen, 2006). As insulin had no effect upon the overall abundance of PKB or NDRG1 (Figure 2A–D), in all subsequent experiments changes to the cellular abundance of the Ser473- and Thr346/356/366-phosphorylated forms of these proteins were assumed to be a reliable biomarker of increased phoshorylation of these residues. All such data were normalized to the phosphoprotein abundances measured in hormone-deprived cells. We did, however, continue to monitor the overall expression of PKB and NDRG1 in all experiments (data not shown).
PI3K-mediated phosphorylation of PKB-Ser473 is an important part of the mechanism that allows hormones to activate this protein kinase (Bayascas and Alessi, 2005; Cohen, 2006); therefore, we also explored the effects of insulin on the phosphorylation of PRAS40-Ser246, an endogenous PKB substrate (Kovacina et al., 2003). Analysis of the data derived from these experiments showed that insulin (20 nM, 30 min) did increase the abundance of Ser246-phosphorylated PRAS40 (Figure 2E,F) but also established that this response coincided with a small (∼20%) fall in the overall expression of PRAS40 (Figure 2E,G). It is therefore possible that the phosphorylation of PRAS40-Ser246 targets this protein for degradation. However, in the present context, the most significant result of this observation is that it implies that changes to the abundance of Ser246-phosphorylated PRAS40 will tend to underestimate the phosphorylation of this residue. We therefore further analyzed these data by normalizing the measured abundance of Ser246-phosphorylated PRAS40 to the corresponding values of overall abundance in order to obtain an indicator of PRAS40-Ser246 phosphorylation. This analysis (Figure 2G), which was used in all subsequent studies, showed that insulin stimulates PRAS40-Ser246 phosphorylation, indicating that it does activate PKB (Kovacina et al., 2003; Cohen, 2006).
Biophysical effects of PI3K inhibitors
Figure 3A shows the results of a series of experiments that used a strictly paired experimental design (see Methods) to explore the effects of wortmannin on the electrometric response to insulin. An examination of the control data indicated that Vt tended to depolarize slightly during the first 30 min of the experiment (initial: −57.0 ± 3.4 mV; 30 min: −50.2 ± 2.1 mV, P < 0.05) and, as Rt was stable (initial: 2.1 ± 0.2 kΩ·cm2; 30 min: 2.2 ± 0.2 kΩ·cm2), this led to an apparently spontaneous (∼15%) decline in IEq (Figure 3A). However, despite this effect, wortmannin consistently inhibited IEq and, after ∼30 min exposure to this substance, this current had decayed to 50.7 ± 2.7% (P < 0.001) of the corresponding control value (Figure 3A). Wortmannin had no effect on Rt over this initial period (initial: 2.0 ± 0.2 kΩ·cm2; 30 min: 2.2 ± 0.2 kΩ·cm2) and this suppression of basal current was therefore due to a depolarization of Vt (initial: −52.6 ± 4.9 mV; 30 min: −21.8 ± 2.4 mV, P < 0.001). While the control data confirmed that insulin normally enhances IEq (Figure 3A) by hyperpolarizing Vt with only a very small effect on Rt, insulin had no effect upon IEq in wortmannin-treated cells (Figure 3A). It is therefore clear that this inhibitor of PI3K abolished the electrometric response to insulin. However, analysis of the raw data recorded from wortmannin-treated cells showed that Vt and Rt declined substantially during exposure to insulin so that, after 60 min exposure to this hormone, these parameters had decayed to −3.6 ± 2.0 mV (P < 0.001) and 0.5 ± 0.2 kΩ·cm2 (P < 0.001) respectively. In contrast, Rt and Vt were normally stable, as the values measured in control cells that had been exposed to insulin for 60 min were 2.2 ± 0.1 kΩ·cm2 and −54.3 ± 5.3 mV respectively. At the end of the experiments all cells were exposed to apical amiloride (10 µM), normally this reduced IEq to −1.4 ± 0.1 µA·cm−2 (>95% inhibition, P < 0.001, Figure 3A) and increased Rt to 3.9 ± 0.6 kΩ·cm2 (P < 0.001). However, although amiloride abolished the small current that persisted in the presence of wortmannin (Figure 3A), this inhibition occurred with no increase in Rt (0.4 ± 0.2 kΩ·cm2). Wortmannin therefore appears to cause loss of epithelial integrity.
Figure 3B,C show data from experiments in which the same protocol was used to explore the effects of PI103 and GDC-0941. Once again, the control data confirm that hormone-deprived cells generate inwardly directed IEq and show that insulin normally increases the magnitude of this current (Figure 3B,C). An examination of the underlying data showed that this control response was due to a hyperpolarization of Vt that was associated with a small fall in Rt. Although PI103 and GDC-0941 did cause slight inhibition of IEq (Figure 3B,C), the currents measured after 30 min exposure to these compounds did not differ significantly from control (PI103: 94.3 ± 5.0% of control; GDC-0941: 94.6 ± 5.1% of control) and the corresponding values of Rt (PI103: 2.1 ± 0.04 kΩ·cm2; GDC-0941: 2.4 ± 0.2 kΩ·cm2) were also essentially identical to control (2.1 ± 0.2 kΩ·cm2 and 2.3 ± 0.2 kΩ·cm2 respectively). These compounds, in contrast to wortmannin, therefore had only very small effects on the electrical properties of hormone-deprived cells. Although insulin slightly increased the magnitude of IEq in cells that had been treated with PI103 (Figure 3B) or GDC-0941 (Figure 3C), the magnitude of these responses were <10% of control. Moreover, the values of Rt measured at the end of these experiments were also similar to control (not shown) and so, although PI103 and GDC-0941 have almost no effect upon the basal IEq, they inhibit (∼90%) the response to insulin without affecting epithelial integrity.
Biophysical effects of rapamycin
Although often referred to as PI3K inhibitors, wortmannin (Brunn et al., 1993) and PI103 (Raynaud et al., 2007) both inhibit TORC1 and we therefore also explored the effects of rapamycin (0.1 µM, 30 min), a selective inhibitor of this signalling complex (Bain et al., 2007), to ensure that effects on TORC1 did not underlie any of the effects reported here. Rapamycin had no effect upon the IEq in hormone-deprived cells and insulin-stimulated cells (Figure 3D) and also had no effect upon Vt and Rt (not shown).
Effects of PI3K-inhibitors on the phosphorylation of endogenous proteins
The data in Figure 4 confirm that insulin (20 nM, 30 min) normally evokes phosphorylation of PKB-Ser473, NDRG1-Thr346/356/366 and PRAS40-Ser246 and show that wortmanin, PI103 and GDC-0941 caused essentially complete dephosphorylation of these residues in both hormone-deprived and insulin-stimulated cells. Because the phosphorylation of this residue is clearly dependent upon PI3K (Bayascas and Alessi, 2005), this result shows that all three compounds essentially cause complete inhibition of this regulatory kinase.
Effects of rapamycin on the phosphorylation of endogenous proteins
Rapamycin did not alter the phosphorylation of PKB-Ser473, NDRG1-Thr346/356/366 and PRAS40-Ser246 in hormone-deprived or insulin-stimulated cells (Figure 4). Additional experiments explored the effects of rapamycin on the phosphorylation of a residue (P70-S6K- Thr389) that has been identified as a TORC1 substrate (Proud, 2007). These studies (n= 4) revealed basal phosphorylation of P70-S6K-Thr389 in hormone-deprived cells and, as anticipated, insulin (20 nM, 30 min) increased the abundance of the Thr389-phosphorylated S6K but had no effect on overall expression. Insulin therefore activates TORC1 in these cells (Proud, 2007). Rapamycin caused essentially complete dephosphorylation of P70-S6K-Thr389 in hormone-deprived and insulin-stimulated cells, indicating that this compound fully inactivates TORC1 (data not shown).
Electrometric effects of GSK650394A
Experiments in which hormone-deprived cells were acutely exposed to GSK650394A (1–10 µM, added to the apical and basolateral baths), an inhibitor of SGK1 (Sherk et al., 2008), showed that this compound had no significant effect upon the IEq when used at 1 µM and 3 µM (Figure 5A). However, when used at 10 µM, GSK650394A rapidly (2–3 min) reduced IEq to a value that was ∼60% of the initial, control value (P < 0.02, Figure 5A). However, this effect was transient because IEq subsequently recovered to a plateau value that was ∼70% (P < 0.05) of that recorded at the onset of the experiment (Figure 5A). Figure 5B–C shows the results of experiments that explored the effects of GSK650394A upon the electrometric response to insulin. These experiments were undertaken using a very strictly paired experimental design in order to ensure that variability between cells at different passage number did not confound data analysis. Each such experiment therefore involved simultaneously recording IEq from four confluent cultures so that we could monitor spontaneously developing and insulin-evoked changes in IEq in both control and GSK650394A-treated cells. Responses to insulin were quantified by subtracting the change in IEq measured in control cells from the change that developed during exposure to insulin and this, in turn, allowed the effects of GSK650394A to be quantified. Data obtained in this way confirm that insulin normally enhances IEq and, while this response did persist in the presence of 1 µM (Figure 5B) and 3 µM (Figure 5C) GSK650394A, this substance did cause some inhibition (1 µM: 44.5 ± 7.6% inhibition, P < 0.01; 3 µM: 54.8 ± 1.5% inhibition, P < 0.001). GSK650394A caused essentially complete block of this response (96.9 ± 5.6% inhibition, P < 0.02) at 10 µM (Figure 5D).
Effects of GSK650394A on the phosphorylation of endogenous proteins
GSK650394A (1–10 µM) had no effect on the overall expression of the NDRG1 protein (not shown) but caused a concentration-dependent decline in NDRG1-Thr346/356/366 phosphorylation in hormone-deprived and insulin-stimulated cells, and this effect was essentially complete at 10 µM (Figure 6A,D). GSK650394A (1–10 µM) also had no effect on the overall expression of PKB (not shown) and did not alter the abundance of Ser473-phosphorylated PKB in hormone-deprived cells (Figure 6B). However, GSK650394A did inhibit the insulin-induced phosphorylation of PKB-Ser473 at 3 µM, and essentially abolished this response at 10 µM (Figure 6B,E) and, because the phosphorylation of PKB-Ser473 is dependent upon PI3K (Bayascas and Alessi, 2005), this finding suggests that GSK650394A may prevent the insulin-induced activation of PI3K. However, despite this effect, GSK650394A (1–10 µM) did not alter the phosphorylation of PRAS40-Ser246 in hormone-deprived cells and did not prevent the insulin-induced phosphorylation of this residue (Figure 6C,F). It thus appears that 10 µM GSK650394A blocks signalling via SGK1 but not PKB (see also Sherk et al., 2008).
Electrometric effects of Akti-1/2
Acutely exposing cells to Akti-1/2 (1–10 µM, apical and basolateral) had no discernible effect upon IEq at 1 and 3 µM (Figure 7A). However, 10 µM Akti-1/2 caused a small (∼25%) but significant (P < 0.01) inhibition of the basal current that became apparent after a latency period of ∼2 min and developed over the following 10–15 min (Figure 7A). Figure 7B–D shows the results of experiments that explored the effects of Akti-1/2 (1–10 µM) on the insulin-induced augmentation of IEq using the method described previously. The control data confirm that insulin consistently increased the magnitude of this current and, although Akti-1/2 appeared to enhance the spontaneous decline in IEq when used at 3 µM and 10 µM, these effects were not statistically significant. Responses to insulin were clearly seen in cells exposed to 1 µM (Figure 7B) and 3 µM Akti-1/2 (Figure 7C), but these were smaller than control (1 µM: 43.4 ± 14.6% inhibition, P < 0.05; 3 µM: 31.3 ± 12.5% inhibition, P < 0.05) and this response was essentially abolished (80.8 ± 7.3% inhibition, P < 0.02) by 10 µM Akti-1/2 (Figure 7D).
Effects of Akti-1/2 on the phosphorylation of endogenous proteins
Exposing cells to Akti-1/2 had no effect on the overall expression of PKB or PRAS40 (not shown) but did cause a concentration-dependent decline in the abundance of Ser473-phopshorylated PKB (Figure 8A,D) and Ser246-phosphorylated PRAS40 (Figure 8B,E) in both hormone-deprived and insulin-stimulated cells. Indeed, these phosphoproteins were almost undetectable after exposure to 10 µM Akti-1/2, indicating essentially complete inactivation of PKB. Akti-1/2 also had no discernible effect upon the overall expression of NDRG1 (not shown) and the data in Figure 8C,F therefore show that Akti-1/2 caused concentration-dependent dephosphorylation of NDRG1-Thr346/356/366 in hormone-deprived and insulin-stimulated cells. Indeed, 10 µM Akti-1/2 essentially abolished the basal phosphorylation of NDRG1-Thr346/356/366 and the response to insulin (Figure 8C and F) and, because these residues are phosphorylated by SGK1 and not by PKB (Murray et al., 2005a,b), these data show that Akti-1/2 blocks signalling via both PKB and SGK1 under the present conditions.
Na+ transport in hormone-deprived cells
Hormone-deprived mpkCCD cells absorb Na+ from the apical bath via an apparently spontaneous mechanism dependent upon ENaC and it is therefore clear that Na+ absorption can occur independently of stimulating hormones, a finding that accords with data from several earlier studies of ASDN-derived cell lines (see e.g. Blazer-Yost et al., 1998; 2003; Record et al., 1998; Paunescu et al., 2000; Alvarez De La Rosa and Canessa, 2003). Analyses of extracted protein showed that hormone-deprived cells displayed basal phosphorylation of PKB-Ser473, establishing (Bayascas and Alessi, 2005) that PI3K is active under these conditions. It is therefore interesting that this regulatory kinase seems to contribute to the control of ENaC function. Early evidence of this came from the observation that LY294002, a PI3K inhibitor, blocks Na+absorption in hormone-deprived cells by progressively reducing the number of active Na+ channels in the apical membrane (Paunescu et al., 2000). Moreover, structural studies (Schild et al., 1996) showed that each ENaC subunit contains a C-terminal PY motif that provides binding sites for the neural precursor cell expressed, developmentally down-regulated protein 4-2 (Nedd-4/2). The binding of Nedd-4/2 (Kanelis et al., 2001) to these motifs targets the ENaC channel complex for internalization and degradation and such Nedd-4/2-mediated internalization/degradation of ENaC seems to limit the rate of Na+ absorption by restricting the Na+ conductance of the apical membrane (GNa) (Debonneville et al., 2001; Wang et al., 2001; Snyder et al., 2004). PI3K contributes to this mechanism by maintaining the catalytic activity of SGK1 (Kobayashi and Cohen, 1999; Park et al., 1999), a protein kinase that prevents the interaction between Nedd-4/2 and ENaC by phosphorylating Nedd-4/2 at Ser342 and Ser428 (Debonneville et al., 2001; Wang et al., 2001; Snyder et al., 2004). PI3K inhibitors such as LY294002 are therefore thought to inhibit Na+ transport by depriving the cell of SGK1 activity and thus facilitating the Nedd-4/2-mediated internalization/degradation of ENaC. However, although widely used, such small molecule kinase inhibitors almost always affect multiple targets and it is now clear that, as well as inactivating PI3K, LY294002 also inhibits TORC1, PLK1 (polo-like kinase 1), PIMK (provirus integration site for Moloney murine leukaemia virus kinase) 1 and 3, HIPK (homeodomain-interacting protein kinase) 2, GSK3 (glycogen synthase kinase 3) and CK2 (casein kinase 2) (Davies et al., 2000; Bain et al., 2007). LY294002 therefore displays poor selectivity for PI3K and its use to block signalling via this kinase is therefore no longer recommended (Bain et al., 2007). Indeed, the fact that LY294002 inhibits CK2 makes it particularly unsuitable for studies of Na+-absorbing epithelia as CK2 does appear to contribute to the control of ENaC activity (Shi et al., 2002; Bachhuber et al., 2008).
The present study therefore explored the effects of wortmannin, PI103 and GDC-0941 as these substances all appear to be relatively selective PI3K inhibitors (Bain et al., 2007; Folkes et al., 2008). The first such studies showed that wortmannin inhibited Na+ absorption in hormone-deprived cells and this is consistent with the idea that PI3K is essential for the maintenance of basal Na+ absorption (Paunescu et al., 2000). However, PI103 and GDC-0941 had very little effect on basal Na+ transport and yet an analysis of extracted protein showed very clearly that wortmannin, PI103 and GDC-0941 all caused essentially complete dephosphorylation of PKB-Ser473, NDRG1-Thr346/356/366 and PRAS40-Ser246. It is therefore clear that all three compounds do inactivate PI3K fully under the present conditions, and also block signalling via SGK1 and PKB, protein kinases that are important downstream targets of PI3K (Kobayashi and Cohen, 1999; Park et al., 1999; Bayascas and Alessi, 2005). The main effect of wortmannin which we now report cannot therefore be attributed to more effective inhibition of PI3K and it is therefore interesting that in vitro studies have shown that this compound can also inactivate PLK1 and smooth muscle myosin light chain kinase (SmMLCK). While we are unaware of any data implicating PLK1 in the control of epithelial Na+ absorption, it has been suggested that SmMLCK might contribute to the regulation of ENaC trafficking (Tokuda et al., 2002) and, because PI103 does not appear to act in this way (Bain et al., 2007), effects on SmMLCK might explain the effect of wortmannin which we now report.
Our data from cells treated with PI103 and GDC-0941 show that signalling via PI3K/SGK1 does not make a major contribution to Na+ absorption in hormone-deprived cells and this contrasts with earlier data (see e.g. Blazer-Yost et al., 1998; 2003; Record et al., 1998; Paunescu et al., 2000; Alvarez De La Rosa and Canessa, 2003). Data from recent studies of H441 human airway epithelial cells are interesting in this context. While these cells absorb Na+ via an ENaC-dependent mechanism, this absorptive phenotype is seen only in glucocorticoid stimulated cells (Clunes et al., 2004; Ramminger et al., 2004) and we have therefore used this cell type as a model system to explore the factors that allow steroid hormones to regulate GNa. It is abundantly clear that glucocorticoids do activate SGK1 in H441 cells and the fact that the physiological effects of glucocorticoid-stimulation are reproduced by transient expression of a constitutively active form of SGK1 suggest strongly that this kinase is involved in this response (Thomas et al., 2004; Brown et al., 2008; Inglis et al., 2009). However, our data also show that SGK1 is active in hormone-deprived cells, despite the fact that GNa is negligible and it is far from clear why this basal activity of SGK1 is not transduced into a Na+-absorbing phenotype (Inglis et al., 2009). Moreover, transient expression of a dominant negative SGK1 mutant suppresses the glucocorticoid-induced activation of the endogenous kinase without blocking the associated rise in GNa and these findings (Inglis et al., 2009), in common with the present data, suggest that signalling via PI3K/SGK1 is not critical to the control of ENaC function.
It is therefore interesting that steroid hormones induce the expression of a number of other proteins that seem to be involved in the hormonal control of GNa. For example, aldosterone evokes expression of the protein encoded by n-myc downstream regulated gene 2 (NDRG2) and this protein has been shown to increase the activity of ENaC expressed in Xenopus oocytes and Fisher rat thyroid cells (Wielpütz et al., 2007). Steroid hormones also induce expression of glucocorticoid-inducible leucine zipper proteins 1-3 (GILZ1-3) and a growing body of evidence implicates these proteins in the control of ENaC function (Soundararajan et al., 2005; 2007). Interestingly, recent work has suggested that GILZ1-3 might act in cooperation with SGK1 (Soundararajan et al., 2009) and established that transient expression of GILZ1 mimics the electrophysiological effects of glucocorticoid stimulation in H441 cells (Watt et al., 2009). It is therefore possible that these mechanisms might contribute to the maintenance of basal Na+ transport in hormone-deprived mpkCCD cells.
Insulin-induced Na+ transport
Insulin acutely stimulated Na+ absorption (see also Wilson et al., 2010) and induced phosphorylation of PKB-Ser473, NDRG1-Thr346/356/366 and PRAS40-Ser246, indicating that this hormone activates PI3K and also increases the activity of the downstream protein kinases SGK1 (see also Wilson et al., 2010) and PKB. While we have assumed that these responses to insulin are mediated via true insulin receptors, we cannot exclude the possibility that these effects may be mediated, at least in part, via receptors for insulin-like growth factor 1 (IGF-1), as the concentration of insulin used here (20 nM) may allow activation of these receptors (Gonzalez-Rodriguez et al., 2007). However, IGF-1 and insulin are thought to control Na+ transport via very similar mechanisms (Gonzalez-Rodriguez et al., 2007) and, while wortmannin, PI103 and GDC-0941 had differing effects upon basal IEq (see above), these compounds all caused essentially complete inhibition of insulin-induced Na+ absorption and abolished the insulin-induced phosphorylation of endogenous proteins. While signalling via PI3K/SGK1 does not seem to be important in the maintenance of basal Na+ absorption (see above), our data suggest strongly that this signalling pathway is critical to insulin-induced Na+ transport. This finding accords well with a number of earlier studies which indicate that insulin stimulates the trafficking of ENaC subunits to the apical membrane via a PI3K-dependent mechanism (see for example Blazer-Yost et al., 1998; 2003; Record et al., 1998; Wang et al., 2001; Arteaga and Canessa, 2006). Indeed, IGF-1 has recently been shown (Gonzalez-Rodriguez et al., 2007) to cause a PI3K-dependent increase in the phosphorylation/expression of SGK1 in mouse cortical collecting duct cells. However, this finding is based upon data obtained by probing Western blots with an antibody against total SGK1 and, under these conditions, changes to the phosphorylation status of this protein are inferred by the appearance of multiple, less mobile bands. While it is highly likely that this apparent phosphorylation SGK1 does lead to an increase in catalytic activity (Kobayashi and Cohen, 1999; Park et al., 1999), it is important to stress that such measurements do not provide any information relating to the catalytic activity of SGK1. In contrast, the present study assessed the activity of SGK1 by monitoring the phosphorylation of an endogenous, SGK1 substrate (NDRG1-Thr346/356/366) and this relatively new approach (Murray et al., 2005a; Inglis et al., 2009) allows us to show unequivocally that insulin-induced Na+ transport is associated with PI3K-dependent activation of SGK1.
Effects of rapamycin
As well as inhibiting PI3K, wortmannin and PI103 also block signalling via TORC1 (Brunn et al., 1993; Raynaud et al., 2007), a kinase activated by insulin that plays an important role in the control of cellular metabolism (Proud, 2007). Because it has been suggested that TORC1 might contribute to the control of SGK1 activity by phosphorylating SGK1-Ser422 (Hong et al., 2008), we also explored the effects of rapamycin, an extremely selective TORC1 inhibitor (Bain et al., 2007). Our data show clearly that rapamycin did not alter the currents generated by hormone-deprived cells; did not modify the electrometric response to insulin, and had no effect upon cellular PI3K, SGK1 and PKB activity in hormone-deprived and insulin-stimulated cells. However, a separate series of experiments confirmed that insulin did evoke P70-S6K-Thr389 phosphorylation, indicating that this hormone does activate TORC1 (Proud, 2007). Moreover, rapamycin caused complete dephosphorylation of P70-S6K-Thr389 in hormone-deprived and insulin-stimulated cells, indicating that this substance fully inactivates TORC1. Our data, in contrast with those presented by Hong et al. (2008), therefore provide no evidence to support the idea that TORC1 is involved in the control of SGK1 activity and it is therefore interesting that recently published data suggest that the apparent rapamycin-sensitive phosphorylation of SGK1-Ser422 reported by Hong et al. (2008) was actually an artefact caused by the inappropriate use of poorly selective antibodies (García-Martínez and Alessi, 2008).
Physiological basis of insulin-induced Na+ absorption
GSK650394A had a relatively modest effect (∼30% inhibition) on Na+ transport in hormone-deprived cells and caused concentration-dependent inhibition of the electrometric response to insulin with essentially complete block at 10 µM. Analyses of extracted proteins showed that GSK650394A caused dephosphorylation of NDRG1-Thr346/356/366 in both hormone-deprived and insulin-stimulated cells and this effect was also essentially complete at 10 µM. The highest concentrations of GSK650394A tested did seem to cause some inhibition of insulin-induced PKB-Ser473 phosphorylation, which raised the possibility that GSK650394A may also cause some inhibition of PI3K. However, GSK650394A had no effect on the phosphorylation of PRAS40-Ser246, even at 10 µM, and it is therefore clear that this substance does not block the insulin-induced phosphorylation of PKB substrates. While this may appear surprising in view of the inhibition of PKB-Ser473 phosphorylation (see above), Logie et al. (2007) have shown that there is considerable extra capacity in the PKB-dependent signalling pathway. Indeed, their studies of liver cells indicated that endogenous PKB activity must be reduced to <10% of the basal level before the phosphorylation of downstream targets is compromised (Logie et al., 2007). The present data therefore confirm (Sherk et al., 2008) that GSK650394A suppresses signalling via SGK1 and not via PKB. The finding that this compound fully suppressed the electrometric response to insulin therefore supports the view that this response is mediated via PI3K/SGK1 (Alvarez De La Rosa and Canessa, 2003; Arteaga and Canessa, 2006) rather than via PI3K/PKB (Lee et al., 2007).
Akti-1/2 also caused concentration-dependent inhibition of the electrometric response to insulin and this effect, in common with the effect of GSK650394A, was essentially complete at 10 µM. Akti-1/2 also caused essentially complete dephosphorylation of PKB-Ser473 and PRAS40-Ser246 in both hormone-deprived and insulin-stimulated cells, which confirms that this compound is an effective inhibitor of PKB (Barnett et al., 2005; Lindsley et al., 2005). However, our data suggest that 10 µM Akti-1/2 is needed for complete inhibition of PKB and this contrasts with data from liver cells, which reported full inhibition at concentrations below 1 µM (Logie et al., 2007). Our findings do, however, accord with earlier work, which suggests that concentrations as high as 20 µM are needed to inhibit PKB fully (Barnett et al., 2005). The present data show that Akti-1/2 also causes dephosphorylation of NDRG1-Thr346/356/366 and this effect, in common with the dephosphorylation of PRAS40-Ser246, was complete at 10 µM. As the phosphorylation of NDRG1-Thr346/356/366 is strictly dependent upon SGK1 (Murray et al., 2005a,b), this result shows that Akti-1/2 inhibits both PKB and SGK1 under the present conditions and it is therefore interesting that 10 µM Akti-1/2 has been shown to cause substantial (∼75%) inhibition of SGK1 in vitro (Logie et al., 2007). Hence, our data indicate that Akti-1/2 is not a selective PKB blocker, and this result highlights the difficulties inherent to all experiments based around such small molecule inhibitors of protein kinases (Bain et al., 2007). All data obtained using Akti-1/2 must therefore be treated with caution and, at present, it is impossible for us to exclude the possibility that PKB may contribute to the control of Na+ transport by acting in concert with SGK1 (Lee et al., 2007).
Significance of the present findings
The present data suggest that signalling via PI3K/SGK1 is not important to the maintenance of basal Na+ transport, because hormone-deprived cells continued to absorb Na+ when PI3K had been fully inactivated using PI103 or GDC-0941. It is therefore understandable that deletion of the sgk1 gene has no overt effect upon renal Na+ handling in animals fed a normal diet (Wulff et al., 2002) and this finding, in common with the present data, suggests strongly that SGK1 is not involved in basal Na+ transport. However, signalling via PI3K does seem to be critical for insulin-stimulated Na+ transport and the finding that GSK650394A abolished the insulin-induced Na+ transport suggests strongly that this response is mediated via SGK1 (see also Record et al., 1998; Wang et al., 2001; Arteaga et al., 2006). Interestingly, sgk1 gene deletion does block the insulin-induced reduction in urinary Na+ excretion (Huang et al., 2006) and it is therefore possible that drugs such as GSK650394A, which selectively inactivate SGK1, might become useful in the treatment of fluid retention/oedema that can complicate the management of type 2 diabetes.
This work was supported by an MRC Project grant (SMW) and by a studentship from the MRC doctoral training scheme. The authors are grateful to Professors Dario Alessi and Phillip Cohen for the provision of many reagents and for valuable advice concerning their use.