Modulation of the Inward Rectifier Potassium Channel IRK1 by the Ras Signaling Pathway*

In this study, we investigated the role of Ras and the mitogen-activated protein kinase (MAPK) pathway in the modulation of the inward rectifier potassium channel IRK1. We show that although expression of IRK1 in HEK 293 cells leads to the appearance of a potassium current with strong inward rectifying properties, coexpression of the constitutively active form of Ras (Ras-L61) results in a significant reduction of the mean current density without altering the biophysical properties of the channel. The inhibitory effect of Ras-L61 is not due to a decreased expression of IRK1 since Northern analysis indicates that IRK1 mRNA level is not affected by Ras-L61 co-expression. Moreover, the inhibition can be relieved by treatment with the mitogen-activated protein kinase/ERK kinase (MEK) inhibitor PD98059. Confocal microscopy analysis of cells transfected with the fusion construct green fluorescent protein-IRK1 shows that the channel is mainly localized at the plasma membrane. Coexpression of Ras-L61 delocalizes fluorescence to the cytoplasm, whereas treatment with PD98059 partially restores the membrane localization. In conclusion, our data indicate that the Ras-MAPK pathway modulates IRK1 current by affecting the subcellular localization of the channel. This suggests a role for Ras signaling in regulating the intracellular trafficking of this channel.

Inwardly rectifying potassium channels play a key role in stabilizing resting membrane potential in both excitable and non-excitable cells. IRK1/Kir 2.1 is a member of this family, showing strong inward rectification properties. It is expressed in a wide variety of tissues and cell types including neurons of the central and peripheral nervous system, glia, muscle, and immune system cells. Phosphorylation of IRK1 protein at both serine/threonine and tyrosine sites modulates its activity. The channel is a substrate of protein kinase A and protein kinase C, and direct activation of these kinases modulates the current (1,2). In vivo, a reduction of IRK1 conductance has been demon-strated after activation of muscarinic (2) and tyrosine kinase receptors (3). Muscarinic m1 receptors modulate IRK1 probably through protein kinase C, and the small GTPase Rho has been implicated in this effect (4). The activation of nerve growth factor receptors leads to tyrosine phosphorylation of IRK1 and to its endocytosis, although it is not yet clear which kinase is involved (5).
In this work, we investigated the role of Ras and of the downstream MAPK 1 pathway on the modulation of IRK1 current. To this purpose and to avoid receptor-mediated effects, we transfected the active form of Ras (Ras-L61) in HEK 293 cells together with the IRK1 channel. We found that activated Ras decreases IRK1 current without modifying the channel properties and that it does so acting through the MAPK kinase pathway. This effect seems to be due to a reduction of channel density at the cell surface, thus suggesting the involvement of the Ras-MAPK pathway in the regulation of IRK1 localization.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-HEK 293 cells and NIH 3T3 cells were grown in Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% heat-inactivated fetal calf serum, 2 mM glutamine, 100 units/ml penicillin, 100 g/ml streptomycin and kept in a 5% CO 2 humidified atmosphere at 37°C. Cells were plated in 35-mm dishes (1.4 ϫ 10 5 cells per dish) and transfected with 2 g of total DNA per dish using LipofectAMINE Plus (Invitrogen). In cotransfection experiments, the different constructs were always used in a ratio of 1:3 for IRK1 and Ras, respectively . An empty vector was used to normalize the total amount of plasmid transfected. Cells were used 48 h after the transfection in all the experiments.
Constructs-The mouse IRK1 (mIRK1) cDNA, originated from a mouse macrophage cell line (6), has been kindly provided by Dr. L. Yan. A deleted version of mIRK1 missing all the 3Ј non-coding region of IRK1, obtained by digestion with BSTX1 and insertion of the 1.7-kb BSTX1 fragment into a pcDNAI vector (3Ј⌬ϪmIRK1), was used as a probe for Northern analysis. The enhanced green fluorescent proteinhuman IRK1 (EGFP-hIRK1) fused construct, EGFP being at the N terminus of hIRK1, was kindly provided by Dr. D. C. Johns (7).
The constitutive active form of Ha-Ras (Ras-L61) fused to a Myc epitope was obtained by PCR modification of 5Ј and 3Ј ends of Ras in pRSV-Ras-L61 in order to create BamHI and EcoRI sites. The fragment was cloned into pBS-Nmyc2 and sequenced, and after excision of Myc-Ras with SalI, it was subcloned into pcDNAIII linearized with XhoI. EGF receptor (EGFR) fused in C-terminal to GFP was kindly provided by Jovin (8).
Electrophysiology-Classical patch clamp methodology in whole cell configuration was used; currents were amplified by an Axopatch-1D amplifier (Axon Instruments, Foster City, CA). We carefully compen-sated pipette capacity, cell capacity, and series resistance before each voltage clamp protocol. Currents were not leakage-subtracted before acquisition. Experimental protocols, data acquisition, and analysis were done using pCLAMP 7 (Axon Instruments) and Origin (Microcal, Northampton, MA) software. All experiments were performed at room temperature, and current traces were filtered at 5 kHz. Currents were measured with pipettes having 3-4 megohm resistance filled with (in mM): 135 sodium aspartate, 0.2 CaCl 2 , 1.6 MgCl 2 , 10 HEPES-KOH, 2 EGTA; the pH was set at 7.35.
The extracellular control solution contained (in mM): 135 NaCl, 4 KCl, 1 MgCl 2 , 2 CaCl 2 , 6 glucose, 10 HEPES-NaOH at pH 7.35. In the test solutions, 40 mM KCl replaced an equivalent amount of NaCl. Cells were normally kept in the control solution, whereas test solutions were superfused by gravity at close proximity of the cell by a rapid solution changer (Warner Instrument Corp., Hamden, CT). Cells transfected with EGFP-hIRK1 displayed much higher currents with respect to those transfected with mIRK1; the control solution, containing 4 mM KCl, was therefore perfused in this case to ensure the correct operation of the patch clamp amplifier (9).
When specified, cells were incubated for 2 h in a 5% CO 2 saturated atmosphere at 37°C with 40 M PD98059 (2Ј-amino-3Ј-methoxyflavone) (Calbiochem-Novabiochem), a selective inhibitor of MEK (10). Leakage, obtained by perfusing the cells with the respective solutions containing 100 M BaCl 2 , was subtracted during off-line analysis, and the resulting currents were normalized to cell capacity.
For electrophysiological experiments in which mIRK1 was used, plasmid containing EGFP was always cotransfected in order to check the efficiency of transfection by visual observation with a fluorescence microscope. Almost 60% of the transfected cells resulted in fluorescence, and 90% of the fluorescent cells displayed the potassium current. For patch clamp experiments, bright fluorescent cells of medium size were chosen. All electrophysiological experiments were done 48 h after transfection at room temperature. All mean values were calculated from not less than three different transfections for each combination of constructs transfected (see figure legends for details). The electrophysiological data, where indicated, were statistically analyzed applying a two-population (independent) Student's t test with significance intervals as specified in the figure legends. Western blots using antibodies against the Myc epitope-tagged protein were performed to check for Ras-L61 expression Northern Blot Analysis-For Northern analysis, HEK 293 and NIH 3T3 cells plated in 100-mm dishes were transfected with 6 g of mIRK1 and 18 g of Myc-Ras-L61 or with 18 g of pCDNA1 vector. 48 h after transfection, total RNA was extracted (11) and subjected to DNase treatment according to published protocol (12). The RNA (20 g for sample) was separated by electrophoresis under denaturing conditions, blotted on a nylon membrane (Amersham Biosciences, Inc.) in 10ϫ SSC overnight, and fixed. Hybridization was performed at 68°C. The DNA template used was the 3Ј⌬ϪmIRK1 digested with HindIII. Riboprobe was synthesized using SP6 RNA polymerase (Roche Molecular Biochemicals) and [␣-32 P]UTP at 40°C for 45 min. After hybridization, the filter was washed at 68°C in 5ϫ SSC for 10 min, 2ϫ SSC for 40 min, 1ϫ SSC for 40 min, and twice in 1ϫ SSC ϩ 0.1% SDS for 45 min. The signal was detected by autoradiography.
Western Blot Analysis-HEK 293 cells plated in 35-mm dishes were transfected with EGFP-hIRK1 and a control plasmid or with EGFP-hIRK1 and Ras-L61 (maintaining the ratio of 1:3 between IRK1 and Ras-L61/control plasmid cDNAs). 48 h after transfection, cells were treated or not with PD98059 for 3 h, and total proteins were extracted in SDS sample buffer. Equal amounts of total proteins (40 g) were separated by SDS-PAGE. Immunoblot analysis was carried out with antibodies raised against hIRK1; polyclonal antibodies (Alomone Laboratories, Jerusalem, Israel); MAPK (Santa Cruz Biotechnology, Santa Cruz, CA); phosphoMAPK (Cell Signaling Technology, Beverly, MA); or Myc (9E10, Berkeley Antibody, Richmond, CA). Bound antibodies were visualized with horseradish peroxidase-conjugated anti-rabbit or antimouse antibodies using the ECL detection system (Amersham Biosciences, Inc.).
Cell Surface Biotinylation-HEK 293 cells plated in 60-mm dishes were transfected with EGFP-hIRK1 with or without Ras-L61 (in a ratio of 1:3). 48 h after transfection, cells were biotinylated using 0.5 mg/ml sulfo-NHS-biotin (Pierce) in phosphate-buffered saline on ice. Cells were lysed in 1% Triton X-100, 0.5% deoxycholate in phosphate-buffered saline, and biotinylated proteins were absorbed using immobilized streptavidin (Sigma). Absorbed material was collected and analyzed by Western blot using either hIRK1 or EGFR (Santa Cruz Biotechnology) antibodies.
Confocal Microscopy-HEK 293 cells plated on glass coverslips were transfected with EGFP-hIRK1 with or without Ras-L61. A parallel set of plates was transfected with EGFR-GFP construct in the presence or not of Ras-L61, maintaining in all cases the ratio of 1:3 as described above. Confocal fluorescence images were obtained from living cells. A Bio-Rad MRC 1024 confocal system (Bio-Rad Laboratories) equipped with a krypton-argon laser and mounted on an upright Zeiss Axiovert microscope (Zeiss, Oberkochen, Germany) was used to acquire the images through a 63X Plan Neofluar oil immersion objective; a ϫ1-ϫ3 zoom factor was applied to the scanned images during acquisition.

Ras-L61 Reduces the Current Density of Ectopically
Expressed mIRK1-The expression of the mIRK1 channel in HEK 293 cells allows the detection of a typical inward rectifying current that is activated at potentials below the potassium equilibrium potential (Fig. 1B) (in our case E K ϭ Ϫ32 mV was calculated from the Nernst equation considering the experimental conditions) and that is completely blocked by 100 M extracellular Ba 2ϩ (Fig. 1C) (6). Cells transfected with a control plasmid did not display either macroscopic endogenous currents or the Ba 2ϩ -sensitive inward current (Fig. 1A).
Cotransfection with Ras-L61, a constitutively active form of Ras, led to a significant reduction of the mean current density (by 43.3% at V test ϭ Ϫ100 mV; p Ͻ 0.01); current kinetics and voltage dependence were not affected (Fig. 2). This effect was specific for the active form of Ras since cotransfection with the same amount of plasmid carrying the cDNA encoding wild type Ras did not modify the current density and its kinetics (data not shown).
To distinguish whether the effect of Ras-L61 on mIRK1 channels depended on Ras itself or on the activation of the MAPK pathway, transfected cells were preincubated with PD98059, a specific inhibitor of MEK, for 2 h before the electrophysiological determinations. As shown in Fig. 2, PD98059 completely reversed the inhibitory effect of Ras-L61 at all membrane potentials tested, although it did not affect the current in cells transfected with the channel alone. This observation suggests that the modulation of the IRK1 current by activated Ras is mediated by the Ras-MAPK cascade. mIRK1 mRNA Level Is Not Affected by Coexpression of Ras-L61-We then verified whether the decrease in current density observed in the presence of Ras-L61 was due to a modification of mIRK1 expression. Cells were transfected with mIRK1 without or with Ras-L61 maintaining at 1:3 the ratio between mIRK1 and Ras-L61. Cells were then treated or not with PD98059 for 3 h. Northern blot analyses for mIRK1 carried out on total RNA are reported in Fig. 3. The expected length of the messenger RNA for mIRK1 is 5.4 kb since the cDNA for mIRK1 includes, in addition to the 1.7-kb open reading frame, a 3Ј Interestingly, when the same construct was transfected in NIH 3T3 fibroblasts (of mouse origin), three different transcripts were detected, and the 5.4-kb band was also highly represented. The finding obtained in NIH 3T3 fibroblasts confirms that, as observed in HEK 293 cells, transfected mIRK1 cDNA gives rise, in addition to the 5.4-kb band, also to shorter RNA products possibly due to alternative polyadenylation and/or early termination of the transcription.
Effects of Ras-L61 on the Human IRK1 Channel-To further analyze the modulatory effect of the Ras/MAPK pathway on IRK1 and to study the cellular distribution of the channel, we utilized an EGFP-hIRK1 construct (7) coding for the green fluorescent protein fused in its C-terminal to the human Kir2.1 channel. Electrophysiological measurements (Fig. 4A) allowed the detection of a Ba 2ϩ -sensitive current, which showed the typical inward rectification properties as already described (13). The kinetics of the whole cell currents exhibited by the human isoform fused to GFP were very similar to those produced by the mouse isoform (compare Figs. 2A and 4A). The differences in shape of the I/V curves are probably due to intrinsic properties of the two channel isoforms; the human channel has a characteristic outward component (13)   As already reported for mIRK1, also for human IRK1 construct, cotransfection with Ras-L61 largely reduced, by 50%, the mean current density (see Table I). A 2-h pretreatment with PD98059 reverted the current density to values close to that observed in cells transfected with EGFP-hIRK1 alone. On the other hand, PD98059 had no significant effect on EGFP-hIRK1-transfected cells. To investigate whether Ras modifies the level of IRK1 protein, equal amounts of protein from cells transfected with EGFP-hIRK1, with or without Ras-L61 and treated or not with PD98059, were analyzed by Western blotting with anti-hIRK1 antibodies. Fig. 5 shows that similar levels of EGFP-hIRK1 protein were present in all conditions. As an internal standard, the expression of the endogenous protein ERK2 was analyzed, confirming comparable loading in all lanes. Immunoblot analysis using anti-Myc antibodies detected a protein with the expected 21 kDa molecular size only in cells transfected with the Ras-L61 construct. The activation state of ERKs was also investigated: antibodies against the phosphorylated form of ERK1 and ERK2 showed that Ras-L61 causes a strong activation of MAPKs and that treatment with the MEK inhibitor completely blocks this activation.
These data clearly indicate that RasL61, which activates ERKs, does not alter the expression of the EGFP-hIRK1 protein. Moreover, a 3-h pretreatment with PD98059 has no effect on the level of EGFP-hIRK1.
Effects of Ras-L61 on the Cellular Localization of EGFP-hIRK1-We then analyzed the subcellular localization of the GFP construct by confocal microscopy. In cells transfected with EGFP-hIRK1 alone, fluorescence was mainly localized at the plasma membrane with rare cytoplasmic spots probably related to a Golgi complex localization (Fig. 6A).
Conversely, cotransfection with Ras-L61 led to a redistribution of the fluorescence signal to the whole cytoplasm (Fig. 6B). In addition, PD98059 reverted this effect since the fluorescence signal returned mainly at the level of the plasma membrane (Fig. 6D). PD98059 treatment of cells transfected with the channel alone (Fig. 6C) did not lead to any change in the fluorescence distribution pattern.
To further verify the specific effect of the Ras-MAPK pathway on IRK1 trafficking, cells were transfected with a construct coding for the EGF receptor (EGFR) fused in C-terminal to GFP (8) that is able to undergo ligand-induced endocytosis (14). As shown in Fig. 6, E and F, coexpression of Ras-L61 did not alter the localization of the fluorescence signal exhibited by EGFR-EGFP, which remained mainly localized at the plasma membrane level.
To confirm the data reported above, we compared the level of surface-associated IRK1 channels and endogenously expressed EGFR under different conditions. Cells were transfected with hIRK1 with or without Ras-L61, and 48 h later, cell surface proteins were biotinylated with sulfo-NHS-biotin. Biotinylated proteins were affinity-purified with immobilized streptavidin and analyzed. Immunoblotting with anti-hIRK1 antibodies shows that the amount of biotinylated (cell surface-associated) hIRK1 is higher in cells transfected with EGFP-hIRK1 alone than in cells cotransfected with Ras-L61. Conversely, no reduction of cell surface-associated EGFR could be detected in the presence of Ras. An equivalent amount of the two proteins was detected in total cell extracts (Fig. 7). These data further support a selective role for the Ras-MAPK pathway on the IRK1 channel without generally affecting membrane trafficking. Ϫ53.2 a Ϯ8.139 EGFP-hIRK1 ϩ Ras-L61/PD98059 Ϫ87 Ϯ13.235 a the mean that is statistically significant from the others (p Ͻ 0.01). Interestingly, our results show that activation of MAPK does not affect endocytosis of the EGFR. We have not analyzed whether MAPK activity alters ligand-induced EGF receptor internalization. DISCUSSION We investigated the role of the Ras-MAPK pathway in regulating IRK1/Kir 2.1 channel by coexpressing in HEK 293 cells Kir 2.1 (either the mouse isoform or the human one fused to GFP) and constitutively active Ras. The expression of the active form of Ras (Ras-L61) reduces IRK1 current, identified as a Ba 2ϩ -sensitive potassium component, without affecting the kinetic properties of the channels. Inhibition of the MAPK cascade by PD98059 restores the level of IRK1 current, suggesting that Ras modulates the current through the downstream pathway.
Our data rule out an inhibitory effect of Ras-L61 on IRK1 at the transcriptional level since mIRK1 mRNA level is not affected by Ras-L61 expression. Even though a 5.4-kb cDNA has been used and endogenous IRK1 has been reported to give rise to a 5.4-kb message (6), we found in HEK 293 cells a predominant 1.7-kb message and a fainter one at 5.4 kb. Moreover, when the same plasmid was transfected in mouse NIH 3T3 fibroblasts, three different transcripts were detected, one of which is 5.4 kb. These results suggest that, depending on the cell type, alternative polyadenylation and/or early termination of the transcription can occur.
To study the effect of Ras-L61 on the level of channel expression and its cellular localization, we used the EGFP-hIRK1 fusion protein. We could demonstrate that neither expression of Ras-L61 nor PD98059 treatment alters the level of the EGFP-hIRK1 protein.
Confocal microscopy shows that although EGFP-hIRK1, when expressed alone, is mainly localized at the cell surface, in the presence of activated Ras, it becomes diffusely distributed in the cytoplasm. This situation can be partially reversed by a 2-h treatment with PD98059. This inhibitor acts mainly on channel redistribution rather than on the de novo synthesis of IRK1 since treatment with cycloheximide together with PD98059 did not significantly alter the fluorescence signal at the plasma membrane (data not shown). These findings, together with the reduced level of biotinylated IRK1 protein found in Ras-L61-transfected cells, indicate that in HEK 293 cells, the Ras-MAPK pathway reduces the channel molecules present on the plasma membrane.
The redistribution of IRK1 does not seem to be due to a general effect of active Ras on membrane trafficking: in fact, we did not detect any significant reduction of an endogenous surface-associated protein (EGFR). Moreover, cotransfection of Ras with EGFR-EGFP (a chimeric construct in which the GFP moiety does not affect the EGFR functions (10)) did not alter its fluorescence distribution.
Our experimental protocols do not allow a determination of how much of the membrane is internalized since capacitance measurements have been done in a steady state condition. However, the determined mean cell capacitance did not show significant differences in the different conditions (data not shown). Thus, the Ras-MAPK pathway appears to act on the trafficking of IRK1 molecules and does not have a generalized effect on membrane proteins.
One of the mechanisms employed by cells to regulate the activity of ion channels is to modulate their localization (15)(16)(17). In fact, cell surface expression is the result of the balance of insertion of de novo synthesized proteins, of internalization through endocytosis, and of recycling (18).
Our results suggest that a MAPK-dependent phosphorylation event rather than protein synthesis is required to reduce the level of cell surface-associated IRK1. We also show that a 2-h treatment with PD 98059 (which completely blocks MAPK activation) is sufficient to inhibit RasL61-mediated effects. This suggests that MAPK activity induces a rapid cycling of the IRK1 channel, although we cannot define whether phosphorylation accelerates internalization or whether it reduces the rate of the surface expression of IRK1 channels. In addition, we do not know whether IRK1 itself is the substrate of this phosphorylation or whether other proteins are involved in this process. However, it is worthwhile to recall that a consensus sequence for MAPKs is present in the C-terminal region of IRK1 that might be important for channel trafficking. Mechanisms for endo-and exocytosis of this channel are yet poorly understood. This point has been studied, for instance, by acute exposure to either extracellular signaling molecules or specific tyrosine phosphatase inhibitors (3). To this regard, Tong et al. (5) have shown that IRK1 endocytosis requires Tyr-242, part of a motif recognized by clathrin adaptor proteins, suggesting that internalization is mediated by clathrin. In addition, recent findings have revealed that forward trafficking (19) of the channel, which is dependent on a C-terminal sequence of IRK1, also contributes to the regulation of the number of surface-associated channel. Further studies are required to fully elucidate the mechanisms of IRK1 channel trafficking.
Modulation of the availability of cell surface ion channels and channel trafficking may be particularly suitable to shape the electrophysiological response (7) and may represent one of the events that control the basal electrical activity of the cell, an issue that seems important for the cell fate, proliferative versus differentiative (20,21). We have previously shown that expression of CDC25 Mm /Ras-GRF1 in the SK-N-BE neuroblastoma, induced to differentiate with retinoic acid, led to a significant increase in the number of cells showing the IRK1-like current (12). The opposite modulation reported here, exerted by the Ras-MAPK pathway on the IRK1 channel, may depend on the cellular system used. The two model lines may differ in the inventory of expressed protein or in the cross-talk of signaling pathways. The occurrence of opposite effects in different systems under the same stimulus is not a new issue. For example, protein kinase A can exert a positive or negative effect on Kir2.1 if the channel is expressed in Xenopus oocytes or in COS7 cells, respectively (1,22).
Ras and MAPK pathway play a pivotal role in cell prolifer- FIG. 7. Ras-L61 expression reduces the level of surface-biotinylated EGFP-hIRK1. HEK 293 cells were transfected with EGFP-hIRK1 with or without Ras-L61 (transfected vector). Cell surface proteins were biotinylated with sulfo-NHS-biotin in phosphate-buffered saline for 1.5 h. Cells were lysed as described under "Experimental Procedures." An equal amount of proteins was either directly analyzed (total protein extract) or absorbed to immobilized streptavidin, and the absorbed material was collected (surface-biotinylated protein) and analyzed by Western blotting (Wb) with antibodies against hIRK1 or EGFR. ation, survival, and differentiation, acting on many different target proteins. Modulation of calcium, potassium, and sodium channels by the Ras signaling pathway has already been demonstrated (23)(24)(25). Moreover, it has been shown that in oligodendrocytes, inhibition of the inward rectifying potassium current, induced by ceramide, is mediated by a Ras-and Raf-1dependent pathway (26). With our present data, we suggest that the Ras-MAPK cascade modulates the inward rectifying potassium channel by reducing the cell surface channel availability. This may be relevant for the ion channel function in the context of its contribution to cell growth activity in non-excitable cell or to the electrical activity in excitable cells.