Cardiac sarcolemmal ATP‐sensitive K+ (KATP) channels, composed of Kir6.2 and SUR2A subunits, are regulated by intracellular ATP and they couple the metabolic status of the cell with the membrane excitability. On the basis of previous studies, we have suggested that glyceraldehyde 3‐phosphate dehydrogenase (GAPDH) may be a part of the sarcolemmal KATP‐channel protein complex. A polypeptide of ∼42 kDa was immunoprecipitated with an anti‐SUR2A antibody from guinea‐pig cardiac membrane fraction and identified as GAPDH. Immunoprecipitation/western blotting analysis with anti‐Kir6.2, anti‐SUR2A and anti‐GAPDH antibodies showed that GAPDH is a part of the sarcolemmal KATP‐channel protein complex in vivo. Further studies with immunoprecipitation/western blotting and the membrane yeast two‐hybrid system showed that GAPDH associates physically with the Kir6.2 but not the SUR2A subunit. Patch‐clamp electrophysiology showed that GAPDH regulates KATP‐channel activity irrespective of high intracellular ATP, by producing 1,3‐bisphosphoglycerate, a KATP‐channel opener. These results suggest that GAPDH is an integral part of the sarcolemmal KATP‐channel protein complex, where it couples glycolysis with the KATP‐channel activity.
ATP‐sensitive K+ (KATP) channels are gated by intracellular ATP and are viewed as a link between cellular metabolism and membrane excitability. In the heart, the opening of these channels has been suggested to protect against myocardial infarction, mediate ischaemic preconditioning and promote survival of cardiomyocytes that are exposed to different kinds of metabolic stresses (Crawford et al, 2003; Budas et al, 2004).
Structurally, the cardiac subtype of KATP channels are heteromultimers composed of Kir6.2 subunit, an inwardly rectifying K+‐channel core primarily responsible for K+ permeance, and SUR2A, a regulatory subunit implicated in ligand‐dependent regulation of the channel gating (Inagaki et al, 1996). A full understanding of how KATP channels sense the metabolic conditions of the cell is still under investigation. From reports of channel regulation until now, the intracellular ATP/ADP ratio would seem to be the most important factor in the regulation of sarcolemmal KATP channels, where ATP and ADP act as endogenous blockers and openers of the channels (reviewed by Dzeja & Terzic, 1998; Jovanović & Jovanović, 2004). In this regard, it has been suggested that there is a close functional relationship between adenylate kinase and creatine kinase phosphotransfer defining the directionality of nucleotide exchange within the sarcolemmal KATP‐channel vicinity. It has been proposed that activation of KATP channels by adenylate kinase (by AMP+ATP↔2ADP) may be counteracted by creatine kinase (by ADP+phosphocreatine↔ATP+creatine), as a system for ATP production and scavenger of the products of adenylate kinase catalysis, thereby keeping KATP channels closed (Carrasco et al, 2001). It has been shown that both adenylate kinase and creatine kinase are physically associated with sarcolemmal KATP channels, where they can synchronously act as anchoring sites for the cellular phosphotransfer network (Carrasco et al, 2001; Crawford et al, 2002a). More recently, it has been shown that, besides ATP and ADP, 1,3‐bisphosphoglycerate, a product of reaction catalysed by the glycolytic enzyme glyceraldehyde 3‐phosphate dehydrogenase (GAPDH), also regulates the activity of KATP channels. It has been suggested that regulation of 1,3‐bisphosphoglycerate could provide a direct link between glycolysis and membrane excitability in the heart independently of ATP (Jovanović & Jovanović, 2005). As 1,3‐bisphosphoglycerate is short‐lived in vivo and abruptly converted into 3‐phosphoglycerate, a compound without direct effects on KATP channels (S.J. & A.J., unpublished results), it is likely that a close spatial relationship between GAPDH and KATP‐channel subunits is required if 1,3‐bisphosphoglycerate‐mediated gating of KATP channels is to be of any physiological importance. A close proximity of two proteins in the intracellular in vivo environment probably requires a physical interaction between these proteins. Consequently, this prompted us to test the hypothesis that GAPDH and KATP‐channel subunits in the heart interact physically with each other and to examine whether this interaction could have a role in the regulation of channel behaviour.
Results and Discussion
Co‐immunoprecipitation assays, if executed carefully and with the required precautions, have the potential to provide strong evidence about physical interaction between proteins (Harlow & Lane, 1999). Here, we have used an antibody raised against the SUR2A KATP‐channel epitope to immunoprecipitate a cardiac membrane fraction. This antibody and strategy were described in our previous reports and found suitable to identify proteins forming sarcolemmal KATP channels (Ranki et al, 2001; Crawford et al, 2002a, 2002b). Coomassie blue staining of the anti‐SUR2A immunoprecipitate showed several proteins, most of which had been previously identified, including lactate dehydrogenase (LDH), Kir6.2, α‐actin, creatine kinase, β‐myosin and SUR2A (Fig 1A; Crawford et al, 2002a, 2002b). One of the proteins that was clearly visible on the gel, but was not identified in previous studies, was p42 (protein of an estimated size of 42 kDa). The appearance of these polypeptide bands was blocked by incubation with the corresponding antigenic peptide, and these proteins did not precipitate if the membrane fraction was probed with a non‐KATP‐channel antibody (Crawford et al, 2002b). To identify the nature of p42, we have used matrix‐assisted laser desorption/ionization time‐of‐flight mass spectrometry (MALDI‐TOF) analysis. This method is based on enzymatic digestion of proteins from Coomassie‐blue‐stained gels and sequencing of the resulting peptide by mass spectrometry (Shevchenko et al, 1996). MALDI‐TOF analysis has identified p42 as GAPDH (Fig 1B), suggesting that GAPDH was immunoprecipitated by an anti‐SUR2A antibody. However, immunoprecipitations could have nonspecific contaminants (Harlow & Lane, 1999), and, therefore, we have used extreme care to provide further evidence that the presence of GAPDH in the immunoprecipitate was not artefactual. First, we have shown that there is no crossreactivity between the anti‐SUR2A antibody and purified GAPDH itself (data not shown), thus excluding the possibility that GAPDH was directly precipitated with our anti‐SUR2A antibody. Second, western blot analysis has shown that GAPDH is present in immunoprecipitate obtained not only with an anti‐SUR2A but also with an anti‐Kir6.2 antibody (Fig 2). As these two antibodies target structurally unrelated proteins, it is not likely that GAPDH would appear in both precipitates as a contaminant. Third, when two proteins are immunoprecipitated with antibodies raised against one or another protein, this is usually considered to be strong evidence of a physical association between these proteins (Harlow & Lane, 1999). Therefore, the reciprocal co‐immunoprecipitation experiment was carried out, in which the anti‐GADPH antibody was used to immunoprecipitate both Kir6.2 and SUR2A subunits from cardiac membrane fraction. As seen in the lower panel of Fig 2, both Kir6.2 and SUR2A could be specifically co‐immunoprecipitated with the anti‐GADPH antibody but not with the control IgG antibody (Fig 2). Taken together, these results strongly suggest that GAPDH associates physically with a protein that is a part of the cardiac KATP‐channel protein complex. As a cardiac KATP‐channel protein complex in vivo is composed of more proteins than Kir6.2 and SUR2A subunits (Carrasco et al, 2001; Crawford et al, 2002a, 2002b), these results per se did not prove that GAPDH has a physical interaction with the KATP‐channel subunits themselves.
However, if GAPDH activity is to have functional significance on the behaviour of KATP channels, it is likely that this would require a close spatial relationship between GAPDH and KATP‐channel subunits and, possibly, a physical interaction between these proteins. To test this hypothesis, we have used genes encoding KATP‐channel subunits and A549 cells as a heterologues expression system. A549 cells are cells that do not contain KATP channels, but express GAPDH (Sundararaj et al, 2004). The idea was to express each of the two KATP‐channel subunits and to test whether GAPDH interacts physically with Kir6.2 or SUR2A when expressed alone in A549 cells. This strategy has been successful previously in detecting proteins that are physically associated with KATP‐channel subunits (Crawford et al, 2002a, 2002b). When the possibility that GAPDH associates with SUR2A was examined, we immunoprecipitated protein extracts of cells expressing the SUR2A subunit with an anti‐SUR2A antibody and probed with an anti‐GAPDH antibody and vice versa. The SUR2A subunit was absent from the anti‐GAPDH precipitate and GAPDH was absent from the anti‐SUR2A precipitate, which suggests that there is no physical interaction between the SUR2A subunit and GAPDH (Fig 3). In contrast, in anti‐Kir6.2 precipitate obtained from cells expressing Kir6.2, GAPDH was detected (Fig 3). Also, Kir6.2 was found in anti‐GAPDH precipitate from cells expressing Kir6.2. These results suggest that Kir6.2 and GAPDH physically associate. Previously, it has been reported that creatine kinase associates physically with the SUR2A subunit and that LDH associates with both Kir6.2 and SUR2A subunits (Crawford et al, 2002a, 2002b). Here, no contact was observed between the amino‐terminal half of Kir6.2 protein and GAPDH (Fig 3), which suggests that the interaction between Kir6.2 and GAPDH is not by means of the N‐terminus. This finding is in accord with previous reports that the N‐terminus interacts with another part of the KATP‐channel protein complex, M‐LDH (Crawford et al, 2002b), and that Kir6.2 and GAPDH interact probably by means of the part of Kir6.2 that is closer to the carboxyl terminus. Thus, these results further confirmed that there is a physical interaction between Kir6.2 and GAPDH. It should be noted that all methods we have used thus far are antibody‐based and that was why we decided to test this hypothesis using a non‐antibody‐based method such as the membrane yeast two‐hybrid system (MbYTH; Stagljar et al, 1998; Thaminy et al, 2003; Iyer et al, 2005). This modified form of the YTH assay is one of several hybrid protein approaches that are capable of detecting interactions occuring at cellular membranes, and its principles have been described in detail elsewhere (Stagljar & Fields, 2002; Stagljar, 2003; Iyer et al, 2005). To test whether Kir6.2 and GAPDH interact in MbYTH, we used full‐length Kir6.2 fused to Cub‐TF as bait and GAPDH fused to NubG as prey (Fig 3D). As shown in Fig 3E, coexpression of Kir6.2–Cub‐TF with a NubG‐tagged GAPDH resulted in the activation of the yeast reporter gene system as exemplified by the strong β‐galactosidase activity. Conversely, Kir6.2–Cub‐TF did not interact with an unrelated human regulator of G protein signalling (Rgs4) protein fused to NubG, indicating that the Kir6.2–GAPDH interaction is specific. As a positive control in MbYTH, the previously described interaction between the human epidermal growth factor receptor 3 (ErbB3) and Rgs4 was used (Thaminy et al, 2003). These results confirm our findings—identified by co‐immunoprecipitation assay—and provide strong evidence that Kir6.2 and GAPDH interact physically. In this regard, GAPDH is the first accessory protein found to associate with the Kir6.2 subunit alone.
The physical association between Kir6.2 and GAPDH may have a significant functional consequence on KATP‐channel behaviour, as this enzyme catalyses the reaction in which 1,3‐bisphosphoglycerate is produced. 1,3‐Bisphosphoglycerate has recently been suggested to be an endogenous KATP‐channel opener that has the ability to activate sarcolemmal KATP channels even in the presence of high levels of intracellular ATP (Jovanović & Jovanović, 2005). Therefore, we have tested whether GAPDH can regulate the activity of sarcolemmal KATP channels by virtue of its enzymatic activity. After excision of a membrane patch from guinea‐pig ventricular cardiomyocyte, an ATP‐sensitive K+ current was observed (Fig 4A). Addition of NAD+ (20 mM) on the intracellular face of the excised membrane patch did not affect the ATP‐induced inhibition of the channel activity (Fig 4A). However, when both NAD+ (20 mM) and glyceraldehyde 3‐phosphate (20 mM) were concomitantly in contact with the intracellular side of sarcolemma, KATP channels opened despite high intracellular ATP (5 mM; Fig 4A). Removal of NAD+ alone resulted in a significant decrease in KATP‐channel activity, indicating that glyceraldehyde 3‐phosphate per se did not activate KATP channels. These results are in accord with the idea that GAPDH is a part of the sarcolemmal KATP‐channel protein complex, where it may activate KATP channels by producing 1,3‐bisphosphoglycerate, a compound that is reported to act as a KATP‐channel opener (Jovanović & Jovanović, 2005). Indeed, when 1,3‐bisphosphoglycerate was added on the intracellular face of the excised membrane patch, it induced activation of KATP channels, which suggests that this intermediate product of glycolysis acts on KATP‐channel proteins or closely associated subunits.
For a long time, it has been proposed that glycolytic ATP may be more important for regulation of KATP channels than ATP obtained by oxidative phosphorylation (Weiss & Lamp, 1987). On the basis of their data, Weiss & Lamp (1987) have suggested that the glycolytic enzymes must be attached physically to the cardiac KATP channels to be able to regulate their behaviour in the described manner. The present study is the first to provide direct evidence in favour of this hypothesis. The existence of a glycolytic signalling pathway in regulating KATP channels probably allows subtlety in the regulation of these channels. A 30 mM portion of glucose was sufficient to activate sarcolemmal KATP channels and this was associated with increased intracellular levels of 1,3‐bisphosphoglycerate (Fig 4C), which suggests that 1,3‐bisphosphoglycerate may have a role in KATP‐channel regulation in vivo. It is possible that glucose metabolism, independently of ATP production or any other factors, may be transduced into changes in membrane excitability, which make the heart responsive to changes in glucose levels irrespective of the conditions of other metabolic pathways and/or total energy status. This idea would be in accord with our previous findings showing that hyperglycaemia may regulate cardiac membrane excitability by stimulation of glucose transport and production of intracellular 1,3‐bisphosphoglycerate (Jovanović & Jovanović, 2005). It seems that physical contact between Kir6.2 and GAPDH secures a close spatial relationship between the pore‐forming subunit and 1,3‐bisphosphoglycerate‐producing enzyme, which, in turn, allows fast sensing of glucose levels.
Methods are presented in the supplementary information online.
This research was supported by grants to A.J. from the British Heart Foundation, Biotechnology and Biological Sciences Research Council, Medical Research Council, Wellcome Trust, TENOVUS‐Scotland and Anonymous Trust. I.S. is funded by grants from the Union Bank of Switzerland (im Auftrag eines Kunden), Gebert Rüf Foundation, Schweizer Bundesamt für Bildung und Wissenschaft BBW no. 01‐0528, the Swiss Cancer League (OCS‐01310‐02‐2003) and the Swiss National Science Foundation (no. 3100A0‐100256/1).
- Copyright © 2005 European Molecular Biology Organization