Coupling of cell energetics with membrane metabolic sensing: Integrative signaling through creatine kinase phosphotransfer disrupted by M-CK gene knockout

Transduction of metabolic signals is essential in preserving cellular homeostasis. Yet, principles governing integration and synchronization of membrane metabolic sensors with cell metabolism remain elusive. Here, analysis of cellular nucleotide fluxes and nucleotide-dependent gating of the ATP-sensitive K+ (K(ATP)) channel, a prototypic metabolic sensor, revealed a diffusional barrier within the submembrane space, preventing direct reception of cytosolic signals. Creatine kinase phosphotransfer, captured by 18O-assisted 31P NMR, coordinated tightly with ATP turnover, reflecting the cellular energetic status. The dynamics of high energy phosphoryl transfer through the creatine kinase relay permitted a high fidelity transmission of energetic signals into the submembrane compartment synchronizing K(ATP) channel activity with cell metabolism. Knock-out of the creatine kinase M-CK gene disrupted signal delivery to K(ATP) channels and generated a cellular phenotype with increased electrical vulnerability. Thus, in the compartmentalized cell environment, phosphotransfer systems shunt diffusional barriers and secure regimented signal transduction integrating metabolic sensors with the cellular energetic network.

Phosphotransfer scanned by NMR spectroscopy - 18 O-assisted 31 P NMR is based on incorporation of 18  NaHCO 3 , 1.2 MgSO 4 , 11 glucose and 20 U/l insulin). Hypoxia was induced with 95%N 2 /5% CO 2 gassed buffer, to reduce partial oxygen pressure to 20-30 mmHg. 18 O labeling was achieved using the buffer supplemented with 40% of 18 O-H 2 O (Isotec) for 30 s. During this time 18 O labeling is still within the initial pseudolinear phase of the labeling kinetic curve, which reaches full saturation only after 2 min following application of 18 O-H 2 O. Hearts freeze-clamped and extracted in 600 mM HClO 4 and 1 mM EDTA. 18 O-induced shifts in 31 P NMR spectra of ATP and creatine phosphate were recorded at 242.9 MHz in a Bruker 14 T spectrometer (42,43), and phosphotransfer fluxes calculated as described (38).
Allosteric model of channel gating -Nucleotide-dependent K ATP channel gating was simulated by an allosteric model where four identical binding sites for ATP and ADP co-exist within the octameric stoichiometry of the K ATP channel complex (16,22). Binding of ATP to the pore-forming Kir6.2 subunit inhibits channel opening (24,25), whereas binding of ADP to the regulatory SUR subunit antagonizes ATP-binding to Kir6.2 (6,26,44). Distribution of channel species (D i ; i=0 to 4) with 0-4 ADP bound molecules was: with the percentage of D i species expressed as a function of ADP concentration: and k ADP the dissociation constant of ADP from SUR, independent from ATP binding. Analogously to Equations 1 and 2, the distribution of channel species (T i ) with 0 to 4 ATP bound molecules was derived as 4 0 , Ti , with k 0 and k 1 representing dissociation constants for ATP binding to Kir6.2 in the absence and presence of ADP at the associated SUR ( Fig. 1A-B). The best fits of experimental data from ATP-induced K ATP channel inhibition in the absence of ADP, at saturating ADP and at below saturating ADP revealed respectively the values for k 0 , k 1 and k ADP , with more than one ATP required to close the channel octamer. Membrane ATP consumption ( J ATPase ) was simulated as a Michaelis-Menten reaction with the Michaelis constant at 0.05 mM (45). Sarcolemmal ATPase activity (1800 nmol/min/g w wt) was derived from total ATPase activity in working hearts measured by 18 O-assisted 31 P NMR (300 nmol/min/mg protein), assuming that 120 mg of protein (with 1 mg of sarcolemmal protein) is contained in 1 g of tissue and that ~5% of total energy is consumed by sarcolemmal ATPases (45). Nucleotide diffusion (with a coefficient D) was calculated according to Fick's law as one-dimensional flux (through total cell area in 1 g of tissue, S) perpendicular to the membrane. Diffusional flux for ATP was:  (Fig. 1D), were defined from:

RESULTS
Metabolic sensing in the submembrane compartment -The defining property of cardiac K ATP channels as membrane metabolic sensors is their overt inhibition by ATP, which can be antagonized by MgADP (Fig. 1A). K ATP channels adopt their highest sensitivity to ATP in the absence of MgADP, and convert to a range of lower ATP-sensitivities with increasing concentrations of MgADP (Fig. 1A). The regulatory SUR2A subunit harbors an intrinsic ATPase activity (36,44)  transitions imparting low or high ATP-sensitivity to the K ATP channel complex (37). MgADP prolongs the lifetime of the conformation associated with reduced sensitivity to ATP (37). Allosteric modeling, which integrated K ATP channel stoichiometry and channel-nucleotide interactions ( Fig. 1 B), demonstrated that on saturating ADP-binding sites (at >100 µM ADP) no further reduction in ATP sensitivity can be achieved (Fig. 1A-B), in accord with the efficacy of ADP to antagonize ATP-induced channel inhibition (47). For MgADP to open at least 1% of K ATP channels, required for significant action potential shortening at 6-10 mM of cytosolic ATP (47)(48)(49)(50), ATP at the channel site needs to be reduced to <3 mM ( Fig. 1C-D). Local drop in ATP could be generated by membrane ATPases (30,51), including ATP hydrolysis by the K ATP channel itself (36,37,44), provided that nucleotide mobility between the cytosol and submembrane is limited (31; Fig. 1C). Calculations, based on membrane ATPase activity and nucleotide gradients between the cytosolic "bulk" and subsarcolemmal space, revealed a strong diffusional hindrance with an apparent diffusion coefficient D=2.3⋅10 -11 cm 2 /s. 1 This value, five orders of magnitude lower than values for nucleotide diffusion in the cytosol (52), is in line with the restricted diffusion of molecules previously observed in the structurally crowded submembrane space of living cells (51,53). Such a diffusional barrier implies virtual confinement of the metabolic sensor within the submembrane zone, impeding direct exposure to cytosolic signals. Thus, integration of K ATP channel activity with cell metabolism requires efficient mechanisms able to shunt diffusional restrictions for proper delivery of energetic signals.
Creatine kinase phosphotransfer synchronized with cellular energy turnover sets nucleotidedependent K ATP channel gating -Creatine kinase molecules are spatially arranged between cellular sites of ATP-production and utilization, providing an integrated network for high-energy phosphoryl conduction (1,3,38,(54)(55)(56). Here, net phosphotransfer flux through the creatine kinase system was captured in intact heart using 18 O-assisted 31 P NMR, and was found to be tightly synchronized with cellular ATP turnover ( Fig. 2A). Labeling of phosphoryl oxygens in creatine phosphate, which reflects 1 D=C⋅∆x/S was derived from C=0.45 cm 3 /min/g w wt (Equation 5, Fig. 1D), assuming a subsarcolemmal space width (∆x) of 10 -5 cm, and a total surface of cells in 1 g of tissue (10 12 µm 3 ) S=3200 cm 2 /g w wt for an average cardiac cell (10×20×100 µm) with a volume of 20000 µm 3 and a surface of 6400 µm 2 . net flux through creatine kinase, paralleled that of γ-phosphate in ATP, an indicator of total cellular high-energy flux ( Fig. 2A). On average, the percentage of 18 O in creatine phosphate was comparable to that of γ-ATP, i.e., 55±4% (n=5) versus 61±3% (n=5), respectively (Fig. 2B). Such a vigorous creatine kinase catalyzed phosphotransfer would rapidly dissipate local nucleotide gradients, and could therefore transmit cellular metabolic signals to membrane metabolic sensors. In fact, creatine kinase catalysis was found high in the cardiac membrane fraction, i.e., 6.2±0.6 µmol/min/mg (n=3), and was sensitive to the conventional creatine kinase inhibitor 2-4-dinitrofluorobenzene (DNFB) which reduced such activity to 0.5±0.1 µmol/min/mg (n=3). To assess the role of creatine kinase flux in regulating K ATP channels as metabolic sensors, channel behavior was measured in the absence and presence of creatine kinase phosphotransfer. To maintain a relative integrity of the cellular infrastructure, cardiomyocytes were permeabilized by local and brief application of digitonin to the region of the cell distal from the patched area ( Fig. 2 C). In such open cell-attached patch configuration, removal of creatine phosphate inactivated creatine kinase phosphotransfer and induced an aberrant sensitivity of K ATP channels towards ATP (Fig. 2D). Indeed, ATP at 100 µM failed to inhibit K ATP channels (Fig. 2 D), a concentration that keeps channels closed in excised membrane patches (Fig. 2E). The reduced ATPsensitivity indicates that, despite clamped "bulk" nucleotide concentrations by continuous cell perfusion, local levels of ATP are decreased and ADP increased in an environment of active membrane ATPases. Activation of creatine kinase phosphotranfer, by addition of creatine phosphate, restored the K ATP channel responsiveness to ATP (Fig. 2D) presumably through scavenging ADP and dissipating the membrane ATPase-induced nucleotide gradient. Creatine phosphate had no significant effect on its own, but secured K ATP channel closure in open cell-attached patches in the presence of low ATP concentrations which in inside-out patches produced only partial channel inhibition ( Fig. 2E and 2F). In the presence of creatine kinase substrates, inhibition of creatine kinase phosphotransfer by DNFB in permeabilized cells uncoupled K ATP channels from phosphotransfer regulation (Fig. 2F). Coupling was restored by providing purified creatine kinase to bypass the irreversible inhibition of endogenous creatine kinase by DNFB (Fig. 2 F). Thus, the creatine phosphate/creatine kinase system is a determinant of the K ATP channel sensitivity to ATP. In permeabilized cells, in the absence of creatine by guest on March 24, 2020 http://www.jbc.org/ Downloaded from kinase flux (J CK = 0), the IC 50 for ATP-induced inhibition was 270±2 µM (n=4; Fig. 2E), close to the value measured in the presence of saturating MgADP in excised patches (Fig. 1A). Active creatine kinase flux (J CK >> 0) significantly reduced the IC 50 to 7±1 µM (n=4), even below the sensitivity of the channel towards ATP seen in excised patches (Fig. 2E). Thus, creatine kinase flux can shunt nucleotide gradients between the "bulk" and subsarcolemmal space, and facilitate delivery of metabolic signals that translate into K ATP channel-dependent sensing of the cellular energetic state.
Creatine kinase phosphotransfer dynamics transduce metabolic stress-induced signals into K ATP channel-driven membrane electrical events -In normoxia, when cardiac K ATP channels are closed, vigorous incorporation of 18 O atoms into creatine phosphate reflected the high creatine kinase phosphotransfer rate of the myocardium (Fig. 3A). Hypoxia markedly reduced creatine kinase flux, from 279±8 nmol CrP/min/mg protein (n=3) to 64±17 nmol CrP/min/mg protein (n=3; p<0.05), indicating a ~75% decrease in creatine kinase phosphotransfer (Fig. 3A). This reduction in creatine kinase phosphotransfer reflects a four-fold drop, from 36±1 to 9±2 nmol/mg protein, in creatine phosphate levels following hypoxia. Thus, under hypoxic stress, creatine kinase has a reduced ability to equilibrate nucleotide levels between the cytosol and subsarcolemma. bring subsarcolemmal nucleotide levels to a range that now lies in the steeper portion defining ATPdependent channel gating, securing effective signal delivery to K ATP channels (Fig. 3C). A drop of 75% in creatine kinase phosphotransfer (J CK :4), observed in hypoxia (Fig. 3A) at cytosolic ATP<6 mM, would translate into activation of >1% of K ATP channels sufficient for significant action potential shortening in a cardiac cell (50), as computed from the allosteric model of nucleotide-dependent channel gating (Fig. 1B) taking into account nucleotide diffusion, membrane ATPase and creatine kinase fluxes (Fig. 3D). Indeed, cardiac action potential duration was significantly decreased from 230±4 ms prior, to 130±7 ms following hypoxic stress (Fig. 3E). Thus, the dynamics of creatine kinase phosphotransfer, governed by the cellular metabolic condition, determine the percentage of open K ATP channels and provide a mediator translating cellular energetic signals into membrane electrical events.
Knockout of creatine kinase disrupts signal delivery to the metabolic sensor -Cytosolic creatine kinase (M-CK) is the major creatine kinase isoform in the heart (1). Deletion of the M-CK gene blunted creatine kinase phosphotransfer, and essentially eliminated creatine kinase activity in the sarcolemma (Fig. 4A). In wild-type cardiomyocytes, K ATP channel activity was highly sensitive to creatine kinase-mediated channel inhibition, revealed from titration with creatine phosphate (IC 50 =94±5 µM, n=5; Fig. 4B-C). This indicates tight integration of creatine kinase-catalyzed energetic signaling with K ATP channel activity. In contrast, in cells lacking M-CK, K ATP channels were uncoupled from the cellular energetic infrastructure, and were no longer sensitive to creatine phosphate regulation ( Fig. 4B-D). Thus, cells lacking the creatine kinase phosphotransfer system display a defective regulation of nucleotide-gated membrane functions.
Fidelity in membrane metabolic sensing lost with deletion of the M-CK gene -Besides creatine kinase, additional phosphotransfer pathways, such as the glycolytic system, have been identified as interrelated components of the cellular energetic network (2,3,5). Here, in wild-type cardiomyocytes, inhibition of phosphoryl delivery through the glycolytic system by deoxyglucose did not elicit a K ATP channel response (Fig. 5A), suggesting privileged control of this metabolic sensor by creatine kinase. In contrast, in cells lacking M-CK, metabolic stress induced with deoxyglucose triggered K ATP channel opening ( Fig. 5B-C). In fact, aberrant coupling of K ATP channels with cellular metabolism in creatine by guest on March 24, 2020 http://www.jbc.org/ Downloaded from kinase-knockout hearts generated a phenotype with electrical instability manifested by premature shortening of action potentials in response to deoxyglucose (Fig. 5D-E). The rate of stress-induced action potential shortening was significantly faster in creatine kinase-knockout (0.042±0.005 min -1 , n=4) than wildtype (0.016±0.004 min -1 , n=4) hearts (p<0.05; Fig. 5D). Accordingly, action potential duration, measured at 90% of repolarization, was essentially unchanged (94±4 % of control value, n=4) in wildtype, but was reduced, by 39±9 % (n=4), in the M-CK knockout following a 7-min long application of deoxyglucose (Fig. 5E). Thus, creatine kinase phosphotransfer is required for proper linkage of cellular energetics with membrane excitability.

DISCUSSION
Metabolic signal transduction governs vital functions that enable cells to respond to metabolic challenges. But how the operation of metabolic sensors is orchestrated to accurately sense the cellular energetic status has remained a long-standing enigma. Using cardiac K ATP channels as prototypic membrane metabolic sensors, we demonstrate that phosphotransfer enzyme-catalyzed metabolic signal delivery synchronizes channel gating with cell energetics. Genetic disruption of the creatine kinase pathway generated a phenotype with increased electrical vulnerability, underscoring the significance of an intact intracellular phosphotransfer network in integrating metabolic signaling.
The extremely low diffusional flux of nucleotides, estimated here in the subsarcolemmal space, indicates that K ATP channels are virtually secluded from cellular "bulk" nucleotide oscillations.
Restricted metabolite mobility in the premembrane area could be due to molecular crowding and reduction in the free diffusional space as previously suggested for different cellular compartments (13)(14)(15)53). In fact due to the "fuzzy space" in the submembrane (31), channel gating would be relegated to local fluctuations of nucleotides ("metabolic background noise"), independent of the cellular metabolic status. Instead of random fluctuations in adenine nucleotides, which would distort energetic signaling, we provide direct evidence that the creatine kinase phosphotransfer system controls exchange of nucleotides securing signal processing between the subsarcolemmal space and cytosolic compartment. Indeed, the dynamics of creatine kinase flux closely followed total ATP turnover, indicating tight coupling between creatine kinase phosphotransfer and the cellular metabolic state (38,43).
Present throughout the cell, creatine kinase reactions form a phosphotransfer relay able to respond to changes in the cellular metabolic state and propagate metabolic waves between cellular compartments (3,57,58). Such catalyzed phosphotransfer can deliver metabolic signals at a rate exceeding simple diffusion (38,58). Accordingly, here, under the normal status of cell metabolism, vigorous creatine kinase phosphotransfer dissipated local nucleotide gradients created by membrane ATPases and diffusional restrictions in the channel environment, keeping K ATP channels predominantly closed. Under metabolic stress, however, reduced creatine kinase phosphotransfer unmasked nucleotide changes in the channel vicinity alerting K ATP channels to adjust membrane excitability. Reduced creatine kinase flux under metabolic insult is known to be associated with concomitant up-regulation of adenylate kinase phosphotransfer, which catalyzes the conversion of ATP to ADP at the channel site (34,57).
Such interplay between phosphotransfer pathways effectively amplifies the metabolic signal translating into K ATP channel opening and ultimately shortening of the cardiac action potential under stress (19,59). Thus, membrane metabolic sensors respond to the dynamics of cellular phosphotransfer flux, reflecting with high fidelity the energetic status of a cell.
Integration of phosphotransfer with K ATP channels appears critical in supporting metabolic signaling. Knockout of the dominant creatine kinase isoform, M-CK, disrupted K ATP channel regulation by creatine phosphate. This produced a cellular phenotype characterized by increased electrical instability, in line with observations that muscles lacking creatine kinase genes display abnormal contractile response and reduced energetic efficiency (1,4,54,55). Moreover, coupling of creatine kinase with Ca 2+ -ATPases of the sarcoplasmic reticulum, which is essential in securing Ca 2+ handling and proper kinetics of intracellular Ca 2+ signals, is compromised following deletion of creatine kinase genes (60). In this regard, creatine kinase can also functionally couple with K ATP channels through direct creatine kinase-dependent regulation of the ATPase catalytic cycle harbored within SUR, the channel regulatory subunit (36,37). The ATP hydrolysis cycle at SUR drives conformational transitions associated with distinct outcomes on channel behavior, with creatine kinase promoting disengagement by guest on March 24, 2020 http://www.jbc.org/ Downloaded from of the MgADP-bound state and K ATP channel closure (37). Thus, nucleotide exchange between cellular phosphotransfer catalyzed by creatine kinase and membrane ATPases, including the channel's own ATPase, provides a mechanistic basis for coupling cell energetics with metabolic signal transduction.
Along with creatine kinase, distinct phosphotransfer systems can also efficiently communicate energetic signals to metabolic signal transducers, and regulate ATP-sensitive cellular components (3,34,61). The interrelationship between intracellular energetic pathways is revealed upon deletion of the M-CK gene, which translated into redistribution of metabolic flux through glycolytic enzymes (56,62).
In accord with the adaptive potential of phosphotransfer pathways (1,5,42), such energetic remodeling was sufficient to maintain apparently normal K ATP channel gating in the absence of metabolic challenge in the M-CK knockout heart. However, knockout of M-CK did produce increased electrical vulnerability manifested by premature action potential shortening in hearts stressed by inhibition of glycolytic enzymes. Thus, an intact phosphotransfer network is a prerequisite for optimal decoding of energetic signals securing adequate function of a metabolic sensor. The significance of these findings is underscored in human disease where compromised creatine kinase phosphotransfer has been associated with cardiac electrical instability (63), and extrapyramidal movement disorders (64). Shortening of action potential duration (APD), in hypoxia recorded by the monophasic action potential electrode. APD was corrected to heart rate using the modified Bazet's formula: