Differential distribution and regulation of mouse cardiac Na+/K+-ATPase α1 and α2 subunits in T-tubule and surface sarcolemmal membranes

Objectives: Two Na+/K+-ATPase (NKA) α-subunit isoforms, α1 and α2, are expressed in the adult mouse heart. The subcellular distribution of these isoforms in T-tubule and surface sarcolemmal (SSL) membranes and their regulation by cAMP-dependent protein kinase (PKA) is unclear.

Methods: We used formamide-induced detubulation of mouse ventricular myocytes to investigate differential functional distribution and regulation by PKA of α1 and α2 in T-tubule versus SSL membranes by measuring NKA current ( I pump) and NKA-mediated Na+ efflux (− d [Na]i/ dt ).

Results: I pump is composed of 88% α1-mediated I pump ( I α1) and 12% α2-mediated I pump ( I α2). α1 and α2 subunits demonstrate distinct ouabain affinities (105±6 and 0.3±0.1 μmol/L respectively) but similar affinity for intracellular Na+ (K1/2Na+ of 16.6±0.8 and 16.7±2.6 mmol/L respectively). Detubulation reduced (i) I pump density (1.42±0.1 to 1.20±0.04 pA/pF), (ii) cell capacitance (181±12 to 127±17 pF), and (iii) I α2 contribution (12 to 6%). Total I pump density was ∼60% higher in T-tubule (1.94 pA/pF, derived) vs. SSL membranes. Although T-tubule membranes represent only 30% of total surface area, they generate ∼70% of I α2 and ∼37% of I α1. I α1 density was substantially higher than I α2 in SSL ( I α1: I α2=16:1) but this was markedly reduced in T-tubules (4:1). In addition to differential localisation, isoprenaline (ISO, 1 μmol/L) significantly increased α1-mediated NKA Na+ affinity (from 16.6±0.8 to 13.3±1.4 mmol/L) and caused a small increase in maximal NKA Na+ efflux rate. ISO had no effect on α2-mediated NKA activity.

Conclusion: These data suggest that NKA α1 and α2 subunits are differentially localised and regulated by PKA in T-tubule and SSL membranes and may have distinct regulatory roles in cardiac excitation–contraction coupling.


Introduction
The cardiac Na + /K + -ATPase (NKA) is the primary mechanism by which intracellular sodium ([Na + ] i ), and hence intracellular calcium [Ca 2+ ] i is regulated in the heart. The NKA establishes and maintains the physiological transmembrane [Na + ] gradient which is essential for a plethora of cellular functions [17,20,26,45] and indirectly controls myocardial contractility by influencing Na + /Ca 2+ exchange (NCX) activity [30,34] and indirectly setting sarcoplasmic reticulum (SR) Ca 2+ load and contractility.
In voltage-clamped guinea-pig ventricular myocytes, Gao et al. [14] demonstrated a clear biphasic relationship between increasing concentration of dihydro-ouabain (DHO) and inhibition of whole-cell Na + /K + pump current (I pump ). This biphasic relationship was due to the presence of both high DHO affinity α 2 pumps and low affinity α 1 pumps.
In physiological terms, it has been proposed that the Na + / K + -ATPase may be specifically tailored for a tissue by differential expression of a mix of functionally different pump isoforms [14]. Studies by Lingrel and colleagues investigated the possibility that α-subunit isoforms are functionally and spatially distinct in the mouse heart. Measurement of cardiac contractility in Langendorff perfused mouse hearts with genetically reduced levels (∼ 50%) of cardiac Na + /K + -ATPase α 1 or α 2 isoforms lead to the proposal of a compartmentalisation model whereby α 2 regulates [Ca 2+ ] i and cardiac contractility within membrane regions (T-tubules) in close proximity to the Ca 2+ regulatory machinery (e.g. Ltype Ca 2+ channels, sarcoplasmic reticulum Ca stores, NCX) and α 1 localises to the surface sarcolemma and plays a general housekeeping role by regulating bulk [Na + ] i [22]. In agreement with this concept, selective inhibition of α 2 activity in mouse astrocytes with genetically modified levels of α 2 subunit expression, increases [Na + ] i and [Ca 2+ ] i (via NCX) in the cytosolic environment between plasma (PM) and endoplasmic reticulum (ER) membranes [16].
Although the validity of the compartmentalisation model has recently been contested [11,37], immunofluorescence studies in guinea-pig ventricular cardiac myocytes suggest that α 1 subunits are predominantly located in the peripheral sarcolemmal whilst α 2 are mainly distributed in the T-tubules [42]. A similar pattern has been reported in primary cultured rat astrocytes, neurons and arterial myocytes [24], but the opposite pattern has been reported in rat ventricular myocytes [29]. Further studies are required to clarify this situation.
Myocyte detubulation enables direct functional measurements of ion channel and transporter function in surface sarcolemma (SSL) vs. T-tubule membranes [3,25]. Detubulation is achieved by subjecting myocytes to osmotic shock which seals off the T-tubules leaving them functionally intact but isolated from the SSL. In detubulated myocytes only currents carried by SSL channels and transporters are accessible. By this method it has been demonstrated that L-type Ca 2+ current (I Ca ) [25], NCX activity, and Na + / K + -ATPase activity [8,47] are preferentially concentrated in the T-tubules of rat ventricular myocytes. This evidence is in favour of a model whereby all the components required for efficient excitation-contraction coupling are localised in the T-tubules and in close proximity to the SR Ca 2+ store. However, detubulation has yet to be used to investigate the distribution of Na + /K + -ATPase α 1 and α 2subunit function in T-tubule and SSL membranes. This may shed light on different physiological roles of α 1 and α 2 in the heart.
In the present study we have assessed the functional distribution of α 1 and α 2 subunits in T-tubule versus SSL membranes by formamide-induced detubulation of mouse ventricular myocytes and measurement of Na + /K + pump current (I pump ) and Na + /K + -ATPase-mediated Na + efflux rate (− d[Na + ] i /dt). We have estimated, (i) the composition of plasma membrane surface area in terms of T-tubule and SSL membranes, (ii) I pump amplitude and density in T-tubule and SSL membranes, (iii) the contribution of Iα 1 and Iα 2 to total I pump , (iv) the Iα 1 :Iα 2 ratio in T-tubule and SSL membrane compartments, and (v) the effect of ISO stimulation on α 1 and α 2 Na + /K + -ATPase activity.

Animals
All animals used in this study received humane care in accordance with the Guide for the Care and Use of Laboratory Animals published by the US National Institute of Health (NIH Publication No. 85-23, revised 1996). This study was also subjected to local ethical review by the Ethical Review Process Committee of King's College London and Loyola University Chicago.

Electrophysiological recording of Na + /K + pump current (I pump )
Mouse ventricular cardiac myocytes were voltageclamped and whole-cell I pump recorded at 35°C using the perforated-patch technique. Electrodes were made from thin-walled (1.5 mm outer diameter, 1.17 mm inner diameter) borosilicate glass capillaries (Harvard Apparatus Ltd, UK) and fire-polished using a three-stage electrode puller (DMG Universal Puller, Zeitz-Instrumente Vetriebs GMBH, Germany). Electrode resistance was 1-2 MΩ when filled with the standard pipette solution. Following gigaohm seal formation, series resistance was monitored with a repetitive 5 mV pulse (− 80 mV holding potential). During membrane permeabilization, series resistance typically fell to 10-15 MΩ within 10 min. Membrane capacitance was recorded after permeabilization by standard techniques [4] by imposing a 25 ms square step from − 80 to −75 mV and integrating the area under the capacitance transient. I pump was recorded continuously at 10 Hz sampling frequency at 0 mV. Pipette and extracellular solutions were designed to inhibit all voltage-gated channels and the Na/Ca exchanger. Standard pipette solution contained (in mmol/L) CsCH 3 O 3 S 90, NaCH 3 O 3 S 35, NaCl 15, CsCl 5, MgCl 2 1, HEPES 10, pH 7.2 at 35°C with CsOH. Amphotericin B (225 μg/mL) (from Streptomyces, Sigma, UK) in DMSO (0.74% v/v) was added to the pipette solution on the day of use. Standard K-containing extracellular solution (5K) contained (in mmol/L) NaCl 140, KCl 5, MgCl 2 1, NiCl 2 2, BaCl 2 1, procaine 0.5, glucose 10, HEPES 10, pH 7.4 at 35°C. K-free solution (0K) was prepared by removing KCl with no correction for osmolarity. In all experiments I pump was defined as that sensitive to the removal of extracellular K and was calculated as the product of steady-state 5K minus 0K current. Ouabain was added to 5K solution on the day of use (from a 10 mmol/L stock) and protected from light.

Measurement of Na + efflux through the Na + /K + pump
Na/K-pump flux was determined as the rate of pumpmediated [Na + ] i decline and dual excitation fluorescence measurements (at 340 and 380 nm; F340 and F380) were performed as previously described [9]. Myocytes were Na +loaded by inhibiting the Na + /K + pump in a K + -free solution containing (mmol/L): 145 NaCl, 2 EGTA, 10 HEPES, and 10 glucose (pH = 7.4). [Na + ] i decline was measured on pump reactivation in solution containing (mmol/L): 140 TEA-Cl, 4 KCl, 2 EGTA, 1 MgCl 2 , 10 HEPES, and 10 glucose (pH = 7.4). Because cell volume does not change with this protocol [10], [Na + ] i decline reflects Na + efflux. The rate of [Na + ] i decline (− d[Na + ] i /dt) was plotted versus [Na + ] i and fitted with: − d[Na + ] i /dt = V max /(1 + (K m /[Na + ] i ) nHill ). In separate experiments, − d[Na + ] i /dt was measured in the presence of 10 mmol/L ouabain to determine Na + pump independent Na + efflux. This was subtracted from Na + /K + pump-mediated efflux. In some experiments, cells were treated with 1 μmol/L ISO during the latter part of pump inhibition and throughout reactivation. Ouabain (10 μmol/L) was used in some experiments to preferentially inhibit high ouabain affinity Na + /K + -ATPase α 2 subunits. All experiments measuring [Na + ] i with SBFI were carried out at room temperature (25°C).

Statistical analysis
Quantitative data are shown as mean ± standard error of the mean (SEM). n values for electrophysiological experiments are given as the number of cells from number of animals. Student t test was used for statistical discriminations, with P b 0.05 considered significant and non-significance indicated (ns).

Cell capacitance and I pump distribution in surface sarcolemmal and T-tubular membrane compartments
In cardiac myocytes the membrane system is composed of surface sarcolemmal (SSL) and T-tubular compartments. T-tubule cell capacitance and localised Na + /K + pump current were calculated by subtracting control values from those recorded in detubulated myocytes (T-tubule = Total − SSL). Membrane capacitance was reduced from 181 ± 12 pF in control to 126 ± 17 pF in detubulated myocytes. These data demonstrate that membrane surface area is composed of 30% T-tubule (54 pF) and 70% SSL (126 ± 17 pF) membranes. I pump amplitude recorded at 0 mV in 5 mmol/L [K] o and 50 mmol/L [Na + ] i was reduced from 257 ± 22 pA in control myocytes (1.42 ± 0.1 pA/pF) to 151 ± 17 pA in detubulated myocytes (1.20 ± 0.04 pA/pF). Therefore, although T-tubules represent only 30% of the membrane surface area, Na + pumps residing there generate ∼ 41% (106 pA) of total I pump (the remaining 59% is generated in the SSL). These data are summarised in Fig. 2. Normalising I pump amplitude data to cell capacitance demonstrates that functional I pump density in T-tubular membranes (1.94 pA/ pF) is 60% higher than in SSL membranes (1.20 pA/pF) (Ttubule:SSL I pump density ratio = 1.6:1).
We also measured the rate of Na + /K + -ATPase-mediated Na + efflux (− d[Na + ] i /dt) in control and detubulated myocytes (Fig. 3). Maximal Na + efflux rate (V max ) was 10.7 ± 1.9 mmol/min in control myocytes and 7.5 ± 0.9 mmol/min following detubulation. Despite this not achieving the level of statistical significance, this ∼ 30% difference in the mean V max values is not incompatible with Fig. 2. Cell capacitance and I pump amplitude/density in SSL and T-tubule membranes recorded in control and detubulated myocytes. (A) Cell capacitance was reduced from 181 ± 12 pF in control (n = 28) to 127 ± 17 pF in detubulated (n = 7) myocytes. Hence 70% of total membrane surface area is localised in the SSL compartment (SSL capacitance = 127 ± 17 pF) and the remaining 30% is localised in the T-tubules (calculated T-tubule capacitance = 54 pF). (B) Average I pump amplitude was 257 ± 22 pA in control (n = 18 cells, 7 hearts) and 151 ± 17 pA in detubulated (n = 5 cells, 4 hearts) myocytes. Hence 151 ± 17 pA (59%) of total recordable I pump is generated by Na + pumps residing in SSL membranes and the remaining 106 pA (41%) by those residing in the T-tubules. (C) Total functional I pump density was 1.42 ± 0.1 pA/pF in control cells, 1.20 ± 0.04 pA/pF in detubulated cells (SSL) and 1.94 pA/pF (calculated) in the T-tubules. the suggestion from the voltage-clamp data that T-tubular Na + /K + -ATPase activity accounts for about ∼ 40% of the total cellular Na + efflux.

I pump composition in control myocytes (SSL and T-tubule membranes)
The contribution of Iα 1 and Iα 2 to total I pump was defined by investigating the dose-dependent inhibition of I pump with ouabain. Iα 1 (low-affinity) and Iα 2 (high-affinity) were defined by their differing sensitivity to ouabain. Maximal I pump inhibition was achieved by exposure to 10 mmol/L ouabain or 0K solution. Recovery from inhibition was rapid and complete with a return to the pre-inhibition level within 3 min. Fig. 4A is a current recording demonstrating dosedependent inhibition of I pump with ouabain. Average data representing the percentage inhibition of I pump by ouabain was fit with a two-site binding hyperbolic function and Iα 1 and Iα 2 determined by curve stripping (Fig. 4B). In control myocytes Iα 1 contributed 88% to the total recordable current with a K d for ouabain of 105 μmol/L and Iα 2 contributed the remaining 12% with a K d of 0.3 μmol/L.
To support the above I pump data, Na + /K + -ATPasemediated Na + efflux was recorded in the presence of low dose ouabain (10 μmol/L) to preferentially inhibit Na + /K + -ATPase α 2 subunits. Curve stripping analysis of our I pump data (Fig. 4B) suggests that 10 μmol/L ouabain, will inhibit 97% of α 2 -mediated Na + /K + -ATPase activity and only 9% α 1 activity. Under these conditions α 1 -mediated Na + efflux predominates due to 11-fold specificity for α 2 inhibition. α 2mediated Na + /K + -ATPase activity was calculated as the difference between total and α 1 -mediated activity. Fig. 5 represents total Na + /K + -ATPase-mediated Na + efflux, and that mediated via α 1 and α 2 subunits. Curve fitting with the Hill equation demonstrated that maximal Na + efflux rate (V max ) was 10.7 ± 1.9 mmol/min in control myocytes. 80% of this Na + efflux capacity was due to α 1 subunits (8.6 ± 1.6 mmol/min) and the remaining 20% via α 2 . These data correlate well with estimates of I pump composition in terms of Iα 1 and Iα 2 . Additionally, these data suggest that α 1 and α 2 subunits have very similar affinity for Na + ions, with K m values of 16.6 ± 0.8 and 16.7 ± 2.6 mmol/L respectively (ns).   3. Na + /K + -ATPase-mediated Na + efflux in control and detubulated myocytes plotted as a function of intracellular Na + concentration ([Na + ] i ). The rate of [Na + ] i decline (− d[Na + ] i /dt) was measured in control (n = 5 cells, 4 hearts) and detubulated (n = 4 cells, 3 hearts) myocytes. Averaged data was fitted with a Hill equation and V max , K m and Hill slope (h) estimated for each data set. Maximal Na + efflux rate (V max ) was reduced 10.7 ± 1.9 mmol/min in control myocytes and 7.5 ± 0.9 mmol/min following detubulation (ns). The K m and slopes of the fitted curves were estimated in control cells to be 16.6± 1.3 mmol/L, and 2.4 ± 0.1 and, in detubulated cells, to be 13.8 ± 3.0 mmol/L (ns) and 2.3 ± 0.1 (ns) respectively.

Iα 1 and Iα 2 amplitude and density in SSL and T-tubular membranes
Having determined the contribution of α 1 and α 2 subunits to I pump in control myocytes we constructed a ouabain doseresponse curve in detubulated myocytes. Under these conditions, Iα 1 contributes 94% and Iα 2 only 6% to total recordable I pump (K d for ouabain of 170 and 0.2 μmol/L respectively) (Fig. 6). These data define the percentage composition of total I pump , in terms of Iα 1 and Iα 2 in the SSL membrane compartment.
In control myocytes cell capacitance was 181 pF and total I pump amplitude (257 pA) was composed of 88% Iα 1 (226 pA) and 12% Iα 2 (31.1 pA). Therefore, Iα 1 density was 1.25 pA/pF, and Iα 2 density was 0.17 pA/pF. In detubulated myocytes (in which only SSL membranes contribute to cell capacitance and only SSL pumps contribute to whole-cell I pump ), total I pump amplitude (151 pA) was composed of 94% Iα 1 (142 pA) and 6% Iα 2 (9.4 pA). After normalising for SSL cell capacitance (127 pF), SSL Iα 1 density was 1.12 pA/pF and Iα 2 density was 0.07 pA/pF. With the above information, T-tubule Iα 1 and Iα 2 amplitude (84 pA and 22 pA respectively) and density (1.54 pA/pF and 0.39 pA/pF respectively) can be determined mathematically. These data are summarised in Table 1, panel A.
and Na + affinity (16.7 ± 2.6 to 11.2 ± 3.4 mmol/L). It is clear from the data in Fig. 7C that the error associated with α 2mediated Na + /K + -ATPase activity is very large. These data are derived mathematically and α 2 represents only a small component of total Na + /K + -ATPase activity.

Discussion
In the present study we have investigated the distribution and function of Na + /K + -ATPase α 1 and α 2 -subunits in SSL and T-tubule membranes in mouse ventricular cardiac myocytes using the technique of myocyte detubulation.
The accuracy of this experimental approach relies on successful and efficient detubulation. The close correlation between the reduction in cell surface area reported here (∼30%) and direct electron microscopic measurements of total sarcolemma in T-tubules in mouse ventricular myocytes (36%) [33] suggests that this method is quantitatively reliable. Furthermore, staining of control myocytes with di-8-ANEPPS revealed a mean T-tubule interval of ∼1.84 μm, which is very similar to that observed in the rat (1.86 μm) [3]. Repetitive Ttubular staining was completely lost following detubulation.
Although α 1 and α 2 isoforms are present in mouse ventricular myocytes, α 1 is the predominant Na + /K + -ATPase isoform. α 1 -mediated I pump (Iα 1 ) contributes 88% to total recordable I pump and α 2 -mediated I pump (Iα 2 ) contributes the remaining 12%. The overall subcellular localisation of Na + / K + -ATPase activity indicates that I pump density is 60% higher in T-tubule vs. SSL membranes (although T-tubule membranes represent only 30% of total membrane area) and that Na + pumps residing there generate ∼ 41% of total I pump ; ∼ 37% of Iα 1 and ∼ 70% of Iα 2 . Moreover, this study provides the first quantitative determination of the relative distribution of α 1 and α 2 -mediated Na + /K + -ATPase activity in mouse ventricular myocytes. We have shown that Iα 1 density predominates over Iα 2 in both SSL and T-tubule membrane compartments. However, the relative ratio of Iα 1 : Iα 2 is markedly different in T-tubule versus SSL membranes. Iα 1 density is substantially higher than Iα 2 in SSL membranes (Iα 1 :Iα 2 density ratio of 16:1), but in T-tubule, the dominance of Iα 1 over Iα 2 is markedly reduced (4:1). Furthermore the T-tubule:SSL Iα 1 ratio (1.4:1) suggests that Iα 1 is relatively uniformly distributed between T-tubule and SSL membranes whereas Iα 2 is ∼5 times more concentrated in the T-tubules (T-tubule:SSL ratio of 5.3:1). These data are summarised in Table 1B. A similar pattern has been reported in rat ventricular myocytes in a recent abstract, with ∼ 4.5 times higher functional density of Iα 2 in the T-tubules and uniform Iα 1 distribution between T-tubule and SSL membranes [6].
Recent studies in detubulated rat myocytes have reported that I Ca , NCX and Na + /K + -ATPase activity are concentrated   in the T-tubules [8,25,47]. Hence, co-localisation of a specific Na + /K + -ATPase α subunit isoform with NCX and the L-type Ca channel in T-tubules would form a structural basis of cardiac excitation-contraction coupling and local control of contractility as described by the compartmentalisation model of James et al. [22]. Recently, the validity of this compartmentalisation model has been contested [11,37], and the authors have now concluded that both α 1 and α 2 isoforms can indirectly regulate cardiac contractility through modulation of forward mode Na/Ca exchange. Our data suggests that α 1 and α 2 subunits are differentially localised. Iα 1 is greater than Iα 2 in both T-tubule and SSL membrane compartments but the predominance of Iα 1 over Iα 2 is lower in T-tubule. Therefore, it is possible that by altering the extent to which α 1 predominates over α 2 (i.e. altering the Iα 1 :Iα 2 balance), previously hidden differential physiological roles for the two isoforms may be revealed. In addition to differential subcellular localisation we have also demonstrated that ISO significantly stimulates α 1mediated Na + /K + -ATPase activity (predominantly via an increase in Na + /K + -ATPase Na + affinity). In agreement, recent data from our laboratory have demonstrated isoformspecific stimulation of Iα 1 in guinea-pig myocytes following PKA stimulation with forskolin [42]. Conversely, in the present study we have shown that ISO has no significant effect on α 2 -mediated Na + /K + -ATPase activity, but due to the small contribution of α 2 to total Na + /K + -ATPase activity coupled with its mathematical derivation, this conclusion should be viewed with caution. Previously published studies from ourselves [42] and others [15] have also suggested that β-stimulation activates α 1 but not α 2 . However, on the basis of the data presented in this present study, it is possible that ISO does influence α 2 -mediated pump function but this is below the limit of detection of this method.
In terms of β-adrenergic stimulation in the heart, many proteins involved in excitation-contraction coupling are targets for PKA phosphorylation (e.g. L-type Ca 2+ channel, phospholamban, ryanodine receptor, troponin I). Until recently the exact mechanism of Na + /K + pump regulation by PKA has remained elusive. It is now clear that this role is played by phospholemman (PLM) [1,7,42], a member of the FXYD family of proteins that are tissue specific regulators of the Na + /K + pump. PLM is the primary sarcolemmal substrate for PKA [35] and PKC [36] and regulates the cardiac Na + /K + pump by applying a tonic inhibition that is relieved by genetic PLM knockout and PLM phosphorylation [1,7]. Previously, we have demonstrated both functional and physical association between PLM and the Na + /K + -ATPase α 1 isoform [42] (but not with α 2 ), and more recently we have demonstrated that the stimulatory effect of ISO on I pump in PLM wildtype voltage-clamped mouse ventricular myocytes is preferentially mediated via stimulation of α 1 -mediated current [1]. With regards to biochemical evidence, co-immunoprecipitation studies have demonstrated association of PLM with the α 2 subunit in rabbit ventricular myocytes and bovine sarcolemmal microsomes [2,5]. However no associ-ation has been reported in guinea-pig ventricular myocytes and rat cardiac homogenates [13,42].
The evidence presented here supports a model whereby α 1 and α 2 subunits demonstrate differential localisation and potentially differential regulation by PKA/β-adrenergic stimulation in mouse ventricular T-tubule and SSL membranes. As β-stimulation leads to elevated [Na + ] i as a direct consequence of positive chronotropy [12,18,43], it may be that the α 1 subunit, located in SSL membranes and regulated by PLM, is primarily involved in controlling bulk [Na + ] i (as suggested by James et al. [22]) and controlling the delicate balance of [Na + ] i by allowing it to rise sufficiently to contribute to the positive inotropic effect of β-stimulation while protecting against the deleterious effects of [Na + ] i and [Ca 2+ ] i overload which may lead to cardiac arrhythmias and diastolic dysfunction.