Complex Rectification of Müller Cell Kir Currents

Abstract

Although Kir4.1 channels are the major inwardly rectifying channels in glial cells and are widely accepted to support K⁺- and glutamate-uptake in the nervous system, the properties of Kir4.1 channels during vital changes of K⁺ and polyamines remain poorly understood. Therefore, the present study examined the voltage-dependence of K⁺ conductance with varying physiological and pathophysiological external [K⁺] and intrapipette spermine ([SP]) concentrations in Müller glial cells and in tsA201 cells expressing recombinant Kir4.1 channels. Two different types of [SP] block were characterized: “fast” and “slow.” Fast block was steeply voltage-dependent, with only a low sensitivity to spermine and strong dependence on extracellular potassium concentration, [K⁺]_o. Slow block had a strong voltage sensitivity that begins closer to resting membrane potential and was essentially [K⁺]_o-independent, but with a higher spermine- and [K⁺]_i-sensitivity. Using a modified Woodhull model and fitting i/V curves from whole cell recordings, we have calculated free [SP]_in in Müller glial cells as 0.81 ± 0.24 mM. This is much higher than has been estimated previously in neurons. Biphasic block properties underlie a significantly varying extent of rectification with [K⁺] and [SP]. While confirming similar properties of glial Kir and recombinant Kir4.1, the results also suggest mechanisms underlying K⁺ buffering in glial cells: When [K⁺]_o is rapidly increased, as would occur during neuronal excitation, “fast block” would be relieved, promoting potassium influx to glial cells. Increase in [K⁺]_in would then lead to relief of “slow block,” further promoting K⁺-influx.

INTRODUCTION

Extracellular fluid in the narrow cleft between neurons and glia has low [K⁺] until neuronal activity (even modest) causes K⁺-efflux from neurons resulting in a considerable rise in [K⁺]_o (Heinemann and Lux, 1977; Karwosky et al., 1989; Krnjevic et al., 1980; Lux and Neher, 1973; Nicholson and Sykova, 1998). If K⁺-regulation fails, pathological development of epileptiform activity, spreading depression, anoxic depolarization, cell swelling and neuronal death may be initiated (Bringmann et al., 2006; Seifert et al., 2006; Somjen, 2001; Wallraff et al., 2006).

Although there is a lack of consensus on how glial cells buffer external [K⁺], net uptake versus K⁺-channels, (Chen and Nicholson, 2000; D’Ambrosio et al., 2002; Kofuji et al., 2000; Newman, 1993; Oakley et al., 1992; Ransom et al., 2000; Walz, 1992, 2000), two models have been suggested by use of K⁺-channels: (i) long distance spatial K⁺-buffering (Gardner-Medwin et al., 1981; Gardner-Medwin, 1983; Orkand et al., 1966; Orkand, 1986) and (ii) short distance K⁺-siphoning (Newman et al., 1984). The presence of long distance (Bykov et al., 1981; Dick et al., 1985) and short distance (Karwoski et al., 1989) spatial K⁺-buffering were demonstrated by simultaneous measurement of potassium ion concentration and electrical potential generated by the glial Müller cells of the retina. Theoretical analysis suggests that long distance K⁺-spatial buffering will be efficient if K⁺-channels are inwardly rectifying (Kir), while short distance K⁺-siphoning will be more effective with channels of a linear conductance (Amedee et al., 1997). Channels with very variable properties have actually been reported in glial cells, and it is unclear how the rectification of glial K⁺ channels differs in respect to location, both from cell type to cell type, and with location on a single cell (Bringmann et al., 2006; Kofuji and Newman, 2004; Reichenbach et al., 1997; Schools et al., 2003; Schopf et al., 2004; Skatchkov et al., 1999, 2006).

Of more than 28 known genes for inwardly rectifying K⁺ channels (Kir), only two, Kir4.1 and Kir6.1, have been consistently demonstrated to be functionally expressed in glial Müller cells (Ishii et al., 1997; Kofuji et al., 2000; Kusaka and Puro, 1997; Skatchkov et al., 2001, 2002). Immunolabel for Kir2.1 was found in these cells by Kofuji et al. (2002), but not by Tian et al. (2003) and these channels were never functionally observed neither in Kir4.1 knock-out mice (Kofuji et al., 2000) nor in conditional knock-out of Kir4.1 mice (Djukic et al., 2007) as well as in rat Müller cells (rev. Bringmann et al., 2006) or in brain glial cells (Pruss et al., 2005). Also, under normal conditions, Kir6.1 channels are likely to be inhibited by intracellular ATP (Skatchkov et al., 2002) and other Kir families were not found immunocytochemically in Müller cells (Skatchkov et al., 2001; Tian et al., 2003), therefore, Kir4.1 is expected to underlie the glial cell inward rectifier current.

Kir4.1 (KCNJ10) is an important component of brain Kir channels (Butt and Kalsi, 2006), and mutations of this channel have been associated with deafness, epilepsy, and seizures (Buono et al., 2004; Ferraro et al., 2004; Lenzen et al., 2005; Rozengurt et al., 2003; Wangemann et al., 2004) in animals and humans. Knock-down of Kir4.1 by siRNA or conditional knock-out of this channel results in impaired glutamate clearance by astrocytes (Djukic et al., 2007; Kucheryavykh et al., 2007a) and in death of mice soon after birth (Djukic et al., 2007). Dramatic down-regulation of Kir4.1 channels and their rectification, specifically in Müller cells, are associated with vitreo- and chorio-retinal pathology including glaucoma (Francke et al., 1997), retinal detachment (Francke et al., 2005), ischemia (Pannicke et al., 2004, 2005), and diabetes (Pannicke et al., 2006).

The Kir channel blocker, spermine, is accumulated in glia rather than in neurons (Biedermann et al., 1998; Laube and Veh, 1997) and localized in the same cellular compartments together with Kir4.1 in Müller glial cells (Skatchkov et al., 2000, 2001). However, the amount of SP in the cytoplasm and the interaction between spermine and potassium with Kir4.1 channels remains an enigma. To begin a systematic study of the rectification properties of Müller cell Kir channels, we have made a direct comparison between Müller cell inward rectifier current and the current generated by recombinant Kir4.1 expressed in tsA201 cells. The results demonstrate very similar properties of the two currents, helping to resolve discrepancies in the literature regarding sensitivity of Kir4.1 to endogenous spermine (Biedermann et al., 1998; Oliver et al., 1998, 2000; Solessio et al., 2001), and pointing to novel features of spermine-induced rectification in Kir4.1 channels. Preliminary data of this study were reported earlier in abstract form (Antonov et al., 2003; Skatchkov et al., 2003).

MATERIALS AND METHODS

Müller Cell Isolation

Experiments were carried out with IACUC approval and in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Adequate measures were taken to minimize pain or discomfort to experimental animals. Müller cells were isolated from retinae of adult frogs (Rana pipiens) and rats (Long Evans or Sprague-Dawley) as previously described (Skatchkov et al., 1995, 2002, 2006). Central portions of retinae were cut in pieces (0.5 × 0.5 mm²) and rinsed in fresh Ca²⁺-Mg²⁺-free phosphate buffered solution (PBS, osmolarity was adjusted for each species), then were put into Ca²⁺-Mg²⁺-free PBS containing the protease Nagarse (Subtilisin, E.C.3.4.21.14, Protease type XXVII, Sigma; 0.01 mg/ml) for 30 min, at 23°C (for frogs) or in PBS-papain solution (papain 24 unit/mL) for 30 min, at 37°C (for mammalian retinae) or 23°C (for amphibian retinae). After washing and trituration in PBS and then in Leibovitz’ medium (L-15, Gibco), the slices were collected in a 1.5 mL microcentrifuge tube and briefly rinsed in L-15 containing 0.005 mg/mL DNase-I (Sigma, D-4263, from bovine pancreas). The cells were then washed in DNase-free L-15 medium, stored for 10 min on ice for sedimentation after which the supernatant was exchanged for fresh L-15. Slow centrifugation (500 rpm for 10 s, Eppendorf 5415 C, USA) was used to separate cells from debris, and the cell suspension was washed with fresh L-15 twice. This procedure yielded many Müller cells with their fine side processes well maintained.

Exogenous Expression of Kir4.1 Tagged with EGFP in tsA201 Cells

Cultures of tsA201 cells (an SV40 transformed variant of the HEK293 human embryonic kidney cell line) were prepared from frozen samples (generously provided by Dr. William Green). tsA201 cells were reconstituted and maintained in Dulbecco’s modified Eagle’s medium supplemented with 10 mmol/L glucose, 2 mmol/L l-glutamine, 10% fetal calf serum, and 200 iU/mL penicillin/200 μg/mL streptomycin at 37°C (5% CO₂ 95% air). Gene constructs: We used two constructs for these experiments: wild-type rat Kir4.1 cDNA and high-expression rat Kir4.1 cDNA with an insertion of the Kir2.1 sequence NSFCYENEVALT into the far C-terminus, between residues P351 and E352 (Stockklausner et al., 2001). Both were expressed in the pEGFP-C1 expression vector (Clontech), which generates EGFP fusion proteins with EGFP on the N-terminus. The included Kir2.1 sequence is in the distal C-terminal portion and affects trafficking of channels (Noel et al., 2005; Stockklausner et al., 2001), but does not line the channel pore and has no effect on pore properties. The Kir4.1 channel subunit clones were transfected into tsA201 cells using the calcium phosphate precipitation technique (Eertmoed et al., 1998). Recordings were made from cells 24–72 h after transfection. The chemicals for cultures and for transfections were purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO). Successfully transfected cells were visualized by EGFP-green fluorescence.

Electrophysiology

A recording chamber containing cells was placed on the stage of an inverted (Nikon 300; Nikon, Japan) or an upright (Zeiss, Jena, Germany) microscope. Membrane currents were measured with the single electrode whole cell patch-clamp technique at room temperature using an Axopatch-200B amplifier and a CV-203BU headstage (Axon Instruments, USA). Recording pipettes were pulled in four steps (using Sutter P-97 puller, USA) or in two steps (using Narashige P-83 puller, Japan) from hard glass (GC-150-10 glass tubing, Clark Electromedical Instruments, England and from GB-150-8P glass type tubing, Science Products GmbH, Germany). In whole-cell recordings, the cytoplasm was dialyzed by use of electrodes filled with intracellular solution (ICS) containing (in mM): KCl (or K-gluconate) 100 for amphibian and 130 for mammals, CaCl₂ 1, EGTA 10, HEPES 10, pH adjusted to 7.1 (with NaOH). The electrodes had resistances of 4–6 MΩ. Spermine 0.1–10 mM was added to ICS from 100 mM stock. The extracellular solution (ECS) contained (in mM): NaCl 115 for frog and 138 for tsA201 cells and for mammal Müller cells, CaCl₂ 2, MgCl₂ 1.9, and HEPES 10; KCl varied from 2.5 to 120 mM (substituted by NaCl to adjust osmolarity that was maintained at 240 mOsm for frog and 308 mOsm for rat and tsA201 cells). The signals were low pass filtered at 2 kHz and digitized at 5 kHz through a DigiData 1200A interface (Axon Instruments). After cell penetration, the access resistance was 10–15 MΩ and serial resistance was compensated by at least 75%. We used only the cells with membrane potential close to Nernstian. To ensure good space clamp in the cells with high Kir4.1 expression level, we used also small outside-out patches and cell attached patches with relatively small absolute currents that could be easily clamped. The pClamp 7 and 9 (Axon Instrument) software packages were used for data acquisition and analysis. To ensure that the analyzed currents were potassium currents, we performed control experiments where KCl was substituted with CsCl in the ICS and K⁺ in ECS was zero or when we used a mixture of bupivacaine (0.1–0.3 mM) and barium (0.1–0.2 mM). Under these conditions, no currents were recorded. The chemicals were purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO, and Taufkirchen, Germany).

Data Analysis

The currents observed with intracellular spermine (SP) in voltage-clamp experiments were described in the framework of the blocking equation

$$\text{O}+\text{SP}\underset{k_{\text{off}}}{\stackrel{k_{\text{on}}}{\leftrightarrow }}\text{OSP},$$ (1)

according to which a SP molecule from the inside can bind and unbind to a single site in the open channel (the state O in Equation 1) with the second-order (M⁻¹s⁻¹) rate constant k_on and the first-order (s⁻¹) rate constant k_off, respectively, and block channel conductance (the state OSP in Equation 1). The equilibrium dissociation constant, K₅₀, is defined as K₅₀ = k_off/k_on with both k_on and k_off being the functions of membrane voltage (V_m)

$$k_{\text{on}}\left(V_{\text{m}}\right)=k_{\text{on}}(0)\exp \left(V_{\text{m}}/a\right),$$ (2)

$$k_{\text{off}}\left(V_{\text{m}}\right)=k_{\text{off}}(0)\exp \left(V_{\text{m}}/b\right),$$ (3)

where k_on(0) and k_off(0) are the rate constants of binding and dissociation at V_m = 0 mV, and a, b are the shifts of membrane potential that induce an e-fold change of corresponding values.

To analyze a pure fraction of current that is sensitive to intracellular SP in the native Müller cells and tsA201 cells expressing Kir4.1 channels, the fraction of K⁺ current resistant to spermine action was subtracted from the total current, assuming a linear background conductance.

Membrane current through unblocked Kir4.1 channels (I_c) was defined as

$$I_{\text{c}}=\left(V_{\text{m}}-V_{\text{rev}}\right)\left\{G_{\text{min}}+\left(G_0-G_{\min }\right)\exp \left(V_{\text{m}}/c\right)\right\},$$ (4)

where V_rev is the reversal potential, G₀ is conductance at 0 mV, G_min is the minimal conductance (normalized to value of 1) and c is a curvature constant. However, in most cases the approximation of the current through the unblocked channels with linear function: I_c = G (V_m − V_rev) was sufficient enough for the adequate data description.

In accordance with Eq. (1), membrane current at equilibrium block (I_b) was defined as

$$I_{\text{b}}=I_{\text{c}}\left\{(1-A)/\left(1+[\text{SP}]/K_{50}\right)+A\right\},$$ (5)

where A is a fraction of unblocked current at infinitely high [SP] (A = 0 in the event of subtraction of SP-insensitive background current) and K₅₀ is a function of voltage defined as

$$K_{50}\left(V_{\text{m}}\right)=K_{50}(0)\exp \left(V_{\text{m}}/V_{\text{e}}\right)$$ (6)

with K₅₀(0) representing the SP equilibrium dissociation constant at 0 mV, and V_e is the voltage shift that causes the e-fold change of K₅₀(0).

In practice, Eq. (4) was fitted to data in the linear portion of the current-voltage (i/V) relationship to set free parameters (G₀ and c) of the function that describes I_c. Then Eqs. (5) and (6) were fit to all data in the i/V curve with K₅₀(0) and V_e as variables. The same fitting procedure was applied to the i/V relationships constructed from currents in response to the “short” and “long” voltage-step protocols to deduce K₅₀(0) and V_e parameters characterizing “fast” and “slow” spermine block.

Additional data analysis consisted in fitting the i/V relationship acquired with ramp voltage-clamp protocol to the product of the linear function describing current through unblocked channels, I_c, and two blocking functions of the type presented by Eq. (5): one with K₅₀(0), V_e, and A parameters for the “fast” spermine block and another one with similar parameters for the “slow” spermine block.

Data were analyzed using CLAMPFIT 7 and 9, Origin 6.1 and SigmaPlot 2001 data analysis fitting and plotting programs. All data are expressed as mean ± standard error of the mean (s.e.m.). We recorded from 196 Müller and 124 tsA201 cells and analyzed only those cells for which we completed the whole stimulating and pharmacological protocols (included in Tables 1 and 2).

TABLE 1.

The Effects of [K⁺]_o on the Parameters of “Fast”, Strongly Voltage-Dependent and “Slow,” Weakly Voltage-Dependent Blocks of the Müuller Cell Kir4.1 Channels by Intracellular SP (mean ± s.e.m.)

[K+]o (mM)	“Fast” block			“Slow” block
	K50(0) (M)	Ve (mV)	n	K50(0) (mM)	Ve (mV)	n
1	3.0 ± 1.6	−14.6 ± 0.5	7	4.0 ± 0.8	−25.9 ± 6.2	7
3	110.5 ± 32.8	−11.9 ± 0.4	17	1.7 ± 1.0	−60.4 ± 7.3	15
5	1,142.1 ± 429.2	−10.5 ± 0.3	11	4.0 ± 1.6	−44.4 ± 18.6	9
10	65,863.4 ± 6,503.7	−9.2 ± 0.7	9	2.7 ± 1.2	−37.6 ± 9.4	6

TABLE 2.

Parameters of the “Fast” and “Slow” Block by Intracellular SP of Kir4.1 Channels Expressed in tsA201 cells (mean ± s.e.m.) and their Dependence on [K⁺]_o

[K+]o (mM)	“Fast” block			“Slow” block
	K50(0) (M)	Ve (mV)	n	K50(0) (mM)	Ve (mV)	n
1	1.1 ± 0.5	−11.3 ± 0.8	4	7.5 ± 4.5	−52.9 ± 12.7	4
3	56.0 ± 7.9	−10.6 ± 0.7	12	4.8 ± 1.1	−55.2 ± 3.9	12
5	423.2 ± 161.4	−8.9 ± 0.9	6	4.1 ± 1.9	−49.9 ± 17.5	7
10	36,784.2 ± 4,302.8	−9.1 ± 1.2	4	3.7 ± 1.8	−61.3 ± 12.6	11

RESULTS

Intracellular SP Blocks a Fraction of Müller Cell K⁺ Channels

We initially examined rectification properties of Müller cells using whole-cell voltage-clamp recordings. It is well known that in retinal center almost 95% of Kir currents (Newman, 1993; Skatchkov et al., 1995) and most of Kir4.1 channel subunits (Pannicke et al., 2004; Skatchkov et al., 2001) of frog, salamander, and rat Müller cells are concentrated in the narrow cell compartment called the endfeet. Therefore, these cells isolated from central, rather than peripheral, retina of amphibians and mammals were clamped exclusively in endfeet to record Kir4.1 currents. Immediately after attainment of the whole-cell configuration, voltage-clamp pulses containing linear ramp portions spanning from −100 to +100 or +160 mV revealed biphasic rectification in both frog and rat Müller cells (see Fig. 1). There was a shallow phase of rectification near physiological potentials (~ −50 mV, Fig. 1, white arrowheads), and then a steep phase above (~ +50 mV, Fig. 1C, black arrowheads). With time, rectification gradually disappeared, such that within about 2 min, the currents were essentially linear, if the cells were dialyzed with zero spermine in the pipette solutions (Figs. 1A,C-left panel). Independent of the type of animals, this biphasic rectification is clearly visible in Müller cells, specifically when linear “leakage channels,” 2P-domain TASK-type, are blocked by bupivacaine (0.2 mM) (Fig 1B, left panel). After co-application of extracellular bupivacaine (0.2 mM) and barium (0.2 mM), there was practically no residual current recorded (Fig. 1B right panel) demonstrating the exclusive K⁺-nature of the currents in these glial cells. In these studies, Ba²⁺ was used to block Kir channels and bupivacaine was used to block TASK channels in Müller cells (Skatchkov et al., 2006). The results demonstrate that the i/V curve in Müller cells after bupivacaine is biphasic (Fig. 1B left panel) and is very similar to the i/V curve recorded from tsA201 cells (Figs. 7C,D) heterologously expressing Kir4.1. In both types of cells, complex SP block of Kir current began, first, near the resting membrane potentials (white arrowheads) and the second phase of SP block took place at depolarization (black arrowheads) (Figs. 1 and 7).

Fig. 1.

Complex rectification of Kir currents in glial Müller cells. A: Recordings from an amphibian (frog, Rana pipiens) retinal glial Müller cell. Rapid loss of rectification (relief of endogenous block) is observed in the Müller cell during time-dependent washout of the cytoplasm by a patch pipette without spermine (SP). The i/V curves have been transformed from double curved shape (white and black arrowheads) to almost linear in ~110 s demonstrating complete block relief by dialysis. The cell had a membrane potential of −60 mV in [K⁺]_o = 5 mM and was clamped at a holding potential of −50 mV. The dotted line shows zero current. Stimulation was achieved by voltage ramps (the protocol is shown). B: Recordings from a Long Evans rat Müller cell. Left panel: Biphasic rectification (black and white arrowheads) in a cell dialyzed with 0.1 mM SP in control solution (control, wash) and in a solution containing the TASK channel blocker bupivacaine (0.2 mM). Stimulation was achieved by voltage ramps (the protocols are shown). Each curve represents an average of four responses to ramp stimuli. Pronounced biphasic rectification denoted by white and black arrowheads was observed in “control” and in “wash” recordings. The biphasic block became more evident when linear “leakage currents” through 2P-domain TASK channels were depressed by bupivacaine. The cell had a membrane potential of −86 mV in [K⁺]_o = 3 mM and was clamped at a holding potential of −50 mV. Dotted lines show zero current. Right panel: Recording from the same cell as in left panel using the same protocol. Nearly complete block of all K⁺-currents were observed in the solution containing bupivacaine 0.2 mM and Ba²⁺ 0.2 mM to block both TASK and Kir channels, respectively. The curve represents ten averaged recordings. C: Recordings from Sprague-Dawley rat Müller cells, dialyzed with 0, 0.03 and 0.3 mM SP in the patch pipette showing that both, the level of rectification and biphasic type of rectification, depend upon the SP concentration in the cytoplasm established after 120 s of cell dialysis. White arrowheads point to residual rectification that is in physiological area, near the resting membrane potential of glial cells. Rectification disappeared after longer dialysis without SP (also shown in Fig. 2). Dotted lines show zero current. The cells (from left to right) had membrane potentials of −77, −70, and −76 mV and were clamped at −50 mV in [K⁺]_o = 5 mM. The protocol is shown and the first records were obtained 30 s after cell opening (ex. black arrowhead) and repeated each 30 s. Note: even 300 μM SP has less blocking effect than endogenous block (first ramp response in right recording, black arrowhead), showing that probably the concentration of free endogenous SP in freshly isolated cells is much higher than 300 μM. The model used to fit i/V curves (see methods) predicts that the free concentration of SP in these cells is 0.80 ± 0.24 mM (n = 6).

Fig. 7.

Biphasic block in tsA201 cells expressing Kir4.1 channels. A: Recordings from a control tsA201 cell which does not express inward K⁺-currents (the currents below the dotted line are negligible). The pattern of the i/V curves remains unchanged during the time of cytoplasmic dialysis by the pipette with SP = 0. These outward currents represent SP-insensitive background (“leakage”) currents which were also insensitive to the Kir-channel blocker Ba²⁺ (0.1 mM, data not shown) and were considered background; they were subtracted in (C) to obtain pure Kir4.1-mediated current in Kir4.1-expressing cells. The cell had a membrane potential of −51 mV in [K⁺]_o = 5 mM and was clamped at a holding potential of −50 mV using a patch pipette that did not contain spermine (SP), Ca²⁺ or Mg²⁺. A voltage ramp protocol was used (insert). B: Whole cell recording from a tsA201 cell transfected with cDNA encoding for Kir4.1 channels. In contrast to control cell in A, robust inward Kir-currents are expressed (below the dotted zero-current line) and outward currents have biphasic rectification shape (denoted by white and black arrowheads). This rectification is “straightening” (as in Müller cells, Fig. 1) when intracellular SP is washed out during cell dialysis in ~120 s using a patch pipette without spermine. The cell had a membrane potential of −78 mV in [K⁺]_o = 5 mM and was clamped at a holding potential of −50 mV using a patch pipette that did not contain SP, Ca²⁺ or Mg²⁺. A voltage ramp protocol was used (see insert). C: Outside-out patch clamp recording from a tsA201 cell transfected with cDNA encoding for Kir4.1 channels; background, Ba²⁺-insensitive (0.1 mM Ba²⁺) current was subtracted. The trace represents an average of 10 responses to ramp stimuli shown in the insert. Pronounced biphasic rectification (denoted by white and black arrowheads) demonstrated two rectification processes that have been previously shown in Müller cells (Fig. 1B, left panel). D: Cell-attached patch-clamp recording from a tsA201 cell (3 mM K⁺ in the bath and pipette) transfected with modified cDNA encoding Kir4.1 channel that was used for high density Kir4.1 expression to reduce the contribution of the background current (see Methods). The graph shows the i/V relationship of Kir4.1 derived from the voltage ramp portion of the current through cell-attached patch (open circles, an average of 9 original traces in the insert, stimulation by voltage ramp as in (C) and its fit to the biphasic SP block model (solid line). The model (see in Methods) uses two sites in the channel with different sensitivity to SP and voltage. The white arrowhead points to the area of high affinity, weakly voltage-dependent SP block (“Slow” block), and the black arrowhead points to the low affinity, steeply voltage-dependent SP block (“Fast” block). The dashed line corresponds to the Goldman-Hodgkin-Katz type of i/V through the unblocked open channel used in the model. The biphasic block demonstrates similar Kir4.1 channel behavior as in (C), however under conditions of undisturbed intracellular environment in tsA201 cell. E: Whole cell recording from a tsA201 cell transfected with cDNA encoding Kir4.1 channels using voltage-steps. The membrane potential was −74 mV in [K⁺]_o = 5 mM. The cell was clamped at a holding potential of −60 mV using patch pipette that did not contain SP, Ca²⁺ and Mg²⁺. The recording was obtained after 120 s of cytoplasmic dialysis. Voltage step protocol: an increment by 5 mV from −60 mV. F: Whole cell recording from a tsA201 cell transfected with cDNA encoding Kir4.1 channels using a patch pipette with 300 μM spermine. The recording was obtained after 120 s of dialysis with SP. The membrane potential was −77 mV and the holding potential was −60 mV. Voltage step protocol: an increment by 5 mV from −60 mV. Block of Kir4.1 outward currents by SP is similar to that seen in Müller cells (see Fig. 2) representing “fast” block.

To examine the possibility that the loss of biphasic rectification is due to gradual depletion of polyamines from the cell interior we dialyzed cells with variable spermine concentrations in the pipette ([SP]_in). As [SP]_in was increased, loss of rectification was retarded and with 300 μM [SP]_in, both phases of rectification were still evident at 2 min of dialysis (Fig. 1C, right panel). Intriguingly, 300 μM [SP]_in in the pipette had less blocking effect than endogenous block (black arrowhead in Fig. 1C-right panel, marking the first ramp response immediately after cell opening), suggesting that the concentration of endogenous SP in freshly isolated rat cells is higher than 300 μM. Using our model (modified Woodhull’s model; see Methods) and fitting i/V curves recorded in freshly opened cells (before dialysis), we allowed SP concentration to vary during the fitting procedure until the experimental and model curves matched. We found that the native free [SP]_in in Müller glial cells was near 800 μM (0.81 ± 0.24 mM; n = 6).

Since the kinetics of the two components of spermine block appeared quite different (white and black arrowheads, Fig. 1B-left panel), we suggest that they represent two separate blocking processes in Kir4.1 channels. Therefore, we further characterized inward rectification of Müller cells using step voltage-clamp pulses (Figs 2–6). With time after dialysis with zero SP in the pipette, the current gradually became essentially linear (Fig. 2A). However, in the presence of even relatively low [SP]_in (0.3 mM), a very steep rectification was induced at depolarization (Fig. 2B), with a saturating time constant of fast current decay, τ_fast,on ~20 ms. Recovery from this rectification was rapid; the “off” transition induced complete unblock of current within ~20 ms (Fig. 2B, left recordings). Using long voltage-clamp pulses (8 s every 10 s), rectification was again apparent in the presence of SP (Fig. 2B, right panel). However, currents did not recover during the 2s interpulse interval, such that the next “on” transition started from a slightly different value of inward current (Fig. 2B, right panel). This indicates a slow blocking and unblocking process, in addition to the fast block and unblock revealed by the short pulse protocol.

Fig. 2.

Different voltage-jump protocols (short and long) induce block of K⁺ current by [SP]_i through frog Müller cell Kir channels that develop with different kinetics. A: Recording after 5 min of cytoplasmic dialysis with zero [SP]_i. Left panel: Traces of K⁺ current induced by brief (80 ms) depolarizing voltage-steps with an increment of 10 mV (the short protocol) that start from −60 mV and reach the maximal depolarization of 280 mV recorded in [K⁺]_o = 10 mM. Complete relief of block is established. Right panel: Traces of K⁺ current induced by long (8 s) depolarizing voltage-jumps with an increment of 10 mV (the long protocol) that start from −100 mV and reach the maximal depolarization of 290 mV recorded in [K⁺]_o = 10 mM. The i/V relationship is near linear. B: Recording in the presence of 0.3 mM [SP]_i. Left panel: Traces of K⁺ current induced by brief (80 ms) depolarizing voltage-steps with an increment of 5 mV depolarizing voltage-steps (the short protocol) that start from −60 mV and reach the maximal depolarization of 240 mV recorded in [K⁺]_o = 10 mM. For illustrative purposes currents recorded in an intermediate voltage range within which the i/V relationship is linear were omitted from this panel. Right panel: Traces of K⁺ current induced by long (8 s) depolarizing voltage-jumps with an increment of 5 mV (the long protocol) that start from −100 mV and reach the maximal depolarization of 140 mV recorded in [K⁺]_o = 10 mM. C: Current-voltage (i/V) relationships measured at the end of depolarizing pulses. Triangles, measurements from recordings illustrated in B (left panel with [SP]_in 0.3 mM). Circles, measurements from recordings illustrated in B (right panel). Notice that both protocols eventually block of the same fraction of total current (dotted line). The residual currents below dotted line are due to spermine insensitive K⁺ current likely via 2-domain pore potassium channels (Skatchkov et al., 2006) and other channels. Black arrowhead represents fast block while white arrowhead points to slow block. D: The i/V relationships shown in B corrected for the fraction of K⁺ current which was not blocked and, presumably determined by K⁺ channels that are not sensitive to intracellular SP, like tandem-pore domain (TASK-1 and TASK-2) K⁺ channels (see Discussion). Triangles show the short protocol measurements and circles show the long protocol measurements. Solid lines are fits to the data with the block parameters K₅₀ (0) and V_e (the value of voltage shift which causes the e-fold change of K₅₀ (0)) of 17966 M and −8.1 mV (low affinity (fast) block, black arrowhead), and 0.47 mM and −37.3 mV (high affinity (slow) block, white arrowhead) for short and long protocols, respectively. All data were obtained from the same experiment.

Fig. 6.

Effects of decreasing intracellular K⁺ concentration ([K⁺]_i) on fast, strongly voltage-dependent and slow, weakly voltage-dependent block by [SP]_i of frog Müller cell Kir channels. The i/V relationships measured by fast (squares) and slow (circles) protocols with 1 mM [SP]_i and 12 mM [K⁺]_i in 1 mM [K⁺]_o. Solid lines are fits to the data. For the fast, strongly voltage-dependent block (squares) K₅₀ (0) equals 0.018 M and V_e equals −12.38 mV. For the slow, weakly voltage-dependent block (circles) K₅₀ (0) equals 0.32 mM and V_e equals −34.38 mV. Slow block is strongly dependent on [K⁺]_i (compare with triangles in Fig. 4B where [K⁺]_i is 120 mM).

The i/V relationships measured at the end of short (triangles) and long (circles) pulse protocols are shown in Fig. 2C. At the most positive voltages, the steady-state currents are the same, indicating that both fast and slow processes block the same fraction of K⁺ current. The data were interpreted within the framework of an open channel block model [Eq. (1)], wherein intracellular SP blocks outward current through a fraction of K⁺ channels with a spermine-insensitive time- and voltage-independent component of K⁺ conductance. This may be due to some intrinsic processes of K⁺-channels, such as TASK (Skatchkov et al., 2006) and small leakage through the unblocked fraction of Kir4.1 (Kucheryavykh et al., 2007b). If such spermine-insensitive time- and voltage-independent component of K⁺ conductance was subtracted, two different components of SP block, fast and slow, were evident (Fig. 2D). Indeed, the kinetics of spermine block is consistent with two separate blocking processes.

“Fast” Block of Müller Cell Kir Channels: Characteristics of SP Block

Figure 3A shows dependences of current amplitude on V_m, as measured by the short protocol, with different [SP]_in. i/V relationships are shown for three different cells, all in the presence of 5 mM [K⁺]_o. Elevation of [SP]_in resulted in a considerable leftward shift of the voltage-dependence of spermine block. Tenfold increase of [SP]_in resulted in ~−40 mV shift in the break-point (arrowed) of the i/V relationship. The K₅₀(0) and V_e values obtained by fitting Eqs. (5) and (6) to the data were very similar (Fig. 3A legend). When conductances in the negative voltage region were normalized (Fig. 3A, insert), the outward current at low positive V_m in the presence of 10 mM [SP]_in was reduced, suggesting that at this higher [spermine], slow block may already contribute to measurements made with the short protocol. Clearly the fast block is [SP]_in dependent, indicating that it is determined by SP interaction with some binding site in the ion pathway. In the simple Woodhull (1973) model of open channel block [Eq. (1)], only the on-rate (k_on) should depend on [SP]_i. The decay of current was fitted by a single exponential function. Block is increased with increasing [SP]_in (from 0.1 to 10 mM, Fig. 3A) and the rates of block onset increase linearly with increasing [SP]_in and with depolarization (Fig. 3B).

Fig. 3.

Dependence of block of K⁺ current through the frog Müller cell Kir channels by intracellular SP induced by the fast protocol. A: i/V relationships normalized to the maximal current (I_max) value obtained in the presence of 5 mM [K⁺]_o and different [SP]_i. Arrows denote the maximal current amplitude in the i/V relationships. Circles, the i/V curve in the presence of 0.1 mM SP. Squares, the i/V curve in the presence of 1 mM [SP]_i. Upward triangles, the i/V curve in the presence of 10 mM SP. Solid lines are fits of the data points with K₅₀ (0) and V_e equal to 712.9 M and −10.0 mV, 324.8 M, and −10.1 mV, and 297.8 M and −11.0 mV for 0.1 mM, 1 mM and 10 mM [SP]_i, respectively. The same i/V curves represented in the absolute current scale (I_m, nA) are shown in the box. Data for each of [SP]_i-s were obtained from different experiments. B: Dependence of the rate of block onset induced by the short protocol on [SP]_i and membrane potential in the presence of 5 mM [K⁺]_o. Each point represents the mean value of two measurements on different cells. Filled circles show measurements at 220 mV; filled downward triangles at 200 mV; filled squares at 180 mV. Straight lines are linear regressions through the data points (the slope of regression lines indicates the rate constant of SP binding to the open channel (k_on)).

Slow Block of Kir Channels in Müller Cell by Intracellular SP

A parallel analysis was performed on currents recorded with the long pulses (see Fig. 4). While block measured by the long protocol in Fig. 4A reveals all of the characteristics (Fig. 4B) described in Fig. 2, a prepulse to −100 mV now induced complete unblock, so that full current amplitudes were resolved during the pulse. Intriguingly, the K₅₀ of slow block is ~5 orders of magnitude more effective than the fast block (Figs. 4C,D) and ~eight times less voltage-dependent (Table 1).

Fig. 4.

Slow, weakly voltage-dependent block of frog Müller cell Kir channels by intracellular SP is not affected by changing the [K⁺]_o: a difference from fast block (see Fig. 5). A: Representative recordings of K⁺ current evoked by the long protocol using 1 mM [SP]_i and 120 mM [K⁺]_i in the presence of 1 mM or 10 mM [K⁺]_o. Depolarizing steps with 5 mV increments were applied from −100 mV to +140 mV. Currents are shown in the same current scale. B: i/V relationships of currents normalized to the maximal current (I_max) shown in A. Downward triangles, the i/V curve in 1 mM [K⁺]_o. Circles, the i/V curve in 10 mM [K⁺]_o. Solid lines are fits to the data with parameters: K₅₀ (0) equals 5.7 mM and V_e equals −28.3 mV for 1 mM [K⁺]_o, and K₅₀ (0) equals 1.7 mM and V_e equals −33.5 mV for 10 mM [K⁺]_o. The same i/V curves represented in the absolute current scale (I_m, nA) are shown in the insert. Slow block is not dependent on [K⁺]_o. C: Comparison of the currents recorded by applying the short (left) and long (right) protocols with 0.1 mM [SP]_i in the presence of 100 mM [K⁺]_o. Fast block is strongly dependent on [K⁺]_o and this is further clarified in Figure 5. The amplitude calibration is the same for the left and right recordings with different time scale. D: i/V relationships measured at the end of recordings shown in C. Squares, measurements performed by applying the short protocol. Diamonds, measurements performed by applying the long protocol. Solid lines are fits to the data. The slow block is characterized by parameters K₅₀ (0) = 0.48 mM and V_e = −37.8 mV.

These data suggest that in Müller cells intracellular SP blocks K⁺ channels, presumably of Kir4.1 type, by two mechanisms. One is a fast, strongly voltage-dependent block during which the SP molecule binds to a low-affinity binding site located deep within the pore. The other is a slow, weakly voltage-dependent block during which the SP molecule interacts with a shallow, but high-affinity binding site. In further experiments, we tested whether these sites differ with respect to their accessibility for K⁺ from the outside and inside.

Different Effects of External and Intracellular K⁺ on Fast and Slow SP Block

In contrast to the fast, strongly voltage-dependent block, the slow, weakly voltage-dependent block was insensitive to [K⁺]_o. Figure 4A shows currents recorded using the long protocol: in both 1 and 10 mM [K⁺]_o, the peak current (break point) was achieved at ~+30 mV. Figure 4B illustrates that whereas the conductance within the linear region of the i/V relationships increases with increased [K⁺]_o, the data overlap within the region of negative slope conductance. In 100 mM [K⁺]_o, slow block still develops even with lowest [SP]_in tested (Fig. 4C right panel), while the fast block was not observed in this case (Fig. 4C, left panel and Fig. 4D, squares). It was practically impossible to calculate K₅₀ (0) in Müller cells for fast block at high [K⁺]_o 100 mM because all cells demonstrated almost complete loss of spermine block at high [K⁺]_o. However, changing [K⁺]_o between 1 and 100 mM had negligible effect on either the K₅₀(0) or the V_e values of SP binding to the shallow blocking site (slow block, Table 1).

Figure 5A shows currents recorded with the short voltage protocol in the presence of 5 mM and 10 mM [K⁺]_o, from the same cell. In 5 mM [K⁺]_o, the current increased with depolarization and maximal current (I_max) was obtained at −90 mV. Further depolarization led to decreasing currents which saturated at ~+140 mV. Increased [K⁺]_o resulted in the expected positive shift of the reversal potential and increase of conductance within the linear portion of the i/V relationship (Fig. 5B). Remarkably, increasing [K⁺]_o in the physiological range led to a considerable weakening of the fast block (Fig. 5B). Increasing [K⁺]_o (between 1 and 10 mM) shifted the reversal potential by about 50 mV toward depolarization, with an even larger (~100 mV) shift in the break point (Fig. 5B), although in this particular experiment, the slope of the negative conductance region of the i/V curves did not change. The relieving effect of external K⁺ was such that when [K⁺]_o was increased to 100 mM, no fast block was observed (Fig. 4C, left panel) with the short voltage protocol, even up to 200 mV depolarization. Table 1 summarizes the data from these experiments for both, fast and slow, blocks. Fast block is characterized by a tremendous increase of the K₅₀(0) value for SP with increasing [K⁺]_o, while the effect on the V_e value is minimal, although there is some tendency of V_e to decrease with elevated [K⁺]_o.

Fig. 5.

Dependence of fast, strongly voltage-dependent block of frog Müller cell Kir channels on [K⁺]_o. A: Representative recordings of K⁺ current evoked by the short protocol using 3 mM [SP]_i in the presence of 5 mM or 10 mM [K⁺]_o. Currents are shown in the same current scale. For illustrative purposes, currents recorded during some intermediate voltage steps within the membrane potential range at which the i/V relationship is linear were omitted from this panel. B: i/V relationships normalized to the maximal current (I_max) obtained with 3 mM [SP]_i in the presence of different [K⁺]_o. Arrows denote the maximal current amplitude obtained in each [K⁺]_o. Downward triangles, the i/V curve in 1 mM [K⁺]_o; upward triangles, the i/V curve in 3 mM [K⁺]_o; squares, the i/V curve in 5 mM [K⁺]_o; circles, the i/V curve in 10 mM [K⁺]_o. Solid lines are fits to the data with K₅₀ (0) and V_e values equal to 0.44 M and −13.8 mV, 34.7 M and −11.3 mV, 574.4 M and −9.6 mV, and 825.6 mM and −11.5 mV in 1 mM, 3 mM, 5 mM, and 10 mM [K⁺]_o, respectively. The same i/V curves represented in the absolute current scale (I_m, nA) are shown in the box. All data were obtained from the same experiment.

We also examined the effect of lowering intracellular K⁺ concentration ([K⁺]_i) on the parameters of fast and slow block. Figure 6 shows an example of the difference between fast and slow blocks in a Müller cell when intracellular potassium is decreased to 12 mM. In cells with high [100 mM (see Fig. 5)] or low [12 mM (see Fig. 6)] intrapipette [K⁺]_i the parameters of fast SP block were similar (compare Fig. 6-squares with Fig. 5B-triangles), indicating no major effect of intracellular potassium on fast block. For example, when [K⁺]_i = 12 mM and [K⁺]_o = 1 mM, fast block has K₅₀ (0) = 21.08 ± 12.68 M and V_e = −12.22 ± 0.12 mV (n = 3) that is similar to the parameters when [K⁺]_i = 100 mM and [K⁺]_o = 1 mM (Table 1).

In contrast, lowering [K⁺]_i resulted in a substantial increase of slow block efficiency, (~25×, circles in Fig. 6): in 12 mM [K⁺]_i the parameters of SP interaction with the shallow site were K₅₀ (0) = 0.163 ± 0.090 mM and V_e = −31.72 ± 1.36 mV (n = 7), compared to K₅₀ (0) = 3.96 ± 0.84 mM and V_e = −25.92 ± 6.22 mV (n = 4) in 100 mM [K⁺]_i. Therefore, during binding to the shallow (cytoplasmic) site, but not the deep binding site, SP molecules apparently sense intracellular K⁺ ions.

Thus, the two blocking processes differ not only with respect to voltage- and SP-sensitivities, but also in respect to their sensitivity to extra- and intracellular K⁺.

Rat Kir4.1 channels expressed in tsA201 cells reveal both types of SP block

The above results were all obtained on Müller glial cells and to determine whether the above K⁺ current blocking processes are reflected in the properties of Kir4.1 channels found in glial cells, we directly examined the effect of intracellular SP on rat Kir4.1 channel currents expressed in tsA201 cells. No endogenous inward rectification was observed in these cells, in fact only small outward currents that did not change significantly with dialysis (Fig. 7A) were observed. In contrast, tsA201 cells expressing Kir4.1exhibited inward rectification (Fig. 7B) that was very similar to that seen in frog (Fig. 1A) and in rat Müller cells (Fig. 1C, left panel). Using the ramp voltage-clamp protocol, this was manifested as a weak rectification around the reversal potential and a steep decline in current above 120 mV (Fig. 7B). With time after dialysis without SP in the intrapipette solution, both components of rectification gradually disappeared (Fig. 7B). Obviously, total cell currents contain Kir4.1 and “leakage” currents together and we further subtracted background “leakage” currents in Kir4.1-tsA201 cells using (i) outside-out patches to improve voltage-clamp and (ii) 0.1 mM barium (to block Kir4.1 channels) to obtain “leakage” current. Thus residual Ba²⁺-insenstitive leakage current was subtracted from the recording of total currents and Fig. 7C shows an average of 10 responses to ramp stimuli (shown in the inset) representing Kir4.1 current isolated from “leakage” currents (ex. demonstrated in Fig. 7A). Figure 7C shows the pronounced biphasic rectification of Kir4.1 channels denoted by white and black arrowheads. This demonstrates the two rectification processes that have been previously shown in Müller cells (Fig. 1B, left panel).

Furthermore, biphasic block is evident when background channels are negligible after high density expression of Kir4.1 (Fig. 7D). We used cell attached recordings: (i) to prevent any intervention of cytoplasmic content and (ii) to minimize voltage clamp problems. With a pipette containing normal ECS ([K⁺]_o = 3 mM), we obtained biphasic block. An insert in left upper corner of Fig. 7D represents 20 averaged responses to repeated ramp stimulations from single cell attached patch clamp recording. Then nine averaged recordings from nine different cells were again averaged (open circles) and fitted using our model with low and high affinity sites for SP (solid line, Fig. 7D). Thus, the biphasic i/V curve in tsA201 cells expressing Kir4.1 channels is very similar to the i/V curve of outside-out patches (Fig. 7C) and of Müller cells (Fig. 1B).

Using the short voltage pulse protocol, Kir4.1 currents were essentially time-independent after dialysis with zero SP (Fig. 7E), but time-dependent Kir4.1-rectification was also similar to Müller cells, when SP was included in the pipette (Figs. 7F and 8A). Following depolarization, the decay of currents was well fitted by a single exponential function (Fig. 8A, bottom panel). As in the case of Müller cells, elevated [K⁺]_o shifted the negative slope conductance region toward more positive potentials (Fig. 8B), suggesting that the block is dependent on [K⁺]_o (Table 2), again without changing the voltage dependence of block (Table 2). The effect of [K⁺]_o was at least partially dependent on k_on, because in 3 mM [K⁺]_o at 85 mV the rate of relaxation to the equilibrium block was substantially lower than in 1 mM [K⁺]_o (Fig. 8A, lower panel). As in the case of Müller cells, recovery from fast block was complete within 100 ms of return to the holding potential. Apart from the slightly slowed kinetics, the parameters and voltage-dependence of fast block in frog Müller cells and in tsA201 cells expressing rat Kir4.1 channels were very similar (Table 2), suggesting that the fast, strongly voltage dependent block of Müller cell currents by intracellular SP reflects an inherent property of Kir4.1 channels.

Fig. 8.

Fast and slow blocks of Kir4.1 channel expressed in tsA201 cells by SP. A: Fast, strongly voltage-dependent block of K⁺ currents were induced in a tsA201 cell expressing Kir4.1 channels by a depolarizing pulse protocol (shown below the records with 5 mV steps) with 10 mM [SP]_i in the presence of 1 mM or 3 mM [K⁺]_o. Currents are shown in the same current scale. Current relaxations from the maximum amplitude to 10% of decay induced by the voltage jump to 85 mV and represented in the semi-logarithmic plot are shown below. Solid line, the decay in 1 mM [K⁺]_o; dotted line, the decay in 3 mM [K⁺]_o. Both decays can be well fitted by a single-exponential function with the decay time constants (τ) of 239.6 ms and 648.6 ms for 1 mM and 3 mM [K⁺]_o, respectively. B: i/V relationships normalized to the maximal current (I_max) value measured at the end of depolarizing pulses from records shown in (A). Arrows denote the maximal amplitude (I_max) of current. Downward triangles, in the presence of 1 mM [K⁺]_o; circles, in the presence of 3 mM [K⁺]_o. Solid lines are fits to the data with the parameters: K₅₀ (0) = 37.4 M and V_e = −5.4 mV, and K₅₀ (0) = 411.6 M and V_e = −6.1 mV for 1 mM and 3 mM [K⁺]_o, respectively. The same i/V curves replotted in an absolute current amplitude scale (I_m (nA)) are shown in box. C: Slow, weakly voltage-dependent, block of K⁺ currents by SP in a tsA201 cell expressing Kir4.1 channels.K⁺ currents were induced in the tsA201 cell by the depolarizing protocol (shown below the records with 5 mV steps) with 10 mM [SP]_i and in the presence of 5 mM [K⁺]_o. D: The i/V relationship (circles) measured at the end of depolarizing pulses from the currents shown in (C). Solid line, fits to the data with K₅₀ (0) equals to 7.0 mM and V_e equals to −24.0 mV demonstrate the similarity of slow block features between tsA201 cells expressing Kir4.1 and Müller cells (Fig. 4 and Table I).

Use of a long protocol revealed an additional slow component of block in Kir4.1-expressing, but not control, tsA201 cells (Fig. 8C). This block also featured the characteristics that were typical for the slow block in Müller cells: it developed slowly even at 10 mM [SP]_i, the unblock was also slow so that a next depolarizing step started from a decreased current amplitude; it was weakly voltage dependent (Fig. 8C) and was characterized by a relatively high affinity for SP. The parameters of slow block were K₅₀(0) = 4.1 ± 1.9 mM with V_e = −49.9 ± 17.5 mV (n = 7)), which were not much different from the characteristics of the slow, weakly voltage dependent block obtained in Müller cells (K₅₀(0) = 4.0 ± 1.6 mM with V_e = −44.4 ± 18.6 mV (n = 9)) taken, for example for external [K⁺] = 5 mM. The complete set of parameters defining “slow” and “fast” block derived from whole-cell current recordings in Kir4.1-expressing tsA201 cells are presented in Table 2. Similar results were obtained from Müller cells and are presented in Table 1.

In summary, rat Kir4.1 channels reveal two types of block by SP that are similar in their basic characteristics to the blocking processes in Müller cells. The observations suggest that these two processes represent an inherent property of Kir4.1, and that the channel may contain two functionally distinct binding sites for SP.

DISCUSSION

Rectification Properties of Müller Glial Cells

Many studies have looked at the rectification properties of the membrane conductance of glial cells in general, and of Müller cells in particular, with widely varying results (Biedermann et al., 1998; Bringmann et al., 2006; Newman, 1993; Skatchkov et al., 1995, 2000, 2006; Solessio et al., 2000, 2001). Early studies (Biedermann et al., 1998; Conner et al., 1985; D’Ambrosio et al., 1998; Ransom and Sontheimer, 1995; Sontheimer and Kettenmann, 1988; Sontheimer, 1994) also reported widely varying rectification properties between different oligodendrocytes, astrocytes and Müller cells. In one of the earliest studies of Müller glial cells from rabbit, Nilius and Reichenbach (1988) reported the presence of differentially rectifying channels in various regions of the cell. Single channel conductance in Müller cells of about 25–40 pS (Biedermann et al., 1998; Ishii et al., 1997; Newman, 1993; Tada et al., 1998) was close to that of Kir4.1 channels (Oliver et al., 1998, 2000; Tada et al., 1998). Indeed, the Kir4.1 subunit is widely expressed in glial cells in the CNS and specifically in Müller retinal glial cells (Kofuji et al., 2000, 2002; Poofpalasundram et al., 2000; Raap et al., 2002; Skatchkov et al., 2001). Relatively strong rectification was recorded by Newman (1985, 1993) and Skatchkov et al. (1995), both in intact amphibian (salamander and frog) Müller cells, and in cells with endfeet removed, although the currents were much larger in endfoot processes. Indeed, Kir4.1 is expressed in endfeet, and a reduced amount of Kir4.1channels was found in Müller glial cells after ischemia (Pannicke et al., 2004, 2005), in diabetes (Pannicke et al., 2006) as well as during human retinal pathology (Francke et al., 1997) showing the extreme importance of this channel type. The expression of Kir4.1 channels, in association with high levels of spermine and spermidine in glial cells (Laube and Veh, 1997; Laube et al., 2002) and particularly in Müller glial cells (Biedermann et al., 1998; Skatchkov et al., 2000), are intriguing features. Kir4.1 and SP are colocalized, and the function of Kir currents correlates with SP content in cell compartments (Skatchkov et al., 2000).

The total concentration of intracellular SP in different cells is estimated at about 3–10 mM (Gilad and Gilad, 1991; Seiler, 1994; Watanabe et al., 1991), although the free SP concentration deduced from functional tests based on its ability to “rectify” receptors and channels tends to be much smaller, ~10–200 μM (Bowie and Mayer, 1995; Fakler et al., 1995; Haghighi and Cooper, 1998). Thus, SP is greatly buffered in the cytoplasm. Phosphates, ATP, GTP (Watanabe et al., 1991), nucleotides, RNA and DNA, and membrane proteins were found to be intracellular buffers (Ingoglia, 1982; Watanabe et al., 1991), but little is known about how SP is liberated from such buffers.

In a study of turtle Müller cells, Solessio et al. (2000, 2001) reported variable rectification, similar to that which we find here, depending on the voltage protocol: With brief pulses, a rather weak rectification which shifted significantly with increasing [K⁺]_o, but with longer voltage protocols, rectification appeared to be much steeper, superimposed on a linear K⁺ leak. This leak is due to spermine- and magnesium-insensitive, noninactivating linear K⁺-currents in Müller cells (Biedermann et al., 1998; Skatchkov et al., 2006; Solessio et al., 2000, 2001). These currents are also not voltage-dependent and not consistent with K_A, K_D, or BK channels as they were recorded in the presence of Kv and BK blockers and most likely represent 2P-domain TASK-type channels. TASK-1 and 2 immunolabel are found in Müller cells and outward currents recorded in these cells are blocked by bupivacaine, a TASK blocker (Fig. 1B and Skatchkov et al., 2006). Indeed, bupivacaine-insensitive robust Kir currents (Fig. 1B) have clear biphasic i/V-curves (similar in shape as in Kir4.1-expressing tsA201 cells (Fig. 7B)). At very positive voltages, the short voltage pulse protocol reveals a steep time-dependent component of rectification (see Fig. 2), which is strictly dependent on pipette [SP] (see Fig. 3) and [K⁺]_o (see Fig. 5). Using the long pulse protocol, we observe a second, more weakly potential-dependent block (Figs. 2 and 4), that is independent of extracellular [K⁺] (see Fig. 4), but is dependent on intracellular [K⁺] (see Fig. 6). As discussed below, these properties are expected to be replicated by the underlying molecular entities expressed in recombinant systems. However, they are not typical of those previously reported for cloned Kir channels studied in inside-out patches.

The Nature of the Müller Glial Cell Inward Rectifier

Perfusion with divalent-ion-free pipette solution caused gradual increase of outward current and loss of biphasic shape (Figs. 1 and 7). Inclusion of spermine in the pipette led to a gradual decrease of outward current, suggesting that the rectification was indeed due to spermine, as we also concluded from experiments with variable [spermine] in the pipette (see Fig. 1). Several reports have now demonstrated expression of Kir4.1 channel subunits in Müller cells (Iandiev et al., 2006; Ishii et al., 1997; Kofuji et al., 2000, 2002; Pannicke et al., 2004, 2005; Poopalasundaram et al., 2000; Raap et al., 2002; Skatchkov et al., 2001). In addition, immunocytochemical evidence for Kir 2.1, has been observed (Iandiev et al., 2006; Kofuji et al., 2002), however, no strongly rectifying (i.e. Kir 2.1-like) currents have been detected (rev. Bringmann et al., 2006) in Müller cells. Furthermore, after knock-out of Kir4.1 protein in rats there were neither significant membrane potential nor conductance remaining to suggest the significant role of Kir2.1 in Müller cells or in astrocytes (Kofuji et al., 2000; Kucheryavykh et al., 2007a). In a direct comparison of Müller cell K⁺ currents and heterologously expressed Kir4.1, Tada et al. (1998) reported identical 25 pS channels in endfeet of rabbit Müller cells and in cells expressing recombinant Kir4.1. Similar conductances were also found by Newman (1993) and Biedermann et al. (1998). Although TASK-type background K⁺-channels are also expressed in Müller cells (Eaton et al., 2004; Skatchkov et al., 2006), these channels are not rectified by SP (Morton et al., 2005; Musset et al., 2006). Thus, Kir4.1 is the primary candidate for generating the Müller cell spermine-sensitive K⁺-current in normal conditions.

Following the cloning of Kir4.1, several studies examined the rectification properties in recombinant cells. Initially it was classified as a strong inward rectifier (Oliver et al., 1998, 2000), with spermine-dependence of the rectification similar to what had been reported for the classic inward rectifiers of the Kir2.x sub-family (Ficker et al., 1994, rev; Nichols and Lopatin, 1997; Lopatin et al., 1994; Panama and Lopatin, 2006; Taglialatela et al., 1995). Such rectification is considerably steeper and more potent than the variable rectification reported in glial cells, and indeed reported in the present study (Figs. 1–6). We therefore examined recombinant Kir4.1 currents in tsA201 cells using similar protocols, and under similar experimental (close to physiological) conditions, to those used for study of the Müller cells. As we now show in Figs. 7–8, the critical properties of the Müller cell Kir channel are indeed replicated using whole cell voltage-clamped tsA201 cells expressing recombinant Kir4.1. In particular, long voltage pulses reveal only weak rectification in physiological [K⁺]_o and membrane potentials (Figs. 8C,D); steep rectification at low extracellular [K⁺]_o is only observed at very positive voltages (> +20 mV, Figs. 8A,B) and the same characteristic lag is observed before the onset of block (Fig. 8A). The difference between the influence of extracellular and intracellular K⁺ on the “slow” and “fast” block by intracellular SP further suggests the existence of two mechanisms of Kir4.1 channel rectification in glial cells, possibly related to two different SP blocking sites. This may be due to: (i) two SP-sensitive sites of Kir channels as has been shown for Kir2.1 channels (Ishihara and Yan, 2007) and/or (ii) the influence of other cations entering the channel pore and interacting with SP and K⁺, as was shown for interaction between Mg²⁺ and SP in Kir2.1 (Yan and Ishihara, 2005). The very high SP constants for “fast” block (Tables 1 and 2) measured in whole cell recordings probably represent unknown interactions between SP and some cytoplasmic ions and molecules.

In this respect, the free concentration of SP may play a key role. The free concentration of SP specifically in glia is not known, however we estimated that the free [SP]_in in freshly isolated rat Müller cells (Fig. 1C, right recording, black arrowhead) is much more than 300 μM. Using our model and fitting i/V curves recorded from Müller cells we calculated that free [SP]_in is 0.8 ± 0.24 mM. This free concentration is lower than K₅₀ (0) of 1.7 mM for slow block in normal 3 mM [K⁺]_o (Table 1) and, thus Kir currents in glia looks weakly rectifying. Impressively, this level in glial cells is about 10 times higher than that previously estimated in neurons (Bowie and Mayer, 1995; Fakler et al., 1995; Haghighi and Cooper, 1998).

The present study establishes that similar rectification can be observed—under similar conditions—in intact Müller cells and in recombinant cells expressing Kir4.1. However, the unusual rectification properties are not readily explained by existing models of spermine block of Kir channels (Kurata et al., 2006; Nichols and Lopatin, 1997; Oliver et al., 2000; Panama and Lopatin, 2006). Our recent finding that spermine may permeate Kir4.1 channels while interacting with K⁺ (Kucheryavykh et al., 2007b) may partially underlie this weak rectification. Little is known about how spermine is allied with other physiological ions and molecules accumulated in glia. Higher SP dissociation constants in whole cell recording as well as in outside-out and cell attached patches at physiological condition (this work) than in inside-out patch recording in symmetrical K⁺ (Kucheryavykh et al., 2007b) may reflect an interaction of SP with the cytoplasmic content. Therefore, future experiments are warranted to characterize in detail the interactions of polyamines and other ions in the regulation of Kir4.1 channels. Such studies are then likely to reveal how the unique properties of this channel contribute to K⁺ spatial buffering in the brain.