Differential expression of Na+/K+-ATPase α-subunits in mouse hippocampal interneurones and pyramidal cells


  • This paper has online supplemental material.

Corresponding author K. S. Richards: INSERM U739, CHU Pitié-Salpêtrière, 105 boulevard de l'Hôpital, 75013 Paris. Email: kat@chups.jussieu.fr


The sodium pump (Na+/K+-ATPase), maintains intracellular and extracellular concentrations of sodium and potassium by catalysing ATP. Three sodium pump α subunits, ATP1A1, ATP1A2 and ATP1A3, are expressed in brain. We compared their role in pyramidal cells and a subset of interneurones in the subiculum. Interneurones were identified by their expression of GFP under the GAD-65 promoter. We used the sensitivity to the cardiac glycoside, ouabain, to discriminate between different α subunit isoforms. GFP-positive interneurones were depolarized by nanomolar doses of ouabain, but higher concentrations were needed to depolarize pyramidal cells. Comparison of pump currents in these cells revealed a current sensitive to low doses of ouabain in interneurones, while micromolar doses of ouabain were needed to suppress the pump current in subicular pyramidal cells. As predicted, nanomolar doses of ouabain increased the frequency but not the amplitudes of IPSPs in pyramidal cells. Immunostaining confirmed a differential distribution of α-subunits of the Na+/K+-ATPase in subicular interneurones and pyramidal cells. In conclusion, these data suggest that while ATP1A3-isoforms regulate sodium and potassium homeostasis in subicular interneurones, ATP1A1-isoforms assume this function in pyramidal cells. This differential expression of sodium pump isoforms may contribute to differences in resting membrane potential of subicular interneurones and pyramidal cells.

The Na+/K+-ATPase is a ubiquitous protein which catalyses 1 molecule of ATP to exchange 3 Na+ ions for 2 K+ ions across the cell membrane (Dobretsov et al. 2003). The primary Na+/K+-ATPase protein complex consists of α and β subunits, in multiple isoforms (Herrera et al. 1987). Three α subunits, ATP1A1, ATP1A2 and ATP1A3, are expressed in the brain, as are three isoforms of the β subunit, ATP1B1, ATP1B2 and ATP1B3 (Eakle et al. 1994; Hasler et al. 1998). Distinct γ subunits which modulate ion and ATP affinities and are now recognized as members of the FXYD family of proteins, are associated with the pump complex (Beguin et al. 1997; Sweadner & Rael, 2000). Cardiac glycosides, such as ouabain, antagonize the activity of all molecular forms of the Na+/K+-ATPase. Complexes containing AT1A3 are more sensitive and those involving ATP1A1 are less sensitive to cardiac glycosides (Urayama & Sweadner, 1988; Munzer et al. 1994).

The exchange of Na+ for K+ by the pump has several implications for neuronal function (Blanco & Mercer, 1998). It assures the homeostasis of potassium and sodium and by controlling intracellular K+ provides an ionic gradient for calcium and chloride transporters (Blaustein & Lederer, 1999; Lingrel et al. 2003). The Na+/K+-ATPase contributes to the regulation of neuronal membrane potential, since transmembrane Na+ and K+ exchange is electrogenic. The resting membrane potential of hippocampal interneurones and pyramidal cells is known to differ (Spruston et al. 1997; Lubke et al. 1998; Jonas et al. 2004). It has been suggested that a lower expression of TASK K-leak channels in interneurones than in pyramidal cells may be involved in this difference (Taverna et al. 2005; Torborg et al. 2006). However, some evidence suggests the Na+/K+-ATPase may function differently in principal cells and interneurones of the dentate gyrus (Ross & Soltesz, 2000). But possible differences in expression of the pump and its subunits in interneurones and principal cells has not been explicitly examined.

In this study we therefore compared Na+/K+-ATPase expression and function in pyramidal cells and interneurones of the subicular formation. We used transgenic mice expressing the green fluorescent protein under the control of the GAD65 promoter (GAD65-eGFP) to identify subicular interneurones. We show that these subicular interneurones and pyramidal cells possess different sensitivities to ouabain. The pump current expressed by interneurones is suppressed by 10 nm ouabain, while a similar current in pyramidal cells is sensitive to 25 μm ouabain. Ouabain sensitive currents contribute to the resting membrane potential of both interneurones and subicular pyramidal cells. Our data suggests that selective protein expression of the α3 subunit of the Na+/K+-ATPase may underlie the pump current in interneurones. In contrast, ouabain sensitive responses of pyramidal cells seem likely to depend on their protein expression of the α1 subunit of the Na+/K+-ATPase.



The GAD65-eGFP transgenic mouse (Brager et al. 2003; Lopez-Bendito et al. 2004) was first generated by pronuclear microinjection of double stranded DNA, with an eGFP construct fused in frame into exon 3 of a GAD65 genomic clone, into fertilized eggs from a CBA mouse (G. Szabó, Institute of Experimental Medicine, Budapest). We obtained the line (from S. Marty, Ecole Normale Superièure, Paris), after it had been backcrossed into C57BL/6 mice and we backcrossed it further into C57BL/6 animals (F5). Slices used in this study were prepared from the progeny of each backcrossed generation. In these mice a subset of GABAergic neurons in brain regions including the hippocampus and neocortex express eGFP (Lopez-Bendito et al. 2004). We recorded from these fluorescent interneurones in the subiculum and pyramidal cells were recorded in slices prepared from either transgenic animals or from C57BL/6 mice (Janvier, Le Genest Saint Isle, France). Experiments were performed in accordance with the European Committee Council Directive of November 24, 1986 (86/89/EEC) and with INSERM guidelines.

In situ hybridization

Figure 1 shows hybridization probes that we designed to distinguish non-overlapping and distinct regions of the ATP1A1 and ATP1A3 genes. The probe used to detect ATP1A1 message was of length 440 bp. It spanned exons 9–13 of the large cytoplasmic loop between the 4th and 5th transmembrane crossing regions as shown in the exon/intron map of Fig. 1. The probe used for ATP1A3 in situ hybridization was of length 857 bp. It was also targeted to the cytoplasmic loop between transmembrane spanning regions encompassing exons 12–14 of ATP1A3. Sense and antisense probes were labelled using a PCR digoxigenin (DIG) probe synthesis kit (Roche, Mannheim, Germany) following the manufacturer's instructions.

Figure 1.

In situ probes
A, model of α1 membrane topology showing the 10 membrane spanning regions in addition to a large intracellular loop between 4th and 5th transmembrane spanning segments that is diverse among ATP1A isoforms, and therefore targeted for probe design (grey box). B, specific locations of the probes designed for this study shown on an intron/exon map of the ATP1A1 and ATP1A3 subunits. The dark grey vertical bars represent exons, and light grey horizontal boxes identify the areas targeted by the probes.

In situ hybridization followed previously described procedures (Schmitt et al. 2002). Brains were dissected, frozen in isopentane (Sigma) and stored at −80°C before sections of thickness 12 μm were cut with with a cryostat and mounted on RNase free slides. Sections were then fixed in 4% formaldehyde (Sigma), transferred to absolute ethanol, rehydrated in a graded series of ethanols and washed in a NaCl–sodium citrate buffer. They were then covered in 10 ng DIG labelled probe (Roche) in a hybridization buffer before incubation with sheep anti-DIG alkaline phosphatase (1 : 500), which was then developed with nitro blue tetrazolium (NBT) and 5-bromo-4-chloro-3-indolyl phosphate (BCIP) (Roche) according to the manufacturer's instructions. Treatment with sense probes revealed no product.


Brains from GAD65-eGFP animals at P17 were removed and fixed in 4% formaldehyde in phosphate buffer for 1 week. Antibodies against cholecystokinin (dilution 1 : 1000, Chemicon, Billerica, MA, USA), neuropeptide Y (1 : 250 Chemicon) and calretinin (1 : 3000, Swant, Bellinzona, Switzerland) were used to examine peptide markers expressed by subicular cells expressing GFP. Horizontal slices of 30 μm were made on cryoprotected mouse brains, after which the slices were washed in saline, blocked in milk powder (5%), bovine serum albumin (2%), and Triton X-100 (0.2%), and incubated in primary antibody. Thereafter slices were washed then incubated in secondary antibodies according to species. the secondary antibodies were Alexa 594 (Invitrogen, Karlsruhe, Germany), and/or Cy2 and Cy3 (Jackson ImmunoResearch Laboratories, Newmarket, UK). Slices were then washed in saline before being mounted on the glass slides using fluorescent mounting media (Vector Laboratories, Burlingame, CA, USA).

For staining of the α3 subunit of the Na+/K+-ATPase, we modified previous protocols (Dobretsov et al. 2003). Tissue from P14 animals was halved, dehydrated in graded ethanols, cleared with xylene, embedded in paraffin, and then sectioned at 3 μm. After washing in xylene and rehydration, the tissue was rehydrated and then pressure-cooked for 5–12 min in 0.06 mm citrate buffer, before blocking and incubation with antibody. We used a monoclonal antibody that recognizes the α3 subunit of the Na+/K+-ATPase (Affinity BioReagents, Golden, CO, USA) at a dilution of 1 : 500. Fluorescent interneurones from eGFP-GAD65 animals were visualized using a rabbit anti-GFP antibody (dilution 1 : 100; Invitrogen, Carlsbad, CA, USA). The Mouse on Mouse kit (M.O.M.) blocking kit was used following manufacturer's instructions (Vector Laboratories, Burlingame, CA, USA). Na+/K+-ATPase subunit immuno-positivity was revealed using a biotinylated secondary antibody and either alkaline phosphatase streptavidin or peroxidase conjugated ABC reagent (Vector Laboratories, Burlingame, CA, USA) as tertiary antibodies. Some experiments were done using the EnVision System HRP mouse kit, and EnVision G/2 System Permanent Red (Dako, Hamburg, Germany). Stained elements were visualized with either the Vector Blue Alkaline Phosphatase Substrate, or the DAB substrate kit for peroxidase (Vector Laboratories, Burlingame, CA, USA).


C57BL6 or GAD65-eGFP transgenic mice aged between P14 and P23 were anaesthetized with ketamine–xylazine (80–100 mg kg−1) and decapitated. Transverse slices including the hippocampus and subiculum were cut at a thickness of 350 μm, in a sucrose-based solution composed of (mm): 10 d-glucose, 1 KCl, 248 sucrose, 26 NaHCO3, 1 CaCl2, 5 MgCl2 and phenol red at 0.01%. They were then transferred to a holding chamber containing a solution of (mm): 11 d-glucose, 2.5 KCl, 26 NaHCO3, 1 NaH2PO4, 124 NaCl, 2 CaCl2 and 3 MgCl2. Solution in the holding chamber was maintained at room temperature, 20–30°C, and was gassed with 95% O2–5% CO2 resulting in a pH close to 7.4.

Slices were transferred to a stage mounted on a Zeiss Axioskop microscope where whole-cell records were made from interneurons and pyramidal cells of the subiculum under IR-DIC video microscopy. The subiculum was defined as the area between the CA1 region and the entorhinal cortex (Fig. 2B), where there was no clearly defined pyramidal cell layer. Interneurones were identified by their green fluorescence, and pyramidal cells were recognized by the orientation and shape of their soma and proximal dendrites. Whole cell current clamp and voltage clamp records were made with borosilicate glass electrodes (Hilgenberg, Malsfeld, Germany). The resistance of pipettes was 2–6 MΩ when they were filled with a recording solution containing (mm): 130 KMeSO4, 10 KCl, 10 Hepes, 1 EGTA, and 4 MgATP, pH 7.4. In experiments to examine the Na+/K+-ATPase pump current, the Na+ concentration in the pipette was increased using a solution containing (mm): 52.7 NaMeSO4, 10 CsCl, 20 TEACl, 60 CsMeSO3, 10 Hepes, 1 EGTA and 4 MgATP, pH 7.4. IPSCs were examined using pipettes containing a higher Cl concentration using a solution containing (mm): 50 KCl, 90 KMeSO4, 10 Hepes, 1 EGTA and 4 MgATP, pH 7.4. The composition of the ACSF was (mm): 11 glucose, 2.5 KCl, 26 NaHCO3, 1 NaH2PO4, 124 NaCl, 2 CaCl2 and 3 MgCl2. It was saturated with 95% O2 and 5% CO2 resulting in a pH of 7.4 and heated to 26–30°C. In some experiments, external Ca2+ and Mg2+ or K+ concentrations were modified as noted. Excitatory or inhibitory synaptic events were isolated in some experiments by suppressing fast GABAergic signalling with bicuculline (20 μm) or by blocking glutamatergic signalling with NBQX (20 μm) and d,l-AP-5 (100 μm). Aliquots of ouabain were stored frozen, and diluted just before use. Chemicals were obtained from Sigma-Aldrich (St Louis, MO, USA) or from Tocris Cookson (NBQX, d,l-AP-5, bicuculline; Bristol, UK).

Figure 2.

Anatomy and physiology of GFP-labelled subicular interneurones
A, GFP positive interneurones labelled by anti-GFP antibodies. Immuno-positive cells typically possessed a large soma with bipolar principal dendrites. Interneurones were often located in the molecular layer. B, diagram of the locations of the soma of recorded cells (grey box). Fluorescent cells were recorded from both the region of pyramidal cell somata (p) and from the molecular layer (r) of the subiculum. C, records from a silent and regularly firing GFP-positive interneurones showing responses to a step hyperpolarization and a family of depolarizing and depolarizing step current injections. D, two examples of CCK staining. Red cell (white star) and GFP-positive (green) cells in the subiculum, where cells coexpressing CCK and GFP are shown in yellow and indicated by white arrowheads. E, examples of subicular staining for NPY (red) and GFP (green) are shown. In the first example, coexpression of both proteins (yellow, white arrowhead) is visible in a field of GFP-only expressing cells. In the second, NPY immuno-positive cell (white star) and a GFP-positive cell are observed. F, subicular cells stained for GFP and also immuno-positive for CR (yellow, white arrowhead), as well as cells that were only positive for CR (white star). G, summary data showing the fraction of GFP positive cells that expressed CCK, NPY or CR. H, the fraction of CCK, NPY or CR positive cells that also expressed GFP is summarized. In all cases scale bar 100 μm.

Electrical signals were recorded using an Axopatch 200B amplifier (Axon Instruments, Union City, CA, USA) and filtered at 10 kHz. They were digitized with pCLAMP 8 software (Axon Instruments), and analysed with either pCLAMP 8 or Mini Analysis software (Synaptosoft, Decatur, GA, USA). Analysis and statistical calculations were made with SigmaPlot 8 or SigmaStat 3.0 (Systat Software Inc., Point Richmond, CA, USA), Microsoft Excel and Origin 6.1 (OriginLab Corp., Northampton, MA, USA). Data are presented as a mean together with the standard error of the mean (mean ±s.e.m.). All data sets were tested for normality before further statistical tests. The significance level was P < 0.05 unless otherwise noted. IPSC amplitude and frequency distributions were compared with the Kolmogorov–Smirnov (K-S) test with a significance level of P < 0.05.


Characteristics of GAD65-GFP fluorescent subicular interneurones

We recorded from fluorescent interneurones in the subiculum of slices prepared from mice of the GAD65-eGFP line. These cells were largely bipolar with somata located either in the molecular layer of the subiculum, or within the pyramidal cell layer (Fig. 2). The firing patterns of fluorescent interneurones were recorded in current clamp with no injected current. Of 14 recorded cells, 6 (43%) did not fire, 7 (50%) discharged action potentials and one cell discharged rapid bursts of 3–6 action potentials at intervals of 10–50 ms separated by longer periods of silence. The mean resting membrane potential of these interneurones was −46.0 ± 1.8 mV (n= 14), and their mean input resistance was 467 ± 54 MΩ (n= 6, range 300–600 MΩ). Figure 2C shows responses of a silent cell and a regularly firing interneurone to current injection. Pyramidal cells were not fluorescent and possessed a pyramidal shaped soma and a principal apical dendrite. Their mean resting membrane potential was −58.8 ± 1.4 mV (n= 9), and their mean input resistance was 131 ± 20.4 MΩ.

While data exists on markers expressed by fluorescent cortical interneurones from this mouse line (Lopez-Bendito et al. 2004; Sugino et al. 2006), there are no equivalent data for the subiculum. We used immunohistochemistry to examine the expression of cholecystokinin (CCK), neuropeptide Y (NPY) and calretinin (CR) in GFP-labelled interneurones. All of these inhibitory cell markers are coexpressed in GFP-labelled cortical cells (Fig. 2DH). CCK was present in 19 ± 2.8% of GPF positive cells (406 GFP positive cells counted; Fig. 2D). GFP positive cells represented 46 ± 5.1% of CCK positive cells (136 CCK positive cells counted; Fig. 2H). We found that 16 ± 5.2 of subicular cells that stained positive for GFP also expressed NPY (206 GFP positive cells counted; Fig. 2E). Conversely 28 ± 7.1% of subicular cells that were immunopositive for NPY also stained for GFP (83 NPY positive cells counted; Fig. 2H). For CR, 34 ± 8.7% of subicular cells that stained positive for GFP also expressed CR (318 GFP positive cells counted; Fig. 2F), and 41 ± 6.7% of subicular cells that showed immunostaining for CR also stained for GFP (196 CR positive cells counted; Fig. 2H). These data are summarized in Fig. 2G and H. They suggest that the GFP positive cells in this study express several peptide markers including CCK, NPY and CR and may have varied firing patterns in the presence and absence of current injection.

The effects of ouabain on interneurones and pyramidal cells

We first examined changes in interneurone and pyramidal cell membrane potential induced by blocking the Na+/K+-ATPase. We used the cardiac glycoside, ouabain, at 25 μm, a concentration which should largely suppress the activity of isoforms containing ATP1A1, ATP1A2 or ATP1A3 subunits (Thomas, 1972; Sweadner, 1985; Shyjan et al. 1990; Stimers et al. 1991). Ouabain applications of duration 10–15 min induced multicomponent responses in both subicular pyramidal cells and interneurones. Ouabain concentrations up to 1 mm evoked similar responses. Figure 3A shows a sequence of two waves of depolarization recorded from a pyramidal cell. The first depolarizing component reached a peak at ∼4 min after the application of ouabain (n= 3), while the second depolarization occurred with a latency of 9–11 min (n= 3). Similar multiphasic responses were observed (Fig. 3B) when action potential generation was suppressed by tetrodotoxin (1 μm, n= 4). Interneurones also responded to prolonged applications of ouabain with a dual or a maintained depolarization (Fig. 3C and D) in the absence and presence of TTX (n= 4). The delayed components of these responses do not then depend on cellular firing or synaptic transmission, but rather may correspond to secondary effects due to the increase in external K+ when pump activity is suppressed (Vaillend et al. 2002). We therefore limited the duration of ouabain applications to 2 min in most experiments, and measured effects on membrane potential or current 1 min later.

Figure 3.

Pyramidal cell responses to prolonged application of ouabain
A, 25 μm ouabain induced a biphasic response consisting of two depolarizing events in this pyramidal cell. B, dual responses were maintained when ouabain was applied in the presence of TTX (1 μm). Measurements were made at the point indicated by the arrow, 3 min after the start of ouabain application corresponding to a 2.5 mV depolarization for the cell shown in A, and 3 mV depolarization for B. C and D, interneurones (n= 4) responded to prolonged applications of 25 μm ouabain with a dual or a maintained depolarization in the absence (C) and in the presence (D) of TTX.

We then compared interneurone and pyramidal cell responses to short ouabain applications (Fig. 4). Previous studies using ouabain to completely block the pump current have used doses in the millimolar range (Dobretsov & Stimers, 1997; Hamada et al. 2003). We examined the effects of 25 μm ouabain and also of a much lower concentration, 10 nm, which should specifically block the activity of isoforms containing the ATP1A3 and -A2 but not the ATP1A1 subunits (Sweadner, 1985; Shyjan et al. 1990). Pyramidal cells and interneurones were recorded with current injected to maintain an initial potential near −60 mV. Pyramidal cell membrane potential was little changed by 10 nm ouabain, with a mean depolarization of 0.4 ± 0.5 mV (n= 5), but was depolarized consistently by 5.4 ± 1.1 mV in response to 25 μm ouabain (n= 5). The depolarization induced by 25 μm ouabain was statistically significant (paired t test P < 0.05). Figure 4A shows pyramidal cell responses and summarizes membrane depolarizations. The depolarization induced by 25 μm ouabain typically caused pyramidal cell firing which sometimes ceased when cells depolarized to potentials where Na+ currents were probably inactivated.

Figure 4.

Current-clamp responses of pyramidal cells and interneurones to 10 nm and 25 μm ouabain
Aa, a subicular pyramidal cell whose membrane potential was not changed by a short application of 10 nm ouabain but was depolarized by 25 μm ouabain. Pyramidal cell holding potential was −64 mV. b, summary of membrane potential responses induced by 10 nm and 25 μm ouabain in 5 pyramidal cells is shown below. The mean values were in control −61.2 ± 1.3 mV, in 10 nm ouabain 60.8 ± 1.5 mV and in 25 μm ouabain −55.8 ± 1.8 mV (n= 5). Ba, an interneurone depolarized by both 10 nm and 25 μm ouabain. The holding potential was −60 mV. b, a summary of membrane potential responses of 14 fluorescent interneurones is shown. The thin black line indicates a 2 min application of 10 nm ouabain (n= 14) and heavy lines show 25 μm ouabain application (n= 12). Averaged membrane potential values were for control −62.4 ± 0.7 mV (n= 14); in 10 nm ouabain −55.2 ± 1.7 mV (n= 14) and in 25 μm ouabain −51.9 ± 2.8 mV (n= 12). Significant differences are indicated by **. C, summary of mean changes in membrane potential of pyramidal cells (black circles) and interneurones (grey circles) elicited by 10 nm and 25 μm ouabain. Significant differences for pyramidal cells are indicated by black ** and for interneurones by grey **.

Figure 4B shows that interneurones were depolarized by 10 nm as well as 25 μm ouabain, in contrast to pyramidal cells. The mean interneurone depolarization from −60 mV was 7.0 ± 1.6 mV for 10 nm ouabain (n= 14 cells) and 10.5 ± 2.3 mV for 25 μm ouabain (n= 12). Three of 14 interneurones exposed to 10 nm ouabain depolarized by less than 3 mV, raising the possibility that membrane potential in some of these cells is less sensitive. The depolarization induced by 10 nm ouabain in all tested interneurones was statistically significant relative to control (paired t test P < 0.05), but the additional depolarization induced by 25 μm ouabain was not. We summarize these data in Fig. 4C.

Differences in the Na+/K+-ATPase pump current in pyramidal cells and interneurones

While changes in potential induced by ouabain provide an insight into the resting activity of the Na+/K+-ATPase, they do not permit comparison of the maximal efficacy of pump molecules expressed by interneurones and pyramidal cells. We approached this question in two ways.

Firstly I–V curves were generated in voltage clamp records from pyramidal cells (Fig. 5A) and interneurones (Fig. 5C) before and after exposure to ouabain (Gadsby & Nakao, 1989; Rakowski et al. 1997). I–V relations were examined using step commands from a holding potential of 0 mV. Recording pipettes used in these experiments contained 52.7 mm Na+ to increase pump activity and Cs+ (70 mm) to suppress K+ currents, while tetrodotoxin (1 μm) was present in the external solution to block voltage-gated Na+ currents. However, since neither calcium current nor Ih was blocked, these I–V traces also reflects the voltage dependence of these currents, and possibly TTX-resistant sodium currents or the Na+/Ca2+ exchanger. One index of the presence of contaminating currents is the estimated value of the reversal potential for the pump. Theoretical values for this reversal potential are close to −230 mV (Glitsch, 2001). Values obtained in experiments on dissociated myocytes (Gadsby & Nakao, 1989; Ishizuka et al. 1996) or cultured dorsal root neurones without processes (Hamada et al. 2003) lie between −100 mV and −50 mV (Gadsby & Nakao, 1989; Ishizuka et al. 1996; Hamada et al. 2003). Values estimated from our data were more positive suggesting the presence of contaminating currents. Data showing pump reversal potentials reverted to more negative levels are presented in online supplemental material, Supplemental Fig. 1. The data shown in Fig. 5 are not intended to provide an accurate estimate of pump reversal potential, but rather to show effects of ouabain over a range of membrane potentials. Despite the presence of the unblocked currents, application of 10 nm ouabain did not change I–V relations in pyramidal cells, but 25 μm ouabain decreased the outward current and increased the slope (Fig. 5A, n= 4). In records from interneurones (n= 4), 10 nm ouabain reduced the outward current in interneurones and increased the slope of the I–V relation (Fig. 5C).

Figure 5.

Pump current in pyramidal cells and interneurones
A, mean I–V relations for subicular pyramidal cells (n= 4) in control conditions (black circles), in the presence of 10 nm (grey circles) and in 25 μm ouabain (grey triangles). The slope of the optimal linear relation was 4.7 ± 0.5 pA mV−1 in control conditions, 6.1 ± 0.6 pA mV−1 in the presence of 10 nm ouabain and 11.4 ± 0.7 pA mV−1 in 25 μm ouabain. The cells were held at 0 mV, the internal Na+ was 50 mm, the external K+ was 2.5 mm and 1 μm TTX and K+ current blockers were present in the electrode, but calcium was not blocked. B, current trace from a pyramidal cell clamped at 0 mV with a pipette containing 52.7 mm Na+. Increasing external K+ from 1 mm (black bar) to 8 mm KCl (grey bar) induced an outward current that was blocked by 25 μm ouabain (white bar). Dotted line indicates the steady state current (32 pA) before increasing K+. Below the data trace is a summary of current responses in 9 pyramidal cells showing a significant difference between the outward currents in 1 mm KCl and 8 mm KCl, as well as between 8 mm KCl and 25 μm ouabain (*). The mean current in control conditions was 37.3 ± 16.6 pA, in 10 nm ouabain it was 74.7 ± 15.6 pA and in 25 μm ouabain 49.7 + 17.8 pA (n = 9), shown in the graph as grey symbols. C, effects of ouabain on I–V curves for interneurones (as in A, n= 4). Ouabain at 10 nm (grey circles) slightly decreased the outward current (slope = 8.7 ± 0.9) and 25 μm ouabain (grey triangles) further decreased the outward current (slope = 13.3 ± 0.9) relative to control (slope = 6.1 ± 0.3, black circles). D, current trace from an interneurone clamped at 0 mV with a pipette containing 52.7 mm Na+. Increasing external K+ from 1 mm (black bar) to 8 mm KCl (grey bar) induced an outward current that was blocked by 10 nm ouabain (white bar). Dotted line indicates the steady current before increasing K+ (35 pA). The current differed significantly between 1 mm KCl and 8 mm KCl, and between 8 mm KCl and 10 nm ouabain (*). The summary data for 7 cells are shown. Average current in control was 52.5 ± 13.2 pA, in 10 nm ouabain 74.4 ± 10.8 pA and in 25 μm ouabain 55.6 ± 10.34 pA (n= 7). They are indicated in the graph by grey symbols.

A second approach to isolate the pump current is to examine responses to increasing external potassium in voltage-clamp records made with an elevated internal sodium concentration. We loaded neurones with Na+ (53 mm) from the recording pipette and clamped at 0 mV (Fig. 5B and D) to measure the pump current induced by changing external K+ from 1 to 8 mm (Thomas, 1972;

Nakamura et al. 1999). In pyramidal cells, increasing external K+ from 1 to 8 mm increased a maintained outward current from 37.3 ± 16.6 pA to 74.8 ± 15.5 pA (Fig. 5B, paired t test P < 0.05, 9 cells). For interneurones, changing K+ from 1 to 8 mm, increased a maintained current from 52.4 ± 13.2 to 74.4 ± 10.8 pA (Fig. 5D, paired t test P < 0.05; n= 7 cells tested). Thus the pump current induced by increasing K+ from 1 to 8 mm was 37.4 ± 7.69 pA in pyramidal cells and 21.9 ± 5.86 pA in interneurones. It was abolished by application of 25 μm ouabain in pyramidal cells (reduction of maintained current to 25.1 ± 9.18 pA, n= 9) and in interneurones by 10 nm ouabain (reduction of maintained current to 18.7 ± 5.61 pA, n= 7).

Effects of ouabain on synaptic signalling

These data show that subicular interneurones were sensitive to lower doses of ouabain than were pyramidal cells, and suggest that the two cell types might express distinct species of Na+/K+-ATPase. Differential expression of the pump on somato-dendritic and axonal membrane might have specific effects on synaptic signalling. We therefore examined the effects of ouabain on inhibitory synaptic events.

Our data suggest low ouabain concentrations should increase interneurone firing and so enhance the frequency of spontaneous IPSCs. We tested this prediction by examining effects of 10 nm ouabain on IPSCs recorded from subicular pyramidal cells in the presence of 10 μm NBQX and 100 μm d,l-APV. Pyramidal cells were voltage clamped at −90 mV with an increased pipette Cl to enhance the amplitude of GABAergic events. We tested the effects of 10 nm ouabain on IPSC amplitude and frequency (Fig. 6). We observed an increase in spontaneous IPSC frequency from 5.3 ± 0.2 to 6.5 ± 0.1 events s−1 (Fig. 6D, n= 8, P < 0.05) providing further indirect evidence that low doses of ouabain depolarize some interneurones (P < 0.05). We determined the cumulative amplitude distribution of the same cell in Fig. 6A and found no significant change in mean IPSC amplitude (Fig. 6B). In all cells tested, ouabain significantly reduced the interval between IPSCs (Fig. 6C, Kolmogorov–Smirnov test applied to each interval distribution, P < 0.05).

Figure 6.

Effect of 10 nm ouabain on IPSCs recorded from pyramidal cells
A, the frequency of spontaneous IPSCs recorded from a pyramidal cell clamped at −90 mV (top trace) was increased by 10 nm ouabain (bottom trace). B, cumulative distribution of the amplitude of spontaneous IPSCs recorded from a pyramidal cell in the absence and presence of 10 nm ouabain (same cell as in A). The amplitude distributions are not statistically different (Kolmogorov–Smirnov, P > 0.05). C, in contrast, the cumulative distribution of intervals between IPSCs in the same cell revealed a significant reduction in event interval between control and 10 nm ouabain (K-S test P < 0.05). D, summary of 8 cells showing an increased IPSC frequency in the presence of 10 nm ouabain (**P < 0.05). E, ouabain (10 nm) did not alter the frequency of miniature IPSCs recorded from pyramidal cells (n= 5), in the presence of 1 μm tetrodotoxin. All records made in the presence of 10 μm NBQX and 100 μm d,l-APV.

We also examined the frequency of miniature IPSCs (mIPSCs), isolated in the presence of 1 μm tetrodotoxin (Fig. 6E), to assess the possibility that pump molecules situated at presynaptic sites might selectively affect inhibitory transmitter release. We found that 10 nm ouabain did not significantly change the frequency of mIPSCs recorded from either principal cells (control frequency 1.3 ± 0.5 s−1, frequency in the presence of 10 nm ouabain 1.3 ± 0.4 s−1, n= 4) or interneurones (control frequency 1.8 ± 0.5 s−1, frequency with 10 nm ouabain 2.1 ± 0.5 s−1, n= 5).

Expression of ATP1A3 and ATP1A1 in the subiculum

The different ouabain sensitivities of both membrane potential and the pump current in subicular interneurones and pyramidal cells might result from a differential expression of Na+/K+-ATPase subunits with distinct properties. We pursued this question using in situ hybridization to recognize α1 and α3 subunits in subicular neurones. Probes for α1 mRNA stained almost all subicular pyramidal cells (96.4 ± 0.4%, n= 445 cells), but few elements in the molecular layer which contains exclusively GABAergic cells (Fig. 7A). The proportion of neurones that were positive for α1 within the molecular layer was 7.3 ± 2.8% (n= 103, n= 3 sections from 2 animals). Anti-α3-labelled probes stained 98.2 ± 1.1% of cells within the pyramidal cell layer (n= 571 cells, n= 5 sections) and 82.5 ± 3.3% of cells (n= 112 number of cells counted, n= 5 sections) within the molecular layer (Fig. 7B and C).

Figure 7.

Distribution of hybridization signals and immunoreactivity for α1 and α3 subunits of the Na+/K+-ATPase in subicular interneurones and pyramidal cells
A, in situ hybridization signals for ATP1A1 (probe as in Fig. 1). Most, but not all (white double arrow heads), cells in the stratum pyramidale of the subiculum showed staining. Very few elements outside this zone generated a positive signal. B, hybridization for the ATP1A3 probe (as in Fig. 1) showing signals from cells in the molecular layer (orange arrow heads) as well as in the stratum pyramidale. C, enlargements of cells in molecular layer stained with ATP1A3 probes. D, immuno-positivity for α3 with long incubation times showing large numbers of cells stained. E, shorter development times were also used which were sufficient to stain surrounding areas of cerebellar Purkinje cells, which are known to densely stain for α3. F, these shorter development times using either DAB or alkaline phosphatase substrates revealed a more selective immunolabelling in hippocampus (double arrowheads) for the α3 subunit of the Na+/K+-ATPase. Ga–d, double immunostaining for GFP as indicated by the red staining, and the α3 subunit represented by brown staining. Some cells show neither α3 nor GFP staining (single arrowhead), while in others both proteins colocalize (double arrowhead). e, in a very few cells we found α3 staining independent of GFP (single heavy arrowhead). Scale 100 μm in A, B, D and E, 50 μm in C and G.

We next asked whether the GFP-positive interneurones examined in this study corresponded to the cells labelled for ATP1A3 mRNA. GFP containing cells were stained with the GFP antibody and α3 expressing neurones were labelled by an antibody against the α3 subunit (Dobretsov & Stimers, 2005). With long DAB or alkaline phophatase development times, immunostaining for α3 was observed in most subicular cells, but some cells were more strongly labelled (Fig. 7D). Shorter development times produced robust staining around cerebellar Purkinje cells (Fig. 7E), which are known for high density α3 staining and so serve as a control for hippocampal staining (Siegel et al. 1984; McGrail et al. 1991; Peng et al. 1997). In the hippocampus, however, shorter development times revealed fewer immuno-positive cells (Fig. 7F). In five thin sections from two animals, 13.5 ± 4.3% of cells were strongly stained. We asked if these lightly stained cells represented the GFP-positive interneurone population. Figure 7G shows examples of neurones double stained with the anti-GFP antibody (red cells), and the anti-α3 antibody of the Na+/K+-ATPase (brown staining). Under these staining conditions, strong labelling for the α3 subunit was present in 70.6 ± 10% of GFP-positive cells. Few GFP-negative cells stained positive for α3 (Fig. 7Ge).


These data reveal a significant difference in the dose of ouabain needed to affect Na+/K+-ATPase function in pyramidal cells and fluorescent subicular interneurones of the GAD65-eGFP mouse line. This suggests that interneurones and pyramidal cells might express distinct isoforms of the pump with different ouabain sensitivities. In situ hybridization and immuno-histochemistry showed that subicular pyramidal cells express the α1 subunit of the Na+/K+-ATPase while both pyramidal cells and inhibitory cells express the α3 subunit. Protein staining suggested more selective staining of the α3 in GFP-expressing interneurones. Combining our data on the diversity of the interneurones used in this study with the strong coexpression between GFP and α3 immunostaining suggests that this low sensitivity to ouabain may represent a general property of interneurones, rather than an attribute of a single cell type.

Identification of interneurones

Interneurones from transgenic mice were identified by their fluorescence (Brager et al. 2003; Lopez-Bendito et al. 2004). In slices from P14–P23 animals, these inhibitory cells were distributed in both pyramidal cell layer and molecular layer of the subiculum (Fig. 2A). Identified cortical interneurones from this animal line at a similar age have been shown to express cholecystokinin, calretinin and neuropeptide Y but usually not parvalbumin (Lopez-Bendito et al. 2004). Our data (Fig. 2DF), show that most fluorescent interneurones in the subiculum express CR, CCK or NPY. However, a significant minority of GFP-positive interneurones seem not to express any of these peptides. The discharge patterns of identified subicular interneurones (Fig. 2C) were heterogeneous with both regularly and irregularly firing cells and some neurones that did not discharge spontaneously. The resting potential of these interneurones was consistently less hyperpolarized, by ∼12 mV, than that of subicular pyramidal cells (Lubke et al. 1998; Ross & Soltesz, 2000). These data show that most GFP-positive cells express one of the peptides we examined and have diverse firing patterns. These cells correspond to a rather heterogeneous population of interneurones (Lopez-Bendito et al. 2004; Sugino et al. 2006).

Measuring responses to ouabain

Prolonged applications of high doses of ouabain induced a multicomponent sequence of membrane potential responses terminating with a maintained depolarization in both pyramidal cells and interneurones (Fig. 3). The later component of these events resembles a spreading depression and presumably results from the suppression of Na+ and K+-homeostasis (Leão, 1944; Haglund & Schwartzkroin, 1990; Vaillend et al. 2002). Prolonged applications of 10 nm ouabain did not induce secondary responses. We attempted to measure primary effects due to suppression of Na+/K+-ATPase activity by high ouabain doses, by limiting the duration of ouabain applications and measuring changes in membrane potential or pump current at a fixed and short latency. This procedure may have underestimated the maximal depolarization induced by ouabain, but since the time course of depolarizations in subicular pyramidal cells and interneurones did not markedly differ, probably did not introduce a systematic bias.

Low doses of ouabain act on interneurones but not pyramidal cells

Our data provide several lines of evidence that most interneurones, but not pyramidal cells, express high affinity ouabain sites. Our data (Fig. 2) suggest that this interneurone population is diverse in both anatomical markers and firing patterns. Differences also seem to exist in responses to 10 nm ouabain: most cells were strongly depolarized. Small responses were observed in a minority of cells tested, and the molecular identity of these weakly responding cells remains uncertain. Nonetheless, some sensitivity to 10 nm ouabain was observed in all GFP-positive cells, thus suggesting that this low dose–response may be a property of diverse inhibitory cell types, as all pyramidal cells tested were insensitive (Fig. 4). Current–voltage relations (Nakao & Gadsby, 1986; Sakai et al. 1996; Senatorov et al. 1997; Kim et al. 2007) were shifted by the lower ouabain concentration only in interneurones (Fig. 5A and C). The pump current detected by applying a step increase in K+ to neurones loaded with Na+ (Thomas, 1972; Gadsby, 1980; Jewell & Lingrel, 1991; Sakai et al. 1996; Nakamura et al. 1999) was abolished by low-dose ouabain in interneurones (Fig. 5B and D). Finally the frequency of spontaneous IPSCs impinging on pyramidal cells was increased by 10 nm ouabain suggesting that activity of some interneurones was increased (Fig. 6). Both low and high affinity ouabain sites have been detected in cells from pineal gland (Shyjan et al. 1990), retina (Shulman & Fox, 1996; Shimura et al. 1998) and heart (Gao et al. 1995; Dobretsov & Stimers, 1997). In dorsal root ganglion cells, dual- component dose–response curves suggest that Kd values for interactions of ouabain with the low and high affinity sites vary by several orders of magnitude (Hamada et al. 2003). Our data suggest that subicular interneurones may express only the high affinity site, as may also be the case at the calyx of Held terminal in the auditory pathway (Kim et al. 2007).

Differences in subunit expression may explain high and low affinity ouabain sites

One explanation for the different efficacy of ouabain in pyramidal cells and interneurones would be a differential expression of distinct pump isoforms. While data vary between expression systems, Na+/K+-ATPase complexes containing α2 or α3 subunits are more sensitive to ouabain than those containing the α1 subunit (Urayama & Sweadner, 1988; Shyjan et al. 1990; Jewell & Lingrel, 1991; Shimura et al. 1998, but see Crambert et al. 2000). In the nervous system, the α2 subunit is expressed by glial elements only (Brines & Robbins, 1993; Cameron et al. 1994; Fink et al. 1996; Hosoi et al. 1997). Thus our data could be consistent with a differential expression of α1 and α3 subunits by subicular interneurones and pyramidal cells.

Differential expression of α1 and α3 subunits

Previous immunocytochemical data suggest that both ATP1A1 and ATP1A3 proteins are expressed in fields occupied by hippocampal pyramidal cell stomata and dendrites (McGrail et al. 1991), but little information exists on interneurone expression of different α subunits of the Na+/K+-ATPase. Using in situ and immunostaining techniques, we therefore searched for differences in expression of α1 or α3 subunits in interneurones and pyramidal cells. Our data suggest that most GFP-positive interneurones express ATP1A3 at perisomatic and dendritic sites, but do not express ATP1A1 (Fig. 7). Both in situ and immunostaining demonstrated the presence of ATP1A1 in pyramidal cells. Message for ATP1A3 was detected in subicular pyramidal cells, but long development times (DAB or AP) were needed to detect ATP1A3 protein suggesting that some labelling may have been non-specific. Double staining for both α3 and GFP revealed a coexpression with tighter correlations between the two proteins, lending support to our physiological findings.

If the high affinity ouabain site is associated with the α3 subunit of the pump, the in situ data raise the question of why pyramidal cells did not respond to 10 nm ouabain. This lack of effect is consistent with previous work suggesting that submicromolar doses of ouabain have no effects on CA1 pyramidal cells (Haglund & Schwartzkroin, 1990) or spiny striatal cells (Calabresi et al. 1995). Taken together our in situ hybridization and immunostaining data could indicate that mRNA for the α3 subunit is present in pyramidal cell cytoplasm, but that the protein is not expressed at the membrane. If so, α3-containing isoforms of the Na+/K+-ATPase might be recruited to the membrane by stimuli which favour its membrane insertion and stabilization (Carranza et al. 1998; Bertorello et al. 2003). Membrane trafficking of the α3 subunit depends in part on intracellular signalling pathways (Yudowski et al. 2000; Ridge et al. 2002), and also on association with specific β subunits (Beguin et al. 1998; Gatto et al. 2001). Alternatively α3-containing, membrane-expressed Na+/K+-ATPase isoforms may be inactive due to their phosphorylation state (Blanco & Mercer, 1998; Therien & Blostein, 2000; Xie & Cai, 2003) although it is not clear why the situation should differ between interneurones and pyramidal cells. Finally, ouabain binding depends on several key amino acids distributed throughout the α3 subunit, and it is conceivable that cell-type specific splice variants might differ between interneurones and pyramidal cells (Schultheis et al. 1993; Croyle et al. 1997).

Functional significance

Our data suggest that the membrane potential of subicular interneurones and pyramidal cells, at resting levels of internal Na+ and K+, is governed in part by the actions of different isoforms of the Na+/K+-ATPase. Measurements, made at a short, fixed delay after ouabain application should reflect primarily actions of the pump on resting potential rather than effects secondary to changes in extracellular K+. The effect of ouabain on membrane potential, measured from a holding potential of −60 mV (Fig. 4), was larger in interneurones than in pyramidal cells even though the pump current was smaller (Fig. 5). Possibly Na+/K+-ATPase actions contribute to differences in resting potential between the two cell types.

We found that low doses of ouabain increased the frequency of spontaneous IPSCs but not that of miniature events, suggesting that the high affinity site is expressed at somato-dendritic sites on interneurons but is either absent from inhibitory terminals or does not affect the GABA release machinery. Higher doses of Na+/K+-ATPase antagonists increase the frequency of miniature IPSCs in CA1 cells (Vaillend et al. 2002), either via alterations in intraterminal Ca2+ (Blaustein, 1993) or due to changes in external K+ when pump activity is suppressed (Fig. 3B). Interactions between the α3 subunit of the pump and intraterminal Ca2+ homeostasis have been described at the calyx of Held terminal (Kim et al. 2007).

The difference in Na+/K+-ATPase contributions to interneurone and pyramidal cell resting potential may permit a modulation of network activity by endogenous ouabain-like compounds in brain cerebrospinal fluid (Halperin et al. 1983; El-Mallakh et al. 2007). If these compounds have a similar efficacy to ouabain, they should excite interneurones at lower concentrations than pyramidal cells. This would enhance GABA release and protect the subicular network against a runaway excitation (Vizi, 1978; Vaillend et al. 2002).



This work was supported by the EC (LSHG-CT-2003-503221), the ANR (P001058) and INSERM.