Adult primary afferent neurons are depolarized by GABA throughout their entire surface, including their somata located in dorsal root ganglia (DRG). Primary afferent depolarization (PAD) mediated by GABA released from spinal interneurons determines presynaptic inhibition, a key mechanism in somatosensory processing. The depolarization is due to Cl− efflux through GABAA channels; the outward Cl− gradient is generated by a Na+,K+,2Cl− cotransporter (NKCC) as first established in amphibians. Using fluorescence imaging microscopy we measured [Cl−]i and cell water volume (CWV) in dissociated rat DRG cells (P0–P21) loaded with N-(ethoxycarbonylmethyl)-6-methoxyquinolinium bromide and calcein, respectively. Basal [Cl−]i was 44.2 ± 1.2 mM (mean ± SE), Cl− equilibrium potential (ECl) was −27.0 ± 0.7 mV (n = 75). This [Cl−]i is about four times higher than electrochemical equilibrium. On isosmotic removal of external Cl−, cells lost Cl− and shrank. On returning to control solution, cells reaccumulated Cl− and recovered CWV. Cl− reaccumulation had Na+-dependent (SDC) and Na+-independent (SIC) components. The SIC stabilized at [Cl−]i = 13.2 ± 1.2 mM, suggesting that it was passive (ECl = −60.5 ± 3 mV). Bumetanide blocked CWV recovery and most (65%) of the SDC (IC50 = 5.7 μM), indicating that both were mediated by NKCC. Active Cl− uptake fell with increasing [Cl−]i and became negligible when [Cl−]i reached basal levels. The kinetics of active Cl− uptake suggests a negative feedback system in which intracellular Cl−regulates its own influx thereby keeping [Cl−]i constant, above electrochemical equilibrium but below the value that would attain if NKCC reached thermodynamic equilibrium.
GABA depolarizes and excites neurons during embryonic and early postnatal developmental periods (Owens and Kriegstein 2002b). As development proceeds, GABA acquires its classic hyperpolarizing and inhibitory actions in mature CNS neurons. Both depolarizing and hyperpolarizing GABA responses are generated by Cl− currents through anion-selective channels coupled to GABAA receptors. The polarity and magnitude of these currents is determined primarily by [Cl−]i, the intracellular-free chloride concentration, which is maintained either above or below electrochemical equilibrium by Cl− transporters and channels expressed in the neuronal plasma membrane (Alvarez-Leefmans 1990, 2001; Payne et al. 2003; Rivera et al. 2005; Yamada et al. 2004; Zhu et al. 2005). The hyperpolarizing shift in EGABA (the reversal potential of GABA-induced currents through GABAA channels) occurring during postnatal maturation is assumed to be due to a downward shift in [Cl−]i during development (Ben-Ari et al. 2007; Kakazu et al. 1999; Owens and Kriegstein 2002b; Stein et al. 2004).
A notable exception to the preceding scheme is that of primary afferent neurons (PANs), the sensory cells that convey virtually all somatic and visceral information to the spinal cord and the brain stem. In these neurons, GABA continues exerting a depolarizing action throughout adulthood (Alvarez-Leefmans et al. 1998; Davidoff and Hackman 1985; Levy 1977). GABA depolarizes the cell bodies (De Groat et al. 1972; Desarmenien et al. 1984; Gallagher et al. 1978; Nishi et al. 1974) as well as the peripheral (Ault and Hildebrand 1994; Bhisitkul et al. 1987; Carlton et al. 1999) and central processes of PANs (Curtis and Lodge 1982; Labrakakis et al. 2003; Rudomin and Schmidt 1999; Willis 1999). When released from spinal dorsal horn interneurons, GABA depolarizes the central terminals of PANs. This primary afferent depolarization (PAD) is a major determinant of presynaptic inhibition in the spinal cord, a pivotal mechanism in the gating and processing of somatosensory information, including nociceptive signaling (French et al. 2006; Rudomin and Schmidt 1999; Willis 1999, 2006). PAD is due to Cl− efflux through anion channels opened by GABA acting on GABAA receptors. PAD amplitude and presynaptic inhibition are expected to be determined by the magnitude of the outward Cl− gradient across the plasma membrane of PANs and hence the importance of understanding the mechanisms controlling the levels of intracellular Cl− in these cells.
Work originated in our laboratory first established that the depolarizing action of GABA is possible because [Cl−]i in PANs is higher than predicted for electrochemical equilibrium due to functional expression of a Na+,K+2Cl− cotransport mechanism (NKCC) that actively accumulates Cl− (Alvarez-Leefmans et al. 1988). Subsequent studies using molecular methods identified it as NKCC1 (Alvarez-Leefmans 2001; Kanaka et al. 2001; Plotkin et al. 1997; Sung et al. 2000). Surprisingly, in spite of its obvious importance for understanding the mechanism of GABA-induced PAD and the modulation of presynaptic inhibition, little is known about Cl− regulation in mammalian PANs, and it is not known if NKCC1 is the only transport mechanism that determines [Cl−]i in these cells. Functional studies of Cl− regulation by direct measurement of changes in [Cl−]i were done in amphibian DRG neurons (Alvarez-Leefmans et al. 1988). However, no comparable studies have been done in mammalian DRG cells in which changes in [Cl−]i are measured directly and not inferred from the reversal of GABA-induced currents. The present study constitutes the first demonstration and functional characterization of an NKCC mechanism in mammalian primary afferent neurons. We used fluorescence imaging microscopy to directly measure [Cl−]i and cell water volume (CWV) in single DRG neurons isolated from newborn and adult rats. We show that irrespective of their phenotype (soma area) or postnatal age, [Cl−]i in DRG neurons is maintained above electrochemical equilibrium primarily through an NKCC mechanism. The rate of active Cl− uptake falls with increasing [Cl−]i and becomes negligible when [Cl−]i reaches its physiological basal level. This saturation kinetics of active Cl− uptake with respect to [Cl−]i suggests a negative feedback system in which intracellular Cl−regulates its own influx thereby keeping [Cl−]i constant, above electrochemical equilibrium but below the value that would attain if NKCC reached thermodynamic equilibrium. Thus kinetic constraints inactivate the cotransporter, even in the presence of a substantial thermodynamic driving force. This suggest that the depolarizing shifts in ECl observed in PSNs in some pathological states like neurogenic inflammation (Valencia-de Ita et al. 2006) and peripheral nerve injury (Pieraut et al. 2007), are thermodynamically feasible by changes in the set point of the feedback system. Some of these results have been published in abstract form (Rocha-Gonzalez et al. 2006).
Isolation and preparation of dorsal root ganglion neurons
Dorsal root ganglion (DRG) neurons were isolated from newborn or juvenile Sprague-Dawley rats (P0–P1 to P21) following published methods (Burkey et al. 2004). The use and handling of animals were approved by the Wright State University Laboratory Animal Care and Use Committee and were in accordance with guidelines provided by the National Institutes of Health. In brief, animals were decapitated and the spinal ganglia dissected in Dulbecco's phosphate-buffered saline without Ca2+ and Mg2+ (GIBCO-Invitrogen, No.14200-075). The ganglia were collected in a 15 ml sterile conical tube containing 10 ml Hanks' balanced salt solution (HBSS) without Ca2+ and Mg2+ (GIBCO-Invitrogen, No. 14170-112) at room temperature. The tube with the ganglia was gently centrifuged at 300 rpm (∼12 g) during 5 min. The supernatant was aspirated and the ganglia were resuspended in 3 ml of DRG growth medium (see following text). The latter was aspirated and replaced by 3 ml of collagenase solution (see following text). The ganglia were incubated for 30 min at 37°C.
The DRG growth medium contained: 88 ml F12 medium (GIBCO-Invitrogen, No. 21700-075) reconstituted as per packet instructions, 10 ml horse serum (heat-inactivated, GIBCO-Invitrogen, No. 26050-088), 1 ml (50 U/ml 50 μg/ml) penicillin/streptomycin (GIBCO-Invitrogen, No. 15070-063), 1 ml (2 mM) l-glutamine (GIBCO-Invitrogen, No. 25030-081), and 250 ng/ml nerve growth factor (Sigma-Aldrich, St. Louis, MO, No. N0513). Nerve growth factor (NGF) was added immediately before use. The Collagenase Solution was prepared as follows: collagenase type 1 (Worthington, No. LS004194) was dissolved in DRG growth medium (without added NGF) at 1.25 mg/ml (0.125%).
Following collagenase incubation the ganglia were gently centrifuged (12 g, 5 min) and the supernatant aspirated. The DRG pellet was resuspended in 3 ml DRG growth medium. Cells were dissociated by mechanical agitation through a 20-gauge 1.5 needle until the suspension of dissociated cells looked homogeneous (∼3 passes through the needle). Cells were plated by adding 0.5 ml of the suspension on 25-mm sterile coverslips (EMS, Hatfield, PA, No. 72196-25) previously coated with poly-d-lysine (BD Bioscience, Bedford, MA, No. 354210, 100 μg/ml) and Laminin (Sigma-Aldrich, No. L-20205, 5 μg/ml). The cell suspension was left undiluted during 15 min to allow initial cell attachment. Then 1.5 ml of fresh DRG growth medium was added to each dish to a final volume of 2 ml. Cells were incubated at 37°C in a 5% CO2-95% air atmosphere. They were used for experiments from 12 to a maximum of 48 h after plating.
The procedure for coating the coverslips with poly-d-lysine and Laminin was as follows: first, the coverslips (EMS, No. 72196-25) were washed with acetone, followed by alcohol and water prior to sterilization. Poly-d-lysine hydrobromide high molecular weight (BD Bioscience, Bedford, MA, No. 354210) was applied (100 μg/ml) to one of the surfaces for each coverslip and left to act for 1 h. Poly-d-lysine was then aspirated and replaced with Laminin (Sigma-Aldrich, No. L-2020). The latter was diluted to 5 μg/ml in sterile deionized water and left to act on the coverslips for 2 h. Laminin was aspirated and the coverslips were allowed to dry in a laminar flow hood at room temperature.
The control isosmotic (ISO) solution contained (in mM) 110 NaCl, 5.5 KCl, 2.5 CaCl2, 1.25 MgCl2, 5 HEPES, and 10 glucose. The pH was adjusted to 7.3 with NaOH. The osmolality was adjusted with sucrose to 290 ± 2 mOsm/kg water. The [Cl−] of this solution was 123 mM, which is within the physiological value of ∼118–123 mM measured in rat cerebrospinal fluid (Reed et al. 1967; Schrock and Kuschinsky 1989). The HEPES concentration was kept at 5 mM in all solutions, including those used for calibration (see following text) because this acid quenches N-(ethoxycarbonylmethyl)-6-methoxyquinolinium bromide (MQAE) fluorescence (cf. Kaneko et al. 2002). This concentration of HEPES was sufficient to buffer the pH of the solutions with minimal and constant dye quenching. To prepare anisosmotic calibration solutions for CWV measurements, the osmolality was adjusted by addition or removal of sucrose to the control ISO solution (Alvarez-Leefmans et al. 2006). This had the advantage of keeping the ionic concentrations constant and at the value of the control ISO solution while changing the osmolality. The latter was expressed as percentage decrement (−10) or increment (+10) with respect to the ISO control. The [Na+] of these solutions (including the ISO control) was ∼112 mM, which is slightly lower than the ∼144 mM measured in rat cerebrospinal fluid (Schrock and Kuschinsky 1989). Because both steady intracellular Cl− levels and active reaccumulation of Cl− following intracellular Cl− depletion depend on the levels of extracellular Na+ (Alvarez-Leefmans et al. 1998; the present study), it was important to test if the slightly lower than physiological [Na+] of the control ISO solution affected [Cl−]i. It was found that neither the steady-state [Cl−]i nor its rate of active accumulation following depletion were affected by varying [Na+] in the range between 100 and 150 mM, indicating that Na+-dependent Cl− transport mechanisms were saturated when [Na+] was ≥100 mM.
The Cl−-free ISO solution (0 Cl−) and the solutions in which external [Cl−] was changed in increments of 20 mM were made by replacing Cl− with gluconate in the ISO-control solution on a mole-for-mole basis. The Na+-free ISO solution (0 Na+) was prepared by mole-for-mole replacement of Na+ with N-methyl-d-glucamine and the pH adjusted with HCl. The K+-free ISO solution was prepared by mole-for-mole replacement of K+ with Na+ and the pH adjusted with NaOH. The other components of the solutions were exactly the same as those in the control ISO. Bumetanide (Sigma-Aldrich, No. B3023) was dissolved in saline solutions to the desired final concentration from a 1,000× stock prepared with DMSO (Sigma-Aldrich). The final concentration of DMSO in these solutions was 0.1%. The final pH of all the solutions was adjusted to 7.3. The osmolality of all these solutions was the same as that of the ISO solution i.e., 290 ± 2 mOsm/kg water.
Measurement of intracellular [Cl−] in MQAE-loaded cells
The fluorescent dye MQAE was used to measure [Cl−]i in DRG neurons following methods similar to those described previously (Koncz and Daugirdas 1994; Maglova et al. 1998; Verkman et al. 1989). When excited at 350 nm, the peak light emission of MQAE occurs at 460 nm. The emitted fluorescence intensity is inversely related to the [Cl−] of the MQAE-containing solution due to quenching by a collisional mechanism with a linear Stern-Volmer relation (1) where F0 is the fluorescence in the absence of Cl−, Ft is the fluorescence in the presence of Cl− at time t, and Ksv is the Stern-Volmer quenching constant for Cl−.
DRG neurons were loaded by incubation during 1–2 h (37°C) in DRG growth medium containing 5 mM MQAE (Invitrogen-Molecular Probes). Each coverslip with dye-loaded cells was mounted in an imaging chamber (RC-21BRW; Warner Instruments, Hamden, CT) and placed on the stage of an epifluorescence inverted microscope (Oympus IX-81, Olympus America, Center Valley, PA) equipped with a fluor oil-immersion lens (Olympus, 40×, NA 1.35) and differential interference contrast (DIC) optics. The dye-loading solution was washed out with ISO control, and the cells were equilibrated for ≥10–15 min in this solution before making initial fluorescence measurements. Experimental solutions were perfused (6 ml/min) by means of two electronic valve systems (VC-6; Warner Instruments). The fluid volume of the chamber (300–400 μl) was exchanged with a half time of <5 s. The temperature of the chamber was 25–26°C; cell viability increased working at this temperature rather than at 37°C. Pilot measurements of basal [Cl−]i done at 37°C were no different to those at 25°C. For all thermodynamic calculations in this study, we used 25°C.
Cell images were captured with a cooled digital CCD camera (ORCA 2-ER C4742-95, Hamamatsu, Hamamatsu City, Japan) using MetaFluor imaging software (Molecular Devices, Sunnyvale, CA). Further details of the imaging setup can be found elsewhere (Alvarez-Leefmans et al. 2006). Recordings of MQAE-emitted fluorescence at 460 ± 25 nm were made from a digital circular region placed at the image plane of each DRG cell body (Fig. 1C, right). The circular regions had 30 pixels in diameter (1 μm = 3.15 pixels). The dye was excited at 350 ± 5 nm using a monochromator (Optoscan, Cairn Research Limited, Faversham, UK) in which both input and output slits were set to the same bandwidth. The excitation light source was a 75-W Xenon Arc lamp. The excitation light passed through a liquid light guide before entering the microscope optical path. A dichroic mirror (400 nm) and a 460 ± 25 nm emission filter cube (Chroma Technology, Rockingham, VT) were positioned underneath the objective lens in the microscope's filter holder.
As reported for other cell types (Bevensee et al. 1997; Kaneko et al. 2001; Koncz and Daugirdas 1994; Maglova et al. 1998; Verkman et al. 1989), the fluorescence signals from MQAE-loaded DRG neurons drifted over the time periods observed (≤2h). This time-dependent decay of MQAE fluorescence has been attributed to photobleaching and dye leakage (Kaneko et al. 2001; Verkman et al. 1989). The degree of drift depended on the total duration (i.e., the time integral) of exposure to the excitation light (Kaneko et al. 2001; Nakamura et al. 1997). To minimize signal drift due to photobleaching, cells were initially identified and photographed with DIC optics avoiding unnecessary UV exposure. Then the illumination was switched to the xenon light source to excite MQAE, and the cells were observed with fluorescence optics. Under these conditions, MQAE signals were sampled with pulses lasting ∼20 ms at a frequency of 0.1 Hz.
A major source of signal drift was identified as drift in the focus of the objective lens. This mechanical drift, which is present in all optical microscopes, varies with environmental temperature and can be wrongly attributed to dye leakage. We corrected this mechanical drift with a nose-piece device (IX2-NPS2, Olympus) that fixes the objective and the stage in such a way that the distance between the objective and the object (focal plane) is always constant, irrespective of drifts in the microscope's mechanic parts. The MQAE signal drift was sometimes linear but more often followed a single exponential time course. Accordingly it was corrected by fitting either a straight line or a mono-exponential function to the data points of the whole fluorescence transient. Irrespective of the function used the correlation coefficient of the fit was >0.98.
To determine net Cl− efflux and influx as well as steady-state [Cl−]i, cells were equilibrated in control ISO solution and then exposed to Cl− free ISO solution (0 Cl−) until they were depleted of Cl−. Cells were assumed to be depleted of Cl− when the MQAE fluorescence signal recorded in the 0 Cl− solution reached a new steady state. The fluorescence signal when the apparent [Cl−]i = 0 mM was taken as F0 (see following text). Intracellular Cl− accumulation was studied by changing the perfusion to a solution containing physiological [Cl−]o, i.e., ISO control, or ISO 0 Na+ followed by ISO control, to determine the Na+-dependent (SDC) and Na+-independent (SIC) components of net Cl− influx, respectively.
MQAE fluorescence was calibrated against [Cl−]i by use of the “double ionophore” technique (Koncz and Daugirdas 1994; Krapf et al. 1988). This involved bathing the cells with a series of calibration solutions containing the ionophores tributyltin (a Cl−/OH− exchanger; Sigma-Aldrich) and nigericin (a K+/H+ exchanger; Sigma-Aldrich). Under steady-state conditions, [Cl−]o and [Cl−]i were assumed to be equal. Cells were exposed to 5 ml of each calibration solution for 3–6 min. The calibration solutions contained (in mM) 0.01 tributyltin, 0.005 nigericin, 10 glucose, 5 HEPES, and 120 K+ and variable NO3− and Cl−. In these solutions, [Cl−] was varied from 0 to 60 mM, keeping the sum [NO3−] + [Cl−] = 120 mM. The osmolality was adjusted with sucrose to 290 mOsm/kg water and the pH to 7.3. The first calibration solution to which the cells were exposed was the one containing 120 mM KNO3 (0 Cl−). Given that NO3− in the range between 0 and 100 mM does not produce quenching of MQAE fluorescence (Kaneko et al. 2002), the fluorescence of each cell in the 120 mM KNO3 solution was taken as a the MQAE fluorescence in 0 Cl− and is denoted as F0(cal). We found that in cells equilibrated with the ISO-0 Cl− solution, which were assumed to be depleted of intracellular free Cl−, the difference between the absolute value of the fluorescence (F0) with respect to F0(cal) was negligible (1.2 ± 0.2 mM, n = 29). This validated the assumption that F0, the steady-state MQAE fluorescence recorded in ISO 0-Cl solution, indeed signals [Cl−]i ≈ 0 mM. Therefore the error in the [Cl−]i reported in this study arising from the difference F0 − F0(cal) was ∼ ±1 mM. The 120 mM KNO3 (0 Cl−) solution was then followed by those containing 20, 40, and 60 mM Cl−. Finally the cells were exposed to a solution containing KSCN (150 mM) to quench the MQAE fluorescence. The KSCN-quenched fluorescence was taken as the background (Fb). Under these conditions, Fb was measured and subtracted point by point from the whole experimental transient and from the fluorescence readings in the calibration solutions.
Fluorescence signals recorded from each individual cell were transformed into [Cl−]i using the following relation derived from Eq. 1 (2) where F0 is the steady-state fluorescence from each cell measured in ISO-0 Cl− solution; Ft is the fluorescence recorded with respect to time during the experimental transients or that measured in the calibration solutions (i.e., fluorescence in the presence of Cl−), and Ksv is the Stern-Volmer quenching constant of MQAE for Cl−. The value of Ksv was calculated for each cell following the four-point calibration procedure outlined in the preceding text. The average Ksv for intracellular calibration of MQAE was 16.3 ± 1 M−1 (n = 75). This value falls within the range of 5–30 M−1 reported for other cell types (Bevensee et al. 1997; Kaneko et al. 2002; Maglova et al. 1998; Martinez-Zaguilan et al. 1994; Verkman et al. 1989).
Measurement of concentration-dependent effect of bumetanide on active Cl− accumulation in MQAE-loaded DRG cells
To study effect of bumetanide on the Na+-dependent component (SDC) of active Cl− accumulation in DRG cells, we followed the protocol illustrated in Fig. 2D. First, cells were depleted of Cl− by exposure to ISO 0 Cl− solution followed by ISO 0 Na+ solution for ∼15 min to assess the magnitude of the Na+-independent component (SIC) of Cl− accumulation. Bumetanide was then added (0.1–100 μm) to the ISO 0 Na+ to determine the interference on MQAE fluorescence produced by this compound (see following text). Cells were exposed to this solution for a period of ∼5 min and then to control ISO solution containing the same bumetanide concentration as in the ISO 0 Na+ solution for ∼30 min. The effect of bumetanide on the amplitude of the SDC was measured 15 min after onset of exposure to the ISO solution containing the inhibitor. Finally cells were exposed to ISO control solution to wash out the bumetanide and measure the level of interference on MQAE fluorescence. This protocol was repeated for each concentration of bumetanide. At physiological pH, bumetanide is a lipophilic anion permeable through the plasma membrane (Alvarez-Leefmans 1990). In addition, at pH 7.4 in PBS, bumetanide is excited by UV light and has maximal absorption and emission at 324 and 414 nm respectively (Fiori et al. 2003). Therefore the effects of bumetanide on MQAE fluorescence are complex and could be due to a mixture of both MQAE quenching and bumetanide fluorescence in a spectral range similar to that of MQAE. Independent of the mechanism, interference of bumetanide on MQAE fluorescence was reversible and reached steady state shortly after (∼2 min) addition or removal of this compound to the bathing solution. This made possible the subtraction of the interference signal from the fluorescence recordings.
Measurement of CWV changes
Changes in CWV were measured in single DRG cells loaded with the fluorescent dye calcein as described in detail elsewhere (Alvarez-Leefmans et al. 1995, 2006; Crowe et al. 1995). In brief, calcein was excited at 495 ± 3 nm and emission was measured at 535 ± 13 nm. Changes in CWV are measured from changes in the dye's fluorescence intensity (F) that in turn results from changes in its intracellular concentration. The changes in F were collected from a small circular region located at the image plane of each dye-loaded cell using wide-field fluorescence microscopy. This circular region had a fixed diameter of 30 pixels (9.5 μm) which on average represented ∼21.7 ± 8.3% (mean ± SD) of the total area of the cells for a given focal plane. As CWV changes, the concentration of the free diffusible calcein molecules (Cf) changes in inverse proportion. As long as the relation between F and Cf is linear and has a positive slope, the emitted F collected through the circular region is directly proportional to Cf and can be quantitatively described by the following expression derived from the Beer-Lambert exponential law (3) where F (for a constant light input) and Cf have been defined, φ is the quantum efficiency of the fluorophore, r is a constant which represents optical instrumental factors, ε is the extinction coefficient, and d is the thickness (optical path length) of the sample. φ, r, and ε are constants, and d can also be considered constant when the depth of field of the objective is less than the cell height (Alvarez-Leefmans et al. 2006).
Statistical analysis was carried out using Student's t-test. Differences were considered significant when P < 0.05. Analysis of bumetanide concentration-response curve was done by a one-way ANOVA followed by Holms-Sidak test (overall significance level = 0.05). Data were expressed as means ± SE unless otherwise stated. Analysis was done using SigmaStat 3.1 software (Systat Software, Point Richmond, CA).
Basal level of [Cl−]i in DRG cells
The basal [Cl−]i measured in 75 DRG cells from newborn or juvenile rats (P0–P1 to P21) was 44.2 ± 1.2 mM with a SD of 9.9 mM (Fig. 1A). This [Cl−]i is three to four times higher than that predicted from a passive distribution which for cells having a resting membrane potential (Em) between −60 and −55 would have been ∼11.9 and ∼14.5 mM respectively as calculated from Eq. 4 (4) where [Cl−]ieq is the intracellular Cl− concentration at electrochemical equilibrium, [Cl−]o is the extracellular Cl− concentration (123 mM), Em was defined in the preceding text and R, T, and ℱ have their usual thermodynamic meaning.
The Cl− equilibrium potential was calculated from Eq. 5 (5)
This is an important parameter inasmuch as the magnitude and polarity of GABA-activated and other Cl− currents are determined by the electrochemical driving force for Cl−, which is the algebraic difference between Em and ECl. The mean ECl calculated from measured intra- and extracellular [Cl−] in the 75 cells was −27.0 ± 0.7 mV (SD 6 mV). Although Em was not measured in these experiments, the wealth of published values for dissociated rat DRG cells recorded at either room temperature (Hong et al. 2004; Petruska et al. 2000; Yu and Kocsis 2005; Zheng et al. 2007) or at 36°C (Ma and LaMotte 2005) indicates that it ranges between −50 and −67 mV. This range of Em values is the same as that recorded in neurons in intact ganglia in vitro at 30–36°C (Desarmenien et al. 1984; Song et al. 2006; Villiere and McLachlan 1996; Zhang et al. 1999), or in vivo (Fang et al. 2005). Thus our results indicate that ECl was 23 to 39 mV positive to Em in agreement with the range of values found in amphibian DRG cells, the only reported example in which intracellular Cl− and Em have been measured directly and simultaneously in the same cells (Alvarez-Leefmans et al. 1988).
There was no correlation between mean [Cl−]i and either postnatal age or cell size (Fig. 1B), suggesting that the higher than passive [Cl−]i is maintained at values above electrochemical equilibrium independently of DRG postnatal developmental stage or cell phenotype. However, the dispersion about the mean [Cl−]i increased with postnatal age. This increase in variability was reflected by an increase in the SD of the mean and the coefficient of variation (CV) with postnatal age for both [Cl−]i and ECl. For instance, at P1, basal [Cl−]i was 45.0 ± 8.05 (SD) mM (n = 9; range: 32–59 mM, CV ≈ 18%); and at P21 was 47.0 ± 16.8 (n = 8; range: 24–66 mM, CV ≈ 36%). Similarly, the mean ECl for the P0–P1 cells was −26.2 ± 4.6 mV (n = 9; range: −19 to −32 mV; CV ≈ 18%); and at P21 was −26.4 ± 10.3 mV (n = 8; range: −16 to −42 mV; CV ≈ 39%). These results suggest that sensory neurons do not undergo a developmental switch in [Cl−]i and ECl, as has been reported in the case of central neurons in which ECl becomes more negative than Em at ∼P7, explaining the hyperpolarizing action of GABA in mature central neurons. In DRG neurons, in contrast, the variability of both [Cl−]i and consequently ECl, increases with postnatal age, whereas ECl remains predominantly more positive than Em, consistent with the fact that these neurons are depolarized by GABA throughout adulthood.
Effect of changes in [Cl−]o on [Cl−]i and CWV
Whether Cl− transport takes place through membrane channels or carriers, it is tightly coupled to CWV; intracellular Cl− affects and is in turn affected by changes in CWV. A notable example of this interdependency is NKCC, which is activated by cell shrinkage and/or a decrease in [Cl−]i, resulting in a net uptake of ions and water thereby restoring CWV (and intracellular Cl−) to normal values (Russell 2000). It has been postulated that NKCC plays a key role in CWV maintenance in various nonneuronal cell types (Hoffmann et al. 2007; Russell 2000). In DRG cells it was predicted, but never experimentally demonstrated, that removal of external Cl− leads to cell shrinkage that reverses by NKCC activation (Alvarez-Leefmans et al. 1988). To characterize the kinetics and mechanisms of Cl− depletion and accumulation and their relation with CWV, DRG cells were transiently exposed to Cl−-free ISO solutions (0 Cl−). On removal of external Cl−, cells lost Cl− and shrank by a final steady-state value of 14.5 ± 1.2% of their initial CWV at initial rates of 3.8 ± 0.3 mM/min (n = 29) and 1.2 ± 0.1%/min (n = 20) respectively. On returning to control ISO, cells actively re-accumulated Cl− and recovered their water volume at initial rates of 7.6 ± 0.6 mM/min and 1.6 ± 0.2%/ min respectively. The rates of CWV and Cl− recovery were significantly faster (P < 0.05) than those of CWV shrinkage and Cl− loss on removal of external Cl−. An example of one of these experiments is shown in Fig. 1, D and E.
Net transmembrane Cl− fluxes
A quantitative functional study aimed at determining the nature of the mechanisms transporting Cl− across the plasma membrane of DRG cells requires measurement of the net Cl− fluxes per unit membrane area. This was done in experiments like those described in the preceding text in which intracellular Cl− depletion and accumulation were measured on isosmotic removal and addition of external Cl− respectively. In the calculation of net Cl− fluxes, it is often conjectured but never demonstrated that CWV changes concomitant with transmembrane Cl− movements are either small and/or too slow to affect the flux measurements. In the present study, measured CWV volume changes on Cl− depletion and accumulation allowed us not only to test these commonly held assumption but also to provide a reasonably accurate estimation of net transmembrane Cl− fluxes in DRG cells. Knowing the cell diameter, the [Cl−]i at time t = 0 and the initial rates of change in cell volume d(Vt/V0)/dt and in intracellular Cl− concentration d[Cl−]i/dt, it is possible to calculate the initial net Cl− flux per unit membrane surface area (JCl) using the following expression derived from the flux equation (Alvarez-Leefmans 1990) (6) where h is the volume-to-surface ratio of the cell (cm). For JCl estimates, DRG cells (n = 29) were assumed to have a spherical shape. The initial net Cl− efflux (JClout) was 30 ± 3 × 10−12mol cm−2s−1, and the net Cl− influx (JClin) was 57 ± 5 × 10−12mol cm−2s−1. This means that JClin was almost twice JClout (P < 0.001). Ignoring the second term of Eq. 3, i.e., assuming that d(Vt/V0)/dt → 0, the apparent JCl estimates were 26 ± 3 × 10−12mol cm−2s−1 for JClout and 53 ± 5 × 10−12mol cm−2s−1 for JClin (P < 0.001). Given that the measured values of d(Vt/V0)/dt were relatively small, it was expected that there would be small differences between net fluxes corrected for volume changes and those not corrected for volume changes. In fact the differences between net fluxes corrected and uncorrected for changes in cell volume were statistically nonsignificant. Therefore we can safely conclude that ignoring CWV changes results only in slight underestimates of JCl as long as the latter is calculated from initial rates of change in [Cl−]i.
External ions requirement for active Cl− accumulation and maintenance of steady-state [Cl−]i
The preceding results show that in rat DRG cells [Cl−]i is maintained above electrochemical equilibrium. We hypothesize that NKCC is one mechanism by which this nonequilibrium distribution of Cl− is generated and maintained. In this mechanism, Cl− is actively accumulated using the energy stored in the combined chemical gradients for Na+, K+, and Cl−. The Na+ and K+ gradients are maintained by the Na+/K+ pump. Ion translocation by the NKCC requires that all three ions (Na+, K+, and Cl−) be simultaneously present on the same side of the membrane. To test these hypotheses, we studied the effects of removal of external Na+ or K+ on intracellular Cl− levels in the steady state and during reaccumulation after cell Cl− depletion. Figure 2A shows that isosmotic removal of external Na+ (0 Na+) resulted in a decrease in basal [Cl−]i that was reversed on returning to the Na+-containing ISO control solution. The average decline in [Cl−]i on isosmotic removal of external Na+ occurred at an initial rate of −1.0 ± 0.1 mM/min (n = 20). On returning to the Na+-containing ISO control solution, cells actively recovered their Cl− at an initial rate of 1.1 ± 0.1 mM/min (n = 20). Similarly isosmotic removal of external K+ (0 K+) resulted in a decrease in [Cl−]i at an initial rate of −1.3 ± 0.5 mM/min (n = 8). On returning to the ISO control solution, [Cl−]i recovered in the same cells at an initial rate of 0.7 ± 0.2 mM/min (Fig. 2B). Further evidence for the role of external Na+ and K+ on intracellular Cl− regulation and maintenance in DRG cells came from experiments in which cells were exposed first to a 0 Na+ solution immediately followed by a 0 K+ solution having normal [Na+]. An example of this is shown in Fig. 2C. Note that the decline in [Cl−]i that began in the 0 Na+ solution proceeded when the cell was exposed to the 0 K+ solution in spite of the fact that the latter solution contained control [Na+]. These observations indicate that both external Na+ and K+ are required not only for maintaining internal Cl− levels above electrochemical equilibrium in the resting state but also for active accumulation of Cl− in isosmotic media.
To further characterize the external Na+ requirement for active accumulation of Cl−, DRG cells were depleted of Cl− by exposure to Cl−-free ISO solution (0 Cl−). This resulted in a decrease in [Cl−]i similar to that shown in Fig. 1. Re-accumulation of Cl− was then monitored for ∼15 min in Na+-free isomotic solution (0 Na+) containing normal [Cl−] (123 mM), and then in the ISO control solution (Fig. 2D). It was found that most cells re-accumulated Cl− in the absence of external Na+. However, the initial rate of Cl− reaccumulation in the Na+ -free solution (3.6 ± 0.4 mM/min; n = 27) was significantly (P < 0.001) slower than that measured in the ISO control solution under similar conditions, i.e., following internal Cl− depletion (7.6 ± 0.6 mM/min; n = 29). The Na-independent component (SIC) of Cl− accumulation reached near steady state at an [Cl−]i of 13.2 ± 1.2 mM (n = 27), and remained stable throughout the duration of exposure to the 0 Na+ solution (≤30 min). The [Cl−]i at which the SIC stabilizes corresponds to an ECl of −60 ± 3 mV. Thus the SIC reached equilibrium at a value close to that expected for Em. The SIC represented 17.5 ± 0.7% (n = 124) of the total JClin and its magnitude was independent of postnatal age.
Following the Na+-free solution, cells were exposed to ISO control solution after which intracellular Cl− accumulation resumed at an initial rate of 3.3 ± 0.3 mM/min (n = 27) until [Cl−]i recovered its initial control values (Fig. 2D). This Na+-dependent component (SDC) of Cl− accumulation started from a baseline [Cl−]i that was close to that predicted for electrochemical equilibrium (i.e., ∼13 mM). Therefore it is likely that the SDC of Cl− accumulation was mediated by an active transport mechanism; it occurred against its electrochemical potential difference across the membrane.
Effect of bumetanide on Na+-dependent Cl− influx
We hypothesized that the SDC of Cl− accumulation was mediated by NKCC activity. A key feature of NKCCs is their pharmacological sensitivity to loop diuretics such as bumetanide (Gamba 2005; Russell 2000). Accordingly, the bumetanide sensitivity of the SDC of the Cl− influx was studied in experiments designed with the same protocol as that illustrated in Fig. 2D and explained in detail in methods. Bumetanide (0.1–100 μM) blocked ∼65% of the SDC of the Cl− influx in a concentration-dependent manner, with an IC50 of 5.7 μM (Fig. 3). We ruled out the possibility that the Na+-dependent Cl− fluxes were mediated through a thiazide-sensitive NaCl cotransporter (NCC); neither chlorothiazide (0.1–1 mM) nor metolazone (10 μM) affected the SDC. These observations, in conjunction with the results presented in previous sections suggest that most of the Na+-dependent Cl− influx was mediated through NKCC.
Cell volume recovery in isomotic media requires the simultaneous presence of external Na+ and Cl− and is blocked by bumetanide
It has been proposed that besides maintaining [Cl−]i above electrochemical equilibrium, NKCC functions to offset osmotically induced cell shrinkage by mediating the net influx of ions and water (Russell 2000). In mammalian cells, the evidence is indirect and comes from experiments in which cell populations are exposed to hyperosmotic solutions. Little is known about the role of NKCC in volume recovery following isosmotic cell shrinkage i.e., the cell volume decrease resulting from loss of intracellular solutes and water in isosmotic media. Isosmotic cell shrinkage is physiologically meaningful; neurotransmitters and hormones can elicit it (Alvarez-Leefmans et al. 1998; Hoffmann 2001; Lee et al. 2007; Russell 2000). We showed that on isosmotic removal of external Cl− DRG neurons loss Cl− and shrank (Fig. 1, D and E). This isosmotic cell shrinkage likely reflects the net loss of intracellular osmolytes (mainly Cl− and K+) and an osmotic equivalent of water. On returning to control ISO solution, cells re-accumulated Cl− and recovered their CWV (Fig. 1, D and E). Re-accumulation of Cl− occurred mainly but not exclusively through a Na+-dependent-bumetanide-sensitive transport mechanism, suggesting that most of the Na+-dependent-Cl− flux was mediated by NKCC (Figs. 2D and 3). The question arises as to whether NKCC mediates volume recovery from isosmotic shrinkage, and if so, to what extent. To investigate this issue, we studied the external Na+ and Cl−- dependence and bumetanide sensitivity of isosmotic volume recovery from shrinkage in calcein-loaded DRG neurons (Fig. 4). In these experiments, cells were equilibrated with ISO control solution, exposed to osmotic calibration solutions (±10% anisosmotic), re-equilibrated with ISO and exposed successively to isosmotic Cl−-free solution (0 Cl−), isosmotic Na+-free solution containing control Cl−concentration (0 Na+) and control (ISO) solution. On isosmotic removal of external Cl− (0 Cl−), neurons (n = 12) shrank by a final steady-state value of 16.2 ± 4.7% of their initial CWV. Exposure to the 0 Na+ solution produced negligible CWV recovery (0.15 ± 0.10%/min) despite the fact that [Cl−]o was the same as that of the control solution, whereas readmission of the ISO control solution produced full CWV recovery (initial rate = 0.9 ± 0.4% min). An example is shown in Fig. 4A. These observations indicate that Na+-dependent mechanisms mediate most of the CWV recovery from steady-state isosmotic shrinkage.
To determine if the Na+-dependent-CWV recovery (SDC in Fig. 4A) from isosmotic shrinkage was mediated by NKCC, we tested the effect of bumetanide. In these experiments, following osmotic calibration pulses, the cells were exposed to ISO 0 Cl− until they reached steady-state shrinkage and then back to control ISO until they recovered their volume. Then they were exposed for a second time to ISO 0 Cl−, and at the end of this second exposure, bumetanide was added. The solution was then switched to ISO with the same concentration of bumetanide (10 μM) for ∼10 min, followed by ISO control solution. This protocol permitted assessing the effect of bumetanide on CWV recovery using the same cell as control and test. The rate of CWV recovery following the control exposure to 0 Cl− was 1 ± 0.2%/min (n = 6), whereas in the presence of bumetanide, it was virtually zero (0.03 ± 0.06%/min). Thus this result shows that bumetanide blocks CWV recovery from shrinkage (Fig. 4B). Volume recovery following bumetanide washout resumed at a very slow rate (0.15 ± 0.07%/min), confirming that the effects of bumetanide are slowly reversible. Pilot experiments ascertained that treating the cells with the same protocol (double exposure to 0 Cl−) but without bumetanide did not affect by itself the rate of CWV recovery in the second exposure. Taken together these results indicate that CWV recovery from isosmotic shrinkage in Cl−-depleted cells requires the simultaneous presence of external Na+ and Cl− and is blocked by bumetanide, hence it is mediated by NKCC.
Relationship between extra- and intracellular [Cl−] and net Cl− influx
In many cell types, there is a tight relationship among NKCC activity, CWV, and [Cl−]i (Hoffmann et al. 2007; Russell 2000). With the possible exception of squid axons, this relationship has been studied only in nonneuronal cell populations, and no direct measurements of [Cl−]i and CWV have been made at the single-cell level. Moreover, intracellular Cl− is a negative feedback regulator of NKCC activity, as was first demonstrated in squid axons (Breitwieser et al. 1990) and later confirmed in nonneuronal mammalian cells [reviewed by (Kahle et al. 2006; Russell 2000)]. To study how intracellular Cl− influences and is influenced by NKCC activity, we began by determining the relation between [Cl−]i, [Cl−]o and CWV in experiments like those illustrated in Fig. 5. Cells were exposed to ISO 0 Cl− solution until they were depleted of intracellular Cl−, and then [Cl−]o was increased in steps of ∼20 mM keeping the osmolality constant between solutions. Intracellular Cl− depletion during exposure to ISO 0 Cl− (Fig. 5A) occurred in parallel with a decrease in CWV (Fig. 5B). Both isosmotic cell shrinkage and intracellular Cl− depletion were reversed in stepwise fashion on exposure to ISO solutions having equal increments in [Cl−]o, suggesting a tight coupling between intracellular Cl− accumulation and CWV recovery. Both CWV and [Cl−]i completely recovered when [Cl−]o was ∼100 mM. Isosmotic increase in [Cl−]o beyond 100 mM did not result in further measurable changes in CWV or [Cl−]i indicating that the mechanisms involved in these processes reached apparent saturation. Note in Fig. 5 that the step changes in both [Cl−]i and CWV (Vt/V0) reached steady-state for each [Cl−]o. That is, for each [Cl−]o, there was a corresponding steady-state [Cl−]i and CWV, suggesting that any mechanisms of Cl− efflux, whether passive (channel-mediated i.e., electrodiffusional) or active (e.g., K+,Cl− cotransport), either balance each other out or contribute little to setting intracellular Cl− levels above electrochemical equilibrium in DRG neurons in isosmotic media.
The relation between steady-state [Cl−]i and [Cl−]o from data obtained in experiments like those detailed in the preceding text was best fitted (r2 = 0.99) by a single exponential. The K0.5 for [Cl−]o was ∼47 mM (Fig. 6A). The possible influence of [Cl−]i on Cl− influx was quantitatively analyzed by measuring the increments in steady-state [Cl−]i (Δ[Cl−]i) and the net Cl− influx (JClin) for ∼20-mM step increments in [Cl−]o, as a function of the initial steady-state [Cl−]i (Fig. 6, B and C). Note (Fig. 6B) that as the initial [Cl−]i increased, Δ[Cl−]i decreased for each 20-mM increment in [Cl−]o. There was also a reciprocal relation between net JClin and [Cl−]i as shown in Fig. 6C. In resting cells, net JClin was 4 ± 1 × 10−12mol cm−2s−1 (n = 11) but increased steeply as [Cl−]i decreased. When initial [Cl−]i was near zero and [Cl−]o was increased from 0 to 20 mM (0→20 in Fig. 6C), the mean net JClin was approximately six times higher (24 ± 6 × 10−12 mol cm−2 s−1) than in resting state, which in this group of cells (n = 11) was 48 ± 2 mM. Extrapolation of the curve describing the relation between net JClin and [Cl−]i indicates that JClin would be negligible when [Cl−]i ≈ 50 ± 5 mM. These results indicate that net JClin falls with increasing [Cl−]i and are consistent with the notion that the level of intracellular Cl− exerts a negative feedback influence on the active Cl− uptake mechanisms. Because most of this active accumulation occurs through a bumetanide-sensitive-Na+-dependent mechanism (i.e. NKCC), the question arises as to whether the final intracellular Cl− levels are determined thermodynamically or kinetically.
Thermodynamic versus kinetic control of steady-state [Cl−]i in DRG neurons
Assuming that [Cl−]i in DRG is determined only by NKCC with a stoichiometry of 1:1:2 (Na:K:Cl) and that the Na+/K+ pump keeps [Na+]i and [K+]i constant, the steady-state [Cl−]i when NKCC attains thermodynamic equilibrium (i.e., without kinetic constraints) is given by the following equation (7)
Figure 7A (○) shows a plot of the predicted steady-state [Cl−]i as a function of [Cl−]o using Eq. 7. The [Na+]i and [K+]i were assumed to be constant at 10 and 135 mM, respectively. There is a linear relation between predicted steady-state [Cl−]i and [Cl−]o. In contrast, the experimentally measured values of [Cl−]i plotted as a function of [Cl−]o yield the monoexponential relation shown with • in Fig. 7A. Comparison of the two graphs in Fig. 7A shows that [Cl−]i in DRG cells is above electrochemical equilibrium but below the value that it would have if NKCC reached thermodynamic equilibrium. The double-headed bracketed-arrows labeled a-c in Fig. 7A indicate: “a”, the difference between the maximal theoretical value of [Cl−]i if NKCC attained thermodynamic equilibrium (83 mM) and the value measured (48.3 ± 2.5 mM) at the physiological [Cl−]o used in this study (123 mM); “b”, the difference between measured [Cl−]i (at [Cl−]o = 123 mM) and the [Cl−]i expected for a passive distribution as defined by Eq. 4, which for Em = −60 mV would be ∼11.9 mM; and “c” the theoretical range of [Cl−]i below electrochemical equilibrium.
NKCC is a secondary active transport mechanism and hence it is expected to be reversible, that is, to mediate ion fluxes into or out of the cell, the net direction of cotransport depending on the overall net free energy (ΔG) of the system. If ΔG is negative, the direction of the cotransport will be inward, favoring uptake. If ΔG is zero, the system is at equilibrium. NKCCs are electroneutral and in mammalian cells they have a 1Na+:1K+:2Cl− stoichiometry (Geck and Heinz 1986; Geck et al. 1980; Hoffmann 2001; Russell 2000). The net free energy ΔG in the combined Na+ + K+ + 2Cl− chemical gradients is given by the following equation (8)
Figure 7B shows a plot of ΔG calculated from Eq. 8 as a function of [Cl−]o. The values of [Cl−]i used in Eq. 8 for calculating ΔG when NKCC is in thermodynamic equilibrium (○) were those obtained by using Eq. 7 and plotted in Fig. 7A (○). However, using the experimentally measured values of [Cl−]i yields the relation shown with •. The arrow labeled “d” in Fig. 7B represents the physiological driving force for cotransport via NKCC. It can be seen from these data that the net energy available for NKCC strongly favors net uptake (ΔG is negative). The actual value of ΔG for the ISO control solution was −2.8 kJ·mol−1.
Basal [Cl−]i in DRG neurons is higher than predicted for electrochemical equilibrium
The present study shows that basal [Cl−]i in rat DRG neurons equilibrated with an extracellular solution having the same [Cl−] of rat cerebrospinal fluid (123 mM) was 44.2 ± 1.2 mM. This [Cl−]i is three to four times higher than predicted for a passive electrochemical distribution which for cells having Em of −60 to −55 mV would have been ∼11.9 and ∼14.5 mM, respectively, as calculated from Eq. 4. On removal of external Cl−, cells not only lost their Cl− but shrank by 14.5 ± 1.2% of their initial CWV. Cells remained shrunken in this steady-state level for the duration of exposure to the ISO 0 Cl− solution. For cells in osmotic equilibrium with the extracellular fluid (290 mOsm/kg water), this steady-state shrinkage indicates a net final loss of ∼42 ± 5 mOsm/kg water of osmotically active particles. Because net Cl− efflux and influx are electroneutral, it is likely that the loss of Cl− is accompanied by an equivalent loss of K+. The latter, however, is expected to be offset by active K+ uptake via the Na+/K+ ATPase, and therefore the net K+ loss must be null at the end of the process, that is, in the steady state [i.e., when d[Cl−]i/dt = 0 and d(Vt/V0)/dt = 0]. Thus assuming that during net electroneutral Cl− efflux the sum of [K+]i and [Na+]i was kept constant by the Na+/K+ pump, the net loss of ∼42 mOsm/kg water of osmotically active particles represents the net loss of an equivalent amount of Cl−, consistent with the mean basal [Cl−]i of ∼44 mM. Therefore the extent of cell shrinkage could be entirely accounted for in terms of measured Cl− depletion.
Basal [Cl−]i is maintained above equilibrium irrespective of postnatal age or neuronal phenotype
All DRG neurons had [Cl−]i above electrochemical equilibrium irrespective of their postnatal age or soma area (Fig. 1B). This suggests that the higher than passive [Cl−]i is maintained above electrochemical equilibrium independently of DRG developmental stage or cell phenotype. We also found that as maturation progresses, the variability in basal [Cl−]i and corresponding ECl values increases. These observations are consistent with earlier results showing that almost all PSNs are depolarized by GABA throughout adulthood via GABAA-activated anion-permeable channels, irrespective of neuronal phenotype (see following text). However, they are in sharp contrast with what has been observed in CNS neurons, which undergo a hyperpolarizing shift in ECl during maturation that renders GABA hyperpolarizing by the middle of the second postnatal week (Owens and Kriegstein 2002a; Rivera et al. 1999, 2005).
The mean ECl of DRG neurons calculated from measured intra- and extracellular [Cl−] was −27.0 ± 0.7 mV (n = 75 cells) and ranged between −16 and −46 mV. The mean Em of rat DRG neurons is approximately −60 mV and ranges between −50 and −67 mV (Fang et al. 2005; Ma and LaMotte 2005; Villiere and McLachlan 1996; Zheng et al. 2007). Hence, ECl was on average 23 to 39 mV positive to Em in agreement with the values reported in amphibian DRG cells (Alvarez-Leefmans et al. 1988). The present results are consistent with early measurements showing that GABA depolarizes virtually all DRG neurons from excised intact ganglia in vitro (Desarmenien et al. 1984). In this early study, [Cl−]i was undisturbed because measurements were done using a single-microelectrode voltage clamp. Neurons were identified as Aβ, Aδ, and C based on conduction velocity and/or cell-diameter measurements. GABA depolarized all Aβ cells, ∼98% of Aδ, and ∼92% of C cells, and EGABA roughly ranged between −10 and −40 mV. Under the assumption that EGABA = ECl, the EGABA values fall within the range of ECl values reported in the present study based on direct Cl− measurements. The results are also consistent with the fact that GABA mediates PAD of Aβ, Aδ, and C intra-spinal terminal axons which results in presynaptic inhibition in the spinal cord (Labrakakis et al. 2003; Rudomin and Schmidt 1999; Willis 1999, 2006).
Mechanisms of Cl− transport and accumulation in DRG neurons
The present results provide functional evidence in support of the notion that the outward Cl− gradient across the plasma membrane of primary sensory neurons is generated and maintained primarily by a Na+,K+,Cl− cotransport mechanism. The evidence can be summarized as follows: 1) brief isosmotic removal of external Na+ and/or K+ resulted in a decrease in basal [Cl−]i that reversed on restoring external Na+ and K+, indicating that these cations are necessary for the maintenance of the Cl− gradient. 2) External Na+ was required for active accumulation of Cl−, i.e., uphill transport of Cl− is Na+-dependent. 3) Most (∼65%) of the Na+-dependent Cl− accumulation was blocked by bumetanide with an IC50 of 5.7 μM (Fig. 3), a value that falls within the upper range reported for NKCC inhibition in other cell types (Russell 2000). Sensitivity to bumetanide varies; it depends on NKCC isoform, whether it is assessed under basal conditions or on activation of the cotransporters (Hannaert et al. 2002), as well as on degree of glycosylation (Paredes et al. 2006). Our results may reflect a mixed population of DRG cells with NKCCs having different bumetanide sensitivities due to one or more of the preceding factors. Further, we considered the possibility that the Na+-dependent bumetanide-insensitive Cl−-uptake, which represents ∼35% of the SDC of the net Cl− flux, could have been mediated by Na+-Cl− cotransport (Gamba 2005). This possibility, however, was ruled out because chlorothiazide (0.1–1 mM) or metolazone (10 μM), potent inhibitors of this cotransport mechanism, did not affect Na+-dependent-Cl−-uptake. Another potential candidate mediating Na+-dependent bumetanide-resistant Cl− uptake is Cl−/HCO3− exchange functionally coupled to Na+/H+ exchange (Cala and Maldonado 1994). Clearly, the identity of the transport pathway that mediates net Cl− influx through Na+-dependent bumetanide-resistant mechanisms is an important issue that requires further investigation.
Besides the Na+-dependent component (SDC) of Cl− accumulation, there was a relatively small Na+-independent component (SIC) of the net Cl− influx. The SIC reached steady-state when [Cl−]i was ∼13 mM, which corresponds to an ECl of ∼−60 mV. That is, the SIC reached equilibrium at a value close to Em, suggesting that it may reflect a passive Cl−-leak conductance. This is intriguing; resting Cl− conductance of DRG cells is known to be negligibly small (Alvarez-Leefmans 1990; Deschenes et al. 1976; Nishi et al. 1974). Moreover, DRG cells lack ClC-2 the ubiquitous inward rectifying Cl− channels that stabilize Em in other neurons (Staley et al. 1996). However, DRG cells are endowed with Ca2+-activated Cl− channels (Frings et al. 2000; Hartzell et al. 2005). Given that the SIC becomes evident on removal of external Na+, and that this maneuver leads to an increase in [Ca2+]i due to reversal of Na+/Ca2+ exchange (Verdru et al. 1997), it is conceivable that a rise in [Ca2+]i activates a Cl− conductance that otherwise would be quiescent. The nature and properties of this putative Cl− conductance in DRG cells are still to be determined.
Another significant feature of the SIC is that once it reached equilibrium at a value close to that expected for Em, it remained stable over the time course of exposure to the Na+-free solution. This suggests that DRG cells are not endowed with mechanisms of active Cl− extrusion in isosmotic medium; if such mechanisms were functionally expressed, in the absence of external Na+, the [Cl−]i should decrease with time, and it does not (Fig. 2D). Consistent with this interpretation is the fact that DRG cells lack the protein and the transcript of KCC2, the neuronal Cl− extruder (Coull et al. 2003; Kanaka et al. 2001; Rivera et al. 1999; Toyoda et al. 2005). Other K+-Cl− cotransporter isoforms (KCC1, KCC3, and KCC4) have been detected in PSNs (Boettger et al. 2003; Pieraut et al. 2007; Toyoda et al. 2005), but their function is unknown. KCC1, KCC3, and KCC4 do not seem to be active in isosmotic media when expressed in Xenopus oocytes, but they are activated by hyposmotic swelling and hence, it is thought that their primary function is cell volume control rather than Cl− homeostasis, although this is controversial (Gamba 2005). Further, CWV measurements revealed that volume changes are negligible during the period in which Na+-independent Cl-influx occurs (0 Na+ in Figs. 2D and 4A). This suggests that the Na+-independent Cl− influx (SIC) does not occur through KCC, otherwise this should be reflected in osmotic cell swelling on restoring Cl− in the absence of external Na+. Moreover, the volume measurements suggest that Cl− entering in the absence of external Na+ must be exchanged for another ion and/or is entering via channels in such a way that there is no net flux of osmotically active particles.
Negative feedback control mechanism of Cl− accumulation
Our results show that net Cl− influx (JClin) falls with increasing [Cl−]i (Fig. 6C) and becomes negligible when [Cl−]i reaches ∼50 mM, i.e., close to the average basal level. Most JClin is active and mediated by NKCC. However, as shown in Fig. 7, A and B, the fall in JClin and the final steady-state level of Cl− are not determined by thermodynamic constraints in Na+,K+,2Cl− cotransport; the calculated net free energy (ΔG) in the combined Na+ + K+ + 2Cl− chemical gradients (−2.8 KJ mol−1) strongly favors net Cl− uptake under conditions in which JClin is maximally inhibited, i.e., in control solution ([Cl−]o = 123 mM). This suggests that a feedback control mechanism that senses and prevents changes in [Cl−]i regulates NKCC transport capacity in DRG neurons. In this kind of control system, the final steady-state concentration to which Cl− is accumulated is expected to vary as [Cl−]o changes and the relation between these two variables is exponential (Jones 1973) as was found to be the case (Fig. 6A). This negative feedback mechanism may keep [Cl−]i at a constant upper limit (set point) in DRG cells under basal conditions. Indeed there is increasing evidence showing that NKCC transport capacity is regulated by a signaling system involving Cl−/volume-sensitive kinase- and phosphatase-mediated phosphorylation/dephosphorylation. In this system, phosphorylation activates and dephosphorylation inhibits NKCC transport activity (Delpire and Gagnon 2006; Kahle et al. 2006). In this Cl− homeostatic system, an increase in [Cl−]i inhibits NKCC activity and vice versa as originally proposed for the NKCC isoform present in squid axons (Breitwieser et al. 1990; Russell 2000). The observation that the rate of Cl− accumulation, JClin and Δ[Cl−]i decreases as the steady-state [Cl−]i increases (Figs. 5A and 6, B and C) is consistent with this model.
It has been proposed that inflammation and other forms of peripheral tissue or nerve injury produce phosphorylation and stimulates NKCC1 in PSNs leading to an increase in [Cl−]i and, consequently, to a depolarizing shift in ECl (Galan and Cervero 2005; Pieraut et al. 2007; Valencia-de Ita et al. 2006). A change in the sensitivity of the regulatory mechanisms of NKCC could alter the “set point” of the negative-feedback mechanism in such a away that intracellular Cl− inhibition occurs at a higher [Cl−]i i.e., closer to the thermodynamic equilibrium of the cotransporter (Fig. 7A). The resulting depolarizing shift in ECl may in turn result in enhancement of GABA-mediated PAD in the intraspinal terminals of nociceptive afferents to levels sufficient to initiate action potentials, i.e., dorsal root reflexes (DRRs) which evoke pain, hyperalgesia and neurogenic inflammation (Valencia-de Ita et al. 2006). The question arises as to whether these shifts in ECl are thermodynamically feasible in PSNs. The present results suggest that they may be. First, in a negative feedback-control system for Cl−, a shift in the set point could change the final steady-state [Cl−]i. There is evidence showing that such shifts in set point occur. For instance, hyperosmotic stress in squid axons stimulates NKCC by resetting the relation between [Cl−]i and transport activity (Breitwieser et al. 1990). Some serum factors in bovine endothelial cells may alter the set point of NKCC1 by reducing the sensitivity of the cotransporter to inhibition by intracellular Cl− (Jiang et al. 2001). Second, the results in Fig. 7A show that [Cl−]i in DRG cells is above electrochemical equilibrium but below the value that it would attain if NKCC reached thermodynamic equilibrium according to Eq. 7, suggesting that the final steady [Cl−]i is determined by kinetic and not thermodynamic constraints on the Cl− accumulating system (Lytle and McManus 2002; Russell 2000). The present results show that basal [Cl−]i measured in cells equilibrated with control solution ([Cl−]o = 123 mM) was 48.3 ± 2.5 mM but [Cl−]i would have been 83 mM if NKCC had attained thermodynamic equilibrium (Fig. 7A). The latter is the upper theoretical limit that [Cl−]i could reach under our experimental (physiological) conditions, assuming that NKCC is the primary mechanism of Cl− accumulation, that the stoichiometry is 1Na+:1K+:2Cl−, that passive leaks are negligible, and that extra- and intra-cellular [Na+] and [K+] are kept constant by the Na+/K+ pump. The difference between these two values of [Cl−]i was 35 mM, thus stimulation of NKCC could have shifted [Cl−]i from ∼48 to 83 mM and ECl from approximately −24 to approximately −10 mV.
Cell volume recovery from isosmotic shrinkage is mediated by NKCC in DRG neurons
It is widely accepted that NKCC plays a central role in cell volume regulation following hyperosmotic cell shrinkage (Hoffmann et al. 2007). Less is known about whether NKCC is involved in restoring CWV on more physiologically relevant isosmotic cell shrinkage. Further, as pointed out by Russell, It has been difficult to determine if “… NKCC is a physiological volume regulatory mechanism or merely a volume-sensitive transport mechanism” (Russell 2000). In this study, using real-time measurements of CWV in single DRG neurons, we demonstrate that isosmotic removal of external Cl− leads to cell shrinkage that reverses on restoring external Cl− and recovery from shrinkage following Cl− depletion in isosmotic media requires the simultaneous presence of external Na+ and Cl− and is almost completely blocked by bumetanide (10 μM). This indicates that volume recovery from isosmotic shrinkage is mediated by NKCC. The tight coupling between [Cl−]i, CWV and NKCC is further evidenced in experiments like those in Fig. 5, suggesting a negative feedback mechanism designed to maintain [Cl−]i and CWV at a constant set point in DRG cells under basal conditions. Altogether, these results indicate that NKCC is an essential component in these feedback mechanisms in PSNs.
What may be the functional significance of these feedback mechanisms in DRG cells? One suggestion is that due to the small volume-to-surface ratio of intraspinal primary afferent terminals, GABA produces a transient decrease in [Cl−]i, in [K+]i (due to the depolarization) and in CWV (due to concomitant loss of intracellular ions and water). This would stimulate NKCC thereby restoring ion and osmotic balance. Decreases in [Cl−]i and/or CWV activate NKCC in many cell types (Russell 2000; Strange et al. 2006). Interestingly, activation of GABAA receptors in immature rat cortical neurons and in olygodendrocytes increases NKCC1 activity (Schomberg et al. 2003; Wang et al. 2003). Decreases in [Cl−]i produced by GABA have been measured in various types of neurons (Ballanyi and Grafe 1985; Chub et al. 2006; Schomberg et al. 2003). Further, the GABA-evoked depolarization in peripheral neurons activates K+ channels resulting in K+ efflux and a decrease in [K+]i (Ballanyi and Grafe 1985; Deschenes and Feltz 1976). Thus activation of GABAA receptors may produce net efflux of K+ and Cl− with consequent osmotic water efflux and cell shrinkage that are reversed by NKCC (Alvarez-Leefmans et al. 1998). GABAA-induced shrinkage has been demonstrated in populations of embryonic brain stem neurons (Momose-Sato et al. 1998) and cultured oligodendrocytes (Wang et al. 2003). Methodological approaches such as two-photon excitation laser scanning microscopy may be used in future studies for measuring changes in CWV and [Cl−]i in fine central axons and terminals of PANs to tests these hypotheses.
This research was supported with funds awarded to F. J. Alvarez-Leefmans by National Institute of Neurological Disorders and Stroke Grant NS-29227 and Research Challenge Grant 666762 from the Ohio Board of Regents.
We are grateful to Dr. Mauricio DiFulvio for comments on the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 by the American Physiological Society