A homeostatic mechanism counteracting K+-evoked choline release in adult brain


Address correspondence and reprint requests to Dr Jochen Klein, Department of Pharmacology, University of Mainz. Obere Zahlbacher Str. 67, D-55101 Mainz, Germany. E-mail: jklein@mail.uni-mainz.de


Choline (Ch) is an essential nutrient as the biosynthetic precursor of acetylcholine (ACh) and phospholipids. Under resting conditions, the intracellular accumulation of Ch (above 10-fold), which is positively charged, is governed by the membrane potential and follows the Nernst equation. Accordingly, in synaptosomes from adult rats during depolarization, we observed a linear relationship between release of free cytoplasmic Ch and KCl concentration (2.7–120 mm). The K+-evoked Ch release was Ca2+-independent and did not originate from ACh or phospholipid hydrolysis. In superfused brain slices of adult rats, however, a K+-induced Ch efflux was absent. Also, under in vivo conditions, 30–60 mm KCl failed to increase the extracellular Ch level as shown by microdialysis in adult rat hippocampus. On the contrary, in brain slices from 1-week-old rats, high K+ as well as 4-aminopyridine evoked a marked Ch efflux in a concentration-dependent fashion. This phenomenon faded within 1 week. Hemicholinium-3 (HC-3, 1␣and 10 µm), a blocker of cellular choline uptake, caused a marked efflux of choline from adult rat slices but no or significantly less release from immature slices. We conclude that depolarization of synaptic endings causes a Ca2+-independent release of free cytoplasmic Ch into the extracellular space. In adult rat brain, this elevation of Ch is counteracted by a homeostatic mechanism such as uptake into brain cells.


AACOCF3, arachidonyl trifluoromethyl ketone










phospholipase A2


phospholipase D.

Choline is a water-soluble, quaternary amine which serves as a precursor of acetylcholine (ACh) and of membrane phospholipids such as phosphatidylcholine and sphingomyelin (Zeisel and Blusztajn 1994). In principle, the␣availability of choline in the brain is guaranteed by an efficient, low-affinity choline uptake system at the blood–brain barrier (Allen and Smith 2001). As this system is highly unsaturated at physiological plasma levels, the uptake of choline into the brain rises parallel to increases of the plasma choline level (Klein et al. 1990). However, powerful homeostatic mechanisms maintain the overall brain levels of free choline within a narrow range (Tucek 1988; Klein et al. 1998a) in spite of dietary fluctuations (5–40 µm; Zeisel 1981), or even marked pharmacological increases, of the plasma choline level (Löffelholz 1998). Homeostatic mechanisms include the rapid removal of choline by cellular uptake, intracellular phosphorylation (Millington and Wurtman 1982) and incorporation into phospholipids (Klein et al. 1992). Surplus bound choline is then gradually released into the extracellular space (Wecker 1989) and returns into the circulation (Klein et al. 1998a). Understanding of these processes is particularly important because of recent observations (Papke et al. 1996; Alkondon et al. 1997) suggesting an agonistic role of choline at α7 nicotinic cholinoceptors.

In view of the positive charge of choline and of the electronegativity on the inside of plasma membranes, it is likely that, at equilibrium, the transmembrane choline gradient follows the Nernst equation. Thus, at a membrane potential of −75 mV and an extracellular choline concentration of 3 µm, an intracellular choline level of 49.3 µm was predicted, which turned out to be almost identical with the biochemically obtained value of 48 µm (Tucek 1985, 1988). Consequently it seemed plausible that depolarization would cause a net release of choline, whereas hyperpolarization would be expected to accelerate the cellular net uptake of choline. However, in preliminary experiments we failed to evoke a release of choline from brain slices of adults rats by high K+-solutions. The aim of the present study was to elucidate this discrepancy. We now show that choline uptake mechanisms are responsible for the ontogenetic development of brain choline homeostasis under depolarizing conditions.

Experimental procedures


Male Wistar rats (Charles River, Sulzfeld, Germany) were kept under standardized light/dark (12 h), temperature (22°C) and humidity (70%) conditions, with rat chow and water available ad␣libitum.

Preparation of synaptosomes and determination of synaptosomal choline release

Synaptosomes were prepared from adult Wistar rat cortex essentially as previously described (Dunkley et al. 1988). Cortices were homogenized in 0.32 m sucrose containing 1 mm Na,Ca–EDTA and 0.25 mm dithiothreitol (pH 7.4) and centrifuged at 1000 g for 10 min. The supernatants were layered on top of a Percoll gradient (3, 10 and 23%) in the same sucrose solution and centrifuged for 5 min at 32 500 g. The fraction containing synaptosomes (between 10 and 23% Percoll) was recovered and washed. Synaptosomal acetylcholinesterase activity was blocked by preincubation of the synaptosomes with 0.1 mm diisopropylfluoro-phosphate (DFP) for 30 min.

For the determination of choline release from synaptosomes we used a modification of the procedure of Israel and Lesbats (1982) in which synaptosomes were incubated together with the choline detection system in the reaction vial of the luminometer. The synaptosomes (50 µL) were first incubated in Tyrode solution (composition: NaCl 137 mm; KCl 2.7 mm; CaCl2 1.8 mm; MgCl2 1.05 mm; NaH2PO4 0.2 mm; NaHCO3 11.9 mm; glucose 5.6 mm) containing 20 mm glycylglycine buffer, pH 8.5, luminol (30 µm), peroxidase (60 µg; 18 units) and choline oxidase (1 mg; 10 units), and the spontaneous reaction was allowed to proceed. Afterwards, depolarizing solutions and/or drugs were added to the vial (total volume: 1 mL), and the light resulting from peroxidase-catalysed oxidation of luminol by hydrogen peroxide was measured at 425 nm. Control experiments included additions of choline and acetylcholine to the incubation medium (Fig. 1), and incubations in the absence of choline oxidase. For determination of total choline release, the areas under the peaks were cut out and weighed (time of incubation: 4 min). Due to the high amount of choline oxidase in the assay system, the choline assay was linear from 10 to 300 pmol of choline.

Figure 1.

Luminometric determination of the release of choline from synaptosomes upon KCl-induced depolarization. Synaptosomes were incubated together with a choline detection system (choline oxidase, peroxidase, luminol) in the reaction vial of a luminometer. After spontaneous reactions were completed, KCl (60 mm) added to the vial at zero-time, triggered a chemiluminescence signal (A) lasting for 3–4 min that was measured at 425 nm. Addition of ACh (1 nmol) failed to cause a signal (B), because AChE had been inactivated by prior incubation of the synaptosomes with DFP. AU, arbitrary units.

Preparation and superfusion of hippocampal slices and luminometric analysis of choline in superfusates

Animals of different ages (7–35 days) were decapitated, and hippocampal slices (400 µm) were prepared and superfused at 35°C with Tyrode solution (0.7 mL/min). In some experiments, CaCl2 was reduced to 0.2 mm (low calcium condition) or omitted. All superfusion solutions were continuously gassed with carbogen (95% O2, 5% CO2). The slices were first incubated with 0.1 mm DFP for 30 min in order to prevent choline release from acetylcholine by the action of acetylcholinesterase. Subsequently, the slices were washed for 40 min, and basal choline efflux was determined. 4-AP (50–5000 µm) and KCl (15–120 mm) were added in Tyrode solution; in the case of KCl, the NaCl concentration was reduced to maintain isoosmolarity. The superfusates were collected at 10-min intervals and analysed for choline content.

Choline from superfusates was determined by a chemoluminescence assay (Israel and Lesbats 1982). Briefly, 10 µL aliquots of the superfusates were given to a reaction mixture consisting of 20 mm Tris buffer pH 8.6, 10 µm luminol, 1 µg (0.3 units) peroxidase and 125 µg (1.25 U) choline oxidase (total volume: 1 mL), and the light resulting from peroxidase-catalysed oxidation of luminol by␣hydrogen peroxide was measured at 425 nm in a LKB-Wallac luminometer. The assay was linear from 1 to 5 pmol choline.

Microdialysis experiment and determination of choline by HPLC

I-shaped, concentric probes were manufactured according to Santiago and Westerink (1990). The probes had an outside diameter of 0.24 mm and an exchange length of 4 mm and were equipped with a dialysis membrane (Filtral AN-69 HF; Hospal, Meyzieu, France) with a molecular weight cutoff of 10 000 Da. For probe implantation, adult rats (12 weeks of age) were anaesthesized with pentobarbital (60–80 mg/kg i.p) and placed in a stereotactic frame. The probe was implanted into the right ventral hippocampus using␣the following coordinates (from lambda): AP + 2.5 mm; L − 5.8 mm; DV − 7.5 mm (Paxinos and Watson 1986). The experiments were carried out on freely moving animals on the first and second day following surgery. The microdialysis probes were perfused at a constant rate of 2.0 µL/min with artificial cerebrospinal fluid (concentrations: NaCl 147 mm; KCl 4 mm; CaCl2 1.2 mm; MgCl2 1.2 mm) containing 10 µm neostigmine. After four samples had been taken to estimate the basal efflux of choline, the perfusion fluid was changed to a solution containing 30 or 60 mm KCl (NaCl was reduced accordingly to maintain osmolarity). Aliquots of the dialysis fluid were collected in 10-min intervals and analysed for choline by HPLC.

Microdialysis samples were injected directly into the HPLC which consisted of a BAS PM-80 pump, polymer column (MF-8904; 350 × 1 mm) and enzyme reactor carrying immobilized choline oxidase (BAS Sepstick), and an electrochemical detector equipped with a platinum electrode operating at + 500 mV (BAS LC4C). The eluent (flow rate: 130 µL/min) consisted of 29 mm NaH2PO4, 22 mm sodium acetate and 0.37 mm EDTA (pH 8.4). The retention time for choline was 13 min, the detection limit 50 fmol/5 µL injection volume. Choline efflux was calculated as per cent of basal efflux which was defined as the average output of four consecutive samples that did not differ by more than 10%.


K+-evoked choline efflux from cortical synaptosomes of adult rats

High K+-solutions were used to analyse the relationship between membrane potential and efflux of free intracellular choline in synaptosomes. For this purpose, we established a modified luminometric assay procedure in which synaptosomes were incubated in the reaction tube together with the detection system for choline. Upon depolarization with K+ (Fig. 1), synaptosomes instantaneously released choline; the signal returned to baseline within 3–4 min. The finding that K+ failed to yield any signal in the absence of choline oxidase demonstrated the specificity of the choline detection system (data not shown). Moreover, ACh was excluded as a source for choline by pre-treatment of the synaptosomes with DFP, an irreversible inhibitor of acetylcholinesterase. Accordingly, addition of exogenous ACh did not cause a choline signal within the first few minutes (Fig. 1). AACOCF3 (10 µm), an inhibitor of intracellular phospholipase A2, had no effect on K+-evoked choline release (data not shown).

Figure 2 shows the linear relationship between increasing K+ concentrations (7 mm up to 120 mm) and synaptosomal efflux of choline, which was not significantly affected by reduction of Ca2+ to 0.2 mm (p > 0.5) or to zero (p > 0.2; Fig. 4a). Likewise, inhibition of Ca2+ channels (verapamil, 30 µm) and of Ca2+/calmodulin kinase II (KN-62, 10 µm) had no effect on the K+-evoked efflux of choline (data not shown). Furthermore, in synaptosomes permeabilized with Staphylococcus aureusα-toxin (Sarri et al. 1998), an increase of the calcium concentration from 0.1 to 15 µm also did not affect choline efflux (data not shown).

Figure 2.

Release of choline from synaptosomes upon KCl-induced depolarisation: concentration–response curve. Synaptosomes were incubated as described in Fig. 1. KCl in various concentrations from 7.5 to 120 mm was added to the incubated synaptosomes, and the light resulting from peroxidase-catalysed oxidation of luminol by hydrogen peroxide was detected. Choline release was calculated from the area under the curve during a 4-min interval after KCl addition. The choline assay was linear from 10 to 300 pmol (see Methods). Means ± SEM of four experiments are presented.

Figure 4.

Ca2+ dependency of K+-induced choline release. (a) Synaptosomes from adult rats were incubated as described in Fig. 1, and K+-induced choline release was determined after depolarization with 60 mm K+ in buffers containing 1.8 mm, 0.2 mm or zero Ca2+. Choline release was quantified as area under curve; 100% was equivalent to 156 pmol/4 min. (b) Hippocampal slices from 7-day-old rats were superfused as described in Fig. 3. At zero-time, the superfusion solutions were switched to solutions containing Tyrode solution (2.7 mm KCl, lower curves) or depolarizing K+ concentration (60 mm KCl, upper curves) containing normal CaCl2 (●, 1.8 mm) or a reduced CaCl2 concentration (○, 0.2 mm). Choline concentrations of the superfusates were determined by luminometry and expressed as percentages of basal choline efflux (see legend of Fig. 3). Means ± SEM of four experiments are presented.

Effects of high K+ on choline efflux from brain slices of adult and young rats

Similar experiments as described for the synaptosomal preparation were carried out using superfused hippocampal slices from 1-week-old, 2-week-old and adult rats which had been pre-incubated with DFP. The basal efflux in slices from adult rats (151 ± 12 nm, n = 12) was only half of that detected in young rats (297 ± 44 nm, n = 9). Moreover, superfusion with high K+ concentrations (15, 30 and 60 mm) failed to increase choline efflux from mature tissue, but dramatically increased the efflux of choline in immature slices in a concentration-dependent manner (Fig. 3). In the latter experiments, reduction of Ca2+ from 1.8. to 0.2 mm had no effect on either basal or K+-evoked choline efflux (Fig. 4b). Figure 5 illustrates the ontogenetic disappearance of the depolarization-induced choline efflux using high K+ (Fig. 5a) or 4-aminopyridine (4-AP), a K+ channel blocker (Fig. 5b). The strong choline efflux observed at post-natal week 1 was drastically reduced from the first to the second postnatal week and was absent in slices from adult rats.

Figure 3.

Effects of KCl-induced depolarisation on choline efflux from hippocampal slices prepared from adult rats and 1-week-old rats. After equilibration (zero-time), the superfusion solution was switched from 2.7 mm to 15, 30 or 60 mm KCl (NaCl content was reduced accordingly to maintain isoosmolarity). The eluates were collected in 10-min intervals. Choline concentrations of the superfusates were determined by luminometry and expressed as increases (in per cent) of basal choline efflux which was 151 ± 12 pmol/mL (106 ± 8 pmol/min) in mature slices (n = 12) and 297 ± 44 pmol/mL (208 ± 31 pmol/min) in immature slices (n = 9). Means ± SEM of four to seven experiments are presented.

Figure 5.

Post-natal fading of the depolarization-induced choline efflux from hippocampal slices prepared from 1-week-old, 2-week-old or 5-week-old rats. After equilibration, (a) high K+ (○, 15 mm; ◆, 30 mm; or ●, 60 mm) or (b) 4-AP (○, 50 μm; ◆, 500 μm; ●, 5000 µm) were added to the superfusion medium. Choline contents of the eluates were determined by luminometry and expressed as increases (in per cent) of the basal choline efflux (see legend of Fig. 3). Presented are means ± SEM of 5–7 experiments.

Effect of HC-3 on choline efflux from brain slices of adult and young rats

HC-3 in a concentration of 1 µm enhanced the efflux of choline only in hippocampal slices from adult rats, but had no effect in slices from 1-week-old rats (Fig. 6). A higher HC-3 concentration (10 µm) enhanced choline efflux also in young rats, but the effect was considerably stronger in slices from the adult animals (p < 0.01).

Figure 6.

Effect of HC-3 on choline efflux from hippocampal slices of adult rats (solid lines) and 1-week-old rats (punctured lines). HC-3 (□,␣1 µm; ▪, 10 µm) was added to the medium at zero-time. Ordinate, changes of the basal efflux of choline (1.51 ± 0.12 pmol/10 µL in adult rats; 2.97 ± 0.44 pmol/10 µL in 7-day-old rats) determined before addition of HC-3. Means ± SEM of six to eight experiments are presented.

In vivo effect of high K+ on the extracellular choline level of the hippocampus

Finally, we utilized the microdialysis technique to determine in vivo changes of the extracellular choline level in adult rat hippocampus (in the presence of 10 µm neostigmine). Neither 30 nor 60 mm K+ caused an increase of the extracellular choline level (Fig. 7). Rather, there was a small but significant decrease of choline efflux during infusion with 60 mm K+ (p < 0.05).

Figure 7.

Effect of high K+ on extracellular choline levels in the brain in␣vivo. Microdialysis probes were implanted into the ventral␣rat␣hippocampus, perfused with artificial CSF containing 10 µm neostigmine, and choline efflux was measured by HPLC and expressed as percentage of basal efflux which was 1.41 ± 0.18 pmol/min (equivalent to 0.7 µm). During the time indicated in grey (0–60 min), the KCl concentration was elevated from 2.7 to 30 mm (○) or 60 mm (●) (NaCl content was reduced accordingly to maintain isoosmolarity). Means ± SEM of five experiments each are presented.


Choline is an ubiquitous molecule with a permanent positive charge which is distributed between extra- and intracellular spaces according to the Nernst equation (see Introduction). Thus, the extracellular brain choline concentration (3–5 µm) is 10-fold lower than the intracellular level (approximately 50 µm; Tucek 1988, 1993; Klein et al. 1993). Unexpectedly, we could not observe a release of choline from brain slices of adult rats by high K+-solutions (Fig. 3) or, under in vivo conditions, during the infusion of high K+-solutions into the hippocampus of adult rats (Fig. 7). We therefore decided to investigate possible homeostatic mechanisms preventing net efflux of choline under depolarizing conditions.

As a first approach, we investigated choline release in synaptosomes, i.e. nerve endings that do not contain neuronal or glial cell bodies. As predicted by the Nernst equation, depolarization of synaptosomes induced a marked release of choline (Fig. 1); we found an almost linear relationship between choline efflux and K+ concentration (Fig. 2). For these experiments, we used a real-time luminometric choline assay (Israel and Lesbats 1982) in which choline released from synaptosomes was immediately consumed by choline oxidase, a process which could be followed continuously by the emission of light due to the peroxidase-catalysed oxidation of luminol by hydrogen peroxide derived from choline oxidation. The assay was modified to allow the detection of large amounts of choline (> 100 pmol; see Methods). The time course of the effect (Fig. 1) suggests that addition of KCl solutions caused changes in membrane potential which led to the new equilibrium of choline distribution within few minutes (Tucek 1988). Released choline was oxidized within 3–4 min, whereas small amounts of choline continued to be released at later time points, possibly due to phospholipid hydrolysis. The most likely source of released choline is free cytoplasmic choline translocated from the synaptosome, a process which likely involves reverse transport of choline through membrane choline carriers (Marchbanks et al. 1981). Synaptosomal uptake of newly released choline is prevented by the presence of the choline-oxidizing system. ACh release did not contribute to choline formation because acetylcholinesterase was inhibited (Fig. 1). A contribution of phospholipid breakdown during the rapid phase of choline release (3–4 min) is unlikely because the phospholipase A2 inhibitor, AACOCF3, was ineffective (not illustrated). Synaptosomal phospholipase D activity does not contribute to choline release because it is inhibited, not activated, during depolarization (Waring et al. 1999). Finally, synaptosomal efflux of choline was not significantly affected by␣manipulations of Ca2+ and Ca2+-dependent processes (Fig. 4a).

In hippocampal slices taken from young and adult rats, we observed a striking ontogenetic dependence of K+-evoked choline release: immature slices taken from 7-day-old rats released large amounts of choline in a calcium-independent manner, whereas this effect was diminished in 14-day-old rats and basically absent in mature tissue taken from 35-day-old rats (Figs 3–5). The most likely explanation for the failure of K+-induced depolarization to enhance choline efflux from mature slices is the existence of a rapid cellular removal pathway. Cellular choline uptake and subsequent phosphorylation may play a prominent role (Francescangeli et al. 1977; Wuttke and Pentreath 1990); a major contribution from glial cells is likely and compatible with our observation of choline release in synaptosomes. The molecular nature of the choline uptake system responsible for cellular removal in adult rats remains speculative. We made an attempt to characterize a hemicholinium-3 (HC-3)-sensitive uptake system in hippocampal slices. HC-3 is known to inhibit choline uptake in synaptic nerve endings with a Ki value of 0.1–1 µm for the high-affinity choline uptake and a Ki value of 50–100 µm for the low-affinity choline uptake (Tucek 1988). From our results with HC-3 (Fig. 6) we conclude that a system with high-affinity for HC-3 is present in adults rats because 1 µm HC-3 increased choline release. The fact that 10 µm HC-3 caused a stronger effect argues for the participation of another choline transporter with lower affinity for HC-3. The recently cloned high-affinity choline uptake (HACU) located at cholinergic nerve endings (Okuda et al. 2000) may partially contribute to these effects because it is highly sensitive to HC-3. The HACU develops post-natally after day 7 (Coyle and Yamamura 1976) and reaches adult levels not before post-natal week 5 (Aubert et al. 1996). The neuronal HACU is known to be Na+-dependent and (at least partially) inhibited during K+ depolarization due to a reduction of the transmembrane sodium gradient (Kuhar and Murrin 1978). The presence of a glial HACU (Massarelli et al. 1974) may be significant in this respect. A non-cholinergic HACU present in oligodendrocytes was recently cloned which showed a high affinity for HC-3 (O'Regan et al. 2000). This system may also guarantee the supply of choline for the synthesis of cell membrane phospholipids (phosphatidylcholine, sphingomyelin) in glial cells.

The molecular nature of the choline uptake with low HC-3 sensitivity is also unclear. Low-affinity uptake of choline in rat brain develops during the brain growth spurt and in parallel with the increase in brain weight and with the post-natal development of glial cells, in particular of astrocytes (Vernadakis 1988). Recent reports from the literature have identified choline transporters with low-affinity for choline which are expressed in the neonatal and mature choroid plexus, respectively (Villalobos et al. 1999; Sweet et al. 2001). Both choline transporters are Na+-independent and pH-sensitive and were inhibited by depolarisation; the organic cation transporter (OCT2) identified in mature ventricular tissue was strongly dependent on membrane potential (Sweet et al. 2001). However, the location of these transporters in the choroid plexus and their very low sensitivity to HC-3 question their role in choline uptake in the mature hippocampal slice as described in the present study.

While the molecular nature of the choline transporters responsible for ontogenetic development remains to be resolved, the present results have bearing on the phenomenon of choline homeostasis. Thus, although choline's major role is as a biosynthetic precursor of ACh and phospholipids (Löffelholz 1998), the substance also possesses agonistic properties at cholinoceptors described several decades ago (Le Heux 1919), and recently it was reported that choline is a selective agonist on α7 nicotinic receptors (Papke et al. 1996; Alkondon et al. 1997). The apparent EC50 of choline for this effect was rather high (1.6 mm) but continuous exposure of the cultured hippocampal neurones to 37 µm choline caused a half-maximal desensitization of the α7 receptors (Alkondon et al. 1997). Our findings in adult rat slices, however, argue against the accumulation of high choline concentrations in the extracellular space (Fig. 3). In our microdialysis experiments, high K+ caused, in fact, a transient decrease of the extracellular choline level in the hippocampus (Fig. 7). This is in agreement with previous work in which we found that, during activation of cholinergic neurones, the rate-limiting step of ACh biosynthesis is shifted from the high affinity uptake (Kd∼1–2 µm) to extracellular choline availability because local choline levels fall below the resting choline concentrations (∼3–5 µm) during stimulation (Köppen et al. 1997). In the microdialysis experiment, the infusion of 30–60 mm KCl likely produced K+ concentrations in the extracellular space (∼5–15 mm) which initiated action potentials rather than sustained depolarization. This experiment therefore is in agreement with a previous study in the heart in which cholinergic nerve stimulation caused a long-lasting decrease of choline efflux due to neuronal choline re-uptake (Lindmar et al. 1980).

Although the slice experiments argue against an accumulation of free choline, our findings with synaptosomes indicate a local release of choline from synaptic endings during depolarization. In the mature brain, however, a local build-up of choline, or an increase of the ambient choline level, is evidently prevented by high-affinity choline uptake mechanisms located on nerve endings or surrounding glial cells, respectively. This conclusion is supported by the finding that the cholinergic innervation of the brain is largely asynaptic (Descarries 1998). Although we cannot exclude short-lasting, locally restricted increases of extracellular choline during neuronal activity, evidence was obtained in the present study that the homeostatic systems controlling free choline include rapid cellular (re)uptake and protect the brain against the consequences of excess extracellular choline concentrations.


This work was supported by a grant from the Deutsche Forschungsgemeinschaft (Kl 598/6–1) and by the Stiftung VERUM.