B R A I N R E S E A R C H 1 4 3 5 ( 2 0 1 2 ) 1 5 – 2 3

Ava i l ab l e on l i ne a t www.sc i enced i r ec t . com

www.e l sev i e r . com/ loca te /b ra i n res

Research Report

Antidepressants inhibit proton currents and tumor necrosisfactor-α production in BV2 microglial cells

Jin-Ho Songa,⁎, William Marszalecb, Li Kaib, Jay Z. Yehb, Toshio Narahashib

aDepartment of Pharmacology, College of Medicine, Chung-Ang University, 221 Heuksuk-Dong, Dongjak-Gu, Seoul 156-756, Republic of KoreabDepartment of Molecular Pharmacology and Biological Chemistry, Northwestern University Feinberg School of Medicine, 303 E. Chicago Avenue,Chicago, IL 60611, USA

A R T I C L E I N F O

⁎ Corresponding author. Fax: +82 2 817 7115.E-mail address: jinhos@cau.ac.kr (J.-H. So

0006-8993/$ – see front matter © 2011 Elseviedoi:10.1016/j.brainres.2011.11.041

A B S T R A C T

Article history:Accepted 18 November 2011Available online 1 December 2011

Proton channels are gated by voltage and pH gradients, and play an important role in themicroglial production of pro-inflammatory cytokines, which are known to be suppressedby antidepressants. In the present study we tested the hypothesis that cytokine inhibitionby antidepressants is due to an inhibitory action on proton currents by comparing their ef-fects on tumor necrosis factor-α production with the effects on the proton currents in BV2murine microglial cells. Imipramine, amitriptyline, desipramine and fluoxetine potentlyand reversibly inhibited proton currents at micromolar concentrations at an intracellular/extracellular pH gradient of 5.5/7.3. Raising extracellular pH to 8.3 sped up the rate and en-hanced the extent of block whereas raising intracellular pH to 6.3 reduced the blocking po-tency of imipramine. These results support a mechanism where the uncharged drug formpenetrates the cell membrane, and the charged form blocks the proton channel from the in-ternal side of membrane. This mode of action was corroborated by an experiment with imi-praminium, a permanently charged quaternary derivative, which showed far less blockcompared to imipramine. The lipopolysaccharide-induced release of tumor necrosisfactor-α was inhibited by imipramine at concentrations comparable to those inhibitingthe proton current. These results support the hypothesis that tumor necrosis factor-α inhi-bition by imipramine is related to its inhibitory effects on proton channels.

© 2011 Elsevier B.V. All rights reserved.

Keywords:AntidepressantImipramineMicrogliaProton channelTumor necrosis factor-α

1. Introduction

Voltage-gated proton channels, first demonstrated in snailneurons (Thomas and Meech, 1982), are found predominant-ly in cells subjected to low pH in their internal environmentsuch as epithelia, leukocytes and related cell lines. The fun-damental function of proton channels is acid extrusionfrom cells (DeCoursey, 2003, 2008; Eder and DeCoursey,2001). In the membrane of phagocytic cells, including micro-glia, activation of NADPH oxidase produces intracellular pro-

ng).

r B.V. All rights reserved.

tons as it transports electrons across the membrane to formextracellular superoxide anions. It also initiates an intracel-lular reactive oxygen species signaling pathway that en-hances the production of pro-inflammatory cytokines suchas tumor necrosis factor-α (Kim et al., 2010). Supporting evi-dences for the role of proton channel and NADPH oxidaseare that zinc, a proton channel inhibitor, and diphenyliodo-nium, a NADPH oxidase inhibitor, have been found to sup-press tumor necrosis factor-α secretion (Qian et al., 2007;Von Bulow et al., 2007).

16 B R A I N R E S E A R C H 1 4 3 5 ( 2 0 1 2 ) 1 5 – 2 3

In an in vitro study, imipramine has been shown to exertneuroprotective effects by suppressing lipopolysaccharide-induced inflammatory processes and modulating down-stream cascades (Peng et al., 2008). Antidepressant drugshave also been found to reduce microglial activation as evi-denced from their inhibition of the production of pro-inflammatory cytokines (Hashioka et al., 2007; Obuchowiczet al., 2006). For instance, imipramine and clomipramine re-duced the lipopolysaccharide-stimulated production of nitricoxide and tumor necrosis factor-α in murine microglial BV-2cells and mouse microglia in primary culture (Hwang et al.,2008).

The neuroprotective effects of antidepressants have beenattributed to the cAMP-dependent protein kinase A pathway(Hashioka et al., 2007) or the p38 mitogen activated protein ki-nase pathway (Hwang et al., 2008), the latter being regulatedby reactive oxygen species. Since the NADPH oxidase activityis supported by proton channels that tend to dissipate protonions accumulation and since NADPH oxidase inhibitor diphe-nyliodonium suppresses lipopolysaccharide-induced tumornecrosis factor-α production in microglia (Qian et al., 2007),the present study was aimed to test the hypothesis that anti-depressants might interfere with microglial activation byinhibiting proton channels. We found that the tested antide-pressants potently inhibited proton currents in BV2microglialcells and that one antidepressant, imipramine, inhibited theproton current to an extent similar to its inhibition of tumornecrosis factor-α production.

Fig. 1 – Effects of antidepressants on voltage-gated protoncurrents in BV2 cells. (A) Currents were evoked by 2-s voltagepulses to +30 mV from a holding potential of −60 mV underthe condition of intracellular pH/extracellular pH (pHi/pHo)=5.5/7.3 (filled circles) and pHi/pHo=5.5/8.3 (open circles).The inset illustrates current traces in response to 10 μMimipramine for 10 min and washout for 10 min under thecondition of pHi/pHo=5.5/7.3. (B) Dose–response relationshipfor antidepressants inhibition of proton currents. Cells weretreated with each drug for 10 min except fluoxetine in whichthey were treated for 20 min under the condition of pHi/pHo=5.5/7.3. The proton current amplitude at the end of thedepolarizing pulse after drug treatment was normalized toan initial control value. Each measurement was carried outfrom a different cell. Data are means±S.E.M. from 6 to 14cells. Curves were drawn after fit with Hill equation.

2. Results

2.1. Antidepressants reversibly inhibit voltage-gated protoncurrents in BV2 cells in a dose-dependent manner

Voltage-gated proton currents in BV2 cells were evoked by 2-sdepolarizing pulses to +30 mV from a holding potential of−60 mV at intracellular pH/extracellular pH (pHi/pHo)=5.5/7.3(Fig. 1A). The currents were activated slowly, reaching apseudo-plateau level during 2-s depolarizations (see currenttraces in the inset). Upon repolarization to the holding poten-tial a long tail current developed, which reflected closure ofthe proton channels, often being referred to as deactivation.As indicated in the time course of inhibition in Fig. 1A, perfu-sion of 10 μM imipramine (a tricyclic antidepressant) slowlyinhibited the proton current, reaching a steady-state level of29±6.0% of the control (n=6) with a half-time of 4 min. Thisinhibition was slowly reversible after washout (Fig. 1A).Thus, proton current inhibition by antidepressants was mea-sured after 10 min drug perfusion.

The dose–response relationship for imipramine inhibitionof proton current was constructed as shown in Fig. 1B. Datawere collected from 6 to 14 different cells for each concentra-tion of imipramine. The mean values of the remaining protoncurrent fraction were fitted with the following Hill equation:

Iimipramine=Icontrol ¼ 1= 1þ imipramine½ �=IC50ð ÞnH� �;

where Iimipramine is the proton current in the presence of imip-ramine, Icontrol is the control proton current before application

of imipramine, [imipramine] is the concentration of imipra-mine, IC50 is the half maximal inhibitory concentration, andnH is the Hill coefficient. Best fits were achieved when IC50

and nH were 5.7 μM and 1.45, respectively. At the same exper-imental condition, amitriptyline, another tricyclic antidepres-sant and structurally very similar to imipramine, showed analmost identical dose–response relationship with an IC50 andnH of 5.8 μM and 1.23, respectively (n=6–7 for each concentra-tion). Desipramine, a demethylated metabolite of imipraminethat retains antidepressant activity reduced the proton cur-rent at 10 μM to 30±1.0% of the control (n=6), comparable tothe effect of 10 μM imipramine.

17B R A I N R E S E A R C H 1 4 3 5 ( 2 0 1 2 ) 1 5 – 2 3

Fluoxetine is a selective serotonin reuptake inhibitorwhich differs from tricyclic antidepressants structurally andpharmacologically. Fluoxetine inhibited proton current evenmore potently than the other three tricyclic antidepressants.However, the inhibition took a longer time to fully developthan did with tricyclic antidepressants. Thus, the effect of flu-oxetine was measured after a 20 min application of the drug.The IC50 and nH values for fluoxetine were 2.1 μM and 1.42, re-spectively (n=6 for each concentration).

2.2. Mechanisms of imipramine block: changes inintracellular pH

A drug of weak base in the bath solution can exist as a neutralform or a cationic charged form and their proportion dependson pKa of the drug and pH of the solution according to theHanderson–Hasselbalch equation. Generally speaking, theneutral form can penetrate the cell membrane while thecharged form cannot. Once inside the cell the neutral drug

Fig. 2 – Lack of effect of imipramine on the reversal potential of pfrom a holding potential of −60 mV and then repolarized to pote(pHi/pHo=5.5/7.3). (A, B) Representative families of proton curren10 min, respectively. Insets show enlarged views of tail currentsrepolarizing pulse from the currents in A and B were measuredpotential was determined to be −82.5 mV, and −82 mV, respectiv(D) Comparison of reversal potentials before and after treatmentindividual cells and filled circles are means±S.E.M (n=10). The d

can pick up internal proton to become the cationic form.This raises the question of whether the drug might sequesterinternal protons to raise intracellular pH, thus lowering theproton gradient for channel activation and reducing the driv-ing force for proton current. Such an effect might reduce theproton current as proposed by DeCoursey (2003). In order toanswer the question whether the inhibitory effect of imipra-mine on proton current was due to a change in intracellularpH, the following two measurements were performed.

2.2.1. Imipramine does not change the reversal potential ofproton currentsIf internal proton concentrations were changed by the accu-mulated imipramine, one would expect a change in the rever-sal potential for the proton current measured before and afterimipramine treatment (Fig. 2). As indicated in this figure, themembrane was depolarized to +50 mV from a holding poten-tial of −60 mV for 2 s to open the proton channels and thenrepolarized to various levels between −50 mV and −100 mV

roton currents. Membrane was depolarized to +50 mV for 2 sntials between −50 and −100 mV in 10-mV decrementsts before and after treatment with 5 μM imipramine for. (C) The tail current amplitudes (Itail) at the beginning of theand plotted against the repolarizing potential. The reversalely, before and during application of 5 μM imipramine.with 5 μM imipramine. Open circles are matched data ofifference was not statistically significant (P>0.2).

Fig. 3 – Effect of imipramine on the activation voltage of proton currents. Currents were evoked by depolarizing pulses topotentials between −50 and +50 mV in 10-mV increments from a holding potential of −60 mV (pHi/pHo=5.5/7.3). (A, B)Representative families of proton currents before and after treatment with 5 μM imipramine for 10 min, respectively. (C, D)Current–voltage curves for proton current at the end of depolarizing pulse and tail current at the beginning of therepolarization, respectively, from the currents in A and B. For the comparison the current after imipramine treatment wasnormalized to the control (crossed filled circles). (E, F) Conductance–voltage curves for proton currents and tail currents,respectively. Conductance was normalized with respect to the conductance at +50 mV and plotted against the test potential(n=9). Curves were drawn after fit with a Boltzmann function.

18 B R A I N R E S E A R C H 1 4 3 5 ( 2 0 1 2 ) 1 5 – 2 3

19B R A I N R E S E A R C H 1 4 3 5 ( 2 0 1 2 ) 1 5 – 2 3

in 10-mV decrements at 15-s intervals with pHi/pHo=5.5/7.3.Examples of the proton current families in the absence andpresence of 5 μM imipramine for 10 min are shown in Figs. 2,A and B, respectively. The tail current amplitude at the begin-ning of repolarization was measured at each repolarizing po-tential and plotted as the function of the repolarizingpotential as illustrated in Fig. 2C. The reversal potential wasdetermined at the potential where the direction of the protoncurrent was changed from inward to outward going direction.In the absence and presence of imipramine the reversal po-tentials were −85.8±0.9 and −83.8±1.6 mV (n=10), respectively(Fig. 2D). The difference was not statistically significant(P>0.1). Even at higher concentrations, imipramine did notappreciably change the reversal potential.

2.2.2. Imipramine does not shift the activation voltage ofproton currentThe activation of proton currents depends on voltage and thepH gradient. Thus, another way to test the possibility thatimipramine alters intracellular pH is to examine the effect ofimipramine on the activation voltage of the proton current.The current–voltage relationship was constructed using a se-ries of 2-s depolarizing pulses in a range between −50 mVand +50 mV in 10-mV increments applied at 10-s intervalsfrom a holding potential of −60 mV at pHi/pHo=5.5/7.3(Fig. 3). Examples of the proton current families in the absenceand presence of 5 μM imipramine for 10 min are shown inFigs. 3, A and B, respectively. The proton current at the endof the depolarizing pulse and the tail currents at the beginningof repolarization following each test potential were plotted toform current–voltage curves (Figs. 3, C and D, respectively).The current activation threshold was approximately −40 mV.Imipramine inhibited both proton current and tail current tothe same extent at all test voltages, as evidenced from theanalysis showing their completely overlapped current–voltagecurves after the imipramine curve was normalized to thecontrol.

To determine the voltage-dependence of proton channelactivation, the proton conductance (G) was calculated. Theproton current for each test pulse potential (Vm) was dividedby the driving force (Vm−Vrev), where Vrev is the reversal po-tential. Normalized proton conductance was plotted againstVm and fitted with a Boltzmann function according to theequation:

G=Gmax ¼ 1= 1þ exp V0:5−Vmð Þ=kð Þð Þ;

where Gmax is the maximal conductance at +50 mV, V0.5 is themembrane potential at which the half-maximal channel openprobability occurs and k is the slope factor that describes thesteepness of voltage dependent activation kinetics. As is im-plied by the current–voltage curves imipramine did not shiftconductance–voltage curves for either proton current or tailcurrent. The control parameters for V0.5 and k for proton cur-rent were calculated to be −6.5±1.2 and 14.2±0.2 mV, respec-tively (n=9). After treatment with imipramine they were−7.1±2.0 and 13.9±0.5 mV, respectively. For tail current theywere calculated to be −10.0±1.6 and 12.0±0.5 mV in the con-trol condition, and were −9.4±2.1 and 12.0±0.9 mV in thepresence of imipramine. In either case the changes in these

parameters were not significant and conductance–voltagecurves overlapped almost completely (Figs. 3, E and F).

The lack of effect of imipramine on the voltage depen-dence of activation of proton currents in combination withits lack of effect on the reversal potential strongly suggeststhat imipramine at 5 μM does not affect intracellular pH.Thus, the inhibitory action of imipramine at the concentra-tion that reduced nearly 50% of the proton current is not dueto a reduction in the intracellular proton ion concentrationnor to an effect on the channel gating.

2.3. Roles of the neutral form and charged form in blockingproton currents

Like many local anesthetics imipramine is a tertiary amineand can exist as either a neutral form or a charged formaccording to its pKa and the pH of the internal and external so-lutions. To test the roles of the neutral and the charged formsin blocking proton currents, we followed the experimentalprotocols pioneered by Narahashi et al. (1970) used to demon-strate that it is the charged form of the local anesthetic thatblocks sodium channels from the intracellular side, whilethe neutral form is the molecular species that crosses themembrane. The gist of this experimental design is changingexternal and internal pHs to alter the neutral form and thecharged form of the drug. Since the activation of the protoncurrent also depends on the pH gradient, both intracellularand extracellular pHs were changed so that the pH gradientwas kept constant.

2.3.1. Imipramine block at a constant pH gradient by changingboth intracellular and extracellular pHsWhen the extracellular pH was increased by 0.5 units to 7.8,the intracellular pH was also raised by 0.5 units to 6.0 inorder to keep the same pH gradient thereby not affecting gat-ing of the proton channel. Indeed, the conductance–voltagecurves and voltage dependence of activation kinetics of theproton currents measured at pHi/pHo=6.0/7.8 were similar tothose seen at the control pHi/pHo=5.5/7.3 (data not shown).When 10 μM imipramine (pKa=9.5) was applied, unchargedimipramine was 0.196 μM at pH 7.8 and 0.063 μM at pH 7.3.But at equilibrium and according to the Handerson–Hassel-balch equation, there were the same intracellular chargedform (630 μM) at both pHi/pHo=6.0/7.8 and pHi/pHo=5.5/7.3.Despite a near 3-fold increase in the external unchargedform at pH 7.8, imipramine exhibited a similar blocking poten-cy with an IC50 of 7.1±0.40 μM (n=6), which is not significantlydifferent from 5.7±0.60 μM obtained at pHi/pHo=5.5/7.3(P>0.2). These results suggest that the neutral form itself isnot the sole determinant in blocking proton current andpoint to a block by intracellular charged imipramine. Thisidea was further tested by the following experiments.

2.3.2. Altering extracellular pH affects potency of imipramineblockIt has been demonstrated that raising extracellular pH in-creases the potency of local anesthetics to block sodium chan-nel currents. This result has been interpreted as being due toan increase in the external neutral form which expeditesdrug entry into the cell. This in turn increases the intracellular

Fig. 4 – Dose–response relationship for imipraminium toblock proton current. Data were fitted by Hill equation withIC50 of 3.37 mM and nH of 0.72 (n=6).

20 B R A I N R E S E A R C H 1 4 3 5 ( 2 0 1 2 ) 1 5 – 2 3

charged species that are responsible for blocking sodium cur-rents. This experimental variation of extracellular pH was re-peated for imipramine in studying its ability to block protoncurrent. Raising extracellular pH from 7.3 to 8.3 increasedimipramine block so that 10 μM imipramine at pH 8.3 gave a92.0±5.00% (n=6) block as compared to the block of 71.0±6.00% obtained at pH 7.3. The onset of block by 10 μM at pH8.3 developed very rapidly, reaching near steady-state blockwith a half-time of 2 min as compared to the slower blockwith a half-time of 4 min at pH 7.3 (see Fig. 1A).

Proton current block produced by 3 μM imipramine at pH8.3 (89.0±7.00%, n=6) was identical to the block observedwith 30 μM imipramine at pH 7.3 (91.0±2.00%, n=6). Nearlyall of the 30 μM imipramine at pH 7.3 is in the charged form.This is almost 10 times the concentration of the chargedform for 3 μM imipramine at pH 8.3. Yet the level of blockwas similar in each case. This tends to rule out the possibilitythat an extracellular charged form of imipramine is responsi-ble for the block. Yet, these two experimental conditions,30 μM at pH 7.3 and 3 μM at pH 8.3, would give the same theo-retical concentration of the neutral form, 0.18 μM, and intra-cellular charged form, ~2 mM, at pH 5.5. Thus, one could notdifferentiate whether it is the neutral form or the intracellularcharged form that is active in blocking the proton current.

Therefore, to test the hypothesis that it is the intracellularcharged form of imipramine that blocks the proton current,two other types of experiments were carried out: one wherealtering internal pH was used to change concentrations ofthe charged form and the other where imipraminium, a per-manently charged form of imipramine, was used to test itsability to block proton currents.

2.3.3. Raising intracellular pH to reduce the charged form ofimipramineTo test the role of the intracellular charged form in blockingthe proton current, the internal pH was raised from 5.5 to 6.3to reduce concentrations of the charged species while exter-nal pH was kept at 7.3. Block by 10 μM imipramine reached26.5±4.11% at pH 6.3 (n=6), which is significantly smallerthan 71.0±6.00% obtained at pH 5.5 (n=6) (P<0.01). Underthese two conditions, since the neutral form was identical,0.063 μM, and since at equilibrium the theoretical intracellularconcentration of the charged form was expected to reach630 μM at pH 5.5 and 100 μM at pH 6.3, these experimental re-sults are consistent with the hypothesis that the charged formof imipramine blocks proton current from inside while theneutral form is essential for crossing the membrane.

2.3.4. Effect of imipraminium on the proton currentThe quaternary derivative of imipramine, imipraminium, wasprepared as described in Section 4.4. Currents were evoked by2-s depolarizing pulses to +30 mV from a holding potential of−60 mV at pHi/pHo=5.5/7.3. When 100 μM imipraminium wasapplied for 10 min, the proton current was only slightly re-duced to 93±2.0% of the control (n=6) (Fig. 4). The degree of in-hibition was dose-dependently increased: to 86±7.0% (n=6) ofthe control by 300 μM, to 69±7.0% (n=6) of the control by 1 mMand to 53±8.0% (n=6) of the control by 3 mM imipraminium.The reversal potential was, however, not changed by imipra-minium. The narrow range of dose–response curve was due

to its limited solubility and its lack of effectiveness. The fitto the limited data gave an IC50 of 3.37 mM with the nH of0.72 (Fig. 4). The low potency of imipraminium suggests thatthe charged form of imipramine on the extracellular side con-tributes negligibly to the 91% blocking action observed by30 μM imipramine, even though the drug is nearly all in thecharged form at pH 7.3.

2.4. Effects of imipramine and imipraminium on tumornecrosis factor-α release

BV2 cells had a low background level of tumor necrosis factor-α, 75±17 pg/ml (n=8), whose release can be enhanced by ap-plying lipopolysaccharide. Following a 24 h treatment of BV2cells with 10 ng/ml lipopolysaccharide, tumor necrosisfactor-α was increased almost 10 fold to 714±11.0 pg/ml(Fig. 5). For the drug treatment groups, BV2 cells were pre-treated with different concentrations of either imipramine orimipraminium for 30 min before adding 10 ng/ml lipopolysac-charide. Imipramine at 10 and 30 μM, dose-dependently de-creased lipopolysaccharide-induced tumor necrosis factor-αrelease: to 78±1.0% and 39±1.0% (n=3, each measured in du-plicate) of the control level, respectively. Under identical con-ditions, imipraminium up to 100 μMhad only a small effect onthe lipopolysaccharide-induced tumor necrosis factor-α re-lease. Neither imipramine nor imipraminium by itself hadany effect on basal tumor necrosis factor-α production, inthe absence of lipopolysaccharide.

3. Discussion

The most important finding of this study is that antidepres-sants imipramine, amitriptyline, desipramine and fluoxetineare potent inhibitors of voltage-gated proton currents in BV2microglial cells with IC50 values of 2.1–5.8 μM. In addition,the inhibition of the proton current might be the basis forthe inhibitory action on lipopolysaccharide-induced tumornecrosis factor-α release. Since these antidepressants areweak bases, they can exist as neutral forms or positivelycharged forms, their roles in inhibiting the proton currents

Fig. 5 – Effects of imipramine and imipraminium on tumor necrosis factor-α release. Tumor necrosis factor-α was measured24 h after treatment of BV2 cells with 10 ng/ml lipopolysaccharide or phosphate buffered saline (PBS). For the drug treatmentgroups, cells were pretreated with different concentrations of either imipramine (A) or imipraminium (B) for 30 min beforeadding lipopolysaccharide (n=3, each measured in duplicate). **P<0.01 and ***P<0.001 compared with lipopolysaccharidetreatment alone.

21B R A I N R E S E A R C H 1 4 3 5 ( 2 0 1 2 ) 1 5 – 2 3

were examined in detail using imipramine as a representativedrug.

Our evidence supports the conclusion that it is the chargedcationic form of imipramine that blocks the proton currentfrom intracellular side of membrane after crossing the mem-brane as the neutral form. Raising external pH, which in-creases the neutral form of imipramine, sped up the onsetand enhanced the degree of block by imipramine (Fig. 1A).However, raising internal pH, which decreases the intracellu-lar charged form of imipramine, reduced the blocking potencyof imipramine. When 10 μM imipramine (pKa=9.5) was ap-plied at pHi/pHo=5.5/7.3, the block reached 71.0±4.00%.When the same concentration of imipramine was applied tothe cell buffered intracellularly to pH 6.3, the block was re-duced to 26.5±4.11%. Despite the same extracellular concen-trations of the neutral form and charged form, theintracellular charged form is greatly different at equilibrium.According to Handerson–Hasselbalch equation, at pHi/pHo=5.5/7.3, the intracellular concentration of the chargedform could reach 0.63 mM, while, at pHi/pHo=6.3/7.3, itwould decrease to 0.10 mM at equilibrium. Thus, the changein the degree of block is more closely related to the concentra-tion of the intracellular charged form. These results are con-sistent with the notion that the charged form is active ininhibiting the proton currents from intracellular side of themembrane.

This notion was further supported by the experiment withimipraminium, a quaternary ammonium of imipramine. TheIC50 value for imipraminium to block the proton current wasestimated to be 3.37±1.00 mM (Fig. 4). The weak proton cur-rent blocking action of imipraminium suggests that thecharged form of external imipramine, roughly 29.81 μM with30 μM imipramine at external pH 7.3, does not contribute tothe 91% block of the proton current.

Our analysis has indicated that the apparent blocking po-tency of externally applied organic bases depends on the ex-perimental conditions used to measure the proton current.For imipramine, the apparent order of blocking potency interms of an externally applied concentration is pHi/pHo=5.5/8.3>pHi/pHo=5.5/7.3>pHi/pHo=6.3/7.3. But the degree of

block under these conditions is ultimately dependent on theconcentration of the intracellular cationic form. Thus, the pHgradient not only affects channel activation but also the dis-tribution of externally applied blocker. This point needs to beconsidered when one compares the blocking potency amongvarious other organic bases that inhibit proton currents suchas amantadine and rimantadine (01–1 mM) (DeCoursey andCherny, 1994), 4-aminopyridine (10 mM), methoxyverapamil(192 μM) (Meech and Thomas, 1987), and nicardipine (a smalleffect at 100 μM) (Thomas, 1989). Thus, under experimentalconditions that maximize proton current recordings (pHi/pHo=5.5/7.3), antidepressants initially appear to be the mostpotent organic inhibitors reported for proton currents withIC50 values of 2–6 μM. However, in terms of the active cationicform needed to block the proton current, imipramine is onlymoderately potent in blocking proton current with IC50s ofsub- to millimolar levels.

The roles of the voltage-gated proton channels in main-taining NADPH oxidase activity and phagocytic functions ofimmune system are well-documented (Morgan et al., 2009).In regard to the inhibition of lipopolysaccharide-stimulatedtumor necrosis factor-α release, the IC50 of imipramine hasbeen estimated at a range from 10 μM (Hwang et al., 2008) to37 μM (Hashioka et al., 2007). We also used imipramine as anexample to compare its effect on cytokine release to its effecton proton current. The IC50 value for the imipramine inhibi-tion of tumor necrosis factor-α release in Fig. 5 is about25 μM. This is similar to that estimated from 26% block by10 μM imipramine of the proton current measured under thecondition of pHi/pHo=6.3/7.3, which is rather closer to thephysiological condition. Since proton efflux uniquely per-forms dual functions of dissipating excess intracellular pro-tons and compensating charge separation, therebymaintaining NADPH oxidase activity (Morgan et al., 2009),the inhibition of this proton current is expected to limitNADPH oxidase activity. These results are consistent withthe expectation that inhibition of the proton current would re-duce NADPH oxidase activity leading to a reduction in thedownstream responses including release of tumor necrosisfactor-α (Qian et al., 2007).

22 B R A I N R E S E A R C H 1 4 3 5 ( 2 0 1 2 ) 1 5 – 2 3

Although the present study was not intended to answerthe question whether this anti-inflammatory effect partlycontributes to their alleviation of depression, the presentdata indicate that the concentrations of imipramine neededfor suppressing the lipopolysaccharide-induced tumor necro-sis factor-α production are on the order of 10 μM. The thera-peutic plasma concentrations of imipramine are at 0.16 to0.54 μM, which could accumulate 32-fold higher in the brainreaching a 10 μM concentration level (DeVane et al., 1984).Thus, it remains to be determined whether the anti-inflammatory action makes a significant contribution to anti-depressant action of imipramine.

4. Experimental procedure

4.1. Cells

BV2 cells, an immortalized mouse microglial cell line(Bocchini et al., 1992), were cultured in a 50:50 mixture of Dul-becco's Modified Eagle's Medium and Ham's F12 (MediatechInc., Manassas, VA) along with 10% fetal bovine serum,100 units/ml penicillin and 100 μg/ml streptomycin. Cellswere grown to near confluence on 100 mm culture dishes,treated with trypsin, harvested, centrifuged, and then resus-pended in 10 ml of media. From this resuspension aliquotsof 10,000 cells each were plated in 6-well culture plates. Eachwell contained 3 ml of media and five 12 mm glass coverslipsthat were previously coated with poly-L-lysine (Sigma-Al-drich, St. Louis, MO). These cells were allowed to grow twodays before using them for electrophysiological experiments.

4.2. Solutions

The standard external solution contained (in mM): N-methyl-D-glucamine aspartate 85, HEPES 100, CaCl2 1, and MgCl2 1.The pHs of the external solutions were titrated to 7.3 or 7.8with 1 N CsOH solution. The buffer N-tris(hydroxymethyl)methyl-3-aminopropanesulfonic acid was used to make anexternal solution with pH 8.3. The standard internal solutioncontained (in mM): 2-(N-morpholino)ethanesulfonic acid 120,N-methyl-D-glucamine aspartate 85, 1,2-bis(2-aminophe-noxy)ethane-N,N,N′N′-tetraacetic acid 1 and MgCl2 3. ThepHs of the internal solutions were titrated to 5.5 or 6.3 with1 N CsOH solution.

Imipramine hydrochloride, amitriptyline hydrochloride,desipramine hydrochloride and fluoxetine hydrochloridewere dissolved in dimethylsulfoxide to form 1000-fold stocksolutions. These stock solutions were stored frozen at −20 °Cin small aliquots and diluted to the desired concentrationsimmediately before use. All chemicals were obtained fromSigma-Aldrich.

4.3. Electrophysiological recording

Individual glass coverslips were removed from the culturedish and placed into a perfusion chamber mounted on thestage of an inverted phase-contrast microscope. The chamberwas continuously perfused with an external solution at a rateof 1 ml/min. The chamber volume was maintained at about

0.3 ml. The patch pipette was made from borosilicate glasscapillaries with filament (BF150-117-10, Sutter Instrument,Novato, CA) with a two-step vertical puller (PP83, Narishige,Tokyo, Japan) and its tip was heat-polished with a microforge(MF83, Narishige). The pipette resistance ranged between 6and 10 MΩ when filled with the internal solution. The refer-ence electrode was an Ag–AgCl wire connected to the bath so-lution through a 3 M KCl-agar bridge. The liquid junctionpotential of 2.6 mV measured between the internal and theexternal solution was corrected before gigaohm-seal forma-tion. Voltage-gated proton currents were recorded in thewhole-cell voltage clamp configuration of the patch clamptechnique. Current signals were amplified using Axopatch200 (Molecular Devices, Sunnyvale, CA), digitized at 1 kHzwith an analog-to-digital converter (Digidata 1200, MolecularDevices) using PCLAMP6 software (Molecular Devices) on apersonal computer. Currents were low-pass filtered at 1 kHz.Data are expressed as means±S.E.M. and n refers to the num-ber of cells examined. Student's t test was used for compari-sons and statistical significance was assessed at P<0.05. Allexperiments were performed at room temperature (22–24 °C).

4.4. Synthesis of quaternary ammonium of imipramine(imipramine methiodide), imipraminium

A total of 400 mg imipramine hydrochloride was dissolved in10 ml of water and the solution was titrated to an alkalinepHwith 1 N NaOH solution. The cloudy solution was extractedwith diethyl ether twice. The combined extract was washedwith water and the top ether layer was collected. The residualwater in the ether fraction was removed bymixing with anhy-drous sodium sulfate. The final ether extract was evaporatedleaving an oily film in the flask. The oily substance waswashed with hexane and then evaporated under reducedpressure. The residue was dissolved in acetone and thenreacted with 0.5 ml of methyliodide for 1 h. Ether was subse-quently added to the reaction medium slowly to initiate crys-tallization which took overnight. The crystals were collectedand washed with hexane. After drying overnight, the crystalsweighed 150 mg giving a yield of 37.5%. The crystal's meltingpoint was determined using a Buchi Melting Point B-540 (Flawil,Switzerland), and was found to be 229–230 °C. The quaternaryammoniumnature of imipraminewas confirmed bymass spec-tra determined at the IMERC facility at NorthwesternUniversity.The relative strength distribution, 100% of 295.2, 21.9% of 296.2and 2.4% of 297.1, is consistent with isotope cluster pattern(Jurasek et al., 1993) expected from the formula C20H27N2, a qua-ternary ammonium of imipramine, imipraminium.

4.5. Effects of imipramine and imipraminium onlipopolysaccharide-induced tumor necrosis factor-α release

Imipramine has been shown to inhibit the release of pro-inflammatory factors induced by lipopolysaccharide or byinterferon-γ (Hashioka et al., 2007; Hwang et al., 2008;Obuchowicz et al., 2006). Here we compared the effect of imip-ramine on tumor necrosis factor-α release induced by lipo-polysaccharide to that of imipraminium. On the first day,4×104 BV2 cells were seeded in a well (24 well plates). Thenext day, imipramine or imipraminium was applied to the

23B R A I N R E S E A R C H 1 4 3 5 ( 2 0 1 2 ) 1 5 – 2 3

cells and 0.5 h later, lipopolysaccharide 10 ng/ml was addedinto the medium. After lipopolysaccharide stimulation for24 h, culture supernatants were collected and centrifuged toremove the particulates. Tumor necrosis factor-α was mea-sured using an ELISA kit (R&D Systems, Minneapolis, MN)according to the instructions of the manufacturer. The absor-bance at 450 nm with the reference wavelength set at 540 nmwas determined using a microplate reader (TECAN, Männedorf,Switzerland) and Magellan software (TECAN). Each experimentwas run in duplicate. The concentration of tumor necrosisfactor-α was calculated according to a standard curve. Data areexpressed as means±S.D. of three separate experiments. Signifi-cance was first determined by a one-way analysis of variance(ANOVA) (P<0.05) followed by a Dunnett's post hoc analysis(P<0.05).

Acknowledgments

The BV2 cells were provided to us by the lab of Dr. Jau-ShyongHong, National Institute of Environmental Health Sciences,National Institutes of Health, Research Triangle Park, NorthCarolina. We thank Dr. S.M. Roy for helping us to synthesizequaternary ammonium of imipramine. We also thankMr. Andrew F. Scheyer and Mrs. Jennifer Whitesides for theirefforts in preparing and maintaining the BV2 cell cultures.This research was supported by the Chung-Ang University Re-search Grants in 2009 (to J.-H.S.).

R E F E R E N C E S

Bocchini, V., Mazzolla, R., Barluzzi, R., Blasi, E., Sick, P., Kettenmann,H., 1992. An immortalized cell line expresses properties ofactivated microglial cells. J. Neurosci. Res. 31, 616–621.

DeCoursey, T.E., 2003. Voltage-gated proton channels and otherproton transfer pathways. Physiol. Rev. 83, 475–579.

DeCoursey, T.E., 2008. Voltage-gated proton channels. Cell. Mol.Life Sci. 65, 2554–2573.

DeCoursey, T.E., Cherny, V.V., 1994. Voltage-activated hydrogenion currents. J. Membr. Biol. 141, 203–223.

DeVane, C.L., Simpkins, J.W., Stout, S.A., 1984. Cerebral and bloodpharmacokinetics of imipramine and its active metabolites inthe pregnant rat. Psychopharmacology (Berl.) 84, 225–230.

Eder, C., DeCoursey, T.E., 2001. Voltage-gated proton channels inmicroglia. Prog. Neurobiol. 64, 277–305.

Hashioka, S., Klegeris, A.,Monji, A., Kato, T., Sawada,M.,McGeer, P.L.,Kanba, S., 2007. Antidepressants inhibit interferon-γ-inducedmicroglial production of IL-6 and nitric oxide. Exp. Neurol. 206,33–42.

Hwang, J., Zheng, L.T., Ock, J., Lee, M.G., Kim, S.H., Lee, H.W., Lee,W.H., Park, H.C., Suk, K., 2008. Inhibition of glial inflammatoryactivation and neurotoxicity by tricyclic antidepressants.Neuropharmacology 55, 826–834.

Jurasek, P., Slimak, M., Kosik, M., 1993. Determination of isotopecluster patterns in mass spectra of GC-MS analyses by achemometric detector. Microchim. Acta 110, 133–142.

Kim, D., You, B., Jo, E.K., Han, S.K., Simon, M.I., Lee, S.J., 2010.NADPH oxidase 2-derived reactive oxygen species in spinalcord microglia contribute to peripheral nerve injury-inducedneuropathic pain. Proc. Natl. Acad. Sci. U. S. A. 107,14851–14856.

Meech, R.W., Thomas, R.C., 1987. Voltage-dependent intracellularpH in Helix aspersa neurones. J. Physiol. 390, 433–452.

Morgan, D., Capasso, M., Musset, B., Cherny, V.V., Rios, E., Dyer, M.J.,DeCoursey, T.E., 2009. Voltage-gated proton channels maintainpH in human neutrophils during phagocytosis. Proc. Natl. Acad.Sci. U. S. A. 106, 18022–18027.

Narahashi, T., Frazier, T., Yamada, M., 1970. The site of action andactive form of local anesthetics. I. Theory and pH experimentswith tertiary compounds. J. Pharmacol. Exp. Ther. 171, 32–44.

Obuchowicz, E., Kowalski, J., Labuzek, K., Krysiak, R., Pendzich, J.,Herman, Z.S., 2006. Amitriptyline and nortriptyline inhibitinterleukin-1β and tumour necrosis factor-α release by rat mixedglial andmicroglial cell cultures. Int. J. Neuropsychopharmacol. 9,27–35.

Peng, C.H., Chiou, S.H., Chen, S.J., Chou, Y.C., Ku, H.H., Cheng, C.K.,Yen, C.J., Tsai, T.H., Chang, Y.L., Kao, C.L., 2008. Neuroprotectionby imipramine against lipopolysaccharide-induced apoptosis inhippocampus-derived neural stem cells mediated by activationof BDNF and the MAPK pathway. Eur. Neuropsychopharmacol.18, 128–140.

Qian, L., Gao, X., Pei, Z., Wu, X., Block, M., Wilson, B., Hong, J.S.,Flood, P.M., 2007. NADPH oxidase inhibitor DPI isneuroprotective at femtomolar concentrations throughinhibition of microglia over-activation. Parkinsonism Relat.Disord. 13, S316–S320.

Thomas, R.C., 1989. Proton channels in snail neurones. Doescalcium entry mimic the effects of proton influx? Ann. N. Y.Acad. Sci. 574, 287–293.

Thomas, R.C., Meech, R.W., 1982. Hydrogen ion currents andintracellular pH in depolarized voltage-clamped snail neurones.Nature 299, 826–828.

Von Bulow, V., Dubben, S., Engelhardt, G., Hebel, S., Plumakers, B.,Heine, H., Rink, L., Haase, H., 2007. Zinc-dependent suppressionof TNF-α production is mediated by protein kinase A-inducedinhibition of Raf-1, IκB kinase β, and NF-κB. J. Immunol. 179,4180–4186.