Dendritic compartmentalization of chloridecotransporters underlies directional responsesof starburst amacrine cells in retinaKonstantin E. Gavrikov*, James E. Nilson†, Andrey V. Dmitriev*, Charles L. Zucker†, and Stuart C. Mangel*‡

*Department of Neuroscience, Ohio State University College of Medicine, Columbus, OH 43210; and †Department of Anatomyand Neurobiology, Boston University School of Medicine, Boston, MA 02118

Edited by John E. Dowling, Harvard University, Cambridge, MA, and approved October 23, 2006 (received for review June 1, 2006)

The mechanisms in the retina that generate light responses selec-tive for the direction of image motion remain unresolved. Recentevidence indicates that directionally selective light responses occurfirst in the retina in the dendrites of an interneuron, i.e., thestarburst amacrine cell, and that these responses are highly sen-sitive to the activity of Na-K-2Cl (NKCC) and K-Cl (KCC), two typesof chloride cotransporter that determine whether the neurotrans-mitter GABA depolarizes or hyperpolarizes neurons, respectively.We show here that selective blockade of the NKCC2 and KCC2cotransporters located on starburst dendrites consistently hyper-polarized and depolarized the starburst cells, respectively, andgreatly reduced or eliminated their directionally selective lightresponses. By mapping NKCC2 and KCC2 antibody staining onthese dendrites, we further show that NKCC2 and KCC2 arepreferentially located in the proximal and distal dendritic compart-ments, respectively. Finally, measurements of the GABA reversalpotential in different starburst dendritic compartments indicatethat the GABA reversal potential at the distal dendrite is morehyperpolarized than at the proximal dendrite due to KCC2 activity.These results thus demonstrate that the differential distribution ofNKCC2 on the proximal dendrites and KCC2 on the distal dendritesof starburst cells results in a GABA-evoked depolarization andhyperpolarization at the NKCC2 and KCC2 compartments, respec-tively, and underlies the directionally selective light responses ofthe dendrites. The functional compartmentalization of interneurondendrites may be an important means by which the nervoussystem encodes complex information at the subcellular level.

direction-selective � GABAergic excitation � interneuron

Neurons that respond selectively to the direction of stimulusmotion are a common feature of the nervous system. In the

vertebrate retina, directionally selective (DS) ganglion cellsrespond well to stimulus motion in one (preferred) direction, butrespond little or not at all to motion in the opposite (null)direction (1). Starburst amacrine cells (SACs), interneuronspresynaptic to the DS ganglion cells, are an essential componentof the mechanism that generates direction selectivity in theretina because selective elimination of this interneuron elimi-nates the directional light responses of DS ganglion cells (2, 3).Moreover, SAC dendrites generate DS light responses (4, 5),providing DS input to DS ganglion cells (5, 6). They hyperpo-larize to stimuli that move centripetally from the periphery totheir proximal dendrites but depolarize to stimuli that movecentrifugally from their proximal dendrites to the periphery (5).However, although the DS light responses of SAC dendrites mayfunction as the core directional process in the retina, themechanisms that underlie them are not known.

In the retina, pharmacological studies have indicated thatactivation of GABAA receptors, which open Cl� channels,mediates direction selectivity (7). The cation-chloride cotrans-porters Na-K-2Cl (NKCC) and K-Cl (KCC) mediate the depo-larizing and hyperpolarizing effects of GABA, respectively (8–11). NKCC transports Cl� into cells so that the Cl� equilibrium

potential (ECl) is more positive than the resting membranepotential, resulting in a GABA-evoked depolarization. In con-trast, KCC extrudes Cl� so that the ECl is more negative than theresting membrane potential, resulting in a GABA-evoked hy-perpolarization. Because selective blockade of NKCC and KCCby the loop diuretics bumetanide (BMN) and furosemide(FUR), respectively, and reduction of the transmembrane Cl�gradient eliminates the directional responses of DS ganglion cellsand SACs (5), we investigated how NKCC and KCC activity andlocalization underlie the directional responses of SACs.

ResultsLight Responses of SACs to Moving Stimuli. SACs express GABAand glutamate receptors along their dendrites (12–14), and theirlight responses exhibit a glutamate-mediated center and aGABA-mediated surround receptive field (RF) organization(15–17). Fig. 1 shows that the light responses of SACs to movingstimuli consist of two components, a slow, GABA-mediatedcomponent that is DS and primarily responsive to surroundillumination, and a faster, glutamate-mediated component thatis responsive to central illumination, but not DS. Fig. 1 Aillustrates that when a slit moved through the RF center andsurround, both the slow DS response component and the faster,non-DS component were produced. Fig. 1B shows that the SACdid not generate the fast depolarization (or the fast OFFhyperpolarization) if the slit did not stimulate the RF center.That is, the SAC produced a slow hyperpolarization to a slitstimulus that moved centripetally from the periphery throughthe surround but did not reach the RF center, and produced aslow depolarization when the slit stimulus reversed direction andmoved centrifugally. The slow hyperpolarization and slow de-polarization to stimulus motion in the surround in the centripetaland centrifugal directions, respectively, represent the DS lightresponse of the SAC dendrite.

Dark-adapted SACs exhibited similar resting membrane po-tentials and responses to moving slit stimuli when recorded withwhole-cell patch-clamp electrodes (Fig. 1C) and with intracel-lular sharp microelectrodes (Fig. 1D), a means of monitoringneurons that does not significantly alter their intracellular mi-lieu. Specifically, the average dark resting potential (see Mate-rials and Methods) and light response amplitudes and dynamics(see Fig. 1 C and D) of SACs were similar when recordings were

Author contributions: K.E.G. and J.E.N. contributed equally to this work; K.E.G., J.E.N.,A.V.D., C.L.Z., and S.C.M. designed research; K.E.G. and J.E.N. performed research; K.E.G.,J.E.N., A.V.D., C.L.Z., and S.C.M. analyzed data; and C.L.Z. and S.C.M. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS direct submission.

Abbreviations: BMN, bumetanide; ChAT, choline acetyltransferase; DS, directionally selec-tive; FUR, furosemide; RF, receptive field; SAC, starburst amacrine cell.

‡To whom correspondence should be addressed at: Department of Neuroscience, OhioState University College of Medicine, 333 West 10th Avenue, Columbus, OH 43210. E-mail:mangel.1@osu.edu.

© 2006 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0604551103 PNAS � December 5, 2006 � vol. 103 � no. 49 � 18793–18798

NEU

ROSC

IEN

CE

Dow

nloa

ded

by g

uest

on

June

18,

202

0

obtained with sharp intracellular and whole-cell patch-clampelectrodes, a finding that strongly suggests that our whole-cellpatch-clamp recording data reveal the electrical characteristicsand light response properties of in vivo SACs.

Starburst Cells Exhibit Both NKCC and KCC Activity and Expression.Although previous results demonstrated that NKCC and KCCcontribute to the DS responses of SACs (5), they did not showwhether the cotransporters are expressed by the SACs them-selves, or alternatively, whether they mediate the directional

responses of SACs through the retinal network. We thus inves-tigated whether BMN and FUR altered the DS responses ofindividual SAC dendrites in the intact rabbit retina when thedrugs were introduced into the cells via the patch pipettes duringwhole-cell recording so that only the cotransporters on the cellunder study were selectively blocked. As shown in Fig. 2, pipetteapplication of either drug eliminated or greatly reduced the DSlight responses of SAC dendrites (A–D) and the initial transienthyperpolarization (C and D), but it did not eliminate the non-DS,RF center light response that is likely glutamate-mediated (Cand D). In addition, FUR (25 �M) depolarized (average depo-larization � 4.2 � 1.5 mV, n � 28) and BMN (10 �M)hyperpolarized (average hyperpolarization � �17.3 � 2.6 mV,n � 13) every cell studied. These results indicate that SACdendrites exhibit both NKCC and KCC activity and that NKCCand KCC are tonically and highly active and work together to

Fig. 1. Light responses of retinal SACs to moving stimuli consist of two com-ponents. (A and B) When a slit stimulus appeared in the periphery �1 mm fromthe cell body and remained stationary and illuminated for 1 sec, it produced atransient hyperpolarization of the SAC (note the initial transient hyperpolariza-tion at the beginning of each record). (A) When the slit stimulus subsequentlymoved centripetally (cptl) from the periphery to the RF center (see diagram toright of recorded trace), the SAC initially generated a slow hyperpolarization thatwas then followed by a relatively fast depolarization as the slit stimulated the RFcenter. When the slit stimulus moved centrifugally (cfgl) from the RF center to theperiphery, the SAC produced a fast depolarization followed by a fast hyperpo-larization, which was likely an OFF response from the cell’s RF center (see below),and finally generated a slow depolarization as the slit moved centrifugallythrough the surround toward the periphery. (B) The SAC produced a slowhyperpolarization to a slit stimulus that moved centripetally from the peripherythrough the surround, but did not reach the RF center (see diagram to right ofrecorded trace), and produced a slow depolarization when the slit stimulusreversed direction and moved centrifugally. The slit stimulus was illuminated andstationary within the RF center (A) or just outside of the RF center (B) during theinterval between centripetal and centrifugal motion. The fast OFF hyperpolariz-ing response did not occur in A when the slit reached the RF center from thecentripetal direction because the slit remained illuminated and stationary overthe RF center during the interval between centripetal and centrifugal motion,andthusdidnotgenerateanOFFresponse. (CandD)Whentheslit stimulus,aftera 1 sec period during which it remained stationary and illuminated in theperiphery (transient hyperpolarization not shown), moved across the retinawithout stopping, the SAC hyperpolarized to centripetal stimulus motion, pro-duced a fast depolarization followed by an OFF hyperpolarization to RF centerstimulation, and then generated a slow depolarization to centrifugal motion.Responses shown are to single centripetal and centrifugal movements (A and B,0.5 mm/sec; C and D, 1.0 mm/sec) of a slit stimulus (A and B, 0.5 � 0.03 mm; C andD, 0.5 � 0.05 mm) obtained with whole-cell patch-clamp (A–C; n � 45) or sharpmicroelectrode intracellular (D; n � 6) recording. (Scale bars: 1 sec.)

Fig. 2. Selective blockade of either NKCC or KCC located on SACs or ofGABAA receptors eliminates the directional light responses of SACs. (A–D) Asdescribed in Fig. 1, during the first minute of whole-cell recording, SACsgenerated a DS response to stimuli that moved centripetally (cptl) and cen-trifugally (cfgl) through the RF surround, but not the RF center (A and B), orgenerated both DS and non-DS responses to stimuli that moved through theRF center and surround (C–E). Ten minutes later after introduction of BMN (10�M; A and C) or FUR (25 �M; B and D) into the cells through the patch pipettes,the DS light responses of SACs (A–D) and the initial transient hyperpolariza-tion to stationary peripheral illumination (C and D) were eliminated, but thenon-DS, RF center light response was not eliminated (C and D). In addition,BMN hyperpolarized and FUR depolarized the SACs. The effects of FUR andBMN were first observed �4 min after whole-cell recording began andreached a maximum �4 min later. (E and F) Bath application of gabazine (100�M) consistently hyperpolarized the cells (E and F) and eliminated the DS lightresponses of SAC dendrites (E) and the hyperpolarizing response to annularstimulation of the RF surround (F), but did not eliminate the glutamate-mediated, non-DS RF center light response (E). Responses shown are to singlecptl and cfgl movements (A and B, 0.5 mm/sec; C–E, 1.0 mm/sec) of a slitstimulus (A, 0.5 � 0.03 mm; B–E, 0.5 � 0.05 mm) or to a single flash (2 sec,horizontal bar beneath traces) of a stationary annulus (F, 1.5 mm i.d. and 2.0mm o.d.). The light response in A was smaller in size than in B–E because thestimulus was narrower in A. The protocol for the slit stimulus in C–E was thesame as that described in Fig. 1 C and D. (Scale bars: 1 sec.) Vm, membranepotential.

18794 � www.pnas.org�cgi�doi�10.1073�pnas.0604551103 Gavrikov et al.

Dow

nloa

ded

by g

uest

on

June

18,

202

0

generate the DS responses of SAC dendrites by driving ECl in thepositive and negative directions, respectively.

Because rabbit SACs express GABAA receptors (12, 14), weinvestigated whether their DS light responses are mediatedby endogenous GABAA receptor activation. Application ofgabazine (SR95331, 50–100 �M), a specific GABAA receptorantagonist, consistently hyperpolarized the cells by an average of6.5 � 2.4 mV (n � 10), eliminated the hyperpolarizing responseelicited by annular stimulation of the RF surround (Fig. 2F) andeliminated their DS light responses, but did not eliminate thenon-DS, RF center light response (Fig. 2E) that is likely gluta-mate-mediated. In contrast, application of TPMPA (50 �M), aspecific GABAC receptor antagonist, had little effect (data notshown).

To identify the specific subtypes of NKCC and KCC expressedby SACs, rabbit retinas were double-labeled with antibodiesagainst choline acetyltransferase (ChAT), which specificallystains SACs (18, 19), and against NKCC (types 1 and 2) or KCC(types 1–4). ChAT-labeled displaced SACs showed specificNKCC2 (Fig. 3C) and KCC2 (Fig. 3F) labeling. Similar NKCC2and KCC2 labeling was also observed on ChAT-labeled SACs inthe amacrine cell layer (data not shown). In contrast, NKCC1,KCC1, KCC3, and KCC4 labeling was not observed on SACs(data not shown).

Differential Dendritic Compartmentalization of NKCC2 and KCC2.Because SACs express both NKCC2 and KCC2 (Fig. 3) andbecause selective blockade of NKCC2 and KCC2 activity elim-inates the DS light responses of SAC dendrites (Fig. 2), weexamined whether NKCC2 and KCC2 are differentially locatedon the proximal and distal portions of SAC dendrites, respec-tively. Individual displaced SACs were injected with Alexa 488,and the retinas were then stained with antibodies against NKCC2or KCC2. The highest density of NKCC2 label was on theproximal dendrites of Alexa-filled SACs with progressive dim-inution of the label more distally (Fig. 4B). In contrast, thehighest density of KCC2 label was on the distal dendrites andvaricosities (Fig. 4D). Quantification of the NKCC2 and KCC2

staining patterns on SAC dendrites indicates that differentdendritic compartments express NKCC2 and KCC2, withNKCC2 predominating proximally and KCC2 predominatingdistally (Fig. 4E).

Because NKCC2 and KCC2 are preferentially located on theproximal and distal portions of SAC dendrites, respectively, the[Cl�]i should be lower in the distal compared with the proximaldendrites; and as a result, EGABA should be more negativedistally than proximally. We thus measured EGABA along boththe proximal and more distal portions of individual SAC den-drites after synaptic block. GABA application onto the proximalportion of the SAC dendritic tree evoked responses that con-sistently reversed at a holding potential �10 mV more positivethan what occurred when GABA was puffed onto a more distalportion of the dendritic tree of the same cell (Fig. 5 A–C).

Because KCC2 is preferentially located on SAC distal den-drites (Fig. 4), selective blockade of KCC2 with FUR should shiftEGABA at the distal dendrite to a more positive value. AverageEGABA at the more distal dendrites was �10 mV more positiveafter application of 25 �M FUR (Fig. 5D). BMN (10 �M)application did not alter EGABA at the distal dendrites (data notshown). The finding that FUR (25 �M) eliminated the proximal-distal difference in EGABA not only indicates that the morenegative EGABA at SAC distal dendrites is mediated by KCC2 butalso that the proximal-distal difference in EGABA does not resultfrom a space-clamp or other measurement artifact. In addition,the finding that FUR (25 �M) did not block the effects of GABA(Fig. 5D), that is, GABA evoked voltage responses of similaramplitude when puffed in the presence or absence of FUR,indicates that FUR did not alter the GABAA receptor-mediatedconductance. This is consistent with the finding that SACsexpress GABAA receptors that do not contain �6 or �4 subunits(12), because FUR only antagonizes GABAA receptors thatcontain �6 and �4 subunits (20). The selective inhibitory effectof FUR (25 �M) for starburst KCC2 compared with NKCC2 isshown by the consistently opposite effects of FUR (25 �M) andBMN (10 �M), a selective inhibitor of NKCC2 at 10 �M (8–11),and can be attributed to the finding that the low concentration(25 �M) of FUR used here is ineffective against many NKCC1and NKCC2 subtypes (8, 9, 11, 21), suggesting that SACs expressa FUR-insensitive NKCC2 subtype.

It is important to note that GABA application to proximaldendrites (Fig. 5A) and onto distal dendrites �100 �m from thesoma (Fig. 5B) produced large amplitude responses, indicatingthat proximal dendrites and the portion of the dendrites 100 �mfrom the soma were adequately current clamped and that theEGABA measurements were accurate, although they were likelyan underestimate of the actual proximal-distal difference inEGABA (see Materials and Methods). However, GABA applica-tion onto dendrites 150–200 �m from the soma produced verysmall amplitude responses (data not shown), suggesting that thismore distal portion was weakly electrically linked with the somaand not adequately clamped. The lack of control of the mem-brane potential of the more distal dendrites probably resultedfrom a significant intracellular resistance that accumulates alongthe dendrites so that the membrane potential of the more distaldendrites is different from that in the soma.

DiscussionWe have demonstrated in this article that (i) SACs express bothNKCC2 and KCC2; (ii) the selective blockade of NKCC2 andKCC2 located on the SAC dendrites eliminates the DS responsesof SACs and shows that the cotransporters work in oppositionand are tonically and highly active; (iii) NKCC2 and KCC2 arepreferentially located on the proximal and distal dendrites,respectively; and (iv) GABA receptor activation at the proximaldendrite produces a depolarization that is mediated by NKCC2,whereas GABA receptor activation at the more distal dendrite

Fig. 3. SACs express both NKCC2 and KCC2. Rabbit retinas were double-labeled with antibodies against ChAT (A and D, shown in red), which specif-ically stains SACs, and against NKCC2 (B) or KCC2 (E) (both shown in green). Asshown by white colocalized pixels, ChAT labeled displaced SACs showedspecific NKCC2 (C) and KCC2 (F) labeling. Moreover, transverse vibratomesections clearly showed NKCC2 labeling on the cell body (*) and proximaldendrites (arrows) of SACs (A–C) and faint KCC2 labeling on SAC cell bodies (*)(F). Images represent single 1-�m-thick optical slices. (Scale bars: 10 �m.)

Gavrikov et al. PNAS � December 5, 2006 � vol. 103 � no. 49 � 18795

NEU

ROSC

IEN

CE

Dow

nloa

ded

by g

uest

on

June

18,

202

0

produces a hyperpolarization that is mediated by KCC2. Thesefindings indicate that the differential distribution of NKCC2 onthe proximal dendrites and KCC2 on the distal dendrites ofSACs results in a GABA-evoked depolarization and hyperpo-larization at the NKCC2 and KCC2 compartments, respectively,and underlies the DS responses of the dendrites. Moreover, theeffects of gabazine on SACs strongly support the view thatendogenous GABA depolarizes the proximal dendrites andhyperpolarizes the distal dendrites. Specifically, the finding thatGABAA receptor blockade hyperpolarized SACs (Fig. 2 E andF) indicates that the primary effect of endogenous GABAAreceptor activation that is detectable at starburst somata is asustained depolarization that depends on NKCC2 activity in thecentral portion of SACs. In contrast, GABAA receptor blockadeeliminated the hyperpolarizing response elicited by annularstimulation of the RF surround (Fig. 2F), indicating that sur-round activation of GABAA receptors hyperpolarizes SACs.

Because synaptic vesicles are located at the distal (but not theproximal) portions of SAC dendrites (22), GABA release (23–25) from the SAC distal dendrite will likely occur when thedendrite is depolarized by light stimuli that move in the centrif-ugal direction. In contrast, GABA release from the distal SACdendrite will be minimal when the dendrite is hyperpolarized bylight stimuli that move in the centripetal direction. The direc-tional release of GABA from SAC dendrites that point in thenull direction of DS ganglion cells and that are located on thenull side of the ganglion cells will thus confer null directioninhibition on DS ganglion cells (5, 6).

Although our results indicate that NKCC2 and KCC2 on SACdendrites work in opposition and are tonically and highly active(Fig. 2), the degree to which they drive local ECl along the

dendrites is uncertain. It is theoretically possible for NKCC2,which utilizes the average transmembrane cation (Na� and K�)gradient, to drive ECl at the proximal dendrite to (ENa � EK)/2(� �31.9 mV in our experimental conditions), and for KCC2,which utilizes the transmembrane K� gradient, to drive ECl at thedistal dendritic tip to EK (� �94.7 mV in our experimentalconditions). The finding that blockade of KCC2 by FUR depo-larized SACs by �4 mV and that blockade of NKCC2 by BMNhyperpolarized SACs by �17 mV (Fig. 2) suggests that thedifference in local ECl between the proximal and distal portionsof the dendrites is at least 21 mV. However, a proximal-distaldifference of 21 mV would occur if SACs only had a Cl�conductance. Because SACs, like other neurons, have significantNa� (26) and K� conductances (16, 27) in addition to a Cl�conductance, the Na� and K� conductances will reduce the sizeof the changes in membrane potential that result when BMN andFUR block the Cl� cotransporters and alter ECl. As a result, theactual proximal-distal dendritic difference in local ECl should begreater than 21 mV. For example, if the Cl� conductance werehalf the total SAC conductance, the proximal-distal differencein local ECl would double to 42 mV.

In addition to providing a physiological–morphological basisof the DS light responses of SACs, our article demonstrates thatdifferent compartments of individual CNS dendrites express twodistinct transporter types so that a single neurotransmitterdepolarizes and hyperpolarizes the different dendritic compart-ments. The processes of interneurons in the retina and elsewherein the CNS may express Cl� cotransporters as well as othertransporter, channel (27), and synaptic proteins in differentfunctional compartments. In fact, it has been reported that singleinterneurons in the midbrain and hippocampus contain com-

Fig. 4. NKCC2 and KCC2 are expressed in different starburst cell dendritic compartments. Individual SACs were injected with Alexa 488 (A–D, shown in green),and the retinas were then stained with antibodies against NKCC2 (A and B) or KCC2 (C and D) (both shown in red in A and C). (B) The highest density of NKCC2label (white colocalized pixels) was on the proximal dendrites of Alexa-filled SACs with sequential diminution of the label more distally. (D) The highest densityof KCC2 label (white colocalized pixels) was on the distal dendrites and varicosities. (E) Average relative distribution (mean colocalized pixels �SEM) of NKCC2(filled bars) and KCC2 (open bars) along individual SAC dendrites. There was a significant difference (P � 0.0001, two-tailed t test) in the relative distributionof NKCC2 and of KCC2 in the proximal and distal compartments with NKCC2 predominating proximally and KCC2 predominating distally. (Scale bars: 20 �m.)

18796 � www.pnas.org�cgi�doi�10.1073�pnas.0604551103 Gavrikov et al.

Dow

nloa

ded

by g

uest

on

June

18,

202

0

partments with different EGABA (28, 29) in a manner that isconsistent with the differential distribution of NKCC2 andKCC2 shown here.

What might be the functional role(s) of GABA-evoked de-polarizing and hyperpolarizing responses in different compart-ments of individual interneuronal processes in the CNS? First,spatial segregation of GABA-evoked depolarizing and hyper-polarizing responses along a neuronal process could enable thatprocess to distinguish between and compare different GABAer-gic inputs. Second, spatial segregation of GABA-evoked depo-larizing and hyperpolarizing responses along a neuronal processcould enable GABA input to differentially control neuronalexcitability at the subcellular level. For example, GABA inputcould facilitate glutamate-evoked excitation along a portion ofa process and inhibit it elsewhere along the same process. Finally,as suggested by our study, the spatial segregation of GABA-evoked depolarizing and hyperpolarizing responses along aneuronal process could provide a mechanism for the directionalrelease of a neurotransmitter from that process because infor-mation flow/neural activity in one direction will depolarize theprocess, but information flow/neural activity in the oppositedirection will hyperpolarize it. The functional compartmental-ization of interneuron dendrites may therefore be an importantmeans by which the CNS encodes complex information at thesubcellular level.

In summary, our findings indicate that the SAC, an interneu-ron with a highly radial and symmetric appearance, possesses aninherent asymmetry in the distribution of Cl� cotransportersalong its dendritic processes, and that this asymmetry underliesthe DS response of its dendrites. The results also show thatsubcellular specializations, such as the location of NKCC and

KCC in different dendritic compartments, can mediate neuralcomputations in the brain.

Materials and MethodsPreparation. Retinal eyecups were obtained after deep general(urethane, 1.5 g/kg, i.p.) and local intraorbital (2% Xylocaine)anesthesia of New Zealand white rabbits (2.5 kg) (5). Animalcare and use followed all federal and institutional guidelines.Superfusate (pH 7.4, 34–35°C) contained 117.0 mM NaCl, 3.1mM KCl, 10.0 mM glucose, 2.0 mM CaCl2, 1.2 mM MgSO4, 30.0mM Na HCO3, 0.5 mM NaH2PO4, and 0.1 mM L-glutamine.SACs were selectively labeled by intraocular injection of 0.3 �gof DAPI 1 day before experiments (5, 19). DAPI-labeled,displaced SACs were identified with brief UV illumination, andinfrared illumination was used for microelectrode manipulation.Biocytin injection confirmed SAC identity.

Electrophysiology. Patch-clamp recording of dark-adapted, dis-placed SACs was used. Whole-cell electrodes, which had resis-tances of 5–6 M�, contained 14.0 mM KCl, 100.0 mM K-gluconate, 5.0 mM EGTA, 5.0 mM Hepes, 3.0 mM MgATP, 0.5mM Na3GTP, 0.5 mM CaCl2, 20.0 mM Na2-phosphocreatine,and 4.1 mM NaHCO3. The average resting potential of SACs was�49.8 � 1.8 mV (n � 45). In control experiments (n � 8) inwhich FUR and BMN were not added to the patch pipettes, theaverage resting potential and the light responses of the cellsremained unchanged for at least 30 min. In control experimentsin which sharp microelectrodes were used for intracellularrecording, the average resting potential of displaced SACs was�52.8 � 3.4 mV (n � 6). Sharp microelectrodes contained 1 MK-acetate and had resistances of �150 M�.

EGABA was measured by pressure ejecting (6 psi, 100 msec)GABA (0.5 mM) from 5- to 6-�m-diameter tip pipettes. GABAwas applied onto SAC proximal and distal dendrites withsynaptic transmission blocked with cobalt (2 mM), while themembrane potential of the cells was shifted between �90 and�40 mV by using constant current pulses. The potential at whichGABA did not evoke a response corresponded to EGABA. Inthese experiments, the superfusate flowed in the same directionas the centrifugal direction of the dendrite under study so thatGABA puffed on the distal dendrite did not reach the proximaldendrite. Dye included in the puff pipette confirmed this (n �4). However, the measured proximal-distal difference in EGABA(see Fig. 5) is likely an underestimate of the actual difference fortwo reasons. First, because GABA puff application at theproximal dendrite necessarily reached the more distal dendrite,the effective distance between the proximal and distal GABAapplications was somewhat less than 100 �m. Second, althoughSAC dendrites 100 �m from the soma were adequately clamped(see Fig. 5), they may not have been as well clamped as the soma,so that the driving force for the Cl� current (Vm � EGABA)elicited by GABA application onto this more distal portion ofthe dendrites may have been slightly attenuated compared withthe Cl� current elicited by GABA application onto the proximaldendrite (30). In addition, due to the cable properties of thedendrites, the response to GABA, which was measured at thesoma, may have been slightly smaller in amplitude when GABAwas applied more distally compared with when it was appliedproximally. It is thus likely that the difference in EGABA betweenthe proximal dendrites and the distal dendrites 150–200 �mfrom the soma is 10 mV and more closely approximates thedifference obtained after application of BMN and FUR (seeDiscussion).

Immunohistochemistry. NKCC2 immunoreactivity was localizedwith a rabbit polyclonal antiserum to a 15-aa C-terminal se-quence of rat NKCC2 (Chemicon AB3562P) (1:200). This aminoacid sequence is common to each of the three known rabbit

Fig. 5. The GABA reversal potential at the starburst distal dendrite is morehyperpolarized than at the proximal dendrite due to KCC2 activity. (A and B)GABA was applied onto the proximal dendrite (A) and onto the distal dendrite(B) �100 �m from the cell body of a SAC in the presence of cobalt (2 mM) toblock synaptic transmission. (C) Average EGABA of the proximal and distaldendrites of SACs were significantly different (P � 0.01, Student’s paired t test,n � 7). Proximal and distal EGABA were measured from each cell. (D) AverageEGABA of distal dendrites before and during bath application of FUR (25 �M)were significantly different (P � 0.01, n � 6). Both control and FUR EGABA datawere obtained from each cell.

Gavrikov et al. PNAS � December 5, 2006 � vol. 103 � no. 49 � 18797

NEU

ROSC

IEN

CE

Dow

nloa

ded

by g

uest

on

June

18,

202

0

NKCC2 splice variants and is located on the cytoplasmic face ofthe membrane (11). All staining in rabbit retina cryostat andvibratome sections was abolished by preincubation of the anti-serum with the NKCC2-specific peptide (Chemicon AG205), aprocedure analogous to Western blot analysis with respect todetermining antibody specificity. The finding that BMN, aselective NKCC inhibitor at 10 �M (8–11), hyperpolarized SACsand eliminated their DS light responses when dialyzed intoindividual cells (see Fig. 2 A) also demonstrates that SACscontain functional NKCC. KCC2 was localized with a rabbitpolyclonal antiserum to an N-terminal His-tag fusion protein(residues 932-1043) of rat KCC2 (Upstate 07-285) (1:150).Specificity of this antiserum in rabbit retina has been shown byboth preabsorption and Western blot analysis (31). ChAT waslocalized with a goat polyclonal antiserum AB 144P (Chemicon)(1:200). Vibratome sections (80 �m) of formaldehyde-fixedtissue (4%, 3 h, 4°C) were labeled for ChAT and NKCC2 orKCC2 to assess their expression on SACs. Incubation in primaryantisera for 7 days (4°C) was followed by a 48-h incubation withFITC donkey anti-rabbit Fab2 and Cy5 donkey anti-goat Fab2(Jackson ImmunoResearch, 711-096-152 and 705-175-147) tovisualize the cotransporter and ChAT antibodies, respectively.Vibratome sections were imaged by using a Zeiss laser confocalmicroscope with a 63 � 1.4-numerical aperture oil-immersionobjective and a 1-�m optical thickness along the z axis.

Determination of NKCC2 and KCC2 Staining Patterns on StarburstDendrites. Iontophoretically filled (Alexa Fluor 488; MolecularProbes A-10440) DAPI-labeled displaced SACs were double-labeled for NKCC2 or KCC2 as above, except that the antiserawere visualized with Alexa Fluor 647 goat anti-rabbit Fab2

(Molecular Probes A11070). A Zeiss laser confocal microscopewas used for image and data acquisition. Images were scannedat a 0.7-�m optical thickness. Image stacks were processed withImageJ (http://rsb.info.nih.gov/ij) and Photoshop 7.0 (Adobe).Colocalized pixels (coincident red and green channels) weredetermined for each slice and recorded as a separate (blue)channel by using the ImageJ RGB colocalization plug-in. z-projections were then generated by using the ImageJ RGBprojection plug-in, and the images were converted to Tiff format.Photoshop was used to assess the percent ratio of colocalizedpixels within proximal, intermediate, and distal compartments,based on unique, previously described SAC morphological char-acteristics (27, 32).

We thank Dr. Christophe Ribelayga, Dr. Yu Cao, Dr. Julie Sandell, andDr. Carter Cornwall for helpful discussions and Dr. Richard Maslandand Dr. Samuel Wu for critical comments on an earlier version of themanuscript. This work was supported in part by National Institutes ofHealth Grants EY014235 (to S.C.M.) and EY007552 (to C.L.Z.).

1. Barlow HB, Hill RM, Levick WR (1964) J Physiol (London) 173:377–407.2. Yoshida K, Watanabe D, Ishikane H, Tachibana M, Pastan I, Nakanishi S

(2001) Neuron 30:771–780.3. Amthor FR, Keyser KT, Dmitrieva NA (2002) Visual Neurosci 19:495–509.4. Euler T, Detwiler PB, Denk W (2002) Nature 418:845–852.5. Gavrikov KE, Dmitriev AV, Keyser KT, Mangel SC (2003) Proc Natl Acad Sci

USA 100:16047–16052.6. Fried S, Munch TA, Werblin FS (2002) Nature 420:411–414.7. Caldwell JH, Daw NW, Wyatt HJ (1978) J Physiol (London) 276:277–298.8. Russell JM (2000) Physiol Rev 80:211–276.9. Delpire E, Mount DB (2002) Annu Rev Physiol 64:803–843.

10. Payne JA, Rivera C, Voipio J, Kaila K (2003) Trends Neurosci 26:199–206.11. Gamba G (2005) Physiol Rev 85:423–493.12. Brandstatter JH, Greferath U, Euler T, Wassle H (1995) Visual Neurosci

12:345–358.13. Marc RE (1999) J Comp Neurol 407:47–64.14. Zhou ZJ, Fain GL (1995) J Neurosci 15:5334–5345.15. Bloomfield SA (1992) J Neurophysiol 68:711–725.16. Taylor WR, Wassle H (1995) Eur J Neurosci 7:2308–2321.17. Peters BN, Masland RH (1996) J Neurophysiol 75:469–480.

18. Vaney DI (1984) Proc R Soc London Ser B 220:501–508.19. Tauchi M, Masland RH (1984) Proc R Soc London Ser B 223:101–119.20. Maksay G, Korpi ER, Uusi-Oukari M (1998) Neurochem Int 33:353–358.21. Hannaert P, Alvarez-Guerra M, Pirot D, Nazaret C, Garay RP (2002)

Naunyn-Schmiedeberg’s Arch Pharmacol 365:193–199.22. Famiglietti EV (1991) J Comp Neurol 309:40–70.23. Vaney DI, Young HM (1988) Brain Res 438:369–373.24. Brecha N, Johnson D, Peichl L, Wassle H (1988) Proc Natl Acad Sci USA

85:6187–6191.25. O’Malley DM, Sandell JH, Masland RH (1992) J Neurosci 12:1394–1408.26. Cohen ED (2001) Visual Neurosci 18:799–809.27. Ozaita A, Petit-Jacques J, Volgyi B, Ho CS, Joho RH, Bloomfield SA, Rudy

B (2004) J Neurosci 24:9335–9343.28. Jarolimek W, Lewen A, Misgeld U (1999) J Neurosci 19:4695–4704.29. Kelsch W, Hormuzdi S, Straube E, Lewen A, Monyer H, Misgeld U (2001)

J Neurosci 21:8339–8347.30. Spruston N, Jaffe DB, Williams SH, Johnston D (1993) J Neurophysiol

70:781–802.31. Vardi N, Zhang L-L, Payne JA, Sterling P (2000) J Neurosci 20:7657–7663.32. Famiglietti EV (1983) Brain Res 261:138–144.

18798 � www.pnas.org�cgi�doi�10.1073�pnas.0604551103 Gavrikov et al.

Dow

nloa

ded

by g

uest

on

June

18,

202

0