Transgenic mice expressing a pH and Cl- sensing yellow-fluorescent protein under the control of a potassium channel promoter

Transgenic mice expressing a pH and Cl± sensing yellow-¯uorescent protein under the control of a potassiumchannel promoter

Friedrich Metzger,1 Vez Repunte-Canonigo,1 Shinichi Matsushita,1 Walther Akemann,1 Javier Diez-Garcia,2

Chi Shun Ho,3 Takuji Iwasato,4 Pedro Grandes,2 Shigeyoshi Itohara,4 Rolf H. Joho3 and Thomas KnoÈpfel11Laboratory for Neuronal Circuit Dynamics, Brain Science Institute, RIKEN, 2±1 Hirosawa, Wako-shi, Saitama 351±0198, Japan2Department of Neurosciences, Faculty of Medicine and Dentistry, Basque Country University, 699±48080 Bilbao, Spain3Center for Basic Neuroscience, The University of Texas South-western Medical Center, Dallas, TX 75390±9111, USA4Laboratory for Behavioural Genetics, Brain Science Institute, RIKEN, 2±1 Hirosawa, Wako-shi, Saitama 351±0198, Japan

Keywords: transgenic mice, green-¯uorescent protein, brain slices, micro¯uorometry, potassium channel

Abstract

During the last few years a variety of genetically encodable optical probes that monitor physiological parameters such as local

pH, Ca2+, Cl±, or transmembrane voltage have been developed. These sensors are based on variants of green-¯uorescent

protein (GFP) and can be synthesized by mammalian cells after transfection with cDNA. To use these sensor proteins in intactbrain tissue, speci®c promoters are needed that drive protein expression at a suf®ciently high expression level in distinct neuronal

subpopulations. Here we investigated whether the promoter sequence of a particular potassium channel may be useful for this

purpose. We produced transgenic mouse lines carrying the gene for enhanced yellow-¯uorescent protein (EYFP), a yellow-green

pH- and Cl± sensitive variant of GFP, under control of the Kv3.1 K+ channel promoter (pKv3.1). Transgenic mouse linesdisplayed high levels of EYFP expression, identi®ed by confocal microscopy, in adult cerebellar granule cells, interneurons of the

cerebral cortex, and in neurons of hippocampus and thalamus. Furthermore, using living cerebellar slices we demonstrate that

expression levels of EYFP are suf®cient to report intracellular pH and Cl± concentration using imaging techniques and conditionsanalogous to those used with conventional ion-sensitive dyes. We conclude that transgenic mice expressing GFP-derived

sensors under the control of cell-type speci®c promoters, provide a unique opportunity for functional characterization of de®ned

subsets of neurons.

Introduction

The green-¯uorescent protein (GFP) from the jelly®sh Aequorea

victoria has been successfully used as a tag for proteins allowing

visualization of their expression level and localization in cell culture

systems and in transgenic animals. More recently, construction of

GFP-based genetically encodable ¯uorescent probes that permit the

measurement of a range of physiologically relevant quantities became

feasible. GFP-based probes that monitor intracellular changes in pH,

Ca2+, Cl± and transmembrane voltage have been described (Kneen

et al, 1998; Elsliger et al., 1999; Miyawaki et al., 1999; Kuner &

Augustine 2000; Sakai et al, 2001). Initially, all these probes were

developed and tested in cell culture systems but they could eventually

also be applied to intact tissues such as the brain in vitro and in vivo.

In particular, optical recordings from de®ned populations of nerve

cells would then be a promising approach to study the functional

interaction of large numbers of neurons. To this end, promoters are

needed that drive expression of the ¯uorescent sensor proteins at

suf®cient high levels in distinct neuronal subpopulations whose

anatomical and physiological functions are known. Several investi-

gators have generated transgenic mice expressing GFP under the

control of promoters driving marker genes characteristic for certain

neuronal subpopulations (Spergel et al., 1999; Oliva et al., 2000), or

in a shotgun approach using the promoter for the lymphocyte and

brain-speci®c protein Thy-1 (Feng et al., 2000). In general, different

neurons often have distinct electrical properties that correlate with a

particular set of voltage-gated ion channels. Voltage-gated potassium

(Kv) channels form a large and diverse class of ion channels and are

particularly important in establishing the electroresponsive properties

of neurons. Hence, we considered it possible that the regulatory DNA

sequences of Kv channels might be good candidates to speci®cally

drive GFP expression in distinct subsets of neurons. Because ion

channels, including Kv channels, are often expressed at relatively low

levels, it was not clear if Kv-speci®c promoter sequences could drive

GFP expression at high enough levels for easy ¯uorescence detection

and quanti®cation.

Here, we tested the concept that promoter sequences of Kv

channels may be used to drive GFP expression in speci®c subsets

of neurons at high enough levels to be used to monitor the

intracellular microenvironment in brain slices and eventually in

the live animal. We chose the regulatory sequences of Kv3.1

(pKv3.1), which had been successfully used to express galacto-

sidase (lacZ) activity in mouse brain (Gan et al., 1999), to drive

the expression of enhanced yellow-¯uorescent protein (EYFP).

Kv3.1 is a non-inactivating, delayed recti®er involved in rapid

Correspondence: Dr Thomas KnoÈpfel, as above.E-mail: [email protected]

Received 17 August 2001, revised 8 November 2001, accepted 19 November2001

European Journal of Neuroscience, Vol. 15, pp. 40±50, 2002 ã Federation of European Neuroscience Societies

repolarization of the action potential (Yokoyama et al., 1989;

Luneau et al., 1991). Kv3.1 subunit-encoding mRNA is expressed

in the adult rat and mouse brain in a subset of cells in cerebral

cortex, striatum and hippocampus, in the reticular nucleus of the

thalamus, in the granule cells and the deep nuclei of the

cerebellum, and in several nuclei of the brainstem (Drewe et al.,

1992; Perney et al., 1992; Rettig et al., 1992; Rudy et al., 1992;

Weiser et al., 1994, 1995; Du et al., 1996; Perney & Kaczmarek,

1997; Sekirnjak et al., 1997). We generated several lines of mice

expressing EYFP under control of pKv3.1 and found that this

promoter drives high levels of EYFP expression in subsets of

neurons. Comparison of different mouse lines revealed a relatively

consistent pattern of expression. It differed, however, in several

aspects from the expected expression pattern based on the

reported expression of the Kv3.1 protein. The expression level

was high enough to characterize the dependence of EYFP

¯uorescence on intracellular pH and intracellular Cl± concentration

in live brain tissue.

Materials and methods

Transgene construction and generation of transgenic mice

A 6.0-kb HindIII fragment of rat genomic DNA (a kind gift of Dr

Leonard Kaczmarek, Yale University) encompassing »5.3 kb regu-

latory sequences upstream of the transcription initiation site of the

Kv3.1 gene followed by the ®rst 739 bp of 5¢ untranslated region

(Gan et al., 1996, 1999) was subcloned in the HindIII site of pEYFP-

1 (Clontech, Palo Alto, CA, USA). This plasmid was cut with

Eco47III and A¯II, and the resulting »7 kb pKv3.1-EYFP fragment

was used to inject the pronuclei of fertilized eggs from C57Bl/

6 3 DBA F1 (BDF1) crosses (Hogan et al., 1994).

Screening and breeding of transgenic mice

Transgenic mice were genotyped using either Southern blot analysis

or a polymerase chain reaction (PCR) protocol with genomic DNA

isolated from mouse tails. For PCR analysis, 1 mL of the genomic

DNA (»500 ng) was applied to every PCR reaction using an EYFP-

speci®c primer set (5¢-GAAGTTCATCTGCACCACCG-3¢ and 5¢-GCGGACTTGAAGAAGTCGTG-3¢) and an internal control-primer

set that spanned part of the gene encoding the NMDAR1 subunit (5¢-

AGCCCTTCAAGTACCAGGGCCTGAC-3¢ and 5¢-AGCGGTCCA-

GCAGGTACAGCATCA-3¢). For Southern blot analysis, the EcoRI-

NotI fragment corresponding to the whole EYFP coding region of the

transgene was directly labelled using the Gene Images Alkaline

Phosphatase direct labeling kit with CDP-star chemiluminescent

detection reagent (Amersham Pharmacia Biotech, Japan). Founder

mice were crossed to wild-type ICR mice to generate F1

heterozygous mice. Subsequent generations of transgenic mice were

obtained by crossing with ICR.

Anatomical analysis

Adult transgenic mice (approximately 30±45 days old) were

anaesthetized with Nembutal sodium solution (Abbott

Laboratories Inc. A, Illinois, USA) and perfused transcardially

with arti®cial cerebrospinal ¯uid (ACSF, for composition see

below) followed by 2% paraformaldehyde in 100 mM phosphate

buffer (pH 8.0). Brains were removed and immersed in the same

®xative for at least 2 h at 4 °C. Fifty micrometer thick coronal

brain slices were prepared using a vibratome. Brain sections were

submerged in a chamber containing 100 mM phosphate buffer at

room temperature. The imaging chamber was positioned on the

stage of an upright laser-scanning microscope (Fluoview,

Olympus, Tokyo, Japan). EYFP ¯uorescence was excited through

water-immersion objectives (103, 203, 403, 603) using the

5‘UTRKv3.1 upstream sequence EYFP SV40poly A

ANHEN

1.0 kb

HEc

~ 0.5 kb

Sp-1

Ap-1

Cre-like

5‘

5‘UTRKv3.1 upstream sequence EYFP SV40poly A

ANHEN

1.0 kb

HEc

~ 0.5 kb

Sp-1

Ap-1

Cre-like

5‘

FIG. 1. DNA construct for targeted expression of EYFP. Approximately6.0 kb genomic rat DNA upstream of the Kv3.1 gene including twoinitiation sites of transcription and the 5¢-untranslated region was fused to amodi®ed GFP gene (EYFP) that shows enhanced ¯uorescence and stabilityin mammalian cells (Cormack et al., 1996). The positions for the consensussites for Sp-1, Ap-1 and for a CRE-like element are shown, and the twotranscription initiation sites mapped by Gan et al. (1996, 1999) are markedby arrows. Abbreviations for restriction enzyme sites are: A, A¯II; Ec,Eco47III; E, EcoRI; H, HindIII; N, NotI.

TABLE 1. EYFP expression in different brain regions was determined by

confocal laser-scanning microscopy

Brain region

EYFP in transgenic line Kv3.1 expression

#27 #29 #41 #44 mRNA Protein

Neocortexcortical layers V > V > II±V > V > VI > II±IV > II±IV >

I±III I±III VI II±IV V±VI V±VIHippocampus

CA1 pyramidal cells + 6 ++ + + ±int. st. rad. +++ ++ ++ ± ++ +int. st. ori. +++ ± ++ ± ++ +CA3 pyramidal cells ± ± 6 ± +++ ±int. st. rad. 6 + ++ ± + +int. st. ori. 6 + ++ 6 + +dentate gyrus 6 ± ++ ++ +++ ±st. gran. 6 ± +++ + + +hilus ± + ++ ± 6 +

Basal nucleisubthalamic nucleus ++ ± + ± +++ +submedius nucleus ++ ± ++ 6

Thalamic nucleiMediodorsal + ± +++ ± 6ventral posterolateral ++ ± ++ 6 ++ventral posteromedial ++ ± ++ ++ ++ventromedial ++ ± + + 6reticular + ± 6 ± +++ +

Brainstemzona incerta ++ ± + ± +++ +

Cerebellummolecular cell layer ++ + ++ + 6 ++Purkinje cells ± ± ± ± ± ±granule cell layer +++ ++ +++ +++ +++ ++

The following scale for expression levels was used: +++ high; ++ moderate; +weak; 6 weak to no ¯uorescence; ± not detectable. The expression patterns ofKv3.1 mRNA and Kv3.1 protein are from Perney et al. (1992) and Weiseret al. (1994, 1995).

EYFP expression controlled by the Kv3.1 promoter 41

ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 40±50

42 F. Metzger et al.


515 nm line of an argon laser. The emitted ¯uorescent light

passed a barrier ®lter (530 nm) and a confocal aperture.

EYFP ¯uorescence as a function of intracellular pH and Cl±

concentration

Parasagittal cerebellar slices were prepared from adult mice follow-

ing previously established techniques (KnoÈpfel et al., 2000). Brie¯y,

the animals were anaesthetized with ether and decapitated. The

cerebellar vermis was removed and placed in ice-cold ACSF

containing (in mM) 118 NaCl, 3 KCl, 1 MgCl2, 1.5 CaCl2, 1

NaH2PO4, 25 NaHCO3, 10 D-glucose. The solution was gassed with

95% O2/5% CO2 resulting at pH 7.4. Parasagittal cerebellar slices

(200 mm thick) were cut using a vibratome and stored for up to 8 h at

25 °C. A single cerebellar slice was transferred to a recording

chamber mounted on an upright ®xed-stage microscope equipped

with a digital imaging system (Axon Imaging Workbench, Axon

Instruments, Inc., CA, USA) or on the stage of a confocal microscope

(Leica TCS, Leica Microsystems Heidelberg GmbH, Germany).

Slices were fully submerged in continuously ¯owing ACSF (26±

28 °C, 2 mL/min) and gassed with 95% O2/5% CO2. EYFP ¯uores-

cence was monitored through a 103 or 403 water-immersion

objective. For epi¯uorescence micro¯uorometry EYFP excitation was

at 510 nm and ¯uorescent light passed through a 530-nm longpass

®lter. Confocal laser scanning microscopy was performed exciting

the slices at 514 nm and recording the ¯uorescence emission at 530±

570 nm.

In a series of experiments slices were ester-loaded with the pH-

indicating ¯uorescent dye SNARF-5F 5-carboxylic acid (SNARF,

20 mM, 60 min at room temperature, Molecular Probes, Eugene,

OR, USA). In combined EYFP and SNARF measurements, slices

were excited at 514 nm; EYFP emission was recorded at

540 6 10 nm and SNARF ¯uorescence was measured at

600 6 10 nm and 640 6 20 nm, respectively. This protocol

permitted us to measure changes in EYFP ¯uorescence simultan-

eously with ratiometric measurement of the intracellular pH. To

calibrate the in situ pH dependence of EYFP ¯uorescence, the

perfusate contained the K+/H+ ionophore nigericin (10 mM) and (in

mM): 10 NaCl, 115 K-gluconate, 2 CaCl2, 1 MgCl2, buffered

either in 25 mM acetate, MES or HEPES (adjusted with KOH to

different pH values). To calibrate the in situ Cl± dependence of

EYFP ¯uorescence, the perfusate initially contained (in mM): 10

Na-gluconate, 115 K-gluconate, 2 CaCl2, 1 MgCl2, 25 HEPES at

pH 7.5. Solutions containing different concentrations of Cl± (0,

25, 62.5, 100, 125 mM at pH 7.5) were prepared by substituting

equimolar concentrations of K-gluconate and Na-gluconate with

KCl and NaCl, respectively. The solutions were added sequen-

tially to the slices along with nigericin (5 mM) and the OH±/Cl±

antiporter tributyltinchloride (100 mM) (Jayaraman et al., 2000).

Fluorescence was recorded as described above.

Calculations and statistics

Individual physiological experiments were performed at least ®ve

times in three different preparations and the data were pooled. Results

were expressed as means 6 standard error of mean (SEM). Individual

data columns were compared for statistical difference using unpaired

t-tests.

Results

Transgenic mice expressing EYFP directed by Kv3.1regulatory sequences

Rat and mouse DNA sequences are nearly identical throughout their

coding regions and are also very similar in parts of their 5¢untranslated regions. Gan et al. (1999) used a 4.3-kb rat genomic

DNA located upstream of the Kv3.1 initiation site of transcription

followed by 739 base pairs 5¢ untranslated region (5¢ UTR) to drive

lacZ (b-galactosidase) expression in Kv3.1-expressing neurons of

transgenic mice. We assembled a somewhat larger DNA construct in

which 6 kb of rat genomic DNA was fused to the coding region of

EYFP (Fig. 1). This 6 kb genomic DNA fragment contains many of

the regulatory elements necessary for neuron-speci®c Kv3.1 expres-

sion (Gan et al., 1996, 1999). It contains ®ve consensus SP-1

elements, one consensus Ap-1 site and one CRE-like site (sites are

highlighted in Fig. 1), followed by two transcriptional start sites

(arrows in Fig. 1). The 6 kb fragment was inserted upstream of the

translation initiation site of the EYFP coding region, and the resulting

plasmid was used for pronuclear injection of fertilized mouse eggs.

We produced six transgenic founder mice (F0) in which the

pKv3.1-EYFP transgene was incorporated in the germ line and

passed on to subsequent generations. The EYFP expression patterns

in F2 animals of four independent lines (lines 27, 29, 41 and 44) were

determined. The results are summarized in Table 1 and are described

in detail below. Although the four lines studied show similar

expression patterns, there are some differences among the different

lines. It appears that all lines show neuron-speci®c EYFP expression;

in particular, we did not detect any ¯uorescence in Bergmann glia in

the cerebellum or in hippocampal astrocytes.

Previous characterization of the expression pattern of Kv3.1

protein and mRNA has established that the granule cell layer in the

cerebellar cortex shows high levels of Kv3.1 mRNA (Drewe et al.,

1992; Perney et al., 1992; Rettig et al., 1992; Weiser et al., 1994) and

that the Kv3.1 protein is expressed in cell bodies, axons (parallel

®bers) and proximal dendrites of granule cells (Weiser et al., 1995;

Sekirnjak et al, 1997). Little or no Kv3.1 protein was detected in

Purkinje, basket or stellate cells. In all transgenic mouse lines

analyzed, EYFP expression was coextensive with the expression

patterns of Kv3.1 mRNA and protein. The granule cell bodies showed

the highest level of ¯uorescence, and it appeared that EYFP had also

diffused extensively throughout the parallel ®ber system as seen by

the ¯uorescence of the molecular layer (Figs 2E±F and 3G±H). EYFP

FIG. 2. Confocal images of brain regions of mouse line 27. (A) A montage showing the EYFP expression patterns in the adult hippocampus. The arrowindicates the stratum lucidum of the CA3 region. (B) Higher magni®cation image of the CA1 region shown in A. The position and shape of the EYFP-expressing cells is characteristic for GABAergic neurons. (C) EYFP-expressing layers of the cingulate cortex. Note the intense ¯uorescence in super®ciallayers. (D) Higher magni®cation images of the cortex. Note that most EYFP-expressing cell bodies are con®ned to layer V EYFP ¯uorescence is also visiblein structures reminiscent of cortical barrels (arrows). (E) EYFP expression in the cerebellum. (F) Higher magni®cation image of E showing intense¯uorescence in the granule cell bodies and in axons (parallel ®bers) in the molecular layer. Arrows indicate non¯uorescent Purkinje cell somata. (G) EYFPexpression in the thalamus. EYFP ¯uorescence can be detected in both cell bodies and in their projections. Scale bars in A and G, 1000 mm; in B and F,50 mm; in C and E, 500 mm; in D, 200 mm. Abbreviations: m, molecular layer of the cerebellum; P, Purkinje cells; g, granule cell layer of the cerebellum;VPM, ventral posteromedial nucleus; VPL, ventral posterolateral nucleus; VM, ventromedial nucleus; ZI, zona incerta; Sth, subthalamic nucleus; MD,mediodorsal nucleus; Sub, submedius thalamic nucleus; RT, thalamic reticular nucleus.





¯uorescence was undetectable in the dendritic arborization and cell

bodies of Purkinje cells, which appeared as a black contrast against

the intense EYFP ¯uorescence (arrows in Figs 2F and 3H). In the

granule cell layer, small non¯uorescent holes reminiscent of the

shape and distribution of Golgi cells were visible. It appeared that no

cell bodies were labeled in the molecular layer, indicating that stellate

or basket cells did not express EYFP.

The expression patterns in cerebral cortex and hippocampus were

not identical to the ones described for Kv3.1-expressing interneurons

(Du et al., 1996; Sekirnjak et al., 1997). Thus, in line 27, most EYFP-

expressing cell bodies in the primary somatosensory cortex were

con®ned to layer V. Diffuse EYFP ¯uorescence was also visible in

structures reminiscent of cortical barrels (arrows in Fig. 2D). In line

41, cell bodies in the somatosensory cortical area showed much

fainter EYFP ¯uorescence; however, the barrel ®eld-like ¯uorescence

was also present (arrows Fig. 3E). The barrel ®elds correspond to the

terminal ®eld of the projections from the ventral posterior lateral

(VPL) and ventral posterior medial (VPM) thalamic nuclei.

Interestingly, many EYFP-¯uorescent cell bodies were seen in

these nuclei (see below) consistent with the idea that the barrel

®eld-like EYFP ¯uorescence emerged from thalamocortical ®bers

(Woolsey & van der Loos, 1970).

Hippocampal expression of EYFP involved features consistent

between lines but also some variations between lines. A high degree

of EYFP ¯uorescence was seen in the stratum lucidum of the CA3

region (arrow in Figs 2A, and 3A and D). High-resolution confocal

image analysis suggested that this ¯uorescence emerged from axonal

arborizations of interneurons. In line 27, strongly EYFP-expressing

cells in CA1 had the localization and characteristic shapes of Kv3.1-

expressing GABAergic interneurons (Fig. 2B). In line 41, unlike line

27, pyramidal cells in CA1 also expressed signi®cant amounts of

EYFP (Fig. 3C). Pyramidal cells in CA3 did not express signi®cant

levels of EYFP in any of the four lines (Fig. 2A and B). Moreover,

the granule cell bodies in the dentate gyrus showed variable levels of

¯uorescence in all lines (Fig. 3B).

In the thalamus, EYFP-¯uorescence patterns were clearly different

from the Kv3.1-expression pattern that has been described (Weiser

et al., 1994, 1995). Although high levels of Kv3.1 mRNA are present

in the reticular nucleus of the thalamus (NRT) (Drewe et al., 1992;

Perney et al., 1992; Rettig et al., 1992; Weiser et al., 1994), only low

levels of EYFP ¯uorescence were seen in the NRT of line 27 and no

EYFP ¯uorescence could be detected in the remaining three lines (29,

41 and 44) (Figs 2G and 3F). We detected distinct EYFP signals in

the ventral posterior medial and posterior lateral nuclei, the ventral

medial nucleus, the zona incerta, the subthalamic nucleus, the medial

dorsal nucleus and the submedius thalamic nucleus. Hence, it appears

that the regulatory sequences used to drive the EYFP transgene

results in an expression pattern that is higher in most thalamic relay

nuclei and lower in NRT than expected from in situ hybridization

studies (Drewe et al., 1992; Perney et al., 1992; Rettig et al., 1992;

Weiser et al., 1994).

EYFP ¯uorescence as an indicator for intracellular pH and Cl±

concentration

EYFP ¯uorescence has been reported to exhibit signi®cant pH

dependence over a broad range of cytoplasmic pH values (Elsliger

et al., 1999). More recently, it has been reported that EYFP also

senses the concentration of halides, such as Cl± (Wachter &

FIG. 3. Confocal images of brain regions of mouse line 41. (A) EYFP expression pattern in the hippocampus. (B±D) Higher magni®cation of the areasdenoted in A. (B) The hilus of the dentate gyrus showing EYFP ¯uorescence in granule cells. (C) Pyramidal cells in the CA1 layer showing ¯uorescentprojections toward stratum radiatum (the dendrites of pyramidal cells). (D) Prominent ¯uorescence of ®bers in the stratum lucidum of the CA3 region. Notethat there is no ¯uorescence in CA3 pyramidal cells. (E) EYFP ¯uorescence in a barrel-like pattern of the cortex without labelling of the cell bodies (arrows).(F) EYFP expression pattern in the thalamus. (G and H) EYFP expression in the cerebellum at low and high magni®cation. Arrows in H indicatenon¯uorescent Purkinje cell somata. Scale bars in A, 1000 mm; in B and C, 50 mm; in D and G, 125 mm; in E and F, 500 mm; in H, 100 mm.

FIG. 4. In vivo sensitivity of EYFP ¯uorescence to intracellular pH andchloride concentration. (A) Titration of EYFP ¯uorescence as a function ofintracellular pH in cerebellar slices. The EYPF-¯uorescence response in thegranule cell layer at different pH values and at [Cl±] = 15 mM is shown asinset. Normalized data points (mean 6 SEM) were ®tted to a theoreticaltitration curve with one titratable group. (B) Titration of EYFP ¯uorescencewith varying [Cl±]i at pH 7.5. The EYFP-¯uorescence response in thegranule cell layer of a cerebellar slice to different [Cl±]i is shown as inset.The data were ®tted to a single-site binding model, Y = [Cl±]/(Kapp + [Cl±]),to calculate the apparent binding constant (Kapp) for Cl±.



Remington, 1999; Jayaraman et al, 2000; Wachter et al., 2000). We

determined the ¯uorescence of EYFP as a function of intracellular pH

(pHi) and intracellular Cl± concentration ([Cl±]i) in granule cells of

acutely prepared cerebellar slices (Fig. 4). The titrations for pH and

[Cl±]i were carried out at conditions when extracellular and

intracellular pH and Cl± concentrations were equilibrated. For

EYFP ¯uorescence, we determined a pKa value of 6.1 (at 15 mM

[Cl±]), which is close to the value reported for recombinant EYFP

(Wachter & Remington, 1999). At pH 7.5, the apparent dissociation

constant for Cl± ions was 168 mM.

In order to demonstrate that the sensitivity of EYFP to pHi and

[Cl±]i could be exploited to monitor the functional responses of

neurons in real time, we exposed live cerebellar slices of adult mice

(line 27) to conditions known to affect pHi, [Cl±]i, or both.

Depolarization of neurons either by elevated external potassium

concentration (high [K+]o) or by glutamate is known to induce a fall

in pHi (Chesler, 1990; Chesler & Rice, 1991). In Fig. 5, the effects of

glutamate and high [K+]o on EYFP ¯uorescence are illustrated.

Application of glutamate (1 mM) for 2 min induced a transient

decrease in EYFP ¯uorescence that was fully reversible (Fig. 5B).

The colour-coded image in Fig. 5C shows that this decrease in

¯uorescence (red colours) is restricted to the granule cell layer. The

molecular layer did not respond in a similar manner suggesting that

the signal was mediated by the action of glutamate at the somato-

dendritic compartment of granule cells. An even stronger decrease of

EYFP ¯uorescence was observed by depolarization of the slice with

high [K+]o (50 mM) for 2 min (Fig. 5D). The quantitative evaluation

of the glutamate and high [K+]o responses in Fig. 5E revealed that the

response to 2 min glutamate application was concentration-dependent

and maximal at 1 mM where the ¯uorescence change was 21 6 3%.

Depolarization of the slice by high [K+]o for 2 min decreased the

¯uorescence by 32 6 2%. Both responses were also dependent on the

duration of application as the ¯uorescence changes were smaller after

1 min superfusion with 8 6 1% for glutamate and 21 6 3% for

[K+]o. Because of the faster recovery kinetics, the shorter applications

were used in all further experiments.

To demonstrate that these optical signals report a fall in pHi, we

performed experiments in slices that where additionally loaded with

the pH-sensitive dye SNARF (see Materials and methods). As EYFP

as well as SNARF are best excited at 514 nm, it was possible to

monitor simultaneously EYFP emission at 540 6 10 nm and SNARF

emission ratiometrically at 600 6 10 and 640 6 20 nm.

Interestingly, loading of cerebellar slices with SNARF resulted in

intense staining of granule cells and interneurons of the molecular

layer. The clear labelling of (presumed stellate) interneurons by

SNARF (Fig. 6B) but not by EYFP (Fig. 6A) con®rms that in the

majority of these neurons the Kv3.1 promoter was inactive.

Combined EYFP and SNARF micro¯uorometry revealed intra-

cellular acidi®cation with application of glutamate (1 mM) and the

GABAA agonist muscimol (50 mM), applied for 1 and 2 min,

respectively (Fig. 6C and D). The acidi®cation with muscimol is

expected from bicarbonate ef¯ux through GABAA receptors (Chesler

& Chen, 1992) but also a small change in [Cl±]i is expected if the

resting membrane potential differs from the equilibrium potential of

Cl±. Indeed, the EYFP signal evoked by muscimol (normalized to the

glutamate signal) was signi®cantly larger (26 6 2%) than the

acidi®cation accessed by SNARF (12 6 1%, P < 0.001). After

superfusion of the slice with picrotoxin (20 mM) for 10 min, the

FIG. 5. EYFP-¯uorescence response to glutamate and high [K+]o depolarization. (A) EYFP ¯uorescence image of a cerebellar slice. (B) Representative¯uorescence signal induced by 1 mM glutamate bath-applied for 2 min. (C) Pseudocolor-coded map of ¯uorescence change obtained during peak of responseshown in B. The colourbar represents the relative ¯uorescence change in percent of baseline ¯uorescence (DF/F) (blue equals +5%, red equals ±30%). (D)Representative EYFP ¯uorescence signal induced by depolarization of the slice with 50 mM [K+]o for 2 min. (E) Quantitative evaluation of the ¯uorescencechanges induced by glutamate (100, 300, 1000 mM, dark bars) or high [K+]o (50 mM, white bars) for 1 or 2 min. Data represent means 6 SEM of 9±15 singleexperiments from three to ®ve different preparations. Scale bar in A, 150 mm.



EYFP and pH response of muscimol but not glutamate was readily

blocked (Fig. 6C and D). These data indicate that the decrease in

EYFP ¯uorescence induced by glutamate application is mainly due to

an intracellular acidi®cation whereas muscimol induces an increase in

[Cl±]i and a fall in pHi.

The small size of the Cl± elevation by muscimol is expected

because the driving force for Cl± is small at resting membrane

potential due to a signi®cant tonic GABAA conductance (Brickley

et al., 1996). We therefore reasoned that [Cl±]i elevated signi®cantly

during high [K+]o-induced depolarization of granule cells because

depolarization increases the driving force for Cl± and a tonic Cl±

conductance would then give rise to a signi®cant in¯ux of Cl±. In

order to test this prediction, we performed experiments with

combined measurement of EYFP and SNARF ¯uorescence in control

and Cl±-free ACSF (Fig. 7). The SNARF ¯uorescence signal was

essentially unaffected by the switch to Cl±-free ACSF. In contrast,

switch to Cl±-free ACSF induced a signi®cant increase in EYFP

baseline ¯uorescence and a decrease in the response to high [K+]o

(Fig. 7A). Quantitative evaluation of these data (Fig. 7B) revealed

that in Cl±-free ACSF the high [K+]o-induced EYFP signal is strongly

reduced to 26 6 3% of the control response (P < 0.001) whereas the

SNARF (pHi) signal was not signi®cantly affected (119 6 12% of

control). The independence of the SNARF signal from [Cl±]o

demonstrates that the increase in EYFP ¯uorescence upon switch to

Cl±-free ACSF is due to a fall in [Cl±]i. Furthermore, depolarization

by high external [K+]o is accompanied by an elevation in [Cl±]i in

control ACSF as the EYFP signal is greatly reduced in Cl±-free

ACSF.

Discussion

We investigated whether a K+ channel promoter is suitable for

expression of EYFP as a representative for a new class of genetically

encoded GFP-derived sensors. For this purpose, we generated

transgenic mouse lines carrying the gene for EYFP under control of

the Kv3.1 K+ channel promoter (pKv3.1). We observed high

expression levels of EYFP in subsets of neurons, at least some of

which are known to express the Kv3.1 channel subunit. We con®rmed

that EYFP reports pHi and [Cl±]i in labeled neurons of these

transgenic mice and demonstrated that these ¯uorescence signals can

readily be exploited in physiological experiments using brain slices.

EYFP expression in subsets of neurons

Kv3.1-expressing neurons represent a minority population making

their identi®cation in electrophysiological recordings dif®cult. It

would be advantageous if Kv3.1-expressing neurons could be easily

identi®ed either among acutely dissociated cells or in brain slices.

The present mouse lines partially satisfy this need. The expression

pattern of EYFP appears to be faithful, i.e. nonectopic EYFP

expression is particularly well preserved in cerebellar cortex in all

mouse lines analyzed. However, the expression pattern exhibited also

features that varied between lines and that are not totally predictable

from the known expression of Kv3.1. This discrepancy may be due to

incompleteness of the regulatory sequences. Gan et al. (1999) showed

that 4.3 kb of regulatory DNA and 739 bp of 5¢ UTR were enough to

direct neuron-speci®c b-galactosidase (lacZ) expression in the brain

of transgenic animals. Although the reported lacZ-expression pattern

SNARF

EYFP

5 min

5 %

10 %

Glu Musc Picrotoxin

A

B

EYFP

SNARF

C

F/F

R/R

EYFP SNARF

20406080

100120

Flu

ores

cenc

esi

gnal

[%of

cont

rol]

D

Glu(control) Picrotoxin

***

GluMusc Musc

******

m P g

m P g

SNARF

EYFP

5 min

5 %

10 %

Glu Musc Picrotoxin

A

B

EYFP

SNARF

C

∆ F

∆ R

EYFP SNARF

20406080

100120

Flu

ores

cenc

esi

gnal

[%of

cont

rol]

D

Glu(control) Picrotoxin

***

GluMusc Musc

******

m P g

m P g

FIG. 6. Glutamate- and muscimol-induced changes in EYFP ¯uorescence and intracellular pH in cerebellar slices. (A and B), Fluorescence image of aSNARF-loaded cerebellar slice at emission wavelengths of 540 6 10 nm (EYFP) and 640 6 20 nm (SNARF), respectively. Note SNARF-loaded interneuronsin the molecular layer (m) that are negative for EYFP. No ¯uorescence is observed in the Purkinje cell layer (P). The granule cell layer (g) is heavily stainedwith EYFP as well as SNARF. (C) Simultaneous optical recording of EYFP and SNARF ¯uorescence. The upper trace shows the ratiometric measurement ofthe SNARF signal (DR/R) and the lower trace the measurement of EYFP (DF/F). Responses to bath application of glutamate (Glu, 1 mM, 1 min) andmuscimol (Musc, 50 mM, 2 min) in the presence or absence of picrotoxin (20 mM) are shown. (D) Quantitative evaluation of the EYFP and SNARF signals.Values are given in percentage of the ®rst glutamate application as control in each experiment. Data represent means 6 SEM of 9±15 single experimentsfrom three to ®ve different preparations. (***P < 0.001).



is qualitatively similar to the EYFP-expression patterns shown here,

there are some differences. The construct that we used to drive EYFP

expression contained an additional 1.0 kb of DNA at the very 5¢ end

compared to the construct that was used to drive lacZ expression. We

detected intense ¯uorescence in the granule cell layer and the

molecular layer of the cerebellum, indicating high-level EYFP

expression in the cell bodies and axons of granule cells; however,

EYFP ¯uorescence could not be detected in Purkinje cells. Hence, the

expression pattern of EYFP in the cerebellum re¯ects accurately the

expression pattern of Kv3.1 mRNA. The use of the shorter 4.3 kb

construct led to lacZ expression in both granule and Purkinje cells

(Gan et al., 1999). It is possible that some of the differences between

the EYFP-expression patterns reported here and the lacZ expression

pattern reported by Gan et al. (1999) are due to the additional 1.0 kb

sequence, although we cannot exclude positional effects due to

different chromosomal integration sites.

The identity of some EYFP-labeled cell types still needs to be

established by, for example immunohistochemical markers prior to

their unequivocal identi®cation in dissociated cultures or in brain

slices based on their intrinsic EYFP ¯uorescence.

Expression level of EYFP expression under control of pKv3.1

The use of GFP-derived genetically encodable probes requires that

the recombinant sensor protein be expressed under a promoter that

leads to high expression levels. It was not clear if Kv-speci®c

promoter sequences could drive GFP expression at high enough

levels for easy ¯uorescence detection and quanti®cation. Considering

the fact that channels, including Kv channels, are often expressed at

relatively low levels we could not positively predict that expression

levels were high. It might be that the Kv3.1 promoter is much

stronger than expected from the expression level of the potassium

channel protein accounting for a high turnover rate of the protein.

EYFP as a reporter of the intracellular neuronal milieu

GFP and its brighter variants have been recognized as useful reporters

not only to detect gene expression in vivo and as fusion tags to

monitor intracellular protein location, but, more recently, applications

have been described that employ the fact that GFP can also act as a

reporter of environmental conditions and cellular activity. While this

concept has been veri®ed previously in cell culture systems, we

explored the application of EYFP as a reporter of pHi and [Cl±]i in

B EYFPEYFPSNARFSNARF

0 [Cl-]oHigh [K+]o

10 %

5 min

A∆R/R

10 %∆F/F

SNARF

EYFPC

hang

ein

∆ F/

For

∆R/R

[%]

*** ***control 0 [Cl-]o wash switch

to 0 [Cl-]o

10

20

0

-10

-20

High [K+]o response

FIG. 7. Depolarization-induced changes in EYFP ¯uorescence and intracellular pH in cerebellar slices. (A) EYFP and SNARF ¯uorescence signals induced byhigh [K+]o (50 mM) in control ACSF and Cl±-free ACSF (0 [Cl±]o). Note that the switch to 0 [Cl±]o ACSF induces a signi®cant increase in EYFP baseline¯uorescence and a decrease in the response to high [K+]o while the pHi-related signals (monitored by SNARF) are unaffected by depletion of [Cl±]o. (B)Quantitative evaluation of data obtained as shown in A. Data represent means 6 SEM of seven independent experiments from three different preparations.***P < 0.001 between the observed columns assessed by unpaired t-test.



speci®c subsets of neurons of transgenic mice. The feasibility of this

approach was not clear because use of EYFP as an in vivo reporter

molecule also required that the indicator function of the protein was

preserved in transgenic animals, i.e. the protein should not undergo

secondary modi®cation such as degradation or sequestration during the

life span of the animal. We characterized the ¯uorescence of EYFP

expressed in the cytosol of living cerebellar granule cells as a function

of intracellular pH and Cl± concentration under conditions that are

normally employed for electrophysiological studies. We found that the

pKa value is similar to that reported for the isolated recombinant EYFP

protein at a given intracellular Cl± concentration (Wachter &

Remington, 1999). We estimated the Kd for Cl± (at pH 7.5) to be

168 mM, a value that is clearly smaller and closer to physiological

[Cl±]i-values than that reported (777 mM) for the isolated recombinant

EYFP protein (Wachter et al., 2000).

Changes in pHi by glutamate or high [K+]o could be readily

resolved by standard confocal imaging techniques. Muscimol-

induced changes in [Cl±]i were modest as expected from the small

driving force for Cl± ions at resting membrane potential in adult

animals. Larger responses of EYFP ¯uorescence (that were not

paralleled by SNARF reported pHi changes) were seen under

depolarizing conditions, i.e. with an increased driving force for Cl±.

These responses contained a large portion of ¯uorescence changes

mediated by Cl± in¯ux as more than 70% of the response was

abolished in Cl±-free superfusate. Recently, a genetically encodable

Cl± sensor based on ¯uorescence resonance energy transfer (FRET)

between two GFP molecules has been introduced (Kuner &

Augustine 2000). Interpretation of measurements using this sensor,

like EYFP used here, has to be careful and requires consideration of

signi®cant changes in pHi during physiological activity. However,

experimental conditions (Cl±-free media, media leading to strong

buffering of pHi) might be selected under which changes in Cl± and

pHi can be separated (Kuner & Augustine, 2000).

We conclude that transgenic mice expressing GFP-derived sensors

under the control of cell-type speci®c promoters provide a unique

opportunity for functional characterization of de®ned subsets of

neurons either using conventional electrophysiological techniques or

multisite micro¯uorometric recordings.

Acknowledgements

Rat genomic DNA encompassing the Kv3.1 promoter sequences was kindlyprovided by Dr L. Kaczmarek (Yale University, New Haven, CT). We thankMs Ayako Takada for expert administrative and secretarial assistance. Thisstudy was supported in parts by grants from the RIKEN Brain ScienceInstitute, Ministerio de Sanidad y Consumo, Fondo de Investigacion Sanitaria,grant number FIS00/0198; Gobierno Vasco, grant number PI-1999-129 andThe Basque Country University, grant number UPV212.327-G24/99. J. Diez-Garcia was in receipt of a fellowship from The Basque Country University(UPV212.327-G24/99).

Abbreviations

EYFP, green-¯uorescent protein; GFP, enhanced yellow-¯uorescent protein.

References

Brickley, S.G., Cull-Candy, S.G. & Farrant, M. (1996) Development of a tonicform of synaptic inhibition in rat cerebellar granule cells resulting frompersistent activation of GABAA receptors. J. Physiol., 497, 753±759.

Chesler, M. (1990) The regulation and modulation of pH in the nervoussystem. Prog. Neurobiol., 34, 401±427.

Chesler, M. & Chen, J.C. (1992) Alkaline extracellular pH shifts generated bytwo transmitter-dependent mechanisms. Can. J. Physiol. Pharmacol., 70(Suppl.), 286±292.

Chesler, M. & Rice, M.E. (1991) Extracellular alkaline-acid pH shifts evokedby iontophoresis of glutamate and aspartate in turtle cerebellum.Neuroscience, 41, 257±267.

Cormack, B. P., Valdivia, R. & Falkow, S. (1996) FACS-optimized mutants ofthe green ¯uorescent protein (GFP). Gene, 173, 33±38.

Drewe, J.A., Verma, S., Frech, G. & Joho, R.H. (1992) Distinct spatial andtemporal expression patterns of K+ channel mRNAs from differentsubfamilies. J. Neurosci., 12, 538±548.

Du, J., Zhang, L., Weiser, M., Rudy, B. & McBain, C.J. (1996) Developmentalexpression and functional characterization of the potassium-channel subunitKv3.1b in parvalbumin-containing interneurons of the rat hippocampus. J.Neurosci., 16, 506±518.

Elsliger, M.A., Wachter, R.M., Hanson, G.T., Kallio, K. & Remington, S.J.(1999) Structural and spectral response of green ¯uorescent protein variantsto changes in pH. Biochemistry, 38, 5296±5301.

Feng, G., Mellor, R.H., Bernstein, M., Keller-Peck, C., Nguyen, Q.T.,Wallace, M., Nerbonne, J.M., Lichtman, J.W. & Sanes, J.R. (2000) Imagingneuronal subsets in transgenic mice expressing multiple spectral variants ofGFP. Neuron, 28, 41±51.

Gan, L., Hahn, S.J. & Kaczmarek, L.K. (1999) Cell type-speci®c expression ofthe Kv3.1 gene is mediated by a negative element in the 5¢ untranslatedregion of the Kv3.1 promoter. J. Neurochem., 73, 1350±1362.

Gan, L., Perney, T.M. & Kaczmarek, L.K. (1996) Cloning andcharacterization of the promoter for a potassium channel expressed inhigh-frequency ®ring neurons. J. Biol. Chem., 271, 5859±5865.

Hogan, B., Beddington, R., Costantini, F. & Lacy, E. (1994) Manipulating themouse embryo: a laboratory manual. 2nd edn. Cold Spring HarborLaboratory Press, Plainview, NY.

Jayaraman, S., Haggie, P., Wachter, R.M., Remington, S.J. & Verkman, A.S.(2000) Mechanisms and cellular applications of a green ¯uorescent protein-based halide sensor. J. Biol. Chem., 275, 6047±6050.

Kneen, M., Farinas, J., Li, Y. & Verkman, A.S. (1998) Green ¯uorescentprotein as a noninvasive intracellular pH indicator. Biophys. J., 74, 1591±1599.

KnoÈpfel, T., Anchisi, D., Alojado, M.E., Tempia, F. & Strata, P. (2000)Elevation of intradendritic sodium concentration mediated by synapticactivation of metabotropic glutamate receptors in cerebellar Purkinje cells.Eur. J. Neurosci., 12, 2199±2204.

Kuner, T. & Augustine, G.J. (2000) A genetically encoded ratiometricindicator for chloride: capturing chloride transients in cultured hippocampalneurons. Neuron, 27, 447±459.

Luneau, C.J., Williams, J.B., Marshall, J., Levitan, E.S., Oliva, C., Smith, J.S.,Antanavage, J., Folander, K., Stein, R.B., Swanson, R., Kaczmarek, L.K. &Buhrow, S.A. (1991) Alternative splicing contributes to K+ channeldiversity in the mammalian central nervous system. Proc. Natl. Acad. Sci.USA, 88, 3932±3936.

Miyawaki, A., Griesbeck, O., Heim, R. & Tsien, R.Y. (1999) Dynamic andquantitative Ca2+ measurements using improved cameleons. Proc. Natl.Acad. Sci. USA, 96, 2135±2140.

Oliva, A.A. Jr, Jiang, M., Lam, T., Smith, K.L. & Swann, J.W. (2000) Novelhippocampal interneuronal subtypes identi®ed using transgenic mice thatexpress green ¯uorescent protein in GABAergic interneurons. J. Neurosci.,20, 3354±3368.

Perney, T.M. & Kaczmarek, L.K. (1997) Localization of a high thresholdpotassium channel in the rat cochlear nucleus. J. Comp. Neurol., 386, 178±202.

Perney, T., Marshall, J., Martin, K.A., Hock®eld, S. & Kaczmarek, L.K.(1992) Expression of the mRNA for the Kv3.1 potassium channel gene inthe adult and developing rat brain. J. Neurophysiol., 68, 756±766.

Rettig, J., Wunder, F., Stocker, M., Lichtinghagen, R., Mastiaux, F., Beckh, S.,Kues, W., Pedarzani, P., SchroÈter, K.H., Ruppersberg, J.P., Veh, R. &Pongs, O. (1992) Characterization of a Shaw-related potassium channelfamily in rat brain. EMBO J., 11, 2473±2486.

Rudy, B., Kentros, C., Weiser, M., Fruhling, D., Serodio, P., Vega-Saenz deMiera, E., Ellisman, M.H., Pollock, J.A. & Baker, H. (1992) Region-speci®cexpression of a K+ channel gene in brain. Proc. Natl. Acad. Sci. USA, 89,4603±4607.

Sakai, R., Repunte-Canonigo, V., Raj, C.D. & KnoÈpfel, T. (2001) Design andcharacterization of a DNA-encoded, voltage-sensitive ¯uorescent protein.Eur. J. Neurosci., 13, 2314±2318.

Sekirnjak, C., Martone, M.E., Weiser, M., Deerinck, T., Bueno, E., Rudy, B. &



Ellisman, M. (1997) Subcellular localization of K+ channel subunit Kv3.1bin selected rat CNS neurons. Brain Res., 766, 173±187.

Spergel, D.J., Kruth, U., Hanley, D.F., Sprengel, R. & Seeburg, P.H. (1999)GABA- and glutamate-activated channels in green ¯uorescent protein-tagged gonadotropin-releasing hormone neurons in transgenic mice. J.Neurosci., 19, 2037±2050.

Wachter, R.M. & Remington, S.J. (1999) Sensitivity of the yellow variant ofgreen ¯uorescent protein to halides and nitrate. Curr. Biol., 9, R628±R629.

Wachter, R.M., Yarbrough, D., Kallio, K. & Remington, S.J. (2000)Crystallographic and energetic analysis of binding of selected anions tothe yellow variants of green ¯uorescent protein. J. Mol. Biol., 301, 157±171.

Weiser, M., Bueno, E., Sekirnjak, C., Martone, M.E., Baker, H., Hillman, D.,Chen, S., Thornhill, W. & Rudy, B. (1995) The potassium channel subunit

Kv3.1b is localized to somatic and axonal membranes of speci®cpopulations of CNS neurons. J. Neurosci., 15, 4298±4314.

Weiser, M., Vega-Saenz de Miera, E., Kentros, C., Moreno, H., Franzen, L.,Hillman, D., Baker, H. & Rudy, B. (1994) Differential expression of Shaw-related K+ channels in the rat central nervous system. J. Neurosci., 14, 949±972.

Woolsey, T.A. & van der Loos, H. (1970) The structural organization of layerIV in the somatosensory region (SI) of mouse cerebral cortex. Thedescription of a cortical ®eld composed of discrete cytoarchitectonic units.Brain Res., 17, 205±242.

Yokoyama, S., Imoto, K., Kawamura, T., Higashida, H., Iwabe, N., Miyata, T.& Numa, S. (1989) Potassium channels from NG108-15 neuroblastoma-glioma hybrid cells. Primary structure and functional expression fromcDNAs. FEBS Lett., 259, 37±42.



Transgenic mice expressing a pH and Cl- sensing yellow-fluorescent protein under the control of a potassium channel promoter

Documents