Neuron, Vol.
14, 211-215, February, 1995, Copyright © 1995 by Cell Press
The Jellyfish Green Fluorescent
Protein: A New Tool for Studying
Ion Channel Expression and Function
John Marshall,* Raymond Molloy,t Guy W. J. Moss,*
James R. Howe,* and Thomas E. Hughestt
*Deparment of Pharmacology
tSection of Neurobiology
SDepartment of Ophthalmology and Visual Science
Yale University School of Medicine
New Haven, Connecticut 06520
Summary
Two methods are described for using the jellyfish
green fluorescent protein (GFP) as a reporter gene for
ion channel expression. GFP fluorescence can be used
to identify the transfected cells, and to estimate the
relative levels of ion channel expression, in cotrans-
fection experiments. A GFP-NMDAR1 chimera can be
constructed that produces a functional, fluorescent
receptor subunit. These methods should facilitate
studies of ion channel expression, localization, and
processing.
Introduction
A problem in transfecting cloned ion channels into mam-
malian cells is identifying the cells that express the chan-
nel and/or localizing the channel within a cell. A variety of
strategies have been employed to overcome this problem,
including generating stable cell lines and using antibodies,
fluorescent enzyme substrates, or the luciferase gene. Al-
though powerful in certain applications, each of these ap-
proaches has limitations.
It has been shown recently that expression of the green
fluorescent protein (GFP) from the jellyfish Aequorea victo-
ria (Prasher et al., 1992) in Escherichia coil or Caenorhab-
ditis elegans produces a protein that is strongly fluores-
cent, and that this fluorescence does not appear to depend
upon exogenous substrates or coenzymes (Chalfie et al.,
1994). Subsequently, GFP was used to construct chimeric
proteins that were functional and fluorescent in Drosophila
melanogaster (Wang and Hazelrigg, 1994). Here we de-
scribe expression of GFP in human embryonic kidney cells
(HEK 293; Graham et al., 1977). It produces a robust sig-
nal, making it possible to identify cells for patch-clamp
studies or to follow chimeric receptor proteins.
Results
Cotransfection of GFP with Cloned Receptors
Identifies the Transfected Cells
GFP expressed under the control of a cytomegalovirus
(CMV) promoter is readily detected in HEK 293 cells 2
days after the transfection (Figures 1A-1C). When GFP
is expressed alone, or in pairwise combination with an ion
channel, the signal produced is a bright green fluores-
cence when viewed with conventional epifluorescence op-
tics for fluorescein isothiocyanate. This signal, which is
Neurotechnique
resistant to photobleaching, makes it possible to combine
both bright-field and epifluorescence illumination so that
the transfected and untransfected cells are visible. The
green fluorescence is apparent in both living cells and
those fixed with paraformaldehyde. Although adjacent
HEK 293 cells are known to form gap junctions, there was
no evidence that GFP spread to surrounding cells. The
expression of GFP does not appear to affect the cells ad-
versely, and strong signals can be found in living cells at
least 5 days after the transfection.
Whole-cell recordings from cotransfected, fluorescent
cells reveal that they invariably express the introduced ion
channels (n = 42). Such expression is absent from cells
that do not exhibit detectable fluorescence. In addition,
there is cell to cell variability in the intensity of the GFP-
associated signal (Figure 1C), and there is a correlation
between the intensity of the fluorescence and the level of
expression of functional ion channels. Figures 2A and 2B
show whole-cell recordings from two cells cotransfected
with cDNAs encoding GFP and the fully edited (R) version
of glutamate receptor 6 (GluR6; Egebjerg et al., 1991).
The recording shown in (A) was made from an intensely
fluorescent cell, whereas the recording in (B) was obtained
from a nearby cell that exhibited a much weaker fluores-
cent signal. Both cells were voltage clamped at -80 mV,
and inward currents were evoked by the application of
10 I~M kainate. The current recorded from the intensely
fluorescent cell is about 40 times larger than the corre-
sponding current in the weakly fluorescent cell. A similar
correlation between the intensity of the fluorescent signal
and functional channel expression was observed for cells
that coexpressed GFP and a Ca2+-activated K+ channel
that was cloned from bovine aorta (bSIo; Moss et al., 1995).
Single-channel currents through Ca2÷-activated K÷ chan-
nels in membrane patches isolated from intensely or
weakly fluorescent cells from the same transfection are
shown in Figures 2C and 2D, respectively. The single-
channel activity recorded at +20 mV is 3-4 times higher
in the patch from the intensely fluorescent cell. It was con-
sistently true that patches from intensely fluorescent cells
showed substantially higher levels of channel activity than
patches isolated from weakly fluorescent cells.
GFP did not appear to alter the macroscopic properties
of currents through GluR6(R) channels, or the properties
of recombinant Ca2÷-activated K÷ channels studied in
patches from GFP-identified cells. These latter channels
were very similar in their properties to the corresponding
native channels expressed in smooth muscle cells from
which the cDNA clone was isolated. We therefore found
no indication that GFP alters the properties of ion channels
with which it is coexpressed.
A NMDAR1-GFP Chimeric Protein Makes It
Possible to Follow Functional NMDA
Receptor Proteins in Living Cells
When GFP is fused to the 3' end of the N-methyi-D-
aspartate (NMDA) receptor subunit NMDAR1, a strongly
 Neuron
212

¢,

Figure 1. GFP Expression in HEK 293 Cells

Cotransfection with plasmids encoding GFP and the Ca2+-activated K÷ channel bSIo produces cells with a strong green fluorescence (A and B).
This signal is present in only a fraction of the cells, as seen in the double exposures (B and C), and is distributed throughout the cell. There is
heterogeneity in the intensity of the signal, producing bright (C, hollow arrow) and dim (C, solid arrows) cells. The chimeric NMDAR1-GFP protein
(D and E) produces labeling in the membrane compartments of the cells and, in cases of particularly strong expression, is concentrated in the
perinuclear organelles (D, hollow arrow). Bar in (A), 16 p.m.
GFP as a Reporter in Mammalian Cells
213
A C
9 -- 200ms
L•A
8--
7--
~ ~ p A
6--
5--
4--
3--
2--
los 1--
O--
200rra
B D
3 --
120pA 12--__ ~
10S
0--
Figure 2. The Intensity of the Fluorescent Signal Correlates with the
Level of Functional Channel Expression
(A and S) Whole-cell patch-clamp recordings from H EK 293 cells co-
transfected with cDNAs encoding GluR6(R) and GFP. Each cell was
voltage clamped at -86 mV, and inward currents were evoked by
superfusing the cell with a solution containing 10 pM kainate (duration
indicated by the bar above each record). The recording in (A) is from
a cell that showed very bright GFP-associated fluorescence, whereas
the recording in (B) is from a cell that gave only a very weak fluorescent
signal. Note that the calibrations in (A) and (B) are different. The re-
cords were low pass-filtered at 200 Hz (-3 dB),
(C and D) Recordings of Ca2+-activated K+ currents in inside-out
patches excised from HEK 293 cells that were cotransfected with
cDNAs encoding a Ca2+-activated K+ channel (bSIo) and GFP. The
currents shown in (C) were recorded in a patch taken from an intensely
fluorescent cell, and the currents in (D) were recorded in a patch from
a cell that was only dimly fluorescent. Each patch was voltage clamped
at +20 mV in symmetrical KCI solutions. The free [Ca2+]of the solution
in contact with the cytoplasmic face of the patches was -0.3 #M.
Single-channel currents are outward and have a unitary amplitude of
-5.5 pA. Current levels indicative of the number of channels open
simultaneously are shown at the left of each record. The records were
low pass-filtered at 500 Hz (-3 dB).
fluorescent protein is produced in transfected cells. In con-
trast to the signal seen with GFP alone, the fluorescent
chimeric protein shows a more restricted distribution
within the cell (Figures 1D and 1E). The signal in weakly
labeled cells appears to be largely at the plasmalemma,
whereas in brighter cells there are also large signals asso-
ciated with intracellular organelles.
Cotransfection of the NMDAR1-GFP chimera with
cDNAs encoding either the NMDAR2-A or NMDAR2-C
subunit results in the expression of functional NMDA-type
ion channels. The macroscopic properties of hetero-
oligomeric channels that contain the NMDAR1-GFP chi-
mera are similar to the corresponding properties of hetero-
oligomeric channels that contain authentic NMDAR1, It is
known that NMDARI/NMDAR2-C heteromeric channels
show substantially slower decay kinetics (Monyer et al.,
1992) and substantially greater sensitivity to glycine (Kut-
suwada et al., 1992) than heteromeric NMDARI/NM DAR2-A
channels. These differences are conserved in heteromeric
channels that contain the chimeric construct. Figure 3
shows a concentration-response curve obtained for gly-
cine modulation of glutamate-activated currents from a
350
300 ~ •
250
200
150
100
50
0
10pA GlycinConcent
e ratio(~M)
n
Figure 3. Heteromeric Channels Containing the NMDARI-GFP Chi-
mera Show Normal Sensitivity to the Coagonist Glycine
Concentration-response data obtained from a HEK 293 cell that was
cotransfected with cDNAs that encode the NMDAR1-GFP chimeric
protein and the NMDAR2-C subunit. The cell was voltage clamped in
the whole-cell patch-clamp configuration, and inward currents were
evoked at-80 mV by the application of I ~M glutamate in the presence
of different glycine concentrations. The steady-state amplitude of the
currents is plotted. The smooth curve is the best fit to the data ac-
cording to a Hill-type equation: I = Im,,([glycine]lEC~o)",l([glycine]l
ECso)°H + 1), which gave ECs0 and n, values of 0.64 I~M and 1.1,
respectively.
cell expressing NMDAR1-GFP and NMDAR2-C. The fit
gives an apparent ECs0 value of 0.64 I~M, which is similar
to the value of 0.7 ~M determined for dimeric NMDARI/
NMDAR2-C channels by Kutsuwada et al. (1992).
Further evidence for the functional integrity of the
N M D A R 1 - G F P chimera was obtained from analysis of sin-
gle-channel currents. Figure 4A shows examples of uni-
tary currents activated by 1 I~M glutamate in an outside-out
patch excised from a HEK 293 cell that coexpressed the
N M D A R 1 - G F P chimera and NMDAR2-A. Varying the
membrane potential of the patch showed that the single-
channel currents reversed near 0 mV. The amplitude of
the currents at - 8 0 mV (Figure 4B) is consistent with a
main conductance level of about 50 pS, although a small
number of openings to a conductance level of about 40
pS were also seen. These results are similar to those of
Stern et al. (1992, 1994) for recombinant N M D A R I /
NMDAR2-A channels expressed in either oocytes or HEK
293 cells. The gating kinetics of channels containing the
chimeric subunit also appear to be largely unaltered.
Dwell-time histograms for channel openings and closings
are shown in Figures 4C and 4D, respectively. The distri-
bution of apparent open durations has been fitted with
three exponential components with time constants of 37
~s, 1.8 ms, and 4.7 ms, respectively. Four exponential
components were detected in the distribution of shut
times. The three briefest components (which likely are
bounded by openings of the same channel molecule) had
respective time constants of 30 ~s, 0.21 ms, and 3.9 ms.
These results are similar to the corresponding results re-
ported for the heteromeric channels expressed after co-
transfection of the cDNAs encoding NMDAR2-A and the
authentic NMDAR1 subunit (Stern et al., 1993).
Neuron
214
lO ~ Inward Current (pA)
lo4 10~ lo`2 10-~ lo~ lo~ lo" lo"
A p p a r e n t O p e n T i m e (see) A p p a r e n t S h u t T i m e (sec)
Figure 4. Unitary Properties of Heteromeric Channels Containing the
NMDAR1-GFP Chimera
(A) Examples of single-channel currents evoked by 1 ~.M glutamate
in an outside-out patch excised from a HEK 293 cell that was co-
transfected with cDNAs encoding the NMDAR1-GFP chimera and
NMDAR2-A. The patch was voltage clamped at -80 mV, and channel
openings are downward. The lower trace shows an expanded portion
of the upper trace taken from above the 100 ms time bar. The records
were low pass-filtered at 1 kHz (-3 dB).
(B) Histogram of the amplitude of individual sample points obtained
from a longer stretch of record from the same patch at -80 mV.
Gaussian functions (smooth curves) have been fitted to the peaks
corresponding to the closed level (left) and the main open level (right).
The fit to the open sample points gives a mean inward current ampli-
tude of 4.0 pA, which corresponds to a single-channel conductance
of 50 pS.
(C and D) Dwell-time histograms for channel openings (C) and closings
(D) in the same patch at -80 mV. For the kinetic analysis, the records
were low pass-filtered (digital Gaussian) at 3 kHz (-3 dB) and sampled
at 20 kHz. Note that log durations are plotted. The distributions of
open times were fitted with three exponential components with time
constants (and relative areas) of 0.037 ms (0.62), 1.8 ms (0.33), and
4.7 ms (0.05). The distributions of closed times were fitted with four
exponential components with time constants (and relative areas) of
0.030 ms (0.52), 0.21 ms (0.18), 3.89 ms (0.22), and 12.5 ms (0.08).
Discussion
T h e e x p r e s s i o n of GFP m a k e s it possible to identify rapidly
cells that have been successfully transfected, and there
is a correlation between the intensity of the o b s e r v e d fluo-
rescence and the relative level of functional ion channel
expression. The c o e x p r e s s i o n of GFP d o e s not a p p e a r to
be deleterious to the cells or to alter, in any obvious way,
the functional properties of either bSIo or GluR6. Thus,
GFP promises to be a useful tool for investigations on
recombinant ion channels.
The chimeric N M D A R 1 - G F P subunit is c a p a b l e of form-
ing hetero-oligomeric receptor complexes, and these com-
plexes exhibit both macroscopic and unitary properties
that are indistinguishable from the c o r r e s p o n d i n g proper-
ties of hetero-oligomers containing N M D A R I . This chime-
ric subunit could be used to visualize the receptor subunit
in transfected cell lines (or p e r h a p s primary neurons) in
o r d e r to a d d r e s s questions related to subunit assembly
or clustering, or to the targeting of subunits to specific
membrane compartments. Because the subunit-associated
fluorescence can be imaged repeatedly in live cells, this
appr oach might of f er significant a d v a n t a g e s over immuno-
cytochemical visualization.
Experimental Procedures
GFP Constructs
To express GFP, the KpnI-EcoRI fragment of plasmid TU #65 (Chalfie
et al., 1994) was inserted into the CMV expression vector pcDNA3
(Invitrogen). For the NMDAR1 chimera, polymerase chain reaction
amplification with Vent polymerase (NEB) was used to introduce Nhel
sites into the 3' end of the NMDAR1 subunit (Moriyoshi et al., 1991)
and the 5' end of the GFP cDNA. These sites were then used to create
a chimera in which the last 13 amino acids of the NMDAR1 subunit
lo*
were replaced with an alanine followed by the GFP sequence coding
for amino acids 2-238. The chimera was then inserted into pcDNA3.
GluR6(R) was in pBK-CMV (Stratagene), NMDAR2-A was in pcDNA-
AMP (Invitrogen), and NMDAR2-C was in a modified version of the
pcB6 vector (supplied by Dr. M. Caplan, Yale University). bSIo was
in the pcDNA3 vector.
Cell Culture and Transfection
Human embryonic kidney cells (HEK 293, ATCC CRL 1573) were
raised in MEM-E supplemented with 10% fetal bovine serum. Expo-
nentially growing cells were plated in 35 mm dishes, and calcium
phosphate transfection (Chen and Okayama, 1987) was carried out
with commercially prepared reagents (Stratagene). For cotransfection
experiments, the total amount of DNA added was - 2 0 I~g.
Microscopy and Electrophysiology
Standard epifluoresence optics (Zeiss cube #10, Olympus cube B)
were used to visualize GFP. Whole-cell recordings and recordings
from membrane patches were made as described by Hamill et al.
(1981) with an EPC-9 patch clamp.
The standard extracellular solution (pH 7.2) contained 150 mM
NaCI, 2.5 mM KCI, 1 mM CaCI2,and 10 mM Na-HEPES. For whole-cell
recording, the standard intracellular solution (pH 7.2) contained 140
mM CsCI, 0.5 mM CaCI2, 5 mM Na-EGTA, and 10 mM Na-HEPES.
The experiments on the Ca2÷-activated K+ channel clone were made
in symmetrical solutions containing 150 mM KCI, 5 mM Na-EGTA,
and 3.8 mM CaCI2buffered with 10 mM Na-HEPES (pH 7.2). Solutions
were applied by local superfusion via a wide-mouthed pipette posi-
tioned within - 1 0 0 t~m of the cell or patch. For the experiments on
homomeric GluR6(R) channels, the cells were pretreated with conca-
navalin A (10-25 p.M) for 10-15 rain to remove desensitization.
Whole-cell and single-channel currents were low pass-filtered (Bes-
sel-type) at 10 kHz (-3 dB) and stored on videotape with a VCR-PCM
recorder at a sampling rate of 94 kHz. For the analysis of single-
channel currents, the data were digitized at 20 kHz and low pass-
filtered (digital Gaussian) at 2-4 kHz (-3 dB). A half-amplitude thresh-
old-crossing routine was used to detect and measure the duration
of channel openings and closings (Colquhoun and Sigworth, 1983).
Histograms of dwell-time distributions were constructed as described
by Sigworth and Sine (1987) and were fitted with two to four exponential
components to estimate the time constant and relative area of each
component.
Acknowledgments
We are indebted to Jim Eberwine for his suggestion to use GFP and
Martin Chalfie for generously sharing GFP. We thank Whit Tingley
and Rick Huganir for NMDAR2-A, Steve Heinemann for GluR6(R), S.
Nakanishi for NMDAR1, Peter Seeburg for NMDAR2-C, and Micro-
Tech Optical, Inc. for the loan of Olympus epifluorescence optics. This
work was supported by the Research to Prevent Blindness Foundation
and NEI R01 EY08362 (-I. E. H.) and NIH R01 NS30996 (J. R. H.).
Fellowship support was provided by The Connecticut American Heart
Association (J. M.) and Donaghue Medical Research Foundation
(G. W. J. M.). J. M. is now at Brown University.
Received November 11, 1994; revised November 23, 1994.
GFP as a Reporter in Mammalian Cells
215

References

Chalfie, M., Tu, Y., Euskirchen, G., Ward, W. W., and Prasher, D. C.
(1994). Green fluorescent protein as a marker for gene expression.
Science 263, 802-805.
Chen, C., and Okayama, H. (1987). High efficiency transformation of
mammalian cells by plasmid DNA. Mol. Cell. Biol. 7, 2745-2752.
Colquhoun, D., and Sigworth, F. J. (1983). Fitting and statistical analy-
sis of single-channel records. In Single-Channel Recording, B. Sak-
mann and E. Neher, eds. (New York: Plenum Press), pp. 191-263.
Egebjerg, J., Bettler, B., Hermans-Borgmeyer, I., and Heinemann, S.
(1991). Cloning of a cDNA for a glutamate receptor subunit activated
by kainate but not AMPA. Nature 351,745-748.
Graham, F. L., Smiley, J., Russell, W. C., and Nairn, R. (1977). Charac-
teristics of a human cell line transformed by DNA from human adenovi-
rus type 5. J. Gen. Virol. 36, 59-74.
Hamill, O. P., Marty, A., Neher, E., Sakmann, B., and Sigworth, F. J.
(1981). Improved patch-clamp techniques for high-resolution current
recording from cells and cell-free membrane patches. PflLigers Arch.
391, 85-100.
Kutsuwada, T., Kashiwabuchi, N., Mori, H., Sakimura, K., Kushiya,
E., Araki, K., Meguro, H., Masaki, H., Kumanishi, T., Arakawa, M.,
and Mishina, M. (1992). Molecular diversity of the NMDA receptor
channel. Nature 358, 36-41.
Monyer, H., Sprengel, R., Schoepfer, R., Herb, A., Higuchi, M., Lomeli,
H., Burnashev, N., Sakmann, B., and Seeburg, P. H. (1992). Hetero-
meric NMDA receptors: molecular and functional distinction of sub-
types. Science 256, 1217-1221.
Moriyoshi, K., Masu, M., Ishii, T., Shigemoto, R., Mizuno, N., and
Nakanishi, S. (1991). Molecular cloning and characterization of the
rat NMDA receptor. Nature 354, 31-37.
Moss, G. W. J., Marshall, J., Morabito, M., Moczydlowski, E. G., and
Howe, J. R. (1995). Cloning of a maxi-Ca-activated K channel from
bovine aortic muscle: sensitivity to a Kunitz inhibitor. Biophys. J., in
press.
Prasher, D. C., Eckenrode, V. K., Ward, W. W., Prendergast, F. G.,
and Cormier, M. J. (1992). Primary structure of the Aequorea victoria
green-fluorescent protein. Gene 111,229-233.
Sigworth, F. J., and Sine, S. M. (1987). Data transformations for im-
proved display and fitting of single channel dwell time histograms.
Biophys. J. 52, 1047-1054.
Stern, P., B~h~, P., $choepfer, R., and Colquhoun, D. (1992). Single-
channel conductances of NMDA receptors expressed from cloned
cDNAs: comparison with native receptors. Prec. R. Soc. Lond. (B)
250, 217-277.
Stern, P., B~he, P., Schoepfer, R., and Colquhoun, D. (1993). Single-
channel kinetics of recombinant NM DA receptors. J. Physiol. 473, 48P.
Stern, P., Cik, M., Colquhoun, D., and Stephenson, F. (1994). Single
channel properties of cloned NMDA receptors in a human cell line:
comparison with results from Xenopus oocytes. J. Physiol. 476, 391-
397.
Wang, S., and Hazelrigg, T. (1994). Implications for bcd mRNA local-
ization from spatial distribution of exu protein in Drosophila oogenesis.
Nature 369, 400-403.