| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Department of Ophthalmology and Visual Sciences, John Moran Eye Center, University of Utah, Health Sciences Center, Salt Lake City, Utah 84132; 2Department of General Zoology and Neurobiology, University of Pécs, Faculty of Natural Sciences, Pécs, H-7601 Hungary; 3Center for Vision Research, State University of New York, Upstate Medical University, Syracuse, New York 13210; and 4Neurological Sciences Institute, Oregon Health and Science University, Beaverton, Oregon 97006
Submitted 30 January 2003; accepted in final form 2 March 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
-conotoxin GVIA did not, suggesting a different function
for the N-type channels. The L-type channels in WFACs are functionally coupled
to a set of calcium-dependent potassium (K(Ca)) channels
to mediate OMPs. The initiation of OMPs depended on penitrem-A-sensitive (BK)
K(Ca) channels, whereas their duration is under
apamin-sensitive (SK) K(Ca) channel control. The
Ca2+ current is essential to evoke the OMPs and
triggering the K(Ca) currents, which here act as resonant
currents, enhances the resonance as an amplifying current, influences the
filtering characteristics of the cell membrane, and attenuates the OMPs via
CDI of the L-type Ca2+ channel. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
) and 29 types in the mouse retina
(Masland 2001). Interposed
between the bipolar cells (BCs) that carry visual information from the outer
retina and the ganglion cells (GCs) at the output stage of the retina, ACs
modify the signals that originate in the outer retina by mediating lateral
interactions (Sakai and Naka
1988).
It is the combination of synaptically gated conductances and membrane
properties, including voltage-activated ion channels, which dictate the AC's
response to input and its resulting output. In addition, modulatory influences
further shape cellular responses. ACs have a particular set of intrinsic
membrane properties which are uniquely specialized for their signal-processing
task. Because many types of ACs possess dendritic arbors spanning several
hundreds of micrometers, the mechanism supporting signal transmission via long
dendrites is of particular importance. It has been shown that some ACs
generate TTX-sensitive, sodium-mediated action potentials
(Bloomfield 1992;
Cook and Werblin 1994;
Feigenspan et al. 1998). The
action potentials in turn may, or may not, contribute to transmitter release
(Bieda and Copenhagen 1999;
Watanabe et al. 2000). Other
populations of ACs are capable of producing membrane potential oscillations in
response to light stimuli and/or in the dark
(Djamgoz et al. 1990;
Sakai and Naka 1988;
Teranishi et al. 1987). The
oscillations were thought to arise as a result of network activity. The role
of these oscillations is not understood in terms of retinal information
processing but could be involved, for example, in information transfer across
large dendritic fields and/or in transmitter release. In other systems,
oscillations spread information over long distances signaling information of
"importance" to other neurons
(Llinas 1988). In coupled
systems, oscillations synchronize the network, and if the cells are
"tuned," broadcast the relevant signal most efficiently
(Gray et al. 1989).
Recently, single GABAergic wide-field ACs (WFAC) isolated from the white
bass retina were shown to generate membrane potential oscillations in response
to extrinsic depolarization (Solessio et
al. 2002). These oscillations also occurred in response to
glutamate analogues; glutamate is the endogenous neurotransmitter impinging on
WFACs. Thus the ability of these WFACs to generate oscillations is an
intrinsic feature of the cell and not due to network properties.
The goal of the present study was to determine the underlying ionic
mechanisms responsible for generating the oscillations. Because the
oscillations may be involved in signal processing
(Gray et al. 1989;
Llinas 1988;
Sakai and Naka 1988), it is
important to understand how they come about to understand the functional
properties of the cell. Much is known about this in other neuronal systems
(Amir et al. 2002;
Hudspeth 1986;
Hutcheon and Yarom 2000), but
the ionic basis for this type of intrinsic oscillatory behavior in amacrine
cells is unknown. We found that the oscillations in WFACs arose as a result of
the complex interplay between voltage-dependent calcium currents and voltage-
and calcium-dependent potassium currents. This is similar to what has been
observed in hair cells in the sacculus
(Hudspeth and Lewis 1988b) and
in CNS neurons (Hutcheon and Yarom
2000) but with some interesting differences.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Isolated WFACs maintained in culture were used for all experiments. The
cell-isolation procedure used was similar to that previously described in
detail (Pfeiffer-Linn and Lasater
1998). All procedures were performed in accordance with the United
States Animal Welfare Act and the National Institutes of Health Guide for the
Care and Use of Laboratory Animals. Briefly, white bass (Morone
chry-sops) were dark-adapted for 2 h and then rapidly killed. The eyes
were enucleated and hemisected, then the retinas were removed from the eyecups
under dim red illumination. This was followed by 20-min incubation in papain
(2.5 U/ml, Worthington, Freehold, NJ) at room temperature followed by washing
several times in fresh modified L-15 (Sigma, St. Louis, MO). The retinas were
then triturated and the cells plated onto cell culture dishes. Cells were
maintained in an incubator (10°C) for 1–5 days in modified L-15 cell
culture medium until used.
Experimental solutions
Before each experiment, the L-15 culture medium in the culture dish was replaced with bass Ringer consisting of (in mM) 130 NaCl, 2.9 KCl, 2.13 CaCl2, 1.23 MgCl2, 0.44 K2PHO4, 10 MOPS, and 8.4 glucose; the pH adjusted to 7.4 with NaOH. The standard patch pipette solution consisted of (in mM) 130potassium gluconate, 4 NaCl, 4 KCl, 0.2 EGTA, 2 MgCl2, and 10 HEPES, and 7.8 µM CaCl2. In experiments studying Ca influx through voltage-gated channels, the normal Ringer solution was modified by increasing CaCl2 concentration to 10 mM and adding 1 µM TTX to block sodium, 5 mM 4-aminopyridine (4-AP) and 10 mM TEA to block potassium currents. Osmolarity was kept constant by removal of equimolar NaCl. Also, in such experiments we used a cesium-based pipette solution to maximally reduce the potassium currents. This solution consisted of (in mM) 106.72 Cs-gluconate, 4 NaCl, 1 CaCl2, 1 MgCl2, 8.4 HEPES, and 0.2 EGTA, supplemented with 4 ATP and 20 phosphocreatine and 50 U/ml creatine phosphokinase. The pH was adjusted with gluconic acid to 7.4. All chemicals were purchased from Sigma. The liquid-junction potentials introduced by the cesium gluconate solution were compensated for off-line.
Solutions containing -Aga-TX (50 nM), -conotoxin GVIA (1
µM), -conotoxin MVIIC (1 µM; all from Alamone Labs, Jerusalem,
Israel), nifedipine (100 µM, Sigma-RBI, Natick, MA), CoCl2 (4
mM), diltiazem (100 µM), and ±Bay-K (1 µM; all from Sigma) were
focally delivered to cells via a 12-reservoir pressure ejection system
controlled by a personal computer (DAD-12, Adams). Because of its light
sensitivity, the container holding nifedipine was shielded from light with
aluminum foil.
Recording procedures and data analysis
Whole cell recordings (Hamill et al.
1981) were made with micropipettes pulled on a two-stage puller
(model PP-83, Narishige Instruments, Tokyo) from borosilicate tubing (Drummond
Scientific, Broomall, PA) and were used unpolished. The electrode tip
resistance was typically 7–9 M
when measured in the bath
solution. Series resistance and capacitance were compensated for
electronically. Voltage and current signals were recorded and low-pass
filtered at 2 kHz with an Axopatch 200A amplifier (Axon Instruments, Foster
City, CA) then sampled at 10 kHz. Data acquisition was controlled by a
personal computer interfaced to a Digidata 1200 or Digidata 1322A (Axon
Instruments) data-acquisition system driven by the pClamp suite of programs
(Axon Instruments). After establishing a whole cell configuration, the cells
were held at a potential (HP) of –70 mV unless indicated otherwise.
Under voltage-clamp conditions, the stimuli consisted of 90-ms-long
incremental and decremental voltage pulses between –120 and +70 mV.
Under current-clamp conditions, the stimuli consisted of a series of
incremental depolarizing pulses ranging from 0 to 1.5 nA. The pulses were 150
ms in duration and applied from the holding potential of the cell unless
indicated otherwise. For clarity, all current-clamp figures except
Fig. 1 show only responses to
0-, 0.5-, 1-, and 1.5-nA depolarizing current pulses.
|
Analysis and plotting of data were performed with Clampfit (Axon Instruments), SigmaPlot (SPSS, Chicago, IL) and Excel. Plots show average responses ± SDs and number of observations is indicated in the figure and/or in the legends. Statistical analysis of the data were performed by using Student's t-test (paired comparison), and the P value of <0.05 was considered to be significant.
Quantification of the oscillations
The oscillations were quantified by using the distance of the
"line" tracing the membrane voltage change during the depolarizing
current steps. The distance was calculated as follows
 |
Intensity (I) represents the duration and the amplitude of the oscillations, T: the observation period, and x: membrane potential at time t. Using this approach, a cell that produced fewer or attenuated oscillations will have a shorter "line" length than one that generated sustained oscillations.
Immunofluorescence techniques
All antibodies against the calcium channel subunits were rabbit polyclonal
obtained commercially from Alamone Labs, except for the 
1F antibody,
which was raised in sheep (Morgans
2001). The monoclonal anti-syntaxin was purchased from Sigma.
Cells were isolated and maintained on con-A-coated coverslips for 2–4
days before staining. Isolated cells were fixed in 4% paraformaldehyde
solution for 15 min, incubated in 10% normal donkey serum in 0.1 M PB 0.5%
Triton X-100 for 1 h at 4°C, and then transferred to either a solution of
a single antibody or a mixture of two primary antibodies. The antibodies were
used at a 1:500 to 1:1,000 dilution. After overnight incubation in the
antibody mix, isolated cells were washed in 0.1 M PB and transferred to a
cocktail of the secondary antibodies [donkey anti-rabbit IgG coupled to
fluorescein (FITC), donkey anti-mouse IgG rhodamine (TRITC) or IgG
fluorescein] at 1:100 dilution in 0.1 M PB 0.5% Triton X-100 for 1 h.
Immunostained sections were examined by confocal microscopy. A Zeiss LSM510
was used to image the cells and to optically section the cells at 1-µm
intervals. Slices were combined to make a composite image or single slices
were analyzed. Control sections were obtained by omitting the primary
antibody.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Cclls were identified as WFACs in culture by their unique, distinctive
morphology (Solessio et al.
2002). WFACs reported on in this study typically had a small,
often triangular shaped cell body (
15 µm or less), with characteristic
long, thick processes extending for ≤250 µm and were typically branched.
Although processes were sometimes truncated during the isolation procedure, it
was not unusual to find cells in culture whose dendritic spread extended
between 300 and 400 µm (see Fig.
5).
|
Basic characteristics of the oscillations
Small, depolarizing current steps evoked passive membrane depolarizations
(Fig. 1, +0.1 nA). With further
increase of the depolarizing current steps, activation of voltage-dependent
mechanisms was observed. Characteristically, oscillation of the membrane
potential was always present if the membrane depolarization reached –43
± 4 mV (n = 10, see Fig.
1, +0.2 nA and up). This is in close agreement with the activation
potentials of the voltage-gated sodium (INa) and
Ca2+ currents (ICa) in WFACs
(Solessio et al. 2002). The
magnitude, duration and frequency of the oscillations increased with membrane
depolarization (see Fig. 1).
Blocking the voltage-dependent Na+ channels with TTX did not
eliminate the response, whereas the Ca2+ channel blocker
Cd2+ or a mixture of K+ channel blockers (TEA
and 4-AP) effectively blocked the oscillations
(Solessio et al. 2002). These
results suggested that oscillatory membrane potentials (OMPs) result from the
interplay of Ca2+ and K+ currents. This
likely includes a Ca2+-dependent potassium current. In
support of this notion, blockage of the voltage-gated
Ca2+ currents with Cd2+
dramatically reduced the outward K+ currents
(Solessio et al. 2002),
strongly indicating the presence of K(Ca) in WFACs. We
went on to study in detail the ionic mechanisms underlying the OMPs with a
focus on the calcium and potassium currents.
Contribution of Ca2+ to the oscillations
Like Cd2+, 4 mM Co2+ blocked
the oscillations (n = 15) (Fig.
2A), except for the first peak. The first peak was
attenuated or absent in the presence of TTX suggesting it was Na+
based (Solessio et al. 2002).
In addition, the application of Co2+ also resulted in a
10- to 15-mV hyperpolarization of the cell. Note also the large increase in
the membrane potential in response to the depolarizing step in the presence of
Co2+ in comparison to the corresponding control traces.
These are due to a block of calcium leak into the cell and an increase in
input resistance, respectively.
|
Next we determined the effect of elevated extracellular Ca2+ concentration on the oscillations. Under control conditions, cells were bathed in normal Ringer containing 2.1 mM Ca2+. High-Ca2+ (10 mM) Ringer solution was flushed over the cells during the recording and then washed away with normal Ringer. Elevation of the extracellular [Ca2+] markedly increased both the duration and the amplitude of the oscillations (n = 17, Fig. 2B). However, the frequency remained unaltered. Interestingly, the threshold potential for the generation of the OMPs did not change significantly (not illustrated). Recovery was achieved within a few minutes of washout. Note the pronounced dampening of the oscillations in 10 mM Ca2+-containing Ringer solution (Fig. 2B). Elevation of the Ca2+-buffer EGTA in the pipette from 0.2 to 10 mM eliminated this dampening (Fig. 2C), suggesting a Ca2+-dependent downregulation of the oscillatory response components.
Based on these results, we concluded that Ca2+ influx through voltage-gated channels was essential to the generation of OMPs in WFACs. On the one hand, elevated extracellular Ca2+ facilitated oscillatory responses in WFACs. On the other hand, Ca2+ entering the cells also triggered a Ca2+-dependent process that dampened the oscillations.
Inactivation of Ca2+ currents
We investigated whether or not the voltage-gated Ca2+
current was subject to Ca2+-dependent
inactivation. With a Cs+-based pipette solution and TTX, TEA, 4-AP,
and 10 mM Ca2+ containing extracellular solution,
50-ms-long depolarizing voltage steps evoked a very large, sustained inward
current (not illustrated). The current amplitude ranged from 1 to 3 nA. For
cells with an input capacitance of 56.0 ± 7.6 pF (n = 13), the
peak current density was 53.7 ± 12.4 pA/pF at –10 mV. To study
the inactivation properties of the Ca2+ currents,
membrane potentials were stepped from –70 to 0 mV for 400 ms. Using long
voltage steps, a marked relaxation of ICa became apparent
over time (Fig. 3A,
control). We were able to dissect two forms of inactivation: a time-dependent
component, which was always present, and a
Ca2+-dependent component, which required
Ca2+ as the charge carrier. These two mechanisms were
distinguished in two ways; first by adding 10 mM
bis-(o-aminophenoxy)-N,N,N',N'-tetraacetic
acid (BAPTA; Fig. 3A,
BAPTA) to the intracellular solution and second by using
Ba2+ as a charge carrier
(Fig. 3A, barium)
(Eckert and Tillotson 1981;
Gleason et al. 1994). For such
an experiment, the tip of the pipettes were filled with 0.2 mM EGTA containing
Cs+-based intracellular solution, then backfilled with this
solution supplemented with 10 mM BAPTA to avoid immediate BAPTA diffusion into
the cells. Control recordings (Fig.
3A, control) were made immediately after breaking into
the cells and then repeated after 5–10 min to allow BAPTA to diffuse
into the cell. Thereafter the cell was washed with Ringer in which
Ca2+ was substituted with equimolar
Ba2+.
|
When only 0.2 mM EGTA was present, the Ca2+ current
declined by 98% at the end of the 400-ms voltage step
(Fig. 3A, control),
but after exposure to BAPTA the decline was only 31% for the same cell
(Fig. 3A). This finding is
similar to that observed in the case of cloned human L-type channels having
only the 1D pore-forming subunit
(Bell et al. 2001). Replacement
of Ca2+ by equimolar Ba2+ in the
extracellular solution resulted in a significant increase in the peak inward
current (Fig. 3A).
That is a well-documented characteristic of calcium channels when
Ba2+ enters into the cell through high-voltage-activated
(HVA)-type channels (Hagiwara and Ohmori
1982; Soldatov et al.
1997). Using Ba2+ left only the
time-dependent portion of inactivation, which was responsible for a 15%
decline in current during the 400-ms voltage step in the cell shown in
Fig. 3A.
Thus in WFACs, for a population of cells the time-dependent decline of the sustained Ca2+ current during a 400-ms voltage step recorded in the presence of Ba2+ averaged around 15% (14.5 ± 5.0%, n = 13, Fig. 3B). On top of this, Ca2+ entering through voltage-dependent channels triggered a more pronounced decline of the Ca2+ current, 62.4 ± 13.1% (n = 13, Fig. 3B, control), which was markedly reduced by 10 mM BAPTA (22.2 ± 20.0%, n = 8, Fig. 3B).
In our current-clamp experiments, 150-ms-long depolarizing pulses were employed. The corresponding decline of ICa at 150 ms for the above population of cells was 5.0 ± 3.2 and 7.0 ± 3.3% for BAPTA and Ba, respectively, whereas it was 19.7 ± 5.9% for control. Hence, in WFACs, significant Ca2+-dependent inactivation of the HVA Ca2+ current occurs and is likely to play an important role in the attenuation of the OMPs.
Pharmacological identification of the voltage-gated Ca2+ currents
Known Ca2+ channel antagonists and agonists were used to characterize the type of calcium channel(s) involved in the generation of the OMPs. After control measurements, drugs were applied for 1 min before their effect was tested on ICa. More than one drug was tested on the same cell only if the currents recovered to within ±10% of the original current amplitude. The data are summarized in Fig. 4, A and B.
|
In WFACs, the relatively minor inactivation of IBa
taken together with an increase in current amplitude when
Ba2+ replaced Ca2+ in the
extracellular solution suggested that a significant portion of
ICa flows through L-type channels. L-type
Ca2+ currents have also been defined pharmacologically
by their sensitivity to low concentrations of 1,4-dihydropyridine (DHP)
antagonists (e.g., nifedipine, nitrendipine) and agonists such as (–)
Bay K 8644 (Nowycky et al.
1985b; Sanguinetti and Kass
1984). In WFACs, Bay K 8644 (1 µM), which prolongs the opening
time of L-type channels (Nowycky et al.
1985a), enhanced the peak ICa by 130 ±
25% (n = 8, Fig.
4B) and shifted the activation of
Ca2+ current by 10 mV toward more negative potentials
(Fig. 4C). This is
consistent with the Bay K effect seen in other preparations
(Avila and Dirksen 2000). The
effectiveness of different L-type specific antagonists varied. A saturating
dose of the DHP nifedipine (100 µM)
(Varming et al. 1997) reduced
the current amplitude only by 21.8 ± 7.3% (n = 10). But
another L-type channel blocker, the benzothiazepine (+) diltiazem (100 µM)
(Kraus et al. 1998) reduced
the current by 51.3 ± 9.1% (n = 21). This pharmacological
profile most closely resembles that of 1D and 1F
subunit-containing Ca2+ channels present in rod and cone
photoreceptors (Kourennyi and Barnes
2000; Wilkinson and Barnes
1996).
The N-type Ca-channel blocker -conotoxin GVIA (1 µM) attenuated
ICa by 29.6 ± 5.5% (n = 14). The less
specific, P/Q/N-type Ca2+-channel blocker
-conotoxin MVIIC (1 µM)
(McDonough et al. 1996)
blocked 22.7 ± 9.9% (n = 7) of the total
Ca2+ current. Because the potent, reversible P/Q-type
blocker -Aga-TK (50 nM) (Teramoto
et al. 1993) exerted no effect on ICa
amplitude in six cells tested (8.7 ± 4.5%, P > 0.05), we
concluded that both the -conotoxins (GVIA and MVIIC) acted on N-type
channels. In support of this notion, GVIA treatment evoked an average block of
the ICa that was not significantly different from the
MVIIC-evoked inhibition (P > 0.05).
When the N-type current component was blocked with 1 µM GVIA, the
remaining portion of ICa was enhanced by 339 ± 2%
(n = 8) by Bay K (Fig.
4B). A mixture of MVIIC (1 µM) and diltiazem (100
µM) reduced the total Ca2+ current by 83.8 ±
11.3% (n = 7). By adding GVIA (1 µM) to the MVIIC/diltiazem
cocktail, we could not increase the inhibition of the total
ICa (79 ± 10.5%, n = 19). This further
supports our notion that -conotoxins GVIA and MVIIC exert their effect
on the same, presumably N-type, Ca2+ channel in WFACs.
The nonspecific Ca2+ channel blocker
Co2+ (4 mM) eliminated ICa almost
completely (97.1 ± 1.9%, n = 11,
Fig. 4A). The
diltiazem and the -conotoxins were used at saturating doses based on
past studies (Hillyard et al.
1992; Kraus et al.
1998; Olivera et al.
1985). Therefore we wondered whether any portion of the
ICa that remained after using the various cocktails of
antagonists reflected T-type channel activity. Nickel
(Ni2+) at 100 µM has been shown in some systems to
selectively block T-type Ca2+ channels
(Carbone and Swandulla 1989).
In WFACs, it seems to be a nonspecific blocker. Ni2+
eliminated 50% of the total ICa in the absence of
other blockers (n = 4, not illustrated). When tested after using the
diltiazem/conotoxin mixture, it eliminated 46.4 ± 9.6% (n = 6)
of the residual ICa. This is not what one would expect if
it were acting on a single type of channel. Moreover, we were unable to
dissect out a transient component by using a stimulus protocol that would
maximally activate or inactivate that current relative to the others
(n = 11). Thus we concluded that T-type channel activity in WFACs is
unlikely.
To summarize, pharmacological investigation of the ICa
in WFACs revealed at least two separate components. The larger portion—
50–60%—was mediated by L-type channels and a smaller, but
still substantial, portion (30–40%) of the voltage gated
Ca2+ enters the WFACs through N-type channels. Based on
the present pharmacological data, it is possible that R-type currents were
responsible for the residual ICa contributing
≤10–15% of the total Ca2+ current in WFACs.
Ca2+ channel immunoreactivity in WFACs
Next, we immunohistochemically localized the voltage-gated
Ca2+ channel subunits present in the WFACs. We used
antibodies directed against the 1A (P/Q-type,
(Restituito et al. 2000),
1B (N-type, (Schiff et al.
2000), and 1E (R-type,
(Saegusa et al. 2000)
pore-forming subunits of the intermediate conductance channels. Antibodies
against the 1C (cardiovascular L-type)
(Kreuzberg et al. 2000),
1D (neuroendocrine L-type) (Jiang
et al. 1999; Yang et al.
1999), and 1F (retinal L-type)
(Morgans 2001) subunits of the
large-conductance Ca2+ channels were also tested. Double
labeling with anti-syntaxin was used to confirm the identity of the cells as
WFACs (Solessio et al.
2002).
In concert with our pharmacological data, WFACs stained positively with
anti-1B and anti-1F (Fig.
5), indicating N-type and retinal L-type
Ca2+ channels. We found that N-type staining localized
in close proximity to the cell body (Fig.
5, C and D), whereas the anti-1F labeled
in a diffuse (Fig. 5, A and
D), punctate pattern
(Fig. 5B) near the
cell body and over the processes. No staining was observed with the
anti-1E nor with the rest of the antibodies tested. Thus the
immunohistochemical data do not support the presence of the P/Q or R-type
channels. Because the specificity of the antibodies used has not been
established previously in fish, these data are still open to
consideration.
Contribution of different types of Ca2+ channels to the OMPs
The purpose of these experiments was to determine if any of the effective
Ca2+ current blockers could inhibit a particular aspect
of the oscillations or selectively change the amplitude and/or frequency.
Neither -conotoxin GVIA (1 µM, n = 15,
Fig. 6A) nor
-conotoxin MVIIC influenced the OMPs (n = 5, not illustrated),
suggesting that N-type channels do not contribute to the mechanism underlying
the oscillatory behavior under our experimental conditions.
|
We found that the L-type Ca2+ channel antagonist diltiazem (100 µM) was almost as potent an inhibitor of the OMPs as Co2+ (Fig. 6B, compare with Fig. 2A). As shown in the preceding text, Bay-K was found to enhance the ICa in WFACs. Nevertheless, it failed to enhance the oscillations and, in some cases, slightly reduced both the duration and the frequency of the OMPs (n = 10, Fig. 5C, see DISCUSSION for details).
Next we asked the question, do pharmacologically distinct Ca2+ channels contributed to the OMPs to different degrees under different conditions? To determine this, we bathed the cells in normal 2.1 mM Ca2+-containing Ringer, then enhanced the oscillations by puffing high (10 mM)-Ca2+-containing Ringer over the WFACs (Fig. 7, control trace). Co-application of 100 µM diltiazem with the high-Ca2+ Ringer eliminated the enhancement. Moreover, puffing 4 mM Co2+ during the long depolarizing step in the presence of diltiazem did not alter the membrane potential (Fig. 7). Thus we concluded that N-type currents did not contribute either to the OMPs enhanced by high extracellular Ca2+ or to the late (flat) phase of the membrane voltage response (n = 5).
|
Large, unattenuated OMPs were observed whenever the cells were superfused
with high extracellular Ca2+ and the solution in the
recording pipette was supplemented with 10 mM BAPTA
(Fig. 8B). As we
demonstrated earlier, this occurred because BAPTA markedly reduced the
inactivation of the Ca2+ current in WFACs
(Fig. 3, A and
B). But the question arises as to which
Ca2+ current component's inactivation is inhibited by
BAPTA? In other words, does the L-type or a non L-type (N-type) voltage-gated
Ca2+ channel provide the Ca2+ that
underlies the nonattenuated OMPs? To determine this, we applied 100 µM
diltiazem with the high Ca2+ Ringer again to WFACs
preloaded with 10 mM BAPTA (Fig.
8C). Diltiazem eliminated the enhanced oscillations
(Fig. 8, C and
E, n = 6). This indicates that the oscillations
in WFACs, whether dampened or not, are evoked by Ca2+
entering into the cells exclusively via 1F subunit containing L-type
channels and that the L-type channels are subject to substantial
Ca2+-dependent inactivation.
|
Role of IK currents
Next, studies were undertaken to determine the role of specific potassium
channel types in the generation of the OMPs. Pharmacological investigation of
the voltage-gated K+ currents in WFACs revealed at least three
pharmacologically distinct components. Extracellular application of the
nonspecific K+ channel blocker TEA (10 mM) reduced the outward
current by 56.2 ± 19.9% (Fig.
9, n = 5), whereas 4-AP (5 mM) did so by 22.6 ±
20% (Fig. 9, n = 5).
Known blockers of calcium-activated potassium channels were then tested. It is
not clear if WFACs express small-conductance (SK) K(Ca)
channels, but apamin (5 µM), a selective (SK) K(Ca)
channel antagonist, produced a 14.8 ± 9.9% reduction of the outward
current. ACs are known to posses large-conductance (maxi or BK)
K(Ca) channels (Mitra
and Slaughter 2002), and the tremorgenic fungal toxin, penitrem A,
selectively blocks (BK) type K(Ca) channels
(Knaus et al. 1994). When
applied to the WFACs, penitrem A (1 µM) reduced K+ currents by
17 ± 8.3% (Fig. 9,
n = 7). Other than TEA, the effect of the preceding potassium channel
blockers on the total potassium current was not impressive.
|
Nonetheless, these agents were tested on the OMPs in an effort to determine whether or not particular potassium channels played a role in OMP generation. When applied prior to evoking OMPs, 4-AP (5 mM) had no effect (Fig. 10B), but a similar application of 10 mM TEA completely eliminated the oscillations in the same cell (Fig. 10C). Extracellular application of 1–5 µM of apamin, to block the (SK) K(Ca) channels, under control conditions only slightly reduced the duration of OMPs in 12 cells tested (Fig. 11A). However, if OMPs were enhanced by washing the cells with 10 mM Ca2+-containing Ringer, the same dose of apamin eliminated the enhancement independent of intracellular Ca2+ buffering (n = 7, see Fig. 11, B and C, respectively). Taking these findings together with the preceding results, it is likely that apamin-sensitive (SK) K(Ca) channels are functionally coupled to L-type Ca2+ channels, and this interaction is responsible for increasing the duration or generating more sustained OMPs in WFACs.
|
|
Application of either 10 mM TEA (Fig. 10C) or 4 mM Co2+ (Fig. 2A) eliminated the OMPs completely, indicating that apamin-insensitive members of the K(Ca) current family play a role in generating the apamin-resistant initial phase of the oscillations. When tested on WFACs under control conditions, penitrem A (500 nM to 1 µM) completely and reversibly eliminated the OMPs (Fig. 11D, n = 10). In addition, the mixture of penitrem A (500 nM) and apamin (5 µM) totally eliminated the OMPs in cells perfused internally with 10 mM BAPTA and bathed in 10 mM calcium containing Ringer (not illustrated).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
). The present set of experiments were undertaken to provide a
detailed insight into the ionic mechanisms underlying the generation and
modulation of these oscillations in an effort to better understand the
functional properties of ACs. Ionic currents mediating the oscillations in WFACs
In this study, we found that WFACs membrane potential oscillations were
mediated by a feed-back loop between voltage-gated Ca2+
and K(Ca) currents (see following text). This is similar
to what has been demonstrated in other systems, for example in saccular hair
cells (Hudspeth 1986;
Roberts et al. 1990), but
different currents seem to be involved. We found that blocking the
voltage-gated Ca2+ channels in WFACs eliminates the OMPs
completely (Fig. 2A).
In an earlier study, we showed that blocking voltage-gated
Ca2+ channels in WFACs also reduced the total
K+ current by 50%
(Solessio et al. 2002),
suggesting current flow through calcium-activated K+ channels. Here
we found that both (SK) and (BK) K(Ca) channels
participate in the generation of the oscillations.
Significantly, we found there is interplay between the two channels to
generate the OMPs: BK channels seem to be necessary to initiate the
oscillations, whereas SK channels control the duration. This pattern of
contribution fits with their biophysical characteristics. The (BK)
K(Ca) channels are gated by Ca2+
with a KD that is steeply voltage-sensitive being in the
nanomolar range at +20 to +40 mV and several micromolar near the resting
potential (Latorre et al.
1989; Weiger et al.
2002). Also, they open rapidly after the activation of
Ca2+ channels but quickly inactivate as membrane
potential and/or intracellular Ca2+
([Ca2+]i) availability drops
(Sah and Davies 2000;
Vergara et al. 1998;
Weiger et al. 2002). In
contrast, the (SK) K(Ca) channels are influenced little by
membrane potential and therefore conduct current dependent only on the
intracellular Ca2+ levels
(Xia et al. 1998).
Most often, K(Ca) channels are coupled to specific
voltage-gated Ca2+ channels (see
Sah and Davies 2000 for
review). We found that OMPs were eliminated by the L-type
Ca2+ channel blocker diltiazem and that was independent
of oscillation duration or dampening (Figs.
6B,
7, and
8C). So in WFACs both
(SK) and (BK) K(Ca) channels must be functionally coupled
to L-type voltage-gated Ca2+ channels to mediate
oscillations (Davies et al.
1996; Marrion and Tavalin
1998; Wisgirda and Dryer
1994). This situation is very similar to that described in
bull-frog saccular hair cells (Hudspeth
and Lewis 1988b). There a calcium current interacts with an
inactivating, transient potassium current and a calcium-activated potassium
current, probably the BK type, to produce dampened oscillations.
Dampening of the oscillations
Including high concentrations of Ca2+ buffers (either
EGTA or BAPTA) in the pipette solution diminished the dampening of the
oscillations (Figs. 2C
and 8B, respectively).
Because the gating of the (SK) K(Ca) channels is
independent of protein phosphatases and kinases including
Ca2+-dependent ones
(Xia et al. 1998),
Ca2+-dependent inactivation of (SK)
K(Ca) chanels in WFACs is not a source for attenuation of
the OMPs. Although the (BK) K(Ca) channels are known to be
subject to multiple modulatory factors, their gating is voltage and
Ca2+ dependent, and Ca2+-dependent
downregulation of the channel has not been reported
(Weiger et al. 2002). Note
though that even in the presence of intracellular BAPTA and 10 mM
extracellular Ca2+, (BK) K(Ca)
currents seems to inactivate based on the oscillatory pattern (it becomes
dampened) when apamin blocked the (SK) K(Ca) channels
(Fig. 11C). In
addition, apamin only slightly reduced the duration of the OMPs under normal
conditions (Fig.
11A). Based on this we concluded that in WFACs, (BK)
K(Ca) channels do inactivate and it is somewhat faster
than the Ca2+-dependent inactivation of the L-type
Ca2+ currents. By the time the (BK)
K(Ca) currents inactivate, the (SK)
K(Ca) currents turn on. However, the inactivation of
ICa is suppressing the OMPs under normal conditions. Taken
together this suggests that the Ca2+-dependent
inactivation of ICa in WFACs is the likely cause of
dampening in the oscillations.
The characteristics and the time scale of the
Ca2+-dependent inactivation in WFACs closely resemble
that seen in cardiac cells (Imredy and Yue
1994), smooth muscle cells
(Giannattasio et al. 1991) and
neurohypophysial nerve endings (Lemos and
Nowycky 1989). That is, it begins without delay after
depolarization and leads to an almost complete inactivation in 500 ms. This is
100-fold faster than the Ca2+-dependent
inactivation of ICa reported in retinal BC synaptic
terminals (Qian et al. 1999)
and 40-fold faster than in rods (Rabl
and Thoreson 2002). This suggests that
Ca2+-dependent inactivation of the L-type current is a
prominent and general feature of different types of retinal neurons,
nevertheless, the kinetics of the process vary, which may reflect different
functional needs. In the case of the WFAC, the frequency of the oscillations
is tied to the time constant of inactivation. It must be rapid for
ICa to participate effectively in the generation of the
oscillations.
It is of interest that in WFACs with high internal BAPTA not only the dampening of the positive oscillatory peaks disappeared, but the dampening of the negative peaks as well. This provides a second line of evidence supporting the notion that the (SK) K(Ca) current only follows the ICa in WFACs because the steady amplitude of the repolarizing peaks must be tied to (SK) K(Ca) currents. If ICa is more sustained, then the hyperpolarization provided by (BK) and more importantly (SK) K(Ca) currents are steady in amplitude as well and does not decline (e.g., Figs. 8, B and D, and 11C). This predicts that any modulatory effect reducing the inactivation of the ICa in WFACs would enhance the duration of the OMPs.
Relationship between L-type Ca2+ and K(Ca) channels in WFACs
In the present experiments, we found that apamin-sensitive (SK)
K(Ca) contributed to the oscillations only if they were
enhanced by high extracellular Ca2+
(Fig. 11, B and
C) or by increasing intracellular buffering with BAPTA
(Fig. 3B).
Interestingly, this suggests that under "normal" conditions (2.1
mM extracellular Ca2+), when WFACs are depolarized, the
[Ca2+]i does not reach a level in the cell
sufficient to fully activate the (SK) K(Ca) current. This
could result from a high Ca2+-buffering capacity due to
intrinsic Ca2+-binding proteins, which are abundant in
the WFACs (Solessio et al.
2002) and/or from the inactivation of the
Ca2+ current (Fig.
3A).
The preceding findings are somewhat puzzling. The calculated free
[Ca2+]i with 10 mM EGTA in the pipette is 6.5
x 10–9 M and about the same when using
BAPTA, which has a similar KD but binds
Ca2+ 100 times faster
(Wang et al. 1997;
Zhang et al. 1995). This
[Ca2+]i is far below the half-activation
concentration (0.31–0.33 µM) for all known members of the (SK)
K(Ca) channel family
(Xia et al. 1998).
Consequently one might expect that at least the
fast-Ca2+ chelator BAPTA could adequately mop up
Ca2+ to prevent K(Ca) current
activation. This in turn would eliminate the oscillations. In fact,
intracellular BAPTA resulted in a diametrically opposite effect; the OMPs were
greatly enhanced. This apparent conflict allows some predictions to be made
about the physical relationship between voltage-gated
Ca2+ and K(Ca) channels in
WFACs.
Several studies have proposed various Ca2+
microdomain hypotheses suggesting that regions of elevated free
Ca2+ are highly localized to the immediate vicinity of
individual Ca2+ channels. But their activation often
times does not lead to a significant increase in the bulk
[Ca2+]i of the cytosol
(Imredy and Yue 1992;
Llinas et al. 1995). In
addition, convincing evidence from other systems supports the notion that
functional coupling between voltage-gated Ca2+ channels
and K(Ca) conductances requires physical co-localization
(Gola and Crest 1993;
Roberts et al. 1990;
Robitaille et al. 1993). This
is in agreement with our finding that intracellular perfusion with BAPTA could
not prevent Ca2+ binding to the
K(Ca) channel's Ca2+
sensor—which is likely to be calmodulin
(Sah and Davies 2000;
Saimi and Kung 2002)—so
the K(Ca) channels opened. Calcium binding by buffers,
either endogenous or introduced, are simply not fast enough to act within a
distance <50 nm of the Ca2+ channels to strongly
influence [Ca2+]i, which can rise to 100
µM in <100 µs (Neher
1998); although in some cases, it has been suggested that BAPTA
exerts some effects as close as 30 nm to the pore
(Budde et al. 2002). In any
case, this suggests that in WFACs, L-type Ca2+ channels
are clustered together with K(Ca) channels, packed next to
each other similar to the arrangement in hair cells
(Roberts et al. 1990) and the
distance between them is <50 nm.
In WFACs both high EGTA and BAPTA eliminated the attenuation of the
oscillations as well as the Ca2+-dependent inactivation
of the L-type Ca2+ currents. In some systems,
Ca2+-dependent inactivation results from calmodulin
activation (Ivanina et al.
2000; Sun et al.
2000). One model of this proposes that calmodulin is tethered to
the Ca2+ channel complex and attaches to the channel
only after Ca2+ binding
(Peterson et al. 1999).
Because high BAPTA and EGTA were found equally potent in eliminating the
attenuation of the OMPs, we propose that the Ca2+ sensor
for the Ca2+-dependent inactivation of L-type
Ca2+ channels in WFACs is localized ≥100 nm away from
the channel. According to Neher
(1998), 100 nm is the
distance from the channel where the "slow" EGTA can exert its
effect and be as potent as BAPTA. This model explains with physical distances
how BAPTA could prevent the (tethered) calmodulin-dependent inactivation of
the L-type channels while it didn't eliminate Ca2+
activation of the calmodulin serving as the putative internal
Ca2+ sensor for the K(Ca) channels
(Saimi and Kung 2002) in
WFACs.
Membrane potential oscillations, cell resonance, and the different roles of ICa in WFACs
As a general rule, membrane potential oscillations can be produced in a
neuron whose membrane exhibits resonant properties
(Hudspeth and Lewis 1988a).
Resonance arises from the interactions between active resonant currents
influenced by the "passive" filtering characteristics of the
membrane (Hutcheon and Yarom
2000). Currents that actively oppose changes in membrane voltage
and that activate slowly relative to the membrane time constant can produce
resonance in a cell. Calcium-dependent K+ currents meet these
criteria (Hudspeth and Lewis
1988a). The Ca2+ dependence of these
channels ensures that they will open with a delay, and because their reversal
potential falls near the foot of their activation curve, they actively oppose
change in membrane potential. To produce oscillations, a resonant cell must
also possess an amplifying current that can interact with the resonant
current. The amplifying current is the opposite of the resonant current; it
activates rapidly and accelerates membrane potential change because its
reversal potential is close to the peak of its voltage-activation relationship
(Hutcheon and Yarom 2000).
Here we have shown that in WFACs ICa fulfills these
criteria.
The Ca2+ current plays a dual role in the generation of WFAC membrane oscillations. First, it provides Ca2+ for the activation of the resonant K(Ca) currents. Second, as described in the preceding text, ICa fits the role of the amplifying current. So blocking the voltage-gated Ca2+ inflow eliminates the OMPs by not only reducing the K(Ca) currents but by eliminating the amplifying conductance as well.
The "passive" filtering characteristics of the membrane are set
by the parallel leak conductance and the capacitance of the membrane that
attenuates responses to inputs at high frequencies. For the WFAC, the membrane
time constant (RC), calculated under conditions where the membrane exhibit
passive responses (e.g., depolarizations below –40 mVs), is on the order
of 25 ms. This value was obtained from measurements of the cell resistance as
inferred from the I-V relationship (R = 500 M)
and using an average capacitive value of 50 pF. This yields a low-pass cutoff
frequency [fL = 1/(2
RC)] of 6 Hz. A similar value
can be obtained by fitting a rising exponential to the voltage response in
current-clamp mode (RC = 20 ms) while the cell is responding passively
(membrane potentials below –40 mVs) before oscillating.
The activation time constant for the potassium channels (pooled) depends on
the Ca2+ concentration. We can estimate an "upper
limit" for this value by fitting an exponential function to the
potassium currents obtained with the Ca2+ and
Na+ channels blocked. This will give a time constant reflecting no
calcium-dependent potassium channels, which is likely to be a little faster
than the fastest time K(Ca) currents can activate
(Sah and Davies 2000). For the
WFAC, this value is 5 ms, which translates into a cutoff for the
high-pass filter of fH = 30 Hz
(Hutcheon and Yarom 2000).
Therefore the resonance characteristics of these cells computed as the product
of the first-order transfer functions for the low-pass (6 Hz) and high-pass
(30 Hz) transfer functions results in a band-pass or notch filter
(Fig. 12A). A problem
with this is immediately evident. The cutoff frequency of the low-pass filter
is lower than that of the high-pass filter. As a result, the system presents a
poorly tuned transfer function that peaks at 10 Hz, nowhere near the 80- to
100-Hz range of the observed oscillations.
|
Alternatively, we can consider that by the time the potassium channels
activate, the Ca2+ currents are already activated
because their activation takes <0.5 ms
(Solessio et al. 2002). So
rather than considering only the passive cell resistance, we must also
consider the Ca2+ conductance in determining the
membrane resistance. Doing so gives a cell conductance of 38 x
10–9 S or equivalently a membrane resistance of 25
M. With a 50-pF membrane capacitance, that makes the RC = 1.3 ms to
give an estimated fL of 120 Hz. The shift of the cutoff to
a higher frequency yields a highly tuned transfer function
(Fig. 12B). The
frequency of resonance under these conditions is 100 Hz, in line with our
observations. At 300–400 Hz, the frequency of the attenuating responses,
the system is working mostly as a low-pass filter and therefore introducing
marked attenuation in the oscillations. So the Ca2+
currents are not only amplifying the oscillations but also contribute to
setting the range of oscillations by its impact on the cell conductance.
Therefore the current can be considered as a band-pass amplifier, much like
what has been seen in hair cells (Hudspeth
and Lewis 1988a).
The changing I-K(Ca) and the Ca2+-dependent modulation of the Ca2+ currents will change the position of the cutoff frequencies and as a result change the oscillatory properties of the cells. For example, the Bay-K doubles the Ca2+ conductance. This should bring a shift in the cutoff frequency of the low-pass component and an increment in the oscillatory response; however, the opposite was observed. This was most likely due to a concomitant decrease in rise time of the I-K(Ca) that raised the value of the cutoff of the high-pass filter, resulting in attenuation of the oscillations. The reduced duration is likely to be the consequence of the increased Ca2+ dependent inactivation of the ICa.
WFACs and oscillations: functional implications
WFACs isolated from the teleost retina possess at least two
phamacologically distinct Ca2+ channels: N-type and
L-type, with differential distributions. N-type channels were found on, and in
close proximity to, the cell body, whereas the L-type channels seemed to be
evenly distributed over the cell surface particularly on the dendrites.
Because the N-type current did not seem to contribute to the oscillations, its
physiological role was not investigated here. Nevertheless, their localization
allows us to make some predictions about their function. In lower vertebrates,
transient WFACs are functionally polarized, so that they are sensitive to
excitatory (glutamatergic) inputs near and at the soma, while they respond to
inhibitory inputs more distally along their dendrites
(Maguire 1999). Thus it is
possible that N-type channels are functionally coupled to glutamate receptors
to boost depolarizations, whereas the L-type currents are functionally coupled
to (BK) and apamin-sensitive (SK) K(Ca) channels to
mediate oscillations as we demonstrated in the preceding text. While (BK)
K(Ca) channels seem to trigger the OMPs, it is the
apamin-sensitive (SK) K(Ca) channel activation that keeps
it going. The characteristic features of the OMPs, frequency, duration, and
amplitude, are greatly modulated by Ca2+. The frequency
of the OMPs is set by the RC properties of the membrane, where R is
influenced by the activation potential and the magnitude of
ICa. The duration is regulated by the
Ca2+-dependent inactivation of the L-type
Ca2+ channels. The amplitude is connected to the L-type
Ca2+ channels as well because first they provide the
Ca2+ to activate the resonant K(Ca)
currents and second, ICa serves as an amplifier of the
resonance leading to self-sustained oscillation
, control; 
, Bay K 8644.
, 10 mM
Ca2+ + diltiazem.
