Báo cáo khoa học: Voltage- and Ca2+-activated potassium channels in Ca2+ store control Ca2+ release docx

Otherwise, a lumen-negative potential is rapidly built up by Ca2+ efflux to such a degree that the negative shift can reach the equilibrium potential for Ca2+ ions, and the Ca2+ efflux is

store control Ca2+ release

Masayuki Yamashita, Miho Sugioka and Yoichi Ogawa

Department of Physiology I, Nara Medical University, Kashihara, Japan

The release of Ca2+from intracellular Ca2+stores is a

pivotal event in Ca2+signaling, which regulates a

vari-ety of cellular activities [1] There are at least two types

of Ca2+ releasing channels that allow Ca2+ efflux

from Ca2+ stores: inositol 1,4,5-trisphosphate (InsP3)

receptor channels and ryanodine receptor channels

These channels are involved in the highly versatile

Ca2+signal systems with various spatial and temporal

dynamics [1] However, the openings of these channels

alone cannot account for the complex patterns of

Ca2+signals It has been known that the Ca2+release

is a ‘quantal’ rather than a continuous process; it

con-sists of a rapid release of a fraction of stored Ca2+

fol-lowed by no or a much slower efflux of Ca2+[2] This

transient and partial release behavior requires a

mech-anism for the attenuation of Ca2+efflux, e.g

inactiva-tion of InsP3 receptor channels However, many

studies have provided evidence for the lack of inactiva-tion or desensitizainactiva-tion of InsP3receptor channels [3–9] The quantal Ca2+ release also occurs following the activation of ryanodine receptors [10,11], and appears

to be a rather general phenomenon [12,13]

To explain the quantal nature of this phenomenon,

it was first proposed that Ca2+ is released in an ‘all-or-none’ fashion from multiple stores with different sensitivities to InsP3(‘all-or-none’ model) [2,14] Irvine [15] has proposed another model, in which the Ca2+ efflux through InsP3 receptor channels is regulated from the luminal side by the concentration of Ca2+in the Ca2+store (‘steady-state’ model) Since this model was proposed, many studies have demonstrated that a reduction in the luminal [Ca2+] concentration attenu-ates Ca2+ efflux [8,9,16,17] Koizumi et al [11] have shown that the Ca2+ release from ryanodine-sensitive

Keywords

Ca2+oscillation; Ca2+release; endoplasmic

reticulum membrane potential;

voltage-sensitive dye; retina

Correspondence

M Yamashita, Department of Physiology I,

Nara Medical University, Shijo-cho 840,

Kashihara 634–8521, Japan

Fax: +81 744 29 0306

Tel: +81 744 29 8827

E-mail: yama@naramed-u.ac.jp

(Received 14 April 2006, revised 6 June

2006, accepted 8 June 2006)

doi:10.1111/j.1742-4658.2006.05365.x

Ca2+ release from Ca2+stores is a ‘quantal’ process; it terminates after a rapid release of stored Ca2+ To explain the quantal nature, it has been supposed that a decrease in luminal Ca2+acts as a ‘brake’ on store release However, the mechanism for the attenuation of Ca2+ efflux remains unknown We show that Ca2+release is controlled by voltage- and Ca2+ -activated potassium channels in the Ca2+ store The potassium channel was identified as the big or maxi-K (BK)-type, and was activated by posit-ive shifts in luminal potential and luminal Ca2+ increases, as revealed by patch-clamp recordings from an exposed nuclear envelope The blockage

or closure of the store BK channel due to Ca2+ efflux developed lumen-negative potentials, as revealed with an organelle-specific voltage-sensitive dye [DiOC5(3); 3,3’-dipentyloxacarbocyanine iodide], and suppressed Ca2+ release The store BK channels are reactivated by Ca2+ uptake by Ca2+ pumps regeneratively with K+ entry to allow repetitive Ca2+ release Indeed, the luminal potential oscillated bistably by
45 mV in amplitude Our study suggests that Ca2+ efflux-induced store BK channel closures attenuate Ca2+release with decreases in counter-influx of K+

Abbreviations

BK channel, big or maxi-K channel; CICR, Ca2+-induced Ca2+release; DiOC 5 (3), 3,3’-dipentyloxacarbocyanine iodide; E Ca , equilibrium potential for Ca 2+ ions; E3, embryonic day 3; ER, endoplasmic reticulum; IK channel, intermediate conductance Ca 2+ -activated K channel; InsP 3 , inositol 1,4,5-trisphosphate; NBS, normal bath solution; SK channel, small conductance Ca 2+ -activated channel.

stores is also regulated by the luminal [Ca2+]

concen-tration Caroppo et al [18] reevaluated this issue by

introducing a membrane-permeant, low affinity Ca2+

chelator [TPEN;

N,N,N¢,N¢-tetrakis(2-pyridylmethyl)-ethylenediamine] to intact BHK-21 cells They have

shown that a reduction in the luminal [Ca2+] does

indeed attenuate the Ca2+release However, it remains

unclear whether the regulation by luminal Ca2+occurs

either directly at the InsP3 receptor channel itself or

indirectly [19,20] Recently, a Ca2+-dependent protein

has been reported to regulate InsP3 receptor channels

from the luminal side [21] However, this protein

affects InsP3 receptor type 1 alone At present, the

underlying mechanism for the attenuation of Ca2+

efflux still remains unknown

The release of Ca2+ from a Ca2+ store should

cause ancillary movements of other ions, such as influx

of K+, to compensate for electrical charge movements

across the store membrane [22,23] Otherwise, a

lumen-negative potential is rapidly built up by Ca2+

efflux to such a degree that the negative shift can

reach the equilibrium potential for Ca2+ ions, and

the Ca2+ efflux is likely to cease with the loss of the

electrochemical driving force for Ca2+ efflux The

counter-movement of K+ has been proposed to

com-pensate for charge movements across the membrane of

sarcoplasmic reticulum and thereby to support rapid

Ca2+ release [24] It may be supposed that the

mem-brane potential of Ca2+store is regulated not only by

Ca2+ releasing channels but also by other channels

in the Ca2+ store such as big or maxi-K (BK)-type

potassium channels [25] At present, however, despite

its functional importance, there is very little direct

experimental information about the membrane

poten-tial of endoplasmic reticulum (ER) or sarcoplasmic

reticulum, as Burdakov et al [26] have pointed out in

a recent review article

The aims of the present study are to reveal dynamic

changes in the membrane potential of the Ca2+ store

(luminal potential) and to estimate the effect of

counter-movements of K+ on Ca2+ release In order

to detect the changes in luminal potential, we applied 3,3’-dipentyloxacarbocyanine iodide [DiOC5(3)], a volt-age-sensitive fluorescent probe for organelle membrane [27,28], to the neuroepithelium of embryonic chick retina, where the activation of G protein-coupled puri-noceptor by ATP causes a robust Ca2+ release and

Ca2+ oscillations occur [29–31] The ATP-induced

Ca2+ mobilization is largest at embryonic day 3 (E3) [29], when almost all cells are undergoing interkinetic nuclear migration along the vertical (outer–inner) axis

of the retina during the cell cycle [32,33] We have already revealed the distribution of ER and nuclear envelope in the E3 chick retina with the DiOC5(3) staining [31] In the present study, we studied the chan-ges in the DiOC5(3) fluorescence intensity in the E3 chick retina to gain an insight into the dynamics of the luminal potential of Ca2+store in intact cells

Results

Luminal potential changes measured with DiOC5(3)

The voltage sensitivity of DiOC5(3) was evaluated

by voltage clamping of an excised membrane patch stained with DiOC5(3) The bath solution (outside the membrane patch) mimicked intracellular solution and the pipette solution (inside the membrane patch) mim-icked the lumen of Ca2+ store in ionic composition (see Experimental procedures) The DiOC5(3) fluores-cence intensity increased with a negative change in the pipette potential and decreased with a positive change (Fig 1A) The rate of change in DiOC5(3) fluorescence intensity against voltage change was )1.3 ± 0.3% ⁄ mV (mean ± SD, n ¼ 8 patches, Fig 1B)

Figure 1C illustrates the vertical plane of the neuro-epithelium of E3 chick retina and Fig 1D shows the horizontal plane of the inner layer, where the somata

of S-phase cells are located The S-phase cell soma

Fig 1 Measurement of the membrane potential of Ca 2+ store with DiOC 5 (3) (A,B) Voltage sensitivity of DiOC 5 (3) An excised membrane patch was voltage-clamped at )10 mV (negative inside the pipette) and stained with DiOC 5 (3) (A) The command voltage (pipette potential) was changed from )10 mV to +30 mV and )70 mV (B) The change in DiOC 5 (3) fluorescence intensity [DF⁄ F 0 DiOC5(3)] is plotted against the pipette potential (Membrane potential) for the same DiOC 5 (3)-stained membrane patch as shown in (A) (C) A schematic drawing of the vertical plane of E3 chick retinal neuroepithelium (total thickness,  40 lm) (D) Nomarski optics (DIC) view of the horizontal plane of the inner layer (5 lm inside of the inner surface) (E) A schematic drawing of a cell with the soma in the inner layer, which is occupied with a nucleus as revealed by DNA staining with SYTO 24 [31] N, nucleus; NE, nuclear envelope (F) A DiOC 5 (3)-stained retinal cell enlarged with a low-Ca2+hypotonic solution Its DiOC 5 (3) fluorescence image (upper) and DIC image (lower) (G) Fluo-4 and DiOC 5 (3) fluorescence responses

to ATP recorded from the inner layer of E3 chick retinae ATP was bath-applied during the bar (500 l M , maximal dose for Ca 2+ response [29]) The measurement area of DiOC5(3) fluorescence was 17 · 17 lm The fluo-4 trace ([Ca 2+ ]i) is an averaged recordings from seven cells

in a different retina D[Ca2+] i is the derivative of the fluo-4 trace (H) Possible movements of Ca2+and K+during Ca2+release and uptake.

A B

D

H G

C

is occupied with a nucleus and hence the nuclear

envelope forms a continuous circular structure in close

apposition to the plasma membrane in the horizontal

plane (Fig 1E) With this structural characteristic,

DiOC5(3) staining shows continuous circular structures

in the inner layer [31] (see also Fig 3E in the

Discus-sion) The circular structure is also labeled with

fluor-escent thapsigargin [31], an ER-specific dye in living

cells [34] As the criterion of the labeling of ER or

nuc-lear envelope with DiOC5(3) is a continuous structure

[28], it seems likely that the circular structure is either

perinuclear ER or a nuclear envelope In fact,

enlarge-ment of the DiOC5(3)-stained cell by applying a

low-Ca2+ hypotonic solution showed the labeling of

nuclear envelope (Fig 1F)

The bath-application of ATP causes a robust Ca2+

release in the E3 retina [29,31] In response to the

bath-applied ATP, a biphasic change occurred in the

DiOC5(3) fluorescence intensity; it increased and then

decreased (Fig 1G, measured in the inner layer) From

the fluorescence–voltage relationship of DiOC5(3)

(Figs 1A,B), it is suggested that the biphasic change

indicates a negative-positive change in the luminal

potential As it is supposed that the luminal potential

changes in a negative direction with Ca2+release, and

in a positive direction with Ca2+ uptake by Ca2+

pumps [22,35], it seems likely that the initial increase

reflects the negative potential change due to Ca2+

release and the subsequent decrease reflects the positive

change by Ca2+ uptake The DiOC5(3) fluorescence

increase appears rather earlier than the increase in the

intracellular Ca2+ concentration ([Ca2+]i) measured

with fluo-4 However, it should be noted here that

Ca2+ fluxes (rates of charge movements by Ca2+, i.e

Ca2+ currents) generate the voltage drops across the

membrane of Ca2+store A rise in [Ca2+]iis the

integ-ration of Ca2+effluxes from the store and hence

devel-ops later than the Ca2+ efflux The Ca2+ flux can be

estimated by the derivative of fluo-4 trace (D[Ca2+]i),

and the efflux of Ca2+precedes the main part of Ca2+

rise (Fig 1G) It is also noted that the DiOC5(3)

fluor-escence increase appears broader than the efflux of

Ca2+ and that the DiOC5(3) fluorescence decrease

returns to the initial level later than the recovery of

[Ca2+]i These temporal characteristics may be due to

the slow time constant of DiOC5(3) fluorescence

response (s 3 s, Fig 1A) and the fact that [Ca2+]iis

reduced by plasma membrane Ca2+ pumps and

intra-cellular Ca2+-binding proteins as well as store Ca2+

pumps, which will continue to work until the store is

replenished Figure 1H illustrates possible movements

of Ca2+ and K+ during Ca2+ release and uptake,

although other ionic movements cannot be excluded

The DiOC5(3) fluorescence change might have reflec-ted contributions from mitochondria However, the application of FCCP [carbonyl cyanide 4-(trifluoro-methoxy)phenylhydrazone] (10 lm, for 60 s), a mitoch-ondrial protonophore, did not change the DiOC5(3) fluorescence (data not shown)

Potassium channel in store membrane

If the potassium conductance of the store membrane regulates the luminal potential, it should depend on the difference in [K+] between the lumen and cytosol To test this idea, a 5 mm-K+solution containing nystatin (an ionophore for monovalent cations and Cl–) was bath-applied to a DiOC5(3)-stained retina It caused an increase in the DiOC5(3) fluorescence (Fig 2A), sug-gesting that the lowering of cytosolic [K+] shifts the luminal potential in a negative direction The subse-quent fluorescence decrease might reflect a positive change due to influx of Na+ or efflux of Cl–; NaCl was replaced with sodium gluconate in the nystatin-containing solution In a steady state without changes

in [K+] or [Ca2+], the membrane potential of Ca2+ store will be determined by the equilibrium potential for Ca2+ ions (ECa), which is lumen-negative, and by the proportion of calcium conductance to potassium conductance (see also Discussion) To test this idea, we applied quinidine (200 lm), a membrane-permeant BK channel blocker [36], to a DiOC5(3)-stained retina It caused an increase in the DiOC5(3) fluorescence (Fig 2B), indicating that a negative shift is caused in the luminal potential The negative shift may be due to

a decrease in the potassium conductance by the block-age of store BK channels, and thus the luminal poten-tial shifts towards the ECa The negative shift could exclude the possibility that quinidine primarily blocks

Ca2+-releasing channels, because a reduction in cal-cium conductance should cause a positive shift

The closing of store BK channels may decrease the counter-influx of K+, and the negative shift in the luminal potential may decrease the driving force for

Ca2+ efflux In accordance with this idea, the ATP-induced [Ca2+]i rise was significantly inhibited by quinidine (Student’s t-test, P < 0.1%); changes in the ratio of fluo-4 fluorescence intensities (DF⁄ F0) were 2.04 ± 0.60 (mean ± sd, n¼ 33 cells) without qui-nidine and 0.54 ± 0.39 (n¼ 32) with quinidine (Fig 2C) Quinidine itself caused no change in the

fluo-4 fluorescence (data not shown) Next, we examined iberiotoxin, a specific peptide blocker for BK channels [37] When iberiotoxin (100 nm) was bath-applied, there was no inhibitory effect on the ATP-induced [Ca2+]i rise (DF⁄ F0: 2.26 ± 0.57, n¼ 32) In contrast, the

introduction of iberiotoxin into the cells by

electropora-tion significantly inhibited it; DF⁄ F0 was decreased

from 1.98 ± 0.46 (n¼ 87) in the control response

(electrical pulses plus vehicle alone) to 1.27 ± 0.46

(n¼ 85, electrical pulses with iberiotoxin, Fig 2D)

We made patch-clamp recordings from an exposed

nuclear envelope in the ‘nucleus-attached’ mode [25]

after removing the plasma membrane (see

Experimen-tal procedures) Recordings in the low-Ca2+ hypo-tonic solution for cell enlargement showed no channel activity (n¼ 9 patches), whereas channel activities were found after perfusion with a Ca2+-increasing solution containing 10 lm ionomycin and 1.1 mm CaCl2 (Supplementary Fig S1) Single channel out-ward currents were activated by positive shifts in the luminal potential (Fig 2E) Because the recordings

D C

H G

Fig 2 K channel in Ca 2+ store (A) DiOC5(3)

response to bath-application of 5 m M -K +

solution containing nystatin (200 lgÆmL)1).

(B) DiOC5(3) response to quinidine (200 l M ).

(C,D) Effects of quinidine (200 l M , C) and

iberiotoxin (IbTx, introduced by

electropora-tion, D) on Ca2+responses to ATP (500 l M )

recorded from somata in the inner layer (E)

Single channel currents recorded from an

exposed nuclear envelope in the

‘nucleus-attached’ mode [25] at pipette potentials

indicated Negative changes mean

lumen-positive changes (F) Immunolabeling of

exposed nuclei with an antibody against BK

channels (upper, fluoro-labeled with Alexa

Fluor 488) and negative control (lower) The

cells were enlarged with a low-Ca2+

hypo-tonic solution and plasma membranes were

removed by perfusion with 0.2% Triton

X-100 (G) DiOC 5 (3) response to Ca2+-free

solution containing thapsigargin (500 n M ).

(H) DiOC5(3) response to 100 m M -K +

solu-tion Measurement area, 13 · 13 lm in the

inner layer (A,B,G,H).

were made with a patch pipette containing a

high-Na+, low-Cl– solution, the outward current would be

carried by K+ ions, although the luminal [K+]

seemed to be considerably lowered by the hypotonic

solution for cell enlargement The exposed nucleus

was immunolabeled with an antibody against BK

channels (Fig 2F)

The above result of Ca2+ dependence suggests that

the store BK channel will be closed by a decrease in

luminal [Ca2+] The luminal [Ca2+] can be lowered by

blocking store Ca2+ pumps with thapsigargin in the

E3 retina [38] When thapsigargin (500 nm) was

applied to a DiOC5(3)-stained retina, the DiOC5(3)

fluorescence was remarkably increased (Fig 2G) This

large negative shift in the luminal potential may be

due to a leak of Ca2+ from the store and also an

increase in the membrane resistance by closure of store

BK channels with the decrease in luminal [Ca2+] The

DiOC5(3) fluorescence was not changed by the

depo-larization of plasma membrane with a 100 mm-K+

solution (Fig 2H), which could suggest that Ca2+

influx through voltage-dependent calcium channels

[31,39] does not induce Ca2+release

Ca2+oscillation and luminal potential oscillation

We observed Ca2+ oscillations ( 6–30 min)1 in

fre-quency), which occurred spontaneously or after

ATP-induced Ca2+ release at rather high temperature

(‡ 28
C) (Fig 3A,B) When ATP was applied to the

retina that was showing spontaneous Ca2+oscillation,

excess increases in [Ca2+]iwere caused with different

on-sets (Fig 3C,D) Nevertheless, the Ca2+oscillation

per-sisted and the frequency of Ca2+oscillation was raised

at the peak level of the agonist-induced [Ca2+]iincrease

(Fig 3D), where the lower level of Ca2+oscillation was

distinctly higher than the upper level of the Ca2+

oscilla-tion before agonist applicaoscilla-tion Thus it seems difficult to

explain the generation of Ca2+ oscillation by [Ca2+]i

-dependent negative feedback models, which predicts

that the Ca2+ oscillation should be arrested by the

excess increase in [Ca2+]i Alternatively, the Ca2+

oscil-lation might have been caused by spontaneous action

potentials independently of the agonist-induced Ca2+

release [40] However, the cells of E3 retina do not fire

action potentials [39], even with current injections

(M Yamashita, unpublished data) Therefore, we

sup-pose that the Ca2+oscillation is caused by the repetition

of quantal Ca2+release that depends on luminal [Ca2+]

[41] or by luminal potential oscillation, as described

below

Bistable oscillations of DiOC5(3) fluorescence were

observed in the inner layer (Fig 3E,F) and the vertical

plane (Fig 3G,H) The magnitude of potential oscilla-tion was estimated to be up to
45 mV (Fig 3F,H) It was noted that the falling phase was more rapid than the rising phase Quinidine lowered the frequency of DiOC5(3) fluorescence oscillation (Fig 4A,B), and then the oscillation ceased (Fig 4C) After washout of quinidine, the DiOC5(3) fluorescence oscillation turned out to be irregular high frequency flickering (Fig 4D) Paxilline (10 lm), another membrane-permeant BK channel blocker [42], also lowered the frequency of DiOC5(3) fluorescence oscillation in a reversible man-ner (data not shown)

Discussion

Changes in the membrane potential across the ER are likely to impact upon both the rate and extent of Ca2+ release from the store, and hence influence the spatial and temporal dynamics of intracellular Ca2+ signals Despite its potential importance, we know very little about how the ER potential can change, or whether this can affect the profile of intracellular Ca2+ signals The major hurdle to our understanding of this area is the technical difficulty in measuring membrane poten-tial changes specific to the ER without contamination from other organelles within living cells In the present study, it was attempted to tackle this issue directly, using a voltage-sensitive fluorescent dye [DiOC5(3)]

We applied DiOC5(3) to the intact cells of embryonic chick retina to measure changes in the membrane potential of Ca2+store The DiOC5(3) fluorescence sig-nal may include contributions not only from the mem-brane of the Ca2+store (nuclear envelope and ER), but also from the plasma membrane or mitochondrial membranes To address this issue, the DiOC5(3)-stained cell was enlarged by applying a low-Ca2+ hypotonic solution Figure 1F clearly shows that the dye is located

in the nuclear envelope or perinuclear ER, and not in the plasma membrane The DiOC5(3) fluorescence was not changed by the depolarization of plasma membrane (Fig 2H) This result may also exclude the contribution from the plasma membrane There was also a concern that the DiOC5(3) fluorescence signal might have reflec-ted contributions from mitochondria To address this issue, we tested FCCP (a mitochondria-specific pro-tonophore), which did not change the DiOC5(3) fluor-escence This result may exclude the contribution from mitochondria We examined the distribution of mito-chondria with rhodamine 123 (a specific fluorescent probe for mitochondria) The fluorescence image showed discrete granular structures, which were quite different from the circular structures (unpublished observation) Thus it is supposed that the DiOC5(3)

fluorescence signal mainly reflects the changes in the

membrane potential of the nuclear envelope or ER

Patch-clamp recordings from an exposed nuclear

envelope showed that channel activities appeared after

perfusion with a Ca2+-increasing solution containing

ionomycin and Ca2+, and that the channel activity

was voltage-dependent Ca2+-activated potassium

channels are classified into three groups [BK-,

inter-mediate-conductance Ca2+-activated (IK)- and

small-conductance Ca2+-activated (SK)-types] according to

voltage dependence and single-channel conductance

[36] Among the three groups, BK-type alone shows

voltage dependence Our patch-clamp recording shows clear voltage dependence We tried to excise a mem-brane patch from the exposed nuclear envelope to esti-mate single-channel conductance with defined K+ concentrations However, it was very difficult to excise

a membrane patch in the low-Ca2+ hypotonic solu-tion, or to maintain a seal at GW-values while chan-ging bath solutions to a high-K+ solution Thus we could not estimate the single-channel conductance in a correct manner Nevertheless, the Ca2+- and voltage-dependence strongly suggests that BK channels are functioning there Our result is in accordance with the

C A

G

E

Fig 3 Ca 2+ oscillation and DiOC5(3) fluorescence oscillation (A) Fluo-4 images of the inner layer at t1 and t2 in (B) (B) Fluo-4 fluorescence changes of the three cells marked in (A) (C,D) Same as (A,B) ATP (500 l M ) was applied during the bar (E–H) DiOC 5 (3) fluorescence oscilla-tions in the inner layer (E,F) and the vertical plane (G,H).

study of Maruyama et al [25], i.e that the BK channel

in nuclear envelope is activated by positive shifts in the

luminal potential and luminal Ca2+increases

Further-more, the immunolabeling of the exposed nuclei with

an antibody against BK channels indicates that BK

channels are present there

Openings of store BK channels tend to nullify the

luminal potential unless a [K+] gradient is formed

across the store membrane As the ECais

lumen-negat-ive, the openings of store BK channels would maintain

the driving force for Ca2+ efflux When Ca2+ is

released, the Ca2+ efflux may be amplified by Ca2+

-induced Ca2+ release (CICR) at the Ca2+ releasing

site (Fig 4E, left) A decrease in luminal [Ca2+] and a

negative potential shift due to the Ca2+ efflux could

close the store BK channels, which decreases the

coun-ter-influx of K+ The increase in store membrane

resistance should enhance the negative potential shift

towards the ECa and the Ca2+ release would decline with a decrease in the driving force for Ca2+ efflux (Fig 4E, right) It should also be considered that the lumen-negative potential could induce blockage of

Ca2+ releasing channels by binding of Mg2+ to sites

in the conduction pathway [43]

The rate of Ca2+uptake by Ca2+ pumps is acceler-ated by a lumen-negative potential [44] and a decrease

in luminal [Ca2+] [45] When the Ca2+-pumping activ-ity is thus raised, the store BK channel would be reac-tivated by an increase in luminal [Ca2+] and a positive shift in the luminal potential due to the uptake of

Ca2+ Capacitative Ca2+ entry [38,46,47] also caused

a positive shift in the luminal potential, even when the

Ca2+ pump was inhibited (unpublished observation) The efflux of Cl– may also contribute to the depolar-ization (Fig 4E, right) The reactivation process of store BK channels should be regenerative, because an

D

E

C

Fig 4 Bistable change in the membrane potential of Ca 2+ store (A–D) Inhibition by quinidine (200 l M ) of DiOC5(3) fluorescence oscillation (measurement area, 13 · 13 lm

in the inner layer) Records before (A) and

3 min after beginning of quinidine applica-tion (B), after applicaapplica-tion for 11 min (C) and after washout for 100 min (D) (E) Two states of Ca 2+ store Left, openings of store

BK channels maintain the driving force for

Ca2+efflux with counter-influx of K+ CICR,

Ca 2+ -induced Ca 2+ release Right, Ca 2+

release declines due to closings of store BK channels.

influx of K+ causes a positive shift in the luminal

potential and this positive shift further activates the

store BK channels Such regenerative activation may

account for the rapid decrease in DiOC5(3)

fluores-cence (i.e positive shift) during oscillation The

DiOC5(3) fluorescence oscillation appears synchronous

in the vertical plane (Fig 3G,H), which could suggest

that the regenerative depolarization rapidly propagates

throughout the Ca2+ store The rapid depolarization

may also give a clue to the explanation for the

syn-chronicity of Ca2+ releases across the oscillating cells

(see Supplementary Doc S1) The oscillatory change

in luminal potential of
45 mV in amplitude could

regulate Ca2+ release, because ECa would be

)50 mV if the luminal [Ca2+] falls below 50 lm [48]

and the peak [Ca2+]iexceeds 1 lm At luminal [Ca2+]

<100 lm, the activation of BK channel may depend

on the membrane potential from 0 to)40 mV [49]

In nonoscillating cells, however, the change in

lumi-nal potential with agonist stimulation may be

underes-timated (Fig 1G), because the agonist-induced Ca2+

releases are not synchronous across the cells

(Fig 3C,D) and the DiOC5(3) fluorescence was

recor-ded from a large area including multiple cells We

recorded the DiOC5(3) fluorescence in such a way

because mechanical drift by changing bath solutions

seriously affected the fluorescence recording if the

DiOC5(3) fluorescence was recorded from a small area

It could be supposed that the local change in luminal

potential at each releasing site is larger than the change

measured from the large area In the all-or-none model

for quantal Ca2+ release, it has been supposed that

Ca2+stores are compartmentalized with different

sensi-tivities to InsP3 to explain dose-dependence of Ca2+

release [2,6,13,14] The heterogeneity in the sensitivity

to InsP3between the compartments may be due to

dif-ferent locations of the compartments or the distance

from the site of InsP3 production [2,50,51] The Ca2+

store may be functionally compartmentalized if

diffu-sion of Ca2+within the lumen of Ca2+store is slowed

by Ca2+-binding substances and⁄ or by narrow luminal

spaces The narrow space raises the luminal resistance,

so that a negative potential change may be locally

developed by Ca2+ efflux at each releasing site Thus

our model would be functioning locally at each Ca2+

-releasing site when an agonist was applied The luminal

potential becomes more positive than the initial level

with the maximal agonist stimulation (Fig 1G) As the

nonoscillating retina was kept at low temperature, the

activity of the Ca2+pump may have been lowered and

store BK channels could not be activated at the initial

potential level after the severe depletion of Ca2+ store

with the maximal stimulation

In summary, the present results show that voltage-and Ca2+-activated potassium (¼ BK-type) channels are present in the Ca2+store The store BK channel is activated by positive shifts in the luminal potential and luminal Ca2+increases The luminal potential shifts in

a negative direction with Ca2+ release and a positive direction with Ca2+uptake, as revealed with DiOC5(3) fluorescence measurements From these findings we propose a new idea that the store BK channel dynam-ically regulates Ca2+ release Figure 5 schematically explains this model When Ca2+ is released (phase 1), the luminal potential rapidly shifts in a negative direc-tion and the luminal [Ca2+] is decreased These chan-ges tend to close the store BK channel (phase 2) Closing of the BK channel decreases the counter-influx

of K+, and the luminal potential becomes more negat-ive, thereby causing more closures of store BK chan-nels according to the voltage dependence of the channel When the luminal potential becomes as negat-ive as the equilibrium potential for Ca2+ions, the dri-ving force for Ca2+efflux declines, which leads to the decline in Ca2+ release (phase 3) This model may account for the attenuation of Ca2+ efflux in the

‘quantal’ process of Ca2+ release The decrease in luminal [Ca2+] and the lumen-negative potential accel-erate the rate of Ca2+uptake by Ca2+ pumps, which increase the luminal [Ca2+] causing a positive shift in the luminal potential by the electrogenic nature of the

Ca2+pump (phase 4) The luminal Ca2+increase and the positive potential shift tend to activate the store

BK channel (phase 5) The influx of K+ shifts the luminal potential in positive direction and further acti-vates the BK channels regeneratively in a voltage-dependent manner to depolarize the luminal potential (phase 6) The luminal depolarization may cause oscil-latory Ca2+release (phase 1) or may allow an agonist-induced Ca2+release (phase 0 to phase 1)

The present model shares some features with the previous models for quantal Ca2+ release (the all-or-none and ‘steady-state’ models) The common feature

of the present model with the steady-state model is that a reduction in luminal [Ca2+] attenuates Ca2+ release The difference between the two models is that the attenuation of Ca2+ efflux is mediated by closures

of store BK channels in the present model, while clo-sures of InsP3receptor channels have been supposed in the steady-state model To explain dose-dependence

of quantal Ca2+release by the present model, it seems necessary to consider compartmentalization of Ca2+ stores as supposed in the all-or-none model The pre-sent model may also account for Ca2+oscillation and the synchronization of oscillatory Ca2+releases across the cells by considering ‘active’ changes in the

membrane potential of Ca2+ store due to the

regener-ative activation of store BK channels

Experimental procedures

Preparation of retinal neuroepithelium

The neural retina (40 lm in thickness) isolated from a

chick embryo incubated for 3 days (E3) at 38
C was

positioned in a recording chamber (volume, 0.2 mL) with

a ring of nylon wire [39] The chamber was mounted on

the fixed stage of an upright microscope (BX51WI with a

water immersion objective, LUMPlanFL60XW or

LUMP-lanFL100XW; Olympus, Tokyo, Japan) and perfused with

a normal bath solution (NBS) containing (in mm): NaCl

137; KCl 5; CaCl2 2.5; MgCl2 1; Hepes 10; glucose 22;

buffered to pH 7.3 by adding NaOH, at 2 mLÆmin)1

Recordings were made at 26–28
C or 28–39
C for Ca2+

oscillations

Confocal fluorescence microscopy

The retina was loaded with DiOC5(3) (2 lm, for 20 min;

Molecular Probes, Eugene, OR, USA) or fluo-4

acetoxy-methyl ester [31] A confocal scanner (CSU10; Yokogawa, Tokyo, Japan) connected with a laser diode (473 nm, HK-5510; Shimadzu, Kyoto, Japan) and a cooled ICCD video camera (Gen IV; Princeton Instruments, Trenton, NJ, USA, and Nippon Roper, Chiba, Japan) was used A fluor-escence cytophotometric system (FC-500; Furusawa labo, Kawagoe, Japan) was used for image recording and analy-sis Details of confocal fluorescence recording are described

in our previous paper [31]

Calibration of DiOC5(3) fluorescence against voltage change

A membrane patch was excised outside out from a dissoci-ated dorsal root ganglion cell of an E9 chick embryo with a patch pipette filled with a solution containing (in mm): KCl 141; CaCl2 0.1; MgCl2 2.05; Hepes 10; buffered to pH 7.3

by adding KOH; electrode resistance, 6–7 MW The mem-brane patch was stained with DiOC5(3) (0.02 lm, for

20 min) in a solution containing (in mm): KCl 141; CaCl2

0.1; MgCl22.3; Hepes 10; EGTA 1; buffered to pH 7.3 by adding KOH, under voltage-clamp at a holding potential of )10 mV The DiOC5(3) fluorescence was measured from the tip of patch pipette

Fig 5 A schematic diagram explaining store

BK channel control of Ca 2+ release Phase 1,

Ca 2+ is released with store BK channel opening Phase 2, BK channel tends to close Phase 3, Ca 2+ release declines with

BK channel closing Phase 4, Ca 2+ pump is activated Phase 5, BK channel tends to open Phase 6, BK channel opens for next release of Ca 2+ It may occur as Ca 2+ oscil-lation when the Ca2+-releasing channel remains open and ⁄ or activated by CICR Phase 0, Ca 2+ -releasing channel is closed It may be activated by inositol 1,4,5-trisphos-phate (+ IP3) with agonist stimulation.