- Email: [email protected]
A comparison of measurements of intracellular Ca by Ca electrode and optical indicators
Biochimica et Biophysica Acta, 805 (1984) 393-404
393
Elsevier BBA 11382
A C O M P A R I S O N OF M E A S U R E M E N T S O F I N T R A C E L L U L A R Ca BY Ca E L E C T R O D E AND O P T I C A L I N D I C A T O R S J. REQUENA *, J. WHITTEMBURY **, T. TIFFERT, D.A. EISNER *** and L.J. MULLINS
Department of Biophysics, University of Maryland School of Medicine, Baltimore, MD (U.S.A.) (Received May 30th, 1984)
Key words: Ca assay," Ca electrode," Optical indicator; Aequorin; Arsenazo III," (Squid axon)
Squid giant axons were injected with aequorin or arsenazo III and impaled with a Ca-sensing electrode. The light output of aequorin or the spectrophotometer output when measuring arsenazo was compared with the voltage output of the electrode when the squid axon was depolarized with high-K solutions, when the seawater was made Na-free, or when the axon was tetanized for several minutes. The results from these treatments were that the optical response rose (as much as 50-fold) with all treatments known to increase Ca entry, while the electrode remained unaffected by these treatments. If axons previously subjected to Ca load are treated with electron-transport poisons such as CN, it is known that [Call rises after a time necessary to deplete A T P stores. In such axons one expects a rise of [Ca] i in axoplasm which does not necessarily have to be uniform although the source of such Ca is the mitochondria and these are uniformly distributed in axoplasm. Under conditions of CN application, the optical signals from aequorin or arsenazo and Ca electrode output do rise together when [Cali is high, but there is a region of ICali concentration where aequorin light output or arsenazo absorbance rises while electrode output does not. Axons not loaded with Ca but injected with apyrase and vanadate have mitoehondria that still retain some Ca and this can be released by CN in a truly uniform manner. The results show that such a release (which is small) can be readily measured with aequorin, but again the Ca electrode is insensitive to such [Ca]i change.
Introduction Two recent reports have appeared where the authors measured the [Ca] of squid axoplasm under a variety of experimental conditions. In the first [1], electrodes of a size suitable for the trans* Present address: Centro de Biociencia, Instituto Internacional de Estudios Avanzados (IDEA), Caracas, Venezuela. ** Present address: Department of Biochemistry and Biophysics, University of Pennsylvania, Philadelphia, PA, U.S.A. *** Present address: Department of Physiology, University College, London, U.K. Abbreviation: Tes, 2((2-hydroxy-l,l-bis(hydroxymethyl)ethyl) amino)ethanesulfonic acid. 0167-4889/84/$03.00 © 1984 Elsevier Science Publishers B.V.
verse insertion into squid axons were employed and the authors found a [Ca]i of 79 nM (range 61-95 nM). In the second study [2], electrodes of a tip diameter of 20 /~m were used and the electrodes were inserted axially into squid giant axons. These authors found a mean [Ca]i of 106 nM (range 66-174 nM) or values in reasonable agreement with the study above. There may be some ambiguity as to what the term [Ca]i actually means, first, because electrodes are generally considered to measure ion activity, while optical indicators such as arsenazo III measure concentration. Since [Ca] in cells is so exceedingly low, it is a good approximation to say that it fits the definition of ideality where the activity coefficient of an ion is unity. The fact that cells of
394
marine invertebrates have a solution ionic strength that approaches 400 mM means that only ionic strength considerations apply. Thus, there are grounds for supposing that concentration and Ca 2 + activity are expected to be reasonably concordant. An earlier study where aequorin was confined to a dialysis capillary in squid axons [3] showed that [Ca]i was approx. 20 nM when a freshly isolated squid axon was kept in 3 mM Ca seawater. This value could be raised by increasing [Ca]o or lowering [Na]o. Similarly, aequorin glow could be reduced by lowering [Ca]o. One reason for comparing Ca electrodes with aequorin is that while the former measure Ca in only a limited region, the latter detect Ca throughout the axon. A comparison of these two indicators might, therefore, yield data on the spatial distribution of [Call, as shown for the axon periphery [4]. Since the electrodes we used were inserted axially, it must be expected that the electrode and aequorin measure [Ca]i at different places within a 500/~m axon (see inset to Fig. 1). On the other hand, it is known (Henkart, personal communication) that mitochondria in squid axoplasm are distributed quite uniformly throughout the axoplasm, so that a release of mitochondrial stored Ca by various inhibitors ought to provide a rise in Ca in axoplasm that is uniform unless mitochondria are differentially loaded with Ca depending on their location in axoplasm. We have, therefore, examined axons poisoned with CN for the response of both a Ca electrode and aequorin, and we find that, for axons only lightly loaded with Ca, a release of this Ca does not lead to a change in [Ca] i as measured with the Ca electrode, although aequorin light emission may rise 50-fold, corresponding to a 7-fold increase in [Ca] if aequorin light emission is as the square of [Ca]. A second sort of experiment that we have carried out involves depolarizing the squid giant axon; this is known to increase Ca entry by at least 100-fold [5]. This treatment produces a large increase in aequorin light emission and no change whatsoever in the reading of the Ca electrode. There are thus reasons for questioning the usefulness of the Ca electrode in connection with the measurements that have been made.
Methods Axons The experiments were performed at the Marine Biological Laboratory, Woods Hole, MA, during the months of M a y - J u n e 1983. Axons were dissected from freshly caught living squid and these were cleaned and mounted in a dialysis-type chamber on glass end cannulas as previously described [6]. Microinjection A horizontal microinjector was used to introduce aequorin or other substances that are not permeable. The usual injection distance was 18 mm and the axon was tested for excitability during and after the microinjection procedure. Total aequorin counts injected were (1-3). 109 counts as measured by treating the axon with 10 mM CaC12 solution (in distilled water) at the end of an experiment. Since resting glow is approx. 103 count/s, aequorin is being consumed a t 10-6//S. After microinjection, the axon was impaled with a capillary pipette of diameter 60 /~m filled with 0.6 M KC1 to record membrane potential with reference to an external bath electrode, and at the other end of the axon a Ca-sensitive capillary electrode was introduced. The electrodes were separated at most by 4-8 ram; this separation was convenient for avoiding optical artifacts when external solutions of differing refractive index were used. The general arrangement of electrodes is shown in Fig. 1. The diagram also shows the area of the axon over which light was collected from aequorin for measurement by photon counting. Light measurements Light measurement was as described by Mullins and Requena [7], with the change that the photomultiplier was cooled to - 2 0 ° C by a thermoelectric cooling device so that background counting was reduced several-fold (15 c o u n t / s in the absence of aequorin). Spectrophometric measurements of arsenazo III were as described by Brinley et al. [8]. Electrical measurements The dialysis chamber in which the axon was mounted contained gold electrodes in its base for
395
Eco
Fm
batelhectrode~~~ connu~ °x°n~2C)~ O
axon~ ~
FiberOptics
vv-..~toPhotomuflp ie il r Co ++ee l ctrode
Fig. 1. The experimental arrangement for measuring fight from an aequorin-injected axon and [Ca] from a Ca electrode. The outputs of Eca and Em were subtracted and the resulting signal fed to a chart recorder. In lower left, the location of the Ca electrode relative to the surface membrane is shown.
both the stimulation of the axon and the recording of extracellular action potentials. Some of the experiments to be reported required the stimulation of axons for prolonged periods of time in order to increase internal N a concentration. Membrane potential measurements were made by connecting the internal capillary electrode via high-input-impedance operational amplifier; the other input to the amplifier was a bath reference electrode. The output of this amplifier was designated E m (membrane potential). Ca electrode measurements were made by using an Orion p H meter with an input impedance of 10 a5 ohm to measure the potential difference between the Ca electrode and the bath reference. This output was designated Eca. Outputs from these two measuring amplifiers were subtracted and fed to a chart recorder and to a Hewlett-Packard magnetic tape recorder. This output is designated ( E c a - E m).
Solutions All solutions were fed to the dialysis chamber via a multichannel p u m p that delivered 1 - 3 m l / m i n . This peristaltic p u m p operated continuously while a Hamilton electric valve was used to change the flow of solutions through the chamber. Chamber washout time was about 15 s. The seawater used in these experiments had a
composition as follows: 440 m M NaC1, 10 m M KC1, 10 m M CaC12, 50 m M MgCI2, 10 m M Tris buffer (pH 7.8), and 0.1 m M EDTA. Variations in the composition of this solution were to replace all the N a by either K or Tris (in some cases choline) to produce respectively high-K or Na-free solutions. Ca-free solutions had 1 m M EGTA. Solutions for internal injection were either aequorin at a concentration of 200 ~tM, or apyrase a n d / o r potassium orthovanadate. The aequorin was injected salt-free so that Ca contamination was minimized, while apyrase and vanadate were made up in 330 m M K-Tes buffer (pH 7.3) and were injected at concentrations of 30 units//~l and 30 mM, respectively.
Preparation and calibration of Ca-sensitive electrodes Glass. Capillary tubing (lead glass) of 1 m m o.d., 0.5 m m i.d. was drawn in a vertical puller to a diameter of 90/~m at the tip and the tubing was cut to 12-cm lengths. The wall thickness was about 7/~m. The tubing was dried in an oven and coated outside with solution of high-density polyvinyl chloride dissolved in tetrahydrofurane, thus obtaining a thickness of polyvinyl chloride of about 5 - 1 0 #m. Sensor. The sensor was the neutral ligand (N, N'-di-l,l-(ethoxycarbonyl)undecyl)-N, N'-4,5tetramethyl-3,6-dioxaoctan-l,8-diamide. The sensor was used at a concentration of 10%, in a solution that also contained the following: 3-8% tetraphenyl arsonium tetrakis(p-biphenyl)borate, 15% polyvinyl chloride, 1% sodium tetraphenyl borate, and the balance, o-nitrophenyl octyl ether (NPOE). The individual components were dissolved in 1.5-3-times their weight in tetrahydrofurane depending on the desired viscosity of the solution. The polyvinyl chloride coated tubing described above was dipped at the tip into the sensor solution and then allowed to dry. The thickness of the dry sensor was 2 - 4 / ~ m . Filling. Tubing coated as described above was then selected by size for filling. The range of sizes considered suitable was from 90 to 110 /tin o.d. Filling pipettes for these capillaries were made from 25 /tl D r u m m o n d micropipettes; these were pulled in a pulling machine to diameters such that
396
they would fit into the sensor-coated pipettes. The filling solution was a pCa 2 (sometimes pCa 6) solution as described by Tsien and Rink [9]. Electrical. A 4 cm long Ag/AgC1 wire of 50/~m diameter was introduced into the glass pipette and the wire was attached to a gold pin connector. The average resistance of such electrodes measured in calibrating solutions was 10-30 M~2. Storage. Electrodes were stored dry in test tubes, and electrodes that had lost sensitivity to Ca could sometimes have this restored by a brief dip in dilute sensor solution followed by drying (restoration). Calibration. We have used the buffer compositions described by Tsien and Rink [9] as calibration solutions. These solutions were developed for mammalian ionic strength and [Na]i and hence some modifications were necessary for working with squid giant axons where ionic strength must be expected to be of the order of 400 m M univalent salt, and where [Na] concentration must be expected to be in the range 20-30 m M and Mg is approx. 4 mM. We have used the buffers as described with the addition of either 300 m M K-Tes or 300 m M K-Tes and 20 m M N a or both in order to bring about conditions realistically paralleling those that obtain in the interior of the axon. These additions must be expected to influence the [Ca] of the buffers in two ways: K will inhibit the dissociation of the buffer complex as an ionic strength effect (thus lowering [Ca]), and it will also have a tendency to be measured by the electrode as [Ca] owing to imperfect electrode selectivity. Sodium, present at much lower concentrations, will simply be read as [Ca] because of electrode selectivity. The actual [Ca] of the buffers we used must, therefore, be considered as nominal and not corrected for the effects outlined above. The electrode was first placed in calibrating solution in a glass capillary, and its responses were recorded, and then inserted into the axoplasm and, at times that varied from 1 to 3 h, it was removed from the fiber and returned to a series of calibrating solutions. The principal observation that resulted from this procedure was that the re-calibration of the electrode seldom corresponded with its initial value and that the plot shown in Fig. 2 is representative in that it shows a substantial shift in
the level of electrode output vs. [Ca]. Another feature of the calibration procedure that deserves comment is the temporal stability of the electrode reading. Most reports in the literature show at best a few minutes of recording of the output of a Ca electrode; the lower part of Fig. 2 is of interest, this output being recorded for a period of 2 h. It shows that while there is a relatively prompt response to a change from a pCa 7 to pCa 6, and that this response has a slope that averaged in our experiments 21 mV per decade, there is a much lower response in going from pCa 7 to pCa 8. Moreover, this change is susceptible to a substantial drift, as is the response to a return to
pCa 7. Since most of the values we recorded were greater than a [Ca]i of 100 nM, our calibration procedure was to reference pCa 7 solutions to a potential of zero and to relate changes in electrode potential to this value. Our results for electrode slope, indicated electrode potential, and aequorin count are shown in Table I, Anions. It seemed necessary to verify that a variety of anions known to exist in axoplasm did not affect the reading of the Ca electrode. Experiments were carried out where 10 m M Tris-ATP, as well as the K salts of sulfate and phosphate also at 10 mM, and the exchange of 100 m M KC1 for 100 m M potassium isothionate were all without effect on the reading of the electrode with any of the Tsien-Rink buffers. We conclude, therefore, that anions of the sort likely to be found in axoplasm are without effect on the Ca sensor. Ionic strength. One effect of increased ionic strength is that the ionization of the C a - E G T A complex is affected. DiPolo et al. [2] have attempted to correct for this effect by recalculating a new dissociation constant for the C a - E G T A complex at 300 m M KC1; a second effect of high ionic strength is that K + must be expected to affect the Ca electrode to the extent that they appear to be permeable cations. This effect has not been corrected for and our own measurements show that if one avoids C a - E G T A buffers and works with 10 /~M Ca there is a + 5 - 7 mV displacement of the Ca electrode output in going from 100 to 350 m M KC1 concentration. Sodium. This ion is a poison for Ca electrodes. If a Ca electrode is placed in seawater, it loses
397
(upper trace), where it can be noted that a substantial change in apparent Ca calibration results from a quite short immersion of the electrode in axoplasm.
3020 °iEc°E :'2;t +30
I
f
I0
tO0
IO00
8
7
6
r {EC°-Em) L
+20[- ~ +10
_
mV ~
pCo7
Results
f
IO" [Co] .M 5
pCo
No 30raM K 300 mM
q 4
~sm,~ ~
i0 [ p C e 6 L¢/I I L I I I I I ' ' ' Z5 50
,pCal,
, , J 75
i i i , J
I00
t ,
, l
125
I
i i
i 150
minutes
Fig. 2. The upper panel shows the Ca electrode response to an initial calibration (e) and a second calibration upon the termination of the experiment some 3 h later (©). Note both the shift in potential and the change in slope. The calibrating buffers were of the Tsien-Rink sort with the addition of 300 m M KCI, 20 m M Na and 5 m M MgCI 2. Electrode No. 18. In the lower panel is a record of the output of a Ca electrode in various buffer mixtures over 150 rain. Note both the fast and slow phases of electrode response, and the electrode drift between 75 and 150 min. Electrode No. 13 bis.
sensitivity to Ca totally and on a time-scale of 12 h irreversibly. This makes it quite important that the glass cannulas to which the axon is tied are carefully rinsed with K-Tes solution and that this be changed before axial Ca electrodes are inserted into axoplasm. Additionally, some correction needs to be made for the [Na] of axoplasm, which is about 20 raM. Our measurements show that adding Na to Tsien-Rink calibrating solutions both reduces the slope of the electrode response in the range pCa 7-8 and that it also displaces the level of output of the electrode• This finding confirms that of DiPolo et al. [2]. Electrode stability. We agree with the experience of Tsien and Rink that one of the difficulties with Ca electrodes is that they sometimes change their calibration with time. Thus an initial calibration, followed by the insertion of the electrode into the cell and then a final calibration may result in measurements that differ by 5-10 mV even though the slopes are the same. This is not a problem of electronics (these are stable at the level of 1 mV for several hours) but a problem of the electrode output. A representative result is shown in Fig. 2
Our results comparing the initial aequorin photon count with the reading of a Ca electrode in the same axon are shown in Table I. The mean [Ca]i was 210 riM, which compares with values in the range of 100 nM as reported by others. It is, however, a value 10-times higher than the value reported using aequorin [3]. We have included some axons with initial aequorin counts of (16-30) • 1 0 3 count/s, which are clearly leaky axons, to demonstrate that, at least in one of these cases, the Ca electrode read only 300 nM. Also included in Table I are a phenol red-injected axon where light is only measured from the periphery of the axon [4], and results with two axons where C a - E G T A / E G T A buffers were injected to clamp the ionized Ca of axoplasm at pCa 7. The final buffer concentration is calculated to be 10 mM yet the Ca electrode gave values of 360 and 1350 riM, respectively. Another series of experiments involved the addition of orthovanadate and the enzyme apyrase to axoplasm to prevent contributions from ATP-dependent systems from influencing the axon; in these cases the Ca electrode read 100 nM in all three axons studied, although one can calculate that normal Ca influx should increase the ionized Ca many-fold over times of the order of 30 rain. Since, as indicated in the Methods section, the Ca electrode was expected to sample [Ca]i from a region of axoplasm that was removed from the periphery, it was of interest to compare the response of aequorin and a Ca electrode to a highly local Ca entry in an axon where depolarization produced a large Ca entry. The result, shown in Fig. 3, is that there was the usual aequorin response of an axon previously loaded with Na by stimulation, in that there was a large increase in the aequorin light emission and that this increase was continuous over a period of more than 25 min. By contrast, the Ca electrode showed no response whatsoever. Now, experiments with a dialysis
398
TABLE I I N I T I A L Ca E L E C T R O D E A N D A E Q U O R I N
READINGS IN AXONS
Values in parentheses have been omitted from the means. Axon
Em
Initial
Calcium electrode
Ref.
(mV)
aequorin count/s
Reading, pCa 7 = 0
Slope, mV/pCa
(mV)
(7 ---, 6)
[Ca] i (nM)
[Ca]o (mM)
260583B
- 57
800
+ 5.0
20.5
175
310583A
- 60
600
+ 3.0
21.0
130
10
300583A
- 57
600
+ 5.5
23.0
173
10
240583A
- 52
240583B 220583
4 000
120
10
+ 2
26.0
(30000) (16 000)
+ 12 + 25
25.0 16.0
302 (3 000)
10 3
3 _ a _ a
(25 000)
+ 22 + 28 26
21.0 23.0 22.0
(1000) (1 600)
10 10 3
_ a _ a
351 (1000) (1000)
10 10 10
190583 230583A 250583B
- 51 - 57
600
310583B 200583A 180583A
- 65 - 57 - 55
4400 1100 1 800
12 21,5 17
22.0 20.0 17.0
290583A 010683A
- 53 - 53
2000 4 500
8 2
19.5 19.0
257 127
10 10
120583A
- 56
1000
8
20.0
251
10
21.0
210
Mean
Notes
1950
_
a
_ a - "
- "
Phenol red axon pCa 7 inj. pCa 7 inj.
020683B
- 55
70
7
17.0
258
3
230583B 230583A 270583
- 51 - 51 - 64
1 400
5 26 0
9.0 23.0 23.0
360 1 350 100
10 10 10
280583A 270583B
- 56 - 57
21 000 16 500
0 0
15.0 23.0
100
10
Apy + VO4
100
10
Apy + VO4
Apy
+
go 4 b
a Calibrated in Na-free Ca buffers. b A p y r a s e 1 unit//tl and vanadate 1 m M final concentration.
capillary containing aequorin but located in about the same place as the Ca electrode [3] showed that a large increase in Ca entry was sensed by aequorin in the center of the axon with a time lag of at best a few minutes, so that there is some reason for doubting that the Ca electrode is in fact measuring the [Ca] as calibrations would suggest that it ought. The axon shown in Fig. 3 was subsequently subjected to a variety of procedures that led to increases in the aequorin light emission, so that from the initial photon count of 103/s emission rose to 105/s (a 100-fold change) with no change in the Ca-sensitive electrode reading. The plot on the right hand side of Fig. 3 is one of Ca electrode reading vs. photon count as indicated by aequorin. To avoid the conclusion that some inadequacy in the Ca electrode itself is responsible for the
results shown in Fig. 3, conditions were arranged whereby the release of Ca was, insofar as we can arrange it, uniform throughout the axoplasm. This was achieved by previously loading the axon with Ca by stimulation and then by depolarizing and applying CN to block electron transport in mitochondria. The results of this experiment were that there was a large increase in both the aequorin light emission and in the reading of the Ca electrode. The plot in the right-hand half of Fig. 3 shows that sensitivity of the Ca electrode began at a count (for aequorin) of about 1 5 0 0 0 0 / s (1.5. 1 0 - 4 is the fraction of aequorin consumed per second). A conclusion is that the Ca electrode does respond to [Ca] in axoplasm but only when aequorin counts are 30-100-times the resting level of light emission.
399
[Ca] i M
count/s X 103
1°"5-- by electrode
I0 450 K.
ECa" Em mV
/
+20
id e
IOCa. /equorin 6
+10 16r
Ca electrode
4
I
i(~el I
4O
i
iOs
I
I
50 60 minutes
I
i
i
i
i i iI
i
I
i
i
i
i i i I
i04 105 Aequorin count/s
i
i
i
i
i i i i i
10 6
I
70
Fig. 3. This axon was given 75000 impulses to raise [Na]i and was then depolarized as indicated with 450 mM K. The aequorin light emission rose from 800/s to 9700/s without any change in the output of the Ca electrode. The electrode had a slope of 22 mV between pCa 7 and 6 and it indicated a steady 160 nM. On the right, a comparison is made between aequorin light emission and electrode indication over the entire range of light emission obtained in the experiment. The application of CN to the axon led, with a delay, to a count that increased to the level of 500000/s; the Ca electrode responded when the count reached 200000 count/s.
Since the results in Fig. 3 indicated that we could only expect Ca electrode responses if aequorin counts were very high, an axon was deliberately highly loaded with Ca by stimulating it in 10 mM Ca seawater at a frequency of 6 0 / s until 145000 impulses had been delivered. The axon was then treated with a solution of 10 mM C a / 4 5 0 mM K / 1 mM cyanide. The result of this treatment (shown in Fig. 4) was a rapid change in aequorin glow (barely noticeable at the low gain used) from 2500 to 50000 c o u n t / s and then the Ca electrode began to respond and the two indicators rose together in the range (50-600)-103 aequorin count and 110 nM to 4500 nM as indicated by the Ca electrode. Upon removal of CN but keeping the axon depolarized with high potassium, both responses fell, with the aequorin response returning to its initial value before that of the Ca electrode. Ultimately, both showed a value
close to that at the start of the experiment. It is useful to note that the aequorin light had fallen to virtually its initial value before the Ca electrode began its decline. The Ca electrode continued falling for 20 min after the aequorin light had reached a steady value and there is the suggestion that the electrode response would continue to fall. This cannot be due to surface pumping by the N a / C a exchange since this was inhibited by the application of a high K solution. When the results for this experiment are plotted (as in Fig. 3) as electrode reading vs. aequorin count over all values reached during the experiment we have the result shown in the lower panel of Fig. 4. Again, electrode sensitivity would appear to start around l0 s aequorin counts. It is of interest to note that there is a region of the curve at a count of around 105 that is linear between Ca and photons, and a region that has a slope of about 2.5
400 mM
¢N ] [ Co
,
I
1o
- - ]
450
ECo -Em(mV)- 45
count/$
40
X I06 1,0
3O 25
0.8 0.6
15 I0 5 0
0.4 0.2 0
75
I00
125
minutes
[Co]i from Electrode M id 6
J
j67
i68
h
103
i
1
[ JiiiiI 10 4
i
i
LIJlII 10 5
J
,
i iJllll 10 6
Aequorin counts/s Fig. 4. This axon had been previously stimulated (145000 impulses) and in this experiment CN and depolarization were applied as indicated. The initial [Ca]i as indicated by the electrode was 110 nM and it rose to 4500 nM at a peak and then fell as CN was removed. The lower chart shows that a sensitivity of the electrode began at about 50000 aequorin counts. Initial aequorin count was 2500/s, while final count (in 450 mM K) was 12300/s.
between 3 • 105 and 6.105 count/s. The results in Fig. 4 are not a peculiarity of aequorin nor of a depolarized fiber. Fig. 5 shows data from a similar experiment where arsenazo III was used to measure Ca i. Increasing external Ca from 3 to 112 mM produced a measurable increase of the arsenazo III signal but no deflection in the Ca electrode trace, However, when CN is added (in 112 m M Cao) both the electrode and the arsenazo III responded (Fig. 5B). Interestingly,
there is a hysteresis between arsenazo III and Ca-electrode response in the same direction as that shown in Fig. 4 between aequorin glow and electrode response. Since the experiments indicated by Figs. 3, 4 and 5 would suggest that there is a limit to the sensitivity of a Ca electrode, even if Ca release is brought about by using mitochondria as the means to effect a rise in [Ca]i, it seemed important to look at what might happen if a mitochondrial release were made more modest and uniform than that in Fig. 3. A second feature is to make ATPdependent buffering as small as possible. Accordingly, a fresh (unstimulated) axon was injected with apyrase + orthovanadate in order both that ATP be made as low as possible and that phosphorylation of membrane proteins be prevented. Such an axon cannot be expected to rely on ATPdependent membrane transport for its Ca flux balance, yet it is clear from Fig. 6 that for 70 min the axon is capable of maintaining a normal and low aequorin light emission and that this emission recovers from a depolarization without the intervention of ATP-dependent processes. The Ca electrode is insensitive to any of the changes measured; the small deflections shown on the electrode trace are in fact junction potential changes in going from Na to Tris in seawater and the result of impedance imbalance between the Ca electrode and its reference. More importantly, when CN is applied to the axon (note change of photon-count scale) and the solution made Ca-free then, on the right hand panel, it is clear that the photon count rises about 20-fold with no real change in the Ca electrode indication. Moreover, the experiment shows that in an axon with total metabolic inhibition so that ATP-dependent mechanisms are suppressed, there is still the ability of the N a / C a extrusion system to reduce [Cain. This occurs after almost 3 h of total inhibition of substrate-driven pumps. In an effort to see something of the time course by which the Ca electrode senses changes in [Ca]i, an axon was kept in a solution that was 450 m M K C I / 1 0 m M CaC12 and had both CN an iodoacetamide. In this solution the Ca electrode in the axon showed a reading of close to 1/~M Ca. Upon changing the solution bathing the axon to one that was CN-free, the reading of the electrode, after a
401 CN
112Ca 3Ca
3Ca
0.02 AA 20 10
[
i lOrein
Fig. 5. On the left, an axon injected with arsenazo III is exposed to high and low Ca seawater and, while the metalchromatic indicator optical signal (upper trace) changed appreciably, the Ca electrode output was steady throughout. On the right hand side, the same axon was subsequently exposed to high Ca o (112 raM) seawater in CN. Notice the change in absorbance scale. The Ca electrode indicated a nominal [Ca]i of 100 nM at the start and this increased to 1.9 ~ M in CN (slope 22 mV between pCa 6 and 7).
Apyrase + Vanadate Injected Axon CN Ca K Na
I
mM I0 '°
440
0
450 I 0 J44(
I0 0 (Choline)
440
J
count/s
count/s x 103
E Co - E re ( m Y )
×,o3
2O
440
1÷,5
I00
1+10 +5
16
12
60 40
O-
./i
i
25
I
I I 35
I I 45
I
i 55
i
i
65
i
1
0
.
J
8.5
a
l
95
J
|
105
l
i
115
i
J
125
i
=
1:55
|
i
145
,
a
155
minutes Fig. 6. An axon injected with 10 U/t~I a p y r a s e + 1 m M vanadate (final concentrations) is capable of maintaining a normal aequorin resting glow for 75 min. The Ca electrode does not sense an increase in aequorin from 2000 to 60000 c o u n t / s . Since Cao = 0 all this increase in Ca is from internal release.
402 CN [,IAA l
Cao I0 Ko 4 5 0 mM
l
.,..~ .
[Ca1i
]1
by
electrode
M ~ ' ~ - -
I0-6
450 K
.,4~ ~ Noo440 /<'l~ CNI N% 440 ~ ~ . ~ CN- f r e e ~ . ~
"' 440 ~°_ free j, T 10.7
[
0
I
I
I
I0
I
20
L
I
30
minutes Fig. 7. An axon previously loaded with Ca and kept in 450 mM K, 10 mM Ca solution containing 1 mM each of CN and iodoacetamide was transferred at the point indicated by 'change solution' to several different solutions and after [Ca]i as indicated by the electrode had reached a steady level, the axon was returned to the original solution and the change to another solution carried out. The results show that the initial rate of decline of [Ca] is about the same whether the axon is in a CN-containing or in a CN-free solution that contains Na. Repolarization is not necessary to bring Ca down to low levels. All recovery solutions contained 10 mM Ca.
delay of 1.6 min for the axoplasmic CN to wash out, fell promptly to a value of about 0.1 #M, as shown in Fig. 7. An interesting feature of the decline is that it is not exponential as are aequorin records of light emission when CN is removed. Instead, the record appears to be truncated at 0.1 /~M as if the electrode had lost sensitivity at this point. The axon was then returned to the initial high-K, CN solution and the electrode returned to close to its initial reading. The solution was now changed to one that was both CN-free and Nacontaining. The decline in Ca concentration with time as indicated by the electrode now took place with less delay (the initial deflection upon solution change is a junction potential effect when the bath solution is changed from K to Na). The conclusion from this experiment is that the Ca electrode is apparently able to sense a change in [Ca] as a result of N a / C a exchange at the surface, since the
only difference between the two solutions tested is that the second contained Na. As a check on this interpretation, the axon was once again returned to its loading solution with CN and, when [Ca]i as indicated by the electrode was stable, the solution was changed to one containing Na and CN. Now the only known mechanism to bring down the Ca in the fiber is surface Ca pumping by N a / C a exchange and in this experiment it would appear that Ca can be brought down to half its initial concentration in a time of the order of 1 min. Another feature of note in this recording is that the time course of decline of [Ca] i has much less of the appearance of being truncated; or that it would appear that the Ca electrode is continuing to measure Ca in axoplasm. A final step, the removal of CN, led only to a small decrease in the indicated level of Ca in axoplasm, and this had the appearance of being truncated. The results shown above are important because they suggest that there is only a small delay before a Ca electrode senses a decrease in [Ca]~ as a result of surface pumping of Ca, and that the electrode functions well in detecting decreases in [Ca]i in the range of from 0.2 to 1.0 /~M. The failure of the electrode to sense increases in [Ca]~ must then mean that the increases such as those shown in Fig. 3 are from a resting level of [Ca]i that is much less than 0.1/sM.
Discussion The present measurements show that a Ca electrode in the middle of the axon is insensitive to changes in Ca entry over times of the order of 1 h. In contrast, the measurements by DiPolo et al. [3], where aequorin was confined in a dialysis capillary in the center of a squid giant axon, showed that the light signal was sensitive to Ca entry with a delay of perhaps 15 min when conditions at the surface of the axon were changed so that Ca entry was either enhanced or diminished. The lack of sensitivity of the Ca electrode cannot, therefore, be attributed to its position and we conclude that it is incapable of measuring [Ca] below a level of 100-150 nM and the [Ca] as it exists in axoplasm is very much lower than this value. The question must then arise: why is it possible to show a response to pCa as low as 8 when the
403
electrode in fact cannot sense this [Ca] in axoplasm? One suggestion is that it is impossible to calibrate electrodes except in C a - E G T A / E G T A buffer mixtures, while such buffers do not exist in axoplasm. One might then imagine that the flux of Ca 2+ to the membrane of the Ca electrode is insufficient to bring about an equilibration of the electrode to [Ca] in the range of 10 nM, while it still has acceptable responses in the 1000 nM range. If this is indeed the explanation for the findings that we report, it would have important applications for electrodes that must measure ion concentrations at very low levels. It would also force a re-examination of claims for measured levels of [Ca]i in many sorts of cells. A second possibility is that, since Ca electrodes are filled with a solution containing Ca, the leak outward of this substance may prevent a proper reading of the electrode in solutions where a Ca buffer is absent and where Ca concentrations are in the range of tens of nanomolar. The proper performance of electrodes under these conditions would require that the solution outside flow quite fast past the electrode surface membrane and this is, of course, not possible with intracellular measurements. It is important to note that the present experiments show that the sensing of [Ca]i by the electrode begins at levels of aequorin count that are about 30-times that of resting glow levels and that these electrode measurements yield an average value of [Ca]i of 200 nM. If aequorin were in its linear range of measurement of cell Ca, then actual [Ca]i in axoplasm would be 200 n M / 3 0 or about 7 nM. If aequorin were responding as the square root, then [Ca]i would be 2 0 0 / 3 ~ = 36 nM. Our previous report [3] was that the value was 20 nM and, allowing for the uncertainties of this measurement, it would appear that a proper value is one that is very much lower than a Ca electrode measures. The experiment shown in Fig. 7 yields useful information on the rate by which [Ca]i declines when external Na is supplied to the axon but CN is maintained in the bathing solutions. Under these conditions one supposes that the decline in [Ca]i is brought about by the reactivation of N a / C a exchange and the rate by which Ca declines is related to the efflux level of Ca at the appropriate level of
[Ca]i and to the buffering that internal Ca undergoes. Thus, the initial slope of the curve for 440 mM Na o, 1 mM CN is 0.15 ~tmol/liter per min and if there were no Ca buffering this would correspond to a Ca efflux of 50 f m o l / c m 2- s. Since Brinley et al. [8] have shown that in the range of [Ca]i around 1 /~M only 5% of the entering Ca goes to increase [Ca]i, it follows that the appropriate Ca flux to bring down the Ca concentration in the fiber is 50 fmol × 20 or a Ca efflux of 1 p m o l / c m 2. s. This value is in reasonable agreement with flux estimates of Ca efflux as obtained in dialyzed axons. A second observation from Fig. 7 is that if CN is removed but a high-K external solution is retained, then one expects that the decline in [Ca]i is the result of Ca sequestration by the mitochondria. The internal rate of decline for such a curve (450 mM K, CN-free) is the same as the initial slope of the curve described in the paragraph above; hence, this finding suggests that the rate by which Ca can be moved from axoplasm to mitochondria and the rate by which Ca can be moved out via N a / C a exchange in the surface membrane are of a comparable magnitude. When [Ca]i is much lower (see the 440 mM N a o / 1 mM CN 1 curve at about 24 min) and of the order of 150 nM, the removal of CN results in an indicated decline of Ca i of about 1 / 1 0 that of the initial slopes of the curves at 1 /LM, corresponding to measured fluxes at this level of [Ca]i.
Acknowledgements This work was aided by grants from the National Institutes of Health (1R01 NS 17718) from the National Science Foundation (BNS-8006271) and from CONICIT S1-1198. The aequorin used in these studies was kindly provided by Dr. John Blinks. We are indebted to the Director and staff of the Marine Biological Laboratory, Woods Hole, Massachusetts for facilities put at our disposal. D.E. was supported by a travel grant from the Wellcome Foundation.
References 1 Baker, P.F. and Umbach, J.A. (1983) J. Physiol. 341, 61 P 2 DiPolo, R., Rojas, H., Vergara, J., Lopez, R. and Caputo, C. (1983) Biochim. Biophys. Acta 738, 311-318
404 3 DiPolo, R., Requena, J., Brinley, F.J., Jr., Mullins, L.J., Scarpa, A. and Tiffert, T. (1976) J. Gen. Physiol. 67, 433-467 4 Mullins, L.J. and Requena, J. (1979) J. Gen. Physiol. 74, 393-413 5 Baker, P.F., Hodgkin, A.L. and Ridgway, E.B. (1971) J. Physiol. 218, 709-755 6 Requena, J., DiPolo, R., Brinley, F.J., Jr. and Mullins, L.J. (1977) J. Gen. Physiol. 70, 329-353
7 Mullins, L.J. and Requena, J. (1981) J. Gen. Physiol. 78, 683-700 8 Brinley, F.J., Jr., Tiffert, T., Scarpa, A. and Mullins, L.J. (1977) J. Gen. Physiol. 70, 355-384 9 Tsien, R.Y. and Rink, T.J. (1981) J. Neurosci. Meth. 4, 73-86
393
Elsevier BBA 11382
A C O M P A R I S O N OF M E A S U R E M E N T S O F I N T R A C E L L U L A R Ca BY Ca E L E C T R O D E AND O P T I C A L I N D I C A T O R S J. REQUENA *, J. WHITTEMBURY **, T. TIFFERT, D.A. EISNER *** and L.J. MULLINS
Department of Biophysics, University of Maryland School of Medicine, Baltimore, MD (U.S.A.) (Received May 30th, 1984)
Key words: Ca assay," Ca electrode," Optical indicator; Aequorin; Arsenazo III," (Squid axon)
Squid giant axons were injected with aequorin or arsenazo III and impaled with a Ca-sensing electrode. The light output of aequorin or the spectrophotometer output when measuring arsenazo was compared with the voltage output of the electrode when the squid axon was depolarized with high-K solutions, when the seawater was made Na-free, or when the axon was tetanized for several minutes. The results from these treatments were that the optical response rose (as much as 50-fold) with all treatments known to increase Ca entry, while the electrode remained unaffected by these treatments. If axons previously subjected to Ca load are treated with electron-transport poisons such as CN, it is known that [Call rises after a time necessary to deplete A T P stores. In such axons one expects a rise of [Ca] i in axoplasm which does not necessarily have to be uniform although the source of such Ca is the mitochondria and these are uniformly distributed in axoplasm. Under conditions of CN application, the optical signals from aequorin or arsenazo and Ca electrode output do rise together when [Cali is high, but there is a region of ICali concentration where aequorin light output or arsenazo absorbance rises while electrode output does not. Axons not loaded with Ca but injected with apyrase and vanadate have mitoehondria that still retain some Ca and this can be released by CN in a truly uniform manner. The results show that such a release (which is small) can be readily measured with aequorin, but again the Ca electrode is insensitive to such [Ca]i change.
Introduction Two recent reports have appeared where the authors measured the [Ca] of squid axoplasm under a variety of experimental conditions. In the first [1], electrodes of a size suitable for the trans* Present address: Centro de Biociencia, Instituto Internacional de Estudios Avanzados (IDEA), Caracas, Venezuela. ** Present address: Department of Biochemistry and Biophysics, University of Pennsylvania, Philadelphia, PA, U.S.A. *** Present address: Department of Physiology, University College, London, U.K. Abbreviation: Tes, 2((2-hydroxy-l,l-bis(hydroxymethyl)ethyl) amino)ethanesulfonic acid. 0167-4889/84/$03.00 © 1984 Elsevier Science Publishers B.V.
verse insertion into squid axons were employed and the authors found a [Ca]i of 79 nM (range 61-95 nM). In the second study [2], electrodes of a tip diameter of 20 /~m were used and the electrodes were inserted axially into squid giant axons. These authors found a mean [Ca]i of 106 nM (range 66-174 nM) or values in reasonable agreement with the study above. There may be some ambiguity as to what the term [Ca]i actually means, first, because electrodes are generally considered to measure ion activity, while optical indicators such as arsenazo III measure concentration. Since [Ca] in cells is so exceedingly low, it is a good approximation to say that it fits the definition of ideality where the activity coefficient of an ion is unity. The fact that cells of
394
marine invertebrates have a solution ionic strength that approaches 400 mM means that only ionic strength considerations apply. Thus, there are grounds for supposing that concentration and Ca 2 + activity are expected to be reasonably concordant. An earlier study where aequorin was confined to a dialysis capillary in squid axons [3] showed that [Ca]i was approx. 20 nM when a freshly isolated squid axon was kept in 3 mM Ca seawater. This value could be raised by increasing [Ca]o or lowering [Na]o. Similarly, aequorin glow could be reduced by lowering [Ca]o. One reason for comparing Ca electrodes with aequorin is that while the former measure Ca in only a limited region, the latter detect Ca throughout the axon. A comparison of these two indicators might, therefore, yield data on the spatial distribution of [Call, as shown for the axon periphery [4]. Since the electrodes we used were inserted axially, it must be expected that the electrode and aequorin measure [Ca]i at different places within a 500/~m axon (see inset to Fig. 1). On the other hand, it is known (Henkart, personal communication) that mitochondria in squid axoplasm are distributed quite uniformly throughout the axoplasm, so that a release of mitochondrial stored Ca by various inhibitors ought to provide a rise in Ca in axoplasm that is uniform unless mitochondria are differentially loaded with Ca depending on their location in axoplasm. We have, therefore, examined axons poisoned with CN for the response of both a Ca electrode and aequorin, and we find that, for axons only lightly loaded with Ca, a release of this Ca does not lead to a change in [Ca] i as measured with the Ca electrode, although aequorin light emission may rise 50-fold, corresponding to a 7-fold increase in [Ca] if aequorin light emission is as the square of [Ca]. A second sort of experiment that we have carried out involves depolarizing the squid giant axon; this is known to increase Ca entry by at least 100-fold [5]. This treatment produces a large increase in aequorin light emission and no change whatsoever in the reading of the Ca electrode. There are thus reasons for questioning the usefulness of the Ca electrode in connection with the measurements that have been made.
Methods Axons The experiments were performed at the Marine Biological Laboratory, Woods Hole, MA, during the months of M a y - J u n e 1983. Axons were dissected from freshly caught living squid and these were cleaned and mounted in a dialysis-type chamber on glass end cannulas as previously described [6]. Microinjection A horizontal microinjector was used to introduce aequorin or other substances that are not permeable. The usual injection distance was 18 mm and the axon was tested for excitability during and after the microinjection procedure. Total aequorin counts injected were (1-3). 109 counts as measured by treating the axon with 10 mM CaC12 solution (in distilled water) at the end of an experiment. Since resting glow is approx. 103 count/s, aequorin is being consumed a t 10-6//S. After microinjection, the axon was impaled with a capillary pipette of diameter 60 /~m filled with 0.6 M KC1 to record membrane potential with reference to an external bath electrode, and at the other end of the axon a Ca-sensitive capillary electrode was introduced. The electrodes were separated at most by 4-8 ram; this separation was convenient for avoiding optical artifacts when external solutions of differing refractive index were used. The general arrangement of electrodes is shown in Fig. 1. The diagram also shows the area of the axon over which light was collected from aequorin for measurement by photon counting. Light measurements Light measurement was as described by Mullins and Requena [7], with the change that the photomultiplier was cooled to - 2 0 ° C by a thermoelectric cooling device so that background counting was reduced several-fold (15 c o u n t / s in the absence of aequorin). Spectrophometric measurements of arsenazo III were as described by Brinley et al. [8]. Electrical measurements The dialysis chamber in which the axon was mounted contained gold electrodes in its base for
395
Eco
Fm
batelhectrode~~~ connu~ °x°n~2C)~ O
axon~ ~
FiberOptics
vv-..~toPhotomuflp ie il r Co ++ee l ctrode
Fig. 1. The experimental arrangement for measuring fight from an aequorin-injected axon and [Ca] from a Ca electrode. The outputs of Eca and Em were subtracted and the resulting signal fed to a chart recorder. In lower left, the location of the Ca electrode relative to the surface membrane is shown.
both the stimulation of the axon and the recording of extracellular action potentials. Some of the experiments to be reported required the stimulation of axons for prolonged periods of time in order to increase internal N a concentration. Membrane potential measurements were made by connecting the internal capillary electrode via high-input-impedance operational amplifier; the other input to the amplifier was a bath reference electrode. The output of this amplifier was designated E m (membrane potential). Ca electrode measurements were made by using an Orion p H meter with an input impedance of 10 a5 ohm to measure the potential difference between the Ca electrode and the bath reference. This output was designated Eca. Outputs from these two measuring amplifiers were subtracted and fed to a chart recorder and to a Hewlett-Packard magnetic tape recorder. This output is designated ( E c a - E m).
Solutions All solutions were fed to the dialysis chamber via a multichannel p u m p that delivered 1 - 3 m l / m i n . This peristaltic p u m p operated continuously while a Hamilton electric valve was used to change the flow of solutions through the chamber. Chamber washout time was about 15 s. The seawater used in these experiments had a
composition as follows: 440 m M NaC1, 10 m M KC1, 10 m M CaC12, 50 m M MgCI2, 10 m M Tris buffer (pH 7.8), and 0.1 m M EDTA. Variations in the composition of this solution were to replace all the N a by either K or Tris (in some cases choline) to produce respectively high-K or Na-free solutions. Ca-free solutions had 1 m M EGTA. Solutions for internal injection were either aequorin at a concentration of 200 ~tM, or apyrase a n d / o r potassium orthovanadate. The aequorin was injected salt-free so that Ca contamination was minimized, while apyrase and vanadate were made up in 330 m M K-Tes buffer (pH 7.3) and were injected at concentrations of 30 units//~l and 30 mM, respectively.
Preparation and calibration of Ca-sensitive electrodes Glass. Capillary tubing (lead glass) of 1 m m o.d., 0.5 m m i.d. was drawn in a vertical puller to a diameter of 90/~m at the tip and the tubing was cut to 12-cm lengths. The wall thickness was about 7/~m. The tubing was dried in an oven and coated outside with solution of high-density polyvinyl chloride dissolved in tetrahydrofurane, thus obtaining a thickness of polyvinyl chloride of about 5 - 1 0 #m. Sensor. The sensor was the neutral ligand (N, N'-di-l,l-(ethoxycarbonyl)undecyl)-N, N'-4,5tetramethyl-3,6-dioxaoctan-l,8-diamide. The sensor was used at a concentration of 10%, in a solution that also contained the following: 3-8% tetraphenyl arsonium tetrakis(p-biphenyl)borate, 15% polyvinyl chloride, 1% sodium tetraphenyl borate, and the balance, o-nitrophenyl octyl ether (NPOE). The individual components were dissolved in 1.5-3-times their weight in tetrahydrofurane depending on the desired viscosity of the solution. The polyvinyl chloride coated tubing described above was dipped at the tip into the sensor solution and then allowed to dry. The thickness of the dry sensor was 2 - 4 / ~ m . Filling. Tubing coated as described above was then selected by size for filling. The range of sizes considered suitable was from 90 to 110 /tin o.d. Filling pipettes for these capillaries were made from 25 /tl D r u m m o n d micropipettes; these were pulled in a pulling machine to diameters such that
396
they would fit into the sensor-coated pipettes. The filling solution was a pCa 2 (sometimes pCa 6) solution as described by Tsien and Rink [9]. Electrical. A 4 cm long Ag/AgC1 wire of 50/~m diameter was introduced into the glass pipette and the wire was attached to a gold pin connector. The average resistance of such electrodes measured in calibrating solutions was 10-30 M~2. Storage. Electrodes were stored dry in test tubes, and electrodes that had lost sensitivity to Ca could sometimes have this restored by a brief dip in dilute sensor solution followed by drying (restoration). Calibration. We have used the buffer compositions described by Tsien and Rink [9] as calibration solutions. These solutions were developed for mammalian ionic strength and [Na]i and hence some modifications were necessary for working with squid giant axons where ionic strength must be expected to be of the order of 400 m M univalent salt, and where [Na] concentration must be expected to be in the range 20-30 m M and Mg is approx. 4 mM. We have used the buffers as described with the addition of either 300 m M K-Tes or 300 m M K-Tes and 20 m M N a or both in order to bring about conditions realistically paralleling those that obtain in the interior of the axon. These additions must be expected to influence the [Ca] of the buffers in two ways: K will inhibit the dissociation of the buffer complex as an ionic strength effect (thus lowering [Ca]), and it will also have a tendency to be measured by the electrode as [Ca] owing to imperfect electrode selectivity. Sodium, present at much lower concentrations, will simply be read as [Ca] because of electrode selectivity. The actual [Ca] of the buffers we used must, therefore, be considered as nominal and not corrected for the effects outlined above. The electrode was first placed in calibrating solution in a glass capillary, and its responses were recorded, and then inserted into the axoplasm and, at times that varied from 1 to 3 h, it was removed from the fiber and returned to a series of calibrating solutions. The principal observation that resulted from this procedure was that the re-calibration of the electrode seldom corresponded with its initial value and that the plot shown in Fig. 2 is representative in that it shows a substantial shift in
the level of electrode output vs. [Ca]. Another feature of the calibration procedure that deserves comment is the temporal stability of the electrode reading. Most reports in the literature show at best a few minutes of recording of the output of a Ca electrode; the lower part of Fig. 2 is of interest, this output being recorded for a period of 2 h. It shows that while there is a relatively prompt response to a change from a pCa 7 to pCa 6, and that this response has a slope that averaged in our experiments 21 mV per decade, there is a much lower response in going from pCa 7 to pCa 8. Moreover, this change is susceptible to a substantial drift, as is the response to a return to
pCa 7. Since most of the values we recorded were greater than a [Ca]i of 100 nM, our calibration procedure was to reference pCa 7 solutions to a potential of zero and to relate changes in electrode potential to this value. Our results for electrode slope, indicated electrode potential, and aequorin count are shown in Table I, Anions. It seemed necessary to verify that a variety of anions known to exist in axoplasm did not affect the reading of the Ca electrode. Experiments were carried out where 10 m M Tris-ATP, as well as the K salts of sulfate and phosphate also at 10 mM, and the exchange of 100 m M KC1 for 100 m M potassium isothionate were all without effect on the reading of the electrode with any of the Tsien-Rink buffers. We conclude, therefore, that anions of the sort likely to be found in axoplasm are without effect on the Ca sensor. Ionic strength. One effect of increased ionic strength is that the ionization of the C a - E G T A complex is affected. DiPolo et al. [2] have attempted to correct for this effect by recalculating a new dissociation constant for the C a - E G T A complex at 300 m M KC1; a second effect of high ionic strength is that K + must be expected to affect the Ca electrode to the extent that they appear to be permeable cations. This effect has not been corrected for and our own measurements show that if one avoids C a - E G T A buffers and works with 10 /~M Ca there is a + 5 - 7 mV displacement of the Ca electrode output in going from 100 to 350 m M KC1 concentration. Sodium. This ion is a poison for Ca electrodes. If a Ca electrode is placed in seawater, it loses
397
(upper trace), where it can be noted that a substantial change in apparent Ca calibration results from a quite short immersion of the electrode in axoplasm.
3020 °iEc°E :'2;t +30
I
f
I0
tO0
IO00
8
7
6
r {EC°-Em) L
+20[- ~ +10
_
mV ~
pCo7
Results
f
IO" [Co] .M 5
pCo
No 30raM K 300 mM
q 4
~sm,~ ~
i0 [ p C e 6 L¢/I I L I I I I I ' ' ' Z5 50
,pCal,
, , J 75
i i i , J
I00
t ,
, l
125
I
i i
i 150
minutes
Fig. 2. The upper panel shows the Ca electrode response to an initial calibration (e) and a second calibration upon the termination of the experiment some 3 h later (©). Note both the shift in potential and the change in slope. The calibrating buffers were of the Tsien-Rink sort with the addition of 300 m M KCI, 20 m M Na and 5 m M MgCI 2. Electrode No. 18. In the lower panel is a record of the output of a Ca electrode in various buffer mixtures over 150 rain. Note both the fast and slow phases of electrode response, and the electrode drift between 75 and 150 min. Electrode No. 13 bis.
sensitivity to Ca totally and on a time-scale of 12 h irreversibly. This makes it quite important that the glass cannulas to which the axon is tied are carefully rinsed with K-Tes solution and that this be changed before axial Ca electrodes are inserted into axoplasm. Additionally, some correction needs to be made for the [Na] of axoplasm, which is about 20 raM. Our measurements show that adding Na to Tsien-Rink calibrating solutions both reduces the slope of the electrode response in the range pCa 7-8 and that it also displaces the level of output of the electrode• This finding confirms that of DiPolo et al. [2]. Electrode stability. We agree with the experience of Tsien and Rink that one of the difficulties with Ca electrodes is that they sometimes change their calibration with time. Thus an initial calibration, followed by the insertion of the electrode into the cell and then a final calibration may result in measurements that differ by 5-10 mV even though the slopes are the same. This is not a problem of electronics (these are stable at the level of 1 mV for several hours) but a problem of the electrode output. A representative result is shown in Fig. 2
Our results comparing the initial aequorin photon count with the reading of a Ca electrode in the same axon are shown in Table I. The mean [Ca]i was 210 riM, which compares with values in the range of 100 nM as reported by others. It is, however, a value 10-times higher than the value reported using aequorin [3]. We have included some axons with initial aequorin counts of (16-30) • 1 0 3 count/s, which are clearly leaky axons, to demonstrate that, at least in one of these cases, the Ca electrode read only 300 nM. Also included in Table I are a phenol red-injected axon where light is only measured from the periphery of the axon [4], and results with two axons where C a - E G T A / E G T A buffers were injected to clamp the ionized Ca of axoplasm at pCa 7. The final buffer concentration is calculated to be 10 mM yet the Ca electrode gave values of 360 and 1350 riM, respectively. Another series of experiments involved the addition of orthovanadate and the enzyme apyrase to axoplasm to prevent contributions from ATP-dependent systems from influencing the axon; in these cases the Ca electrode read 100 nM in all three axons studied, although one can calculate that normal Ca influx should increase the ionized Ca many-fold over times of the order of 30 rain. Since, as indicated in the Methods section, the Ca electrode was expected to sample [Ca]i from a region of axoplasm that was removed from the periphery, it was of interest to compare the response of aequorin and a Ca electrode to a highly local Ca entry in an axon where depolarization produced a large Ca entry. The result, shown in Fig. 3, is that there was the usual aequorin response of an axon previously loaded with Na by stimulation, in that there was a large increase in the aequorin light emission and that this increase was continuous over a period of more than 25 min. By contrast, the Ca electrode showed no response whatsoever. Now, experiments with a dialysis
398
TABLE I I N I T I A L Ca E L E C T R O D E A N D A E Q U O R I N
READINGS IN AXONS
Values in parentheses have been omitted from the means. Axon
Em
Initial
Calcium electrode
Ref.
(mV)
aequorin count/s
Reading, pCa 7 = 0
Slope, mV/pCa
(mV)
(7 ---, 6)
[Ca] i (nM)
[Ca]o (mM)
260583B
- 57
800
+ 5.0
20.5
175
310583A
- 60
600
+ 3.0
21.0
130
10
300583A
- 57
600
+ 5.5
23.0
173
10
240583A
- 52
240583B 220583
4 000
120
10
+ 2
26.0
(30000) (16 000)
+ 12 + 25
25.0 16.0
302 (3 000)
10 3
3 _ a _ a
(25 000)
+ 22 + 28 26
21.0 23.0 22.0
(1000) (1 600)
10 10 3
_ a _ a
351 (1000) (1000)
10 10 10
190583 230583A 250583B
- 51 - 57
600
310583B 200583A 180583A
- 65 - 57 - 55
4400 1100 1 800
12 21,5 17
22.0 20.0 17.0
290583A 010683A
- 53 - 53
2000 4 500
8 2
19.5 19.0
257 127
10 10
120583A
- 56
1000
8
20.0
251
10
21.0
210
Mean
Notes
1950
_
a
_ a - "
- "
Phenol red axon pCa 7 inj. pCa 7 inj.
020683B
- 55
70
7
17.0
258
3
230583B 230583A 270583
- 51 - 51 - 64
1 400
5 26 0
9.0 23.0 23.0
360 1 350 100
10 10 10
280583A 270583B
- 56 - 57
21 000 16 500
0 0
15.0 23.0
100
10
Apy + VO4
100
10
Apy + VO4
Apy
+
go 4 b
a Calibrated in Na-free Ca buffers. b A p y r a s e 1 unit//tl and vanadate 1 m M final concentration.
capillary containing aequorin but located in about the same place as the Ca electrode [3] showed that a large increase in Ca entry was sensed by aequorin in the center of the axon with a time lag of at best a few minutes, so that there is some reason for doubting that the Ca electrode is in fact measuring the [Ca] as calibrations would suggest that it ought. The axon shown in Fig. 3 was subsequently subjected to a variety of procedures that led to increases in the aequorin light emission, so that from the initial photon count of 103/s emission rose to 105/s (a 100-fold change) with no change in the Ca-sensitive electrode reading. The plot on the right hand side of Fig. 3 is one of Ca electrode reading vs. photon count as indicated by aequorin. To avoid the conclusion that some inadequacy in the Ca electrode itself is responsible for the
results shown in Fig. 3, conditions were arranged whereby the release of Ca was, insofar as we can arrange it, uniform throughout the axoplasm. This was achieved by previously loading the axon with Ca by stimulation and then by depolarizing and applying CN to block electron transport in mitochondria. The results of this experiment were that there was a large increase in both the aequorin light emission and in the reading of the Ca electrode. The plot in the right-hand half of Fig. 3 shows that sensitivity of the Ca electrode began at a count (for aequorin) of about 1 5 0 0 0 0 / s (1.5. 1 0 - 4 is the fraction of aequorin consumed per second). A conclusion is that the Ca electrode does respond to [Ca] in axoplasm but only when aequorin counts are 30-100-times the resting level of light emission.
399
[Ca] i M
count/s X 103
1°"5-- by electrode
I0 450 K.
ECa" Em mV
/
+20
id e
IOCa. /equorin 6
+10 16r
Ca electrode
4
I
i(~el I
4O
i
iOs
I
I
50 60 minutes
I
i
i
i
i i iI
i
I
i
i
i
i i i I
i04 105 Aequorin count/s
i
i
i
i
i i i i i
10 6
I
70
Fig. 3. This axon was given 75000 impulses to raise [Na]i and was then depolarized as indicated with 450 mM K. The aequorin light emission rose from 800/s to 9700/s without any change in the output of the Ca electrode. The electrode had a slope of 22 mV between pCa 7 and 6 and it indicated a steady 160 nM. On the right, a comparison is made between aequorin light emission and electrode indication over the entire range of light emission obtained in the experiment. The application of CN to the axon led, with a delay, to a count that increased to the level of 500000/s; the Ca electrode responded when the count reached 200000 count/s.
Since the results in Fig. 3 indicated that we could only expect Ca electrode responses if aequorin counts were very high, an axon was deliberately highly loaded with Ca by stimulating it in 10 mM Ca seawater at a frequency of 6 0 / s until 145000 impulses had been delivered. The axon was then treated with a solution of 10 mM C a / 4 5 0 mM K / 1 mM cyanide. The result of this treatment (shown in Fig. 4) was a rapid change in aequorin glow (barely noticeable at the low gain used) from 2500 to 50000 c o u n t / s and then the Ca electrode began to respond and the two indicators rose together in the range (50-600)-103 aequorin count and 110 nM to 4500 nM as indicated by the Ca electrode. Upon removal of CN but keeping the axon depolarized with high potassium, both responses fell, with the aequorin response returning to its initial value before that of the Ca electrode. Ultimately, both showed a value
close to that at the start of the experiment. It is useful to note that the aequorin light had fallen to virtually its initial value before the Ca electrode began its decline. The Ca electrode continued falling for 20 min after the aequorin light had reached a steady value and there is the suggestion that the electrode response would continue to fall. This cannot be due to surface pumping by the N a / C a exchange since this was inhibited by the application of a high K solution. When the results for this experiment are plotted (as in Fig. 3) as electrode reading vs. aequorin count over all values reached during the experiment we have the result shown in the lower panel of Fig. 4. Again, electrode sensitivity would appear to start around l0 s aequorin counts. It is of interest to note that there is a region of the curve at a count of around 105 that is linear between Ca and photons, and a region that has a slope of about 2.5
400 mM
¢N ] [ Co
,
I
1o
- - ]
450
ECo -Em(mV)- 45
count/$
40
X I06 1,0
3O 25
0.8 0.6
15 I0 5 0
0.4 0.2 0
75
I00
125
minutes
[Co]i from Electrode M id 6
J
j67
i68
h
103
i
1
[ JiiiiI 10 4
i
i
LIJlII 10 5
J
,
i iJllll 10 6
Aequorin counts/s Fig. 4. This axon had been previously stimulated (145000 impulses) and in this experiment CN and depolarization were applied as indicated. The initial [Ca]i as indicated by the electrode was 110 nM and it rose to 4500 nM at a peak and then fell as CN was removed. The lower chart shows that a sensitivity of the electrode began at about 50000 aequorin counts. Initial aequorin count was 2500/s, while final count (in 450 mM K) was 12300/s.
between 3 • 105 and 6.105 count/s. The results in Fig. 4 are not a peculiarity of aequorin nor of a depolarized fiber. Fig. 5 shows data from a similar experiment where arsenazo III was used to measure Ca i. Increasing external Ca from 3 to 112 mM produced a measurable increase of the arsenazo III signal but no deflection in the Ca electrode trace, However, when CN is added (in 112 m M Cao) both the electrode and the arsenazo III responded (Fig. 5B). Interestingly,
there is a hysteresis between arsenazo III and Ca-electrode response in the same direction as that shown in Fig. 4 between aequorin glow and electrode response. Since the experiments indicated by Figs. 3, 4 and 5 would suggest that there is a limit to the sensitivity of a Ca electrode, even if Ca release is brought about by using mitochondria as the means to effect a rise in [Ca]i, it seemed important to look at what might happen if a mitochondrial release were made more modest and uniform than that in Fig. 3. A second feature is to make ATPdependent buffering as small as possible. Accordingly, a fresh (unstimulated) axon was injected with apyrase + orthovanadate in order both that ATP be made as low as possible and that phosphorylation of membrane proteins be prevented. Such an axon cannot be expected to rely on ATPdependent membrane transport for its Ca flux balance, yet it is clear from Fig. 6 that for 70 min the axon is capable of maintaining a normal and low aequorin light emission and that this emission recovers from a depolarization without the intervention of ATP-dependent processes. The Ca electrode is insensitive to any of the changes measured; the small deflections shown on the electrode trace are in fact junction potential changes in going from Na to Tris in seawater and the result of impedance imbalance between the Ca electrode and its reference. More importantly, when CN is applied to the axon (note change of photon-count scale) and the solution made Ca-free then, on the right hand panel, it is clear that the photon count rises about 20-fold with no real change in the Ca electrode indication. Moreover, the experiment shows that in an axon with total metabolic inhibition so that ATP-dependent mechanisms are suppressed, there is still the ability of the N a / C a extrusion system to reduce [Cain. This occurs after almost 3 h of total inhibition of substrate-driven pumps. In an effort to see something of the time course by which the Ca electrode senses changes in [Ca]i, an axon was kept in a solution that was 450 m M K C I / 1 0 m M CaC12 and had both CN an iodoacetamide. In this solution the Ca electrode in the axon showed a reading of close to 1/~M Ca. Upon changing the solution bathing the axon to one that was CN-free, the reading of the electrode, after a
401 CN
112Ca 3Ca
3Ca
0.02 AA 20 10
[
i lOrein
Fig. 5. On the left, an axon injected with arsenazo III is exposed to high and low Ca seawater and, while the metalchromatic indicator optical signal (upper trace) changed appreciably, the Ca electrode output was steady throughout. On the right hand side, the same axon was subsequently exposed to high Ca o (112 raM) seawater in CN. Notice the change in absorbance scale. The Ca electrode indicated a nominal [Ca]i of 100 nM at the start and this increased to 1.9 ~ M in CN (slope 22 mV between pCa 6 and 7).
Apyrase + Vanadate Injected Axon CN Ca K Na
I
mM I0 '°
440
0
450 I 0 J44(
I0 0 (Choline)
440
J
count/s
count/s x 103
E Co - E re ( m Y )
×,o3
2O
440
1÷,5
I00
1+10 +5
16
12
60 40
O-
./i
i
25
I
I I 35
I I 45
I
i 55
i
i
65
i
1
0
.
J
8.5
a
l
95
J
|
105
l
i
115
i
J
125
i
=
1:55
|
i
145
,
a
155
minutes Fig. 6. An axon injected with 10 U/t~I a p y r a s e + 1 m M vanadate (final concentrations) is capable of maintaining a normal aequorin resting glow for 75 min. The Ca electrode does not sense an increase in aequorin from 2000 to 60000 c o u n t / s . Since Cao = 0 all this increase in Ca is from internal release.
402 CN [,IAA l
Cao I0 Ko 4 5 0 mM
l
.,..~ .
[Ca1i
]1
by
electrode
M ~ ' ~ - -
I0-6
450 K
.,4~ ~ Noo440 /<'l~ CNI N% 440 ~ ~ . ~ CN- f r e e ~ . ~
"' 440 ~°_ free j, T 10.7
[
0
I
I
I
I0
I
20
L
I
30
minutes Fig. 7. An axon previously loaded with Ca and kept in 450 mM K, 10 mM Ca solution containing 1 mM each of CN and iodoacetamide was transferred at the point indicated by 'change solution' to several different solutions and after [Ca]i as indicated by the electrode had reached a steady level, the axon was returned to the original solution and the change to another solution carried out. The results show that the initial rate of decline of [Ca] is about the same whether the axon is in a CN-containing or in a CN-free solution that contains Na. Repolarization is not necessary to bring Ca down to low levels. All recovery solutions contained 10 mM Ca.
delay of 1.6 min for the axoplasmic CN to wash out, fell promptly to a value of about 0.1 #M, as shown in Fig. 7. An interesting feature of the decline is that it is not exponential as are aequorin records of light emission when CN is removed. Instead, the record appears to be truncated at 0.1 /~M as if the electrode had lost sensitivity at this point. The axon was then returned to the initial high-K, CN solution and the electrode returned to close to its initial reading. The solution was now changed to one that was both CN-free and Nacontaining. The decline in Ca concentration with time as indicated by the electrode now took place with less delay (the initial deflection upon solution change is a junction potential effect when the bath solution is changed from K to Na). The conclusion from this experiment is that the Ca electrode is apparently able to sense a change in [Ca] as a result of N a / C a exchange at the surface, since the
only difference between the two solutions tested is that the second contained Na. As a check on this interpretation, the axon was once again returned to its loading solution with CN and, when [Ca]i as indicated by the electrode was stable, the solution was changed to one containing Na and CN. Now the only known mechanism to bring down the Ca in the fiber is surface Ca pumping by N a / C a exchange and in this experiment it would appear that Ca can be brought down to half its initial concentration in a time of the order of 1 min. Another feature of note in this recording is that the time course of decline of [Ca] i has much less of the appearance of being truncated; or that it would appear that the Ca electrode is continuing to measure Ca in axoplasm. A final step, the removal of CN, led only to a small decrease in the indicated level of Ca in axoplasm, and this had the appearance of being truncated. The results shown above are important because they suggest that there is only a small delay before a Ca electrode senses a decrease in [Ca]~ as a result of surface pumping of Ca, and that the electrode functions well in detecting decreases in [Ca]i in the range of from 0.2 to 1.0 /~M. The failure of the electrode to sense increases in [Ca]~ must then mean that the increases such as those shown in Fig. 3 are from a resting level of [Ca]i that is much less than 0.1/sM.
Discussion The present measurements show that a Ca electrode in the middle of the axon is insensitive to changes in Ca entry over times of the order of 1 h. In contrast, the measurements by DiPolo et al. [3], where aequorin was confined in a dialysis capillary in the center of a squid giant axon, showed that the light signal was sensitive to Ca entry with a delay of perhaps 15 min when conditions at the surface of the axon were changed so that Ca entry was either enhanced or diminished. The lack of sensitivity of the Ca electrode cannot, therefore, be attributed to its position and we conclude that it is incapable of measuring [Ca] below a level of 100-150 nM and the [Ca] as it exists in axoplasm is very much lower than this value. The question must then arise: why is it possible to show a response to pCa as low as 8 when the
403
electrode in fact cannot sense this [Ca] in axoplasm? One suggestion is that it is impossible to calibrate electrodes except in C a - E G T A / E G T A buffer mixtures, while such buffers do not exist in axoplasm. One might then imagine that the flux of Ca 2+ to the membrane of the Ca electrode is insufficient to bring about an equilibration of the electrode to [Ca] in the range of 10 nM, while it still has acceptable responses in the 1000 nM range. If this is indeed the explanation for the findings that we report, it would have important applications for electrodes that must measure ion concentrations at very low levels. It would also force a re-examination of claims for measured levels of [Ca]i in many sorts of cells. A second possibility is that, since Ca electrodes are filled with a solution containing Ca, the leak outward of this substance may prevent a proper reading of the electrode in solutions where a Ca buffer is absent and where Ca concentrations are in the range of tens of nanomolar. The proper performance of electrodes under these conditions would require that the solution outside flow quite fast past the electrode surface membrane and this is, of course, not possible with intracellular measurements. It is important to note that the present experiments show that the sensing of [Ca]i by the electrode begins at levels of aequorin count that are about 30-times that of resting glow levels and that these electrode measurements yield an average value of [Ca]i of 200 nM. If aequorin were in its linear range of measurement of cell Ca, then actual [Ca]i in axoplasm would be 200 n M / 3 0 or about 7 nM. If aequorin were responding as the square root, then [Ca]i would be 2 0 0 / 3 ~ = 36 nM. Our previous report [3] was that the value was 20 nM and, allowing for the uncertainties of this measurement, it would appear that a proper value is one that is very much lower than a Ca electrode measures. The experiment shown in Fig. 7 yields useful information on the rate by which [Ca]i declines when external Na is supplied to the axon but CN is maintained in the bathing solutions. Under these conditions one supposes that the decline in [Ca]i is brought about by the reactivation of N a / C a exchange and the rate by which Ca declines is related to the efflux level of Ca at the appropriate level of
[Ca]i and to the buffering that internal Ca undergoes. Thus, the initial slope of the curve for 440 mM Na o, 1 mM CN is 0.15 ~tmol/liter per min and if there were no Ca buffering this would correspond to a Ca efflux of 50 f m o l / c m 2- s. Since Brinley et al. [8] have shown that in the range of [Ca]i around 1 /~M only 5% of the entering Ca goes to increase [Ca]i, it follows that the appropriate Ca flux to bring down the Ca concentration in the fiber is 50 fmol × 20 or a Ca efflux of 1 p m o l / c m 2. s. This value is in reasonable agreement with flux estimates of Ca efflux as obtained in dialyzed axons. A second observation from Fig. 7 is that if CN is removed but a high-K external solution is retained, then one expects that the decline in [Ca]i is the result of Ca sequestration by the mitochondria. The internal rate of decline for such a curve (450 mM K, CN-free) is the same as the initial slope of the curve described in the paragraph above; hence, this finding suggests that the rate by which Ca can be moved from axoplasm to mitochondria and the rate by which Ca can be moved out via N a / C a exchange in the surface membrane are of a comparable magnitude. When [Ca]i is much lower (see the 440 mM N a o / 1 mM CN 1 curve at about 24 min) and of the order of 150 nM, the removal of CN results in an indicated decline of Ca i of about 1 / 1 0 that of the initial slopes of the curves at 1 /LM, corresponding to measured fluxes at this level of [Ca]i.
Acknowledgements This work was aided by grants from the National Institutes of Health (1R01 NS 17718) from the National Science Foundation (BNS-8006271) and from CONICIT S1-1198. The aequorin used in these studies was kindly provided by Dr. John Blinks. We are indebted to the Director and staff of the Marine Biological Laboratory, Woods Hole, Massachusetts for facilities put at our disposal. D.E. was supported by a travel grant from the Wellcome Foundation.
References 1 Baker, P.F. and Umbach, J.A. (1983) J. Physiol. 341, 61 P 2 DiPolo, R., Rojas, H., Vergara, J., Lopez, R. and Caputo, C. (1983) Biochim. Biophys. Acta 738, 311-318
404 3 DiPolo, R., Requena, J., Brinley, F.J., Jr., Mullins, L.J., Scarpa, A. and Tiffert, T. (1976) J. Gen. Physiol. 67, 433-467 4 Mullins, L.J. and Requena, J. (1979) J. Gen. Physiol. 74, 393-413 5 Baker, P.F., Hodgkin, A.L. and Ridgway, E.B. (1971) J. Physiol. 218, 709-755 6 Requena, J., DiPolo, R., Brinley, F.J., Jr. and Mullins, L.J. (1977) J. Gen. Physiol. 70, 329-353
7 Mullins, L.J. and Requena, J. (1981) J. Gen. Physiol. 78, 683-700 8 Brinley, F.J., Jr., Tiffert, T., Scarpa, A. and Mullins, L.J. (1977) J. Gen. Physiol. 70, 355-384 9 Tsien, R.Y. and Rink, T.J. (1981) J. Neurosci. Meth. 4, 73-86