9E-V) curves are shifted along the voltage axis when the external calcium. squid giant nerve fibre shows a sequence of specific changes in sodium and

Transcription

1 76 J. Physiol. (I958) 143, MEMBRANE CURRENTS IN ISOLATED FROG NERVE FIBRE UNDER VOLTAGE CLAMP CONDITIONS By F. A. DODGE* AND B. FRANKENHAEUSER From the Nobel Institute for Neurophysiology, Karolinska Institutet, Stockholm 60 (Received 14 March 1958) With the voltage clamp technique it has been found that the membrane of the squid giant nerve fibre shows a sequence of specific changes in sodium and potassium conductances when the membrane is depolarized (Hodgkin & Huxley, 1952a-d; Hodgkin, Huxley & Katz, 1952). A step depolarization is associated with a rapid transient increase in sodium conductance followed by a slower but lasting increase in potassium conductance. The magnitudes and the rates of change of these permeability changes vary continuously with the membrane potential so that they are larger and more rapid at large cathodal polarizations than at small. The mechanism underlying these permeability changes is mainly unknown, but in further experiments with the voltage clamp technique it was found that the conductance-membrane potential (gns-v and 9E-V) curves are shifted along the voltage axis when the external calcium concentration is altered (Frankenhaeuser & Hodgkin, 1957). It is not clear to what extent this analysis may be applied to the myelinated nerve fibre. The resting potential is mainly determined by the external potassium concentration, whereas the action potential changes with the external sodium concentration, as it would if the membrane were mainly permeable to sodium at the peak of the action potential (Huxley & Staimpfli, 1951). The effect of changing the external calcium concentration is very much that which is expected on the basis of the squid experiments (Frankenhaeuser, 1957 b). This supports the view that specifit changes in sodium and potassium conductances are the basis for activity in myelinated nerve as they are in squid nerve. On the other hand, from experiments where an action potential is interrupted with an anodal pulse, Tasaki (1956) concludes: 'An unsuccessfu-l attempt was made to interpret these experimental results in terms of the sodium theory of nerve excitation', referring to the squid voltage clamp analysis. * Graduate fellow of the Rockefeller Institute for Medical Research, New York.

2 VOLTAGE CLAMP ON MYELINATED NERVE 77 Previously, from measurements of the membrane resistance during the action potential, Tasaki & Freygang (1955) concluded that the myelinated nerve does not undergo a change in potassium conductance during the action potential. The voltage clamp technique is, however, to be preferred for resolving the changes in the specific ionic conductances of the membrane, associated with depolarizations. Proper stabilization of the membrane potential in voltage clamp experiments requires a highly reliable potential-measuring system. A system for the myelinated nerve fibre has been developed using negative feedback to meet such requirements (Frankenhaeuser, 1957 a) and incorporated into a voltage clamp arrangement (for preliminary description, see Frankenhaeuser & Persson, 1957). A different method for clamping the nodal membrane potential has recently been described (del Castillo, Lettvin, McCulloch & Pitts, 1957). In this method the membrane potential is measured as the longitudinal current when the external impedance is made high in an air gap (cf. Tasaki & Frank, 1955). The method of Frankenhaeuser & Persson, involving two feedback amplifiers, has been used with some modifications in the present experiments. The first amplifier, with negative series feedback, provided an instrument with high input impedance for recording changes in membrane potential, whereas the second, with negative parallel feedback, stabilized or clamped the nodal membrane potential. The method, its limitations, and the ionic currents associated with step polarizations, are described in this paper. The quantitative analysis of these phenomena will be dealt with in a subsequent paper. The nomenclature introduced by Hodgkin et al. (1952) for the experiments on the squid fibre is followed as far as possible, with the exception that the signs of potentials are taken in the sense of axis cylinder potential minus potential of outside medium. The action potential is therefore a change in positive direction, outward membrane current is consequently given as positive. METHODS Preparation. The experiments were done on large single myelinated fibres dissected from the sciatic nerve of the frog (Rana esculenta and R. temporaria). Most of the experiments were carried out at about + 30 C. Solution&. The Ringer's solution used had the following composition (mm): NaCl 111-3, KCI 2-5, CaCl2 2-0, NaHCO3 2-5 Voltage clamp. Two feedback amplifiers, A1 and A2, were connected to the fibre as shown in Fig. 1. Since changes in the potential across the membrane of node No, as measured with amplifier system A1, appeared as changes in the potential difference between pools B and A, the input of amplifier A2 was effectively connected across the membrane of node No (Frankenhaeuser, 1957 a). Part of the current from the output of A2 flowed through the membrane of node No via the internode N-1 - No the remainder passing through the seal EA and having no important effect on the system. Since both the input and the output of A2 were effectively across the membrane of node No and since the feedback was negative, A2 worked as a voltage stabilizer, clamping the membrane potential to a value determined by a reference potential modified to undergo rectangular steps. This system was thus equivalent to that used in voltage clamp experiments on the squid axon.

3 78 F. A. DODGE AND B. FRANKENHAEUSER Both feedback amplifiers were similar in design, two-stage differential amplifiers with singlesided output at mean level of earth potential and with adequate balance controls. The voltage gain was somewhat more than 1500 and constant from d.c. to more than 200 kc/s. Great care was taken to avoid drift, especially in the input stages of amplifier A1. All power supplies including a d.c. heater supply were well stabilized, wire-wound resistors used, the input valves were selected for minimum drift, and the grids of the input cathode followers were grounded when not connected to the preparation. Amplifier As was biased with external rectangular pulses to change the membrane potential in sudden steps as required. Out In Out I E A B C Im Fig. 1. Arrangement for clamping the nodal membrane potential. N-1, No and N+1 nodes of Ranvier. A, B, C and E pools with Ringer's solution. Electrodes, cathode followers, inputs and outputs are denoted in the text with the letter of the pool to which they are connected. Shaded regions are petroleum jelly seals around the fibre. A1 feedback amplifier for measuring membrane potential of node No. As feedback amplifier for stabilizing membrane potential of node No. Recording instruments for recording membrane potential (Vm) and membrane current (Im) are indicated. Seals EA, AB and BC 150,u each, pool A 200 t and pool B 230ju. Electrodes. Silver-silver chloride electrodes, used in the earlier experiments (Frankenhaeuser & Persson, 1957), were, unfortunately, rather unstable. Drift in the recording electrodes in pools B and C (Fig. 1) necessitated rebalancing amplifier A1 often (small errors, i.e. 1-2 mv, had an appreciable effect on the magnitude of the inward currents during a clamp). Polarization of the current-carrying electrodes in pools E, A, and B resulted in erroneous and variable balance settings. More elaborate calomel half-cells made with 3 M-KCI and with a contact surface area between the mercury and the calomel of about 25 mm2, showed no appreciable polarization (<1 mv) for currents about 100 times larger than the currents in the experiments and these were also found to be sufficiently drift free. The calomel cells were connected to the recording cell by bridges, made of glass capillaries filled with 3 m-kcl in a kaolin suspension, closed with plugs of balsa wood. Recording electrodes and current-carrying electrodes were separated to eliminate from the recording systems the small errors due to a potential drop across the current-carrying electrodes. The current electrode in pool B had to be of rather low resistance (less than about loooq), otherwise amplifier Al went into oscillations at a frequency of approximately 1 Mc/s due to the Vm

4 VOLTAGE CLAMP ON MYELINATED NERVE 79 small stray capacitance between pools A and B. The other electrodes were not so critical in this respect. Recording cell. A Perspex cell with four pools separated by three partitions was used. The fibre was mounted in the cell so that the node under investigation was in pool A. Petroleum jelly seals were applied to the fibre with a syringe at the three partitions. In preliminary experiments it was found that the dimensions of the cell are rather critical in two respects: (a) wide pools and wide seals limit the high frequency gain which can be used in amplifier A,, because the phase lag introduced by the fibre increases with the distance between seals EA and BC; (b) a very narrow pool B results in changes in membrane potential being recorded with attenuation (see p. 81). The dimensions of the cells used, which are given in the legend of Fig. 1, were chosen as a compromise between these two effects. Electrical shielding. Although the recording cell and the electrodes were enclosed in a brass block through which cold water was circulated, some additional electrical shielding was necessary, since serious discrepancies between true membrane potential and recorded membrane potential resulted when shielding was insufficient. Pool C and electrode C were provided with a separate shield connected to the cathode of the cathode follower C, in order to reduce the stray capacity to ground. Other electrodes were surrounded by grounded brass plates. The efficiency of shielding was tested with the recording cell placed as in the experiments, but the fibre not extending to pool E. When a rectangular pulse of 10 V amplitude applied to pool E evoked a transient output signal of less than 20 mv in amplifier A,, screening was considered satisfactory. Recording of membrane current. If no current flows through the fibre between D and C (D is a point in the axis cylinder at No), the change in membrane current per square centimetre membrane associated with a step depolarization is: Im=AVZD, ANED (1) where VED is the change in potential difference between pool E and point D in the axis cylinder at node No (see Fig. 1); ZED is the impedance from pool E to the axis cylinder at node N.; and Ax is the surface area of node No. The feedback amplifier A1 (Fig. 1) prevented any appreciable flow of current in the internode No - N+1, whereas the feedback amplifier A, supplied the current required to stabilize the potential across the membrane of node No, that is, the membrane current associated with step depolarizations flowed almost entirely through the internode N-1 - No. It was previously shown (Franken. haeuser, 1957a) that the axis cylinder close to node No is held by amplifier A1 at a potential nearly constant relative to ground; changes in membrane potential therefore appear as changes of the potential of the external pool A relative to the grounded reference electrode in pool B. VED is therefore almost equal to VEB which was recorded. The membrane current could therefore be calculated from the potential VEB and the impedance ZED. This impedance could not be measured in the experiments. It is quite clear that any 'activity ' or rectification in node N-1would make it difficult to calculate the membrane currents. Since the membrane currents from node No flow through node N-1 and are of a magnitude far exceeding the threshold of node N-1, activity in node N-1 was prevented by replacing the Ringer's solution in pool E either with cocaine Ringer's solution or with isotonic KCI solution (change in membrane potential compensated). A serious drawback of the cocaine method was that the membrane currents developed potentials across the node N-,, which were sufficient to damage the membrane. This did not happen when potassium chloride was used, since the node in KCI solution had a much lower resistance. The current path was thus made a 'passive' network which was mainly resistive, axis cylinder and membrane, but partly capacitive, membrane and myelin sheath. The effect of the capacitive part of the current path was to attenuate somewhat the high frequency component of the response, but this seemed inappreciable for the recording of the ionic currents. The membrane currents were linearly proportional to the recorded potential VEB. To calculate a scale for the current measurements according to equation 1, a value for the product ANZzD Was required. Neither of these factors was measured. When KCI was applied to node N-1 it was

5 80 F. A. DODGE AND B. FRANKENHAEUSER reasonable to assume that the membrane resistance of node N-1 was negligible in comparison with the internodal resistance of the axis cylinder. For these cases a value of 10 Qcm2 was somewhat arbitrarily chosen for the product ANZED, on the basis of probable values (fibre diameter 10-15,u, nodal gap 05-1/l, internodal resistance MQl, cf. Tasaki, 1955; Stampfli, 1952). A considerable error in the scale for the membrane currents is likely to be introduced by the factor 10, the order of magnitude ought, however, to be correct. A relative scale for the membrane currents is sufficient for the major part of the present analysis. Procedure to balance amplifiers. The feedback amplifiers were balanced in either of the following manners. (a) When pool E contained cocaine Ringer's solution, and pool C Ringer's solution. Amplifier Al was balanced so that the potential VBA was zero. The output of amplifier A2 was connected to the preparation and balanced so that the potential VBA remained zero. (b) When pool E and pool C contained isotonic KOl solution. Amplifier A1 was balanced so that the potential VBA was - 70 mv, amplifier A2 so that the potential VBA remained - 70 mv when the amplifier was connected to the preparation. The assumptions underlying these procedures were that the resting membrane potential was the same in the nodes in Ringer's solution and in cocaine Ringer's solution, and that a node was depolarized 70 mv when the outside solution was isotonic KCI. The balance condition chosen to represent the 'resting potential' was evidently influenced by these approximations. In case a the conditions of nodes N-1 and N+1 were of importance, but not in case b. Procedure b was used in most of the experiments. The most correct ways of balancing would be to balance A1 to zero output when feedback loop is open and pool C is short-circuited to B, or to balance A1 to zero potential VBC. Both these methods were found impracticable since they complicated the experiments too much. Preliminary experiments The voltage clamp technique here described was developed for quantitative experiments. A number of technical and experimental difficulties arose, oscillatory behaviour, attenuation in the voltage recording system, electrode and amplifier drift, etc. A large number of tests on a nerve model and on about 200 isolated fibres were performed while the technique was being developed. The aim of these experiments was to find the best method of stabilizing the membrane potential and to determine the reliability of the technique. Speed of stabilization. It was found that a more rapid stabilization of the membrane potential could be achieved by narrowing the seals and the pools in the recording cell (see Fig. 1). 'Ringing' at the step was of higher frequency in experiments with the narrow cell, and the membrane currents affected the membrane potential less. Currents were clearly regenerative when a wide cell was used. This was most pronounced at steps of mv, which is the region where the peak current-membrane potential curve was steepest and, therefore, the region where the tendency of regeneration was largest. From a number of experiments with recording cells of various dimensions it was concluded that a recording cell with very short pools and seals would permit most rapid stabilization. There was, however, a limit, partly because it was difficult in practice to manipulate the seals in the very narrow cells and partly because the membrane potential was recorded with some attenuation when the cell was too narrow (see p. 81). The phase lag limiting the maximum useable gain in amplifier A2 was determined partly by the cable-like properties of the fibre and partly by the amplifier systems. Pools E, A and B (Fig. 1) were all connected to low impedances, so stray capacitances from these to ground were insignificant. The input C was sensitive to stray capacitances to A and E, but somewhat less sensitive to a stray capacity to ground (see Frankenhaeuser, 1957 a). A grounded shield round the large calomel cell at C was found to introduce too much stray capacity; pool C and electrode C were therefore screened separately with a shield connected to cathode C. With these precautions the performance of the system was probably limited by the fibre and not by residual inadequacies in the amplifiers.

6 VOLTAGE CLAMP ON MYELINATED NERVE 81 Attenuation in voltage recording 8y8tem. When a cell with a very narrow pool B (Fig. 1) was used, it was observed that an action potential was elicited in node N+1 at large depolarizations (about mv) of node No. A related effect is that when pool B was very narrow, the recorded action potential was mv, instead of mv as with a wide pool. A flow of current through the internode No to N+1 and through the node N+1 was of necessity required to excite node N+1. The only return path available for such a current was through seal BC, where this current must cause a potential drop across the seal resistance, which was about 5 MQ. As a direct consequence of current through seal BC the recorded potential change VBA must differ from the membrane potential by an amount determined by the current and by the impedance from D to pool C. A current through seal BC of sufficient magnitude could possibly arise in the following ways: (a) amplification in amplifier A1 was not sufficient to prevent flow of longitudinal current; (b) pool B was not sufficiently equipotential, the current through seal AB developing a potential in the solution between the recording electrode in B and the point where the fibre entered seal BC in pool B; and (c) current spread in a conducting path between the myelin and a resistive layer surrounding the fibre. That condition a contributed little to the current was clear from the observations that the pulse amplitude (VBA) necessary to excite node N+1 was virtually independent of the amplification in Al (i.e. the ratio VBA: VBC) over the range , while the potential VBC was decreased by a factor of 10. Correspondingly, the amplitude of the recorded action potential decreased only a few per cent, when the amplification was decreased from 1500 to 100. A current through seal BC which was independent of the amplification in A1 was therefore required to explain the artifacts. In conditions b and c the potential outside the myelin in pool B at seal BC would be slightly different from the potential of the measuring electrode in B. Although approximate calculations suggested that condition b was unlikely to account for more than a few per cent of the potential (with seal impedances of about 5 MQ), the relative contributions of these two conditions are more clearly seen from the results of the following experiment. A fibre was mounted in the cell with a narrow pool B so that about 300ti of fibre was in B. The geometry of the seals relative to the recording cell was the same as in the earlier experiment with the narrow B. The seals were relative to the fibre, however, as in the wider cell. In this case the action potential was about 120 mv, and the large pulses did not excite node N+1. The seal currents were approximately the same in the two experiments, the main difference being the length of fibre in pool B. It was therefore concluded that the current spread within the fibre was the cause of these artifacts; probably this current flowed in a conducting path between the myelin and a resistive layer outside the myelin. The characteristic length of this outside 'cable' must have been short, since attenuation was negligible with about 300 j. of fibre in pool B and about 30% with 200 I. The histological structure which may account for the effect is the Schwann cell layer. From these experiments it was not possible to decide where the longitudinal conductor was located relative to the Schwann cell. Replacing the Ringer's solution in pool B with potassium chloride solution had no appreciable effect on the attenuation, indicating that the resistance of the outside layer was unaffected by potassium chloride. The 'attenuation artifact' was negligible with a long length of fibre in B (>300,u) as in the earlier experiments (Frankenhaeuser, 1957a). A long length of fibre in B, however, resulted in some phase lag in the potential recording system. This phase lag was quite negligible for recording the membrane action potential, but limited seriously the possible maximum gain at high frequencies in the voltage clamp amplifier (A2). The attenuation could be estimated, (a) by measuring the membrane action potential, or (b) by measuring the critical depolarization of node No which just excited node N+1. Discussion of limitation8 in the method. The limitations and difficulties in the method may be summarized in the following groups: (1) high frequency oscillations, (2) attenuation in potential recording system, (3) polarization and drift of electrodes, (4) deterioration of the fibre. 6 PHYSIO. CXLTIII

7 82 F. A. DODGE AND B. FRANKENHAEUSER (1) A feedback amplifier which stabilizes the membrane potential with little time lag is liable to oscillations. Oscillations of this kind were one of the difficulties encountered with the present technique, a considerable number of fibres being destroyed. The following steps were taken in order to overcome the difficulty. (a) In the design and construction of the feedback amplifiers attention was given to the phase lags. (b) The major phase lag appeared in that part of the fibre which was between pools E and C. The feedback amplifier A1 encountered the phase lag in the fibre between seals EA and BC. In order to minimize the total phase lag the recording cell was made small. I~~~~~~~~0 mv Vm 0 5 msec Fig. 2. Membrane current (Im) and membrane potential (Vm). Membrane potential was to some extent influenced by flow of membrane current at short times after step. This effect was less pronounced at other values of depolarization. Temp. 30 C. Note crossing of beams. The lower practical limit for this was that the pools should be wide enough to be efficiently equipotential throughout, so that the 'seal currents' should have little effect. A number of cells with different dimensions were tested. The dimensions of the cell which were found most suitable are given in Fig. 1. (c) In the experiments the gain of the feedback amplifiers had to be increased with considerable care to avoid high frequency oscillations (about 500 kc/sec). It was possible'to use a voltage gain in the amplifier A1 of about The voltage gain of amplifier A2 could be increased to about 400 with a substantial high frequency cut. This gain was sufficient to stabilize the membrane potential so that regenerative activity in node No was prevented. Large membrane currents did affect the membrane potential to a small extent ( <5mV), so itwas necessaryto record the membrane potential simultaneously with the membrane currents (Fig. 2). (d) The system showed some 'ringing' at the beginning and end of the step polarizations. The experiments were therefore carried out at low temperature, about 30 C, where the permeability changes were slower and thus the major part of the rising phase of the inward current was not affected by the ringing. (2) As described above an appreciable (i.e. up to 30 %) attenuation of the membrane potential occurred in the measuring system when a short length of fibre was in pool B. Attenuation of this kind must be accounted for in the measurements of the membrane potential. (3) Errors in setting the balance of the amplifiers were minimized by the use of calomel electrodes and by the stability of the amplifiers (see p. 78). (4) The fibres deteriorate to some extent during the course of an experiment lasting some hours. This effect was observed as a decrease of the maximum inward current and a shift of the potential at which the initial current reversed from inward to outward. The voltage clamp technique may have been rather damaging to the isolated fibre, first because three petroleum jelly seals were applied to the fibre close to the node under investigation, and secondly, because the membrane

8 VOLTAGE CLAMP ON MYELINATED NERVE 83 currents in voltage clamp experiments were much larger than the currents during normal impulse activity. There was some indication that the condition of the frogs may have an influence on how well the fibres survive the experiments. The slow deterioration of the preparation should not be confused with the very sudden destruction of the fibre caused by high frequency oscillations in the amplifiers. RESULTS Ionic currents The ionic membrane currents associated with step polarizations are illustrated in Fig. 3. The currents strikingly resemble the membrane currents in squid fibre voltage clamp experiments (cf. Hodgkin et at Fig. 12; note that our records show cathodal polarizations and outward currents as positive values). The very early part of the records was complicated by the fact that the membrane potential was stabilized to the new value accompanied by some overshoot and ringing at the beginning of the step. This was due to the limitations of the method imposed by the phase lags in the fibre and in the amplifiers at high frequencies. The currents were small and outward at small cathodal polarizations. At intermediate depolarizations the records showed an initial a mv -36 _ mv _ 38 m 57-57M b ma/cm2l 129--' I I I I I l ma/cm msec 0 5 msec Fig. 3. Membrane currents associated with step polarizations of amplitudes indicated to the left of trace, these values probably 10-20% too small, owing to attenuation (see text). Outward current as upward defiexion. Current calibration calculated from eqn. 1 with ANZED = 10QcmS. a, Membrane potential before pulse was resting potential minus about 15 mv (i.e. 15 mv anodal polarization); fast time base to show initial current. b, Experiment on another fibre with slow time base to show delayed outward current; potential before pulse was resting potential plus about 5 mv (i.e. 5 mv cathodal polarization). Temp. 30C. Capacity current is not visible in the records. 6-2

9 84 F. A. DODGE AND B. FRANKENHAEUSER inward current followed at long times by an outward current. With larger pulses the initial current decreased in amplitude and reversed its direction to outward current at about 120 mv. The change in step polarization from 38 to 57 mv in Fig. 3a was too large to show how the initial current changed with the mv t 50 ma/cm] 0 5 msec Fig. 4. Membrane currents in same experiment as Fig. 3a. Value of step polarization changed with 2 mv between frames to show that currents were smoothly graded with depolarization. Temp. 30C. membrane potential in this region. In Fig. 4 records are given at a number of intermediate steps from the same experiment. From these records and from Fig. 5, where the peak initial current is plotted against the membrane potential during the step, it is seen how the initial currents were smoothly graded with the membrane potential. The late outward currents increased in amplitude and approached a maximum value more rapidly as the pulse was increased in amplitude (Fig. 3b). The effect of the membrane potential on the ionic currents was investigated by changing the membrane potential in two steps (Fig. 6). Experiments of this type revealed that the initial peak currents were larger when the second pulse was preceded by anodal polarization of the membrane. The peak initial current-membrane potential curve (Fig. 7) was S-shaped like the squid fibre 'inactivation' curve. In this experiment the test pulse was applied some 80 msec after the onset of the conditioning polarization; this time being necessary for the effect to develop fully as shown in Fig. 8 (for comparison

10 VOLTAGE CLAMP ON MYELINATED NERVE 85 with the squid fibre see Hodgkin & Huxley, 1952c). The membrane potential at which the initial ionic current reversed from inward to outward was independent of the conditioning polarization. ma/cm2 +40~ ~ mV I I., I, I I IjI I I ziii Fig. 5. Peak initial current density plotted against membrane potential during step polarization. From same experiment as Figs. 3a and 4. Resting membrane potential marked as zero. Step polarizations superposed on 15 mv hyperpolarization. Temp. 30 C. A msec Fig. 6. Membrane currents associated with a sudden cathodal polarization to a final value of 42 mv from the resting potential. The step polarization was chosen so as to be associated with maximum inward current. A, Step from 5 mv cathodal polarization. B, Step preceded by 15 mv anodal polarization. Temp. 300.

11 86 F. A. DODGE AND B. FRANKENHAEUSER The reversible nature of the conductance changes was investigated by repolarizing the membrane during the time of the initial inward ionic current (Fig. 9; cf. Hodgkin & Huxley, 1952b) _ 58 mv 0--- V mv Fig. 7. Ordinate, peak initial membrane current associated with sudden cathodal polarizations of 58 mv from the resting potential, expressed relative to the maximum current. Abscissa, membrane potential preceding step polarization. Currents and potentials measured as shown on inset. Resting potential marked zero. Temp. 30 C. 1*01 VI- 20 mv 0*5 o l Tm se)30 40 S0 msec Time (msec) Fig. 8. Ordinate, peak initial current (relative to the maximum) plotted against (abscissa) time interval between onset of preceding polarization (V1) and cathodal test step. Example shown in inset. Temp. 30 C.

12 VOLTAGE CLAMP ON MYELINATED NERVE 87 I J I 0 1msec. Fig. 9. Membrane currents associated with a step cathodal polarization of about 50 mv inter. rupted at the peak of the inward current. Note the 'tail' of inward current after pulse. (Currents at the steps complicated by 'ringing'.) Temp. 3 C. DISCUSSION A voltage clamp technique with sufficient reliability for quantitative experiments has been developed for the myelinated nerve fibre. The quantitative treatment of such data is a rather time-consuming task. It seemed, therefore, appropriate to present some results at this stage in a qualitative manner, since a number of questions concerning the applicability of the analysis of the squid nerve to the properties of the myelinated nerve are illuminated by a qualitative treatment of the present experiments. The membrane currents recorded in the present investigation were very similar to the squid fibre voltage clamp currents. The records of the initial surge of capacitive current were, however, complicated by ringing of the amplifier and by the complex nature of the impedance ED. The initial ionic currents in the frog fibre had the following characteristic properties: (a) the currents were smoothly graded with the step depolarization; (b) the time course of the current was S-shaped; (c) the current reached its peak more quickly with large pulses than with small; (d) the initial inward current reached a maximum value at depolarizations of mv from resting potential; (e) the initial current reversed from inward to outward at a definite potential; (f) the potential for current reversal was unaffected by conditioning polarization; (g) the potential for current reversal was shifted when the external sodium concentration was changed; (h) the initial current was larger when the test pulse was preceded by anodal polarization. There are naturally some differences between the present results and those

13 88 F. A. DODGE AND B. FRANKENHAEUSER obtained on the squid axon. The more obvious points are: a, The ionic current densities appear to be much larger. b, The curve relating the initial current and the pulse amplitude was steeper at medium-size pulses and showed some divergence from a straight line at the region of the largest depolarizations. c, The outward 'potassium current' does not decline at long times. The decline was attributed by Frankenhaeuser & Hodgkin (1956) to the presence of an external barrier tentatively identified as the Schwann cell. It may be significant that Robertson (1957) reports a much more open arrangement of the Schwann cell processes at the node than has been seen in the squid fibre (Geren & Schmitt, 1954). These points are consistent with the idea that the initial ionic current was a sodium current just as in the squid fibre. 'Oscillations' and 'all-or-none' regenerative responses, such as those reported by Tasaki & Bak (1957) for the toad myelinated fibre, were seen, but were regularly traced to insufficient stabilization of the membrane potential, and were therefore considered artifacts. With the present method these regenerative responses could be fully suppressed even in hyperpolarized fibres where the tendency to regeneration was much stronger, and the ionic currents were smoothly graded with time and with membrane potential. The initial current was regularly followed at all cathodal polarizations by a delayed outward current. This current showed great similarities with the potassium current in the squid fibre. We have never observed the absence of delayed outward current, such as has been described by del Castillo et al. (1957, Fig. 2). The only cases in which we have seen relatively small outward currents have been in fibres that clearly had been damaged, and hyperpolarized before the test pulse. These fibres had shown 'normal' currents before the damage. All the properties of both the initial and the delayed currents were qualitatively the same as those of the squid fibre currents; it seems therefore safe to conclude that the myelinated nerve, when depolarized, undergoes specific conductance changes very similar to those described for the squid nerve. There are differences between the two nerves. At the moment it seems as if these differences were quantitative rather than qualitative. The conclusion of Tasaki (1956) and Tasaki & Hagiwara (1957), that the phenomenon of the abolition of an action potential with an anodal pulse is inconsistent with the voltage clamp analysis, is surprising in view of the facts that: (a) the abolition phenomenon during the rising phase has been previously illustrated and discussed (Hodgkin, Huxley & Katz, 1949); and (b) the mathematical formulation of the squid analysis permits a test of the hypothesis. We have calculated the effect of an anodal pulse, applied at the time of the falling phase of the theoretical membrane action potential as described by the equations (Hodgkin & Huxley, 1952d) and found that the equations predict

14 VOLTAGE CLAMP ON MYELINATED NERVE 89 an interruption of the action potential by the anodal pulse. The phenomenon is therefore fully consistent with the squid fibre analysis. Tasaki & Freygang (1955) concluded that the myellnated nerve does not undergo a change in potassium conductance during the action potential. That this conclusion, based on indirect evidence, could be correct seems doubtful because of the delayed currents described in the present paper. SUMMARY 1. A method of stabilizing the membrane potential of an isolated frog nerve fibre, and its limitations, are described. This method involves two feedback amplifier systems, the first for measuring the membrane potential and the second for stabilizing the membrane potential at desired values. 2. A qualitative description of the membrane currents associated with step polarizations is presented. The currents have the following properties: (a) The currents are inward and small at anodal polarizations. (b) At small cathodal polarizations the currents are outward. (c) At cathodal polarizations of about, mv an initial transient inward current and a delayed lasting outward current were observed. (d) The initial current reversed its direction from inward to outward at a definite potential. (e) Both the initial current and the delayed outward current reached their maximum values more quickly at larger depolarizations. (f) The currents were continuously graded with the membrane potential. (g) The initial current was larger when the pulse was preceded by anodal polarization. The effect of preceding polarization reached a limiting value at about 30 mv. (h) Repolarization of the membrane during inward current flow was associated with a large, rapidly declining 'tail' of inward current. 3. There is a striking similarity between the membrane currents described here and those observed in the squid fibre. It is therefore concluded that the frog nerve, when depolarized, undergoes specific changes in membrane conductance which are very similar to those in the squid giant nerve fibre. This investigation was supported by a grant from Stiftelsen Therese och Johan Anderssons Minne. REFERENCES DEL CASTILLO, J., LETTVIN, J. Y., MCCULLOCH, W. S. & PITTS, W. (1957). Membrane currents in clamped vertebrate nerve. Nature, Lond., 180, FREANKNHAEuISER, B. (1957a). A method for recording resting and action potentials in the isolated myelinated frog nerve fibre. J. Physiol. 135, EuANKENHAiUSER, B. (1957b). The effect of calcium on the myelinated nerve fibre. J. Physiol. 137, FRANKENHAiUSER, B. & HODGKN, A. L. (1956). The after-effects of impulses in the giant fibres of Loligo. J. Physiol. 131,

15 90 F. A. DODGE AND B. FRANKENHABUSER FRANKENAKEUsER, B. & HODGKIN, A. L. (1957). The action of calcium on the electrical properties of squid axons. J. Physiol. 137, FRANKENHAEUSER, B. & PERssoN, A. (1957). Voltage clamp experiments on the myelinated nerve fibre. Acta phy8iol. 8cand. 42, Suppl. 145, 45. GEREN, B. B. & ScHmrrr, F. 0. (1954). The structure of the Schwann cell and its relation to the axon in certain invertebrate nerve fibres. Proc. nat. Acad. Sci., Wash., 40, HODGKIN, A. L. & HuXLEY, A. F. (1952a). Currents carried by sodium andpotassium ionsthrough the membrane of the giant axon of Loligo. J. Phy8iol. 116, HODGKIN, A. L. & HUXLEY, A F. (1952b). The components of membrane conductance in the giant axon of Loligo. J. Physiol. 116, HODGKIN, A. L. & HUXLEY, A. F. (1952c). The dual effect of membrane potential on sodium conductance in the giant axon of Loligo. J. Phyajol. 116, HODGKIN, A. L. & HUxLEY, A. F. (1952d). A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Phyaiol. 117, HODGKIN, A. L., HUXLEY, A. F. & KATZ, B. (1949). Ionic currents underlying activity in the giant axon of the squid. Arch. Sci. phy8iol. 3, HODGKaN, A. L., HUXLEY, A. F. & KCATZ, B. (1952). Measurements of current-voltage relations in the membrane of the giant axon of Loligo. J. Physiol. 116, HUxLEY, A. F. & STAMPFLI, R. (1951). Effect of potassium and sodium on resting and action potentials of single myelinated nerve fibres. J. Phy8iol. 112, ROBERTSON, J. D. (1957). The ultrastructure of nodes of Ranvier in frog nerve fibres. J. Physiol. 137, 8-9P. STXMPFLI, R. (1952). Bau und Funktion isolierter markhaltiger Nervenfasern. Ergebn. Phy8iol. 47, TASAx, I. (1955). New measurements of the capacity and the resistance of the myelin sheath and the nodal membrane of the isolated frog nerve fiber. Amer. J. Phy8iol. 181, TASAKI, I. (1956). Initiation and abolition of the action potential of a single node of Ranvier. J. gen. Phy8iol. 39, TASAKI, I. & BAK, A. (1957). Oscillatory membrane currents of squid axon under voltage-clamp. Science, 126, TAsAX, I. & FRANK, K. (1955). Measurement of the action potential of myelinated nerve fiber. Amer. J. Phyaiol. 182, TAsAmx, I. & FREYGANG, W. H. (1955). The parallelism between the action potential, action current, and membrane resistance at a node of Ranvier. J. gen. Phy8iol. 39, TAsAxI, I. & HAGIwAPA, S. (1957). Demonstration of two stable potential states in the squid giant axon under tetraethylammonium chloride. J. gen. Physiol. 40,