ABSTRACT The gastropod mollusc gastropod Strombus luhuanus, prey to some Conus snails, exhibits a morphologically unique extendable foot utilized for fast escape response. The pedal ganglion of specimens was identified using in situ stimulation of radiating axon bundles and observing pedal response. Once the two lobes were removed, neurons were dissociated using glass teasing needles and studied under whole-cell patch clamp. Outward potassium currents were observed to have complex activation kinetics and biphasic inactivation. These kinetics were characterized and used to develop a pre-pulse protocol for current separation. Two distinct voltage-gated potassium currents were identified and isolated using this pre-pulse procedure. The fast activating and inactivating current resembled a classic A-current and could reach peak current within 2 milliseconds after pulse initiation for large depolarizations, and inactivated completely in about 30 ms. The second component of potassium current resembled a delayed rectifier and exhibited slow and incomplete inactivation, with peak current occurring within 5 or 6 milliseconds for large depolarizations and partial inactivation over the much longer period of about 150 milliseconds. The kinetic properties of each individual current were analyzed and described and a model based on Hodgkin-Huxley was used to compute conductance time course as a function of voltage. INTRODUCTION The development of the voltage-clamp technique (Cole, 1949; Hodgkin, Huxley and Katz, 1949 and 1952) allowed measurement and description of the various currents due to voltage-gated ion channels in a neuronal membrane. Studies of the properties of excitable membranes have led to the identification of many ion channels controlled by several mechanisms, but voltage-gated sodium and potassium channels form the backbone of excitation and action-potential propagation (Hodgkin and Huxley, 1952) Many different current types arise from these various channels within the membranes of neurons, some sharing similar enough characteristics to be classified into subgroups. "Transient inward" currents are those usually involved in the rising phase of an action potential, and are carried predominantly by sodium ions. "Delayed rectifier has been used to describe the repolarizing potassium currents involved in the falling phase of the action potential. Another porassium current type, deemed "A-current" by its original describers (Conner and Stevens, 1971a-c), plays a unique role that may not be directly involved in the falling phase of the action potential. A-type currents have the general form of an outward potassium current that activates at quite negative voltages and inactivates rapidly and completely, also at fairly negative voltages. The physiological significance of this category of current is not entirely agreed upon within the literature-perhaps mostly due to the large number of transient outward currents that have been given its title or the current lack of an identified molecular subtype. However there is general consensus that A-currents allow stimulus-intensity encoding through frequency modulation (Hille, 1992). Membranes lacking such currents generally respond to a steady stimulus with high, constant frequency of firing or a single spike followed by constant hyperpolarization. With A-current channels, membranes can respond to stimuli of varying intensity and encode this information in a graded frequency response. More current types than those described here exist (Lancaster and Pennefather, 1986), and many studies have sought to describe the physiological purposes of each. After Conner and Stevens work, a paper by Stephen Smith (1978) showed the role of A¬ current in the bursting nature of neurons in Tritonia diomedia. The currents in neurons of the gastropod mollusc Strombus luhuanus have not been previously described, and unique and rapid escape response displayed by this snail formed the basis for the present work. Strombus uses a large muscular foot to jump away from possible danger, a very unusual and fast behaviour, atypical of slow-moving gastropods. Most organsisms, such as the squid, (a cephalopod mollusc), make use of high conduction speed to initiate and coordinate fast escape responses, and the ion channels underlying activity in the relevant neurons appear to have fast kinetics (Gilly et al., 1997). The present description of the potassium currents within neurons from the Strombus pedal ganglion leads to the identification of two current types. One is a delayed rectifier with slow inactivation, the other an A-type current with fast and complete inactivation qualities. METHODS Preparation Strombus snails were obtained from American Samoa and maintained in oxygenated tanks at room temperature (about 24 °C). Dissection was carried out under light microscope using magnifications from 6x to 50x. The ganglia of the Strombus encircle the esophagus and were located with incisions near the base of the eye stalks. The twisted nature of the snail makes direct identification of the pedal ganglion difficult, however it is distinguishable by having only two fused lobes and a large number of nerves radiating into the foot musculature. The other ganglion complex has two primary lobes, each with a smaller auxiliary lobe attached. Functional identification of the pedal ganglion was accomplished during each dissection using direct stimulus of the radiating motor nerves and observation of subsequent contractions of the foot. Once removed, the pedal ganglion from each organism was placed in LIS-based medium that had been osmotically balanced to seawater and mixed with antibiotics, (Gilly et al., 1990) Dissociation Dissected ganglia were treated for 45 minutes in filtered seawater containing 7 mg of bacterial protease type XIV per mL, and desheathed using sharp forceps and a razor- blade knife. Under 50x magnification round neurons were distinguishable in desheathed areas, and were subsequently teased onto glass cover slips coated with Con A, following procedures previously described (Gilly et al. 1990). The teasing procedure used glass micro-pipettes made from type 7052 glass. After dissociation, the cover slips with cells were left undisturbed for 45 minutes or longer without motion to allow the neurons to adhere. Cells were then incubated overnight at 12 °C to allow recovery, (most neurons appear swollen immediately after dissociation.) The bathing medium was replaced every other day as the cell stock was used, and most cells would not produce reliable seals after one week post-dissection. Patch-Clamp Whole-cell, voltage clamp recordings were carried out at 20 °C as described in detail elsewhere (Gilly and Brismar, 1989). The largest cells were selected for recording. as they often expressed more sodium current and were consequently the putative motor neurons. The internal solution (electrode) contained 20 KCl, 305 K-glutamate, 50 KF, 180 glycine, 115 sucrose, 1 EFTA, I EDTA, 10 lysine, 10 Hepes, with a pH of 7.8. The external (bathing) solution contained 470 tetra-methyl ammonium (TMA), 10 KCl, 10 Cachz, 20 MgCl, 20 MgSO4, 10 Hepes, also with a pH of 7.8. RESULTS General observations Preliminary experiments were carried out to examine the potassium currents in the isolated neurons of the Strombus pedal ganglion. A series of voltage steps from the holding potential of-80 mV (Fig. 1.) showed currents with little or no inactivation at small depolarizations, complicated activation kinetics at intermediate depolarizations (e.g. late peaks, arrows in Fig. 1), and fast inactivation at large depolarizations. These current characteristics were found in a sample of three cells, and suggest the presence of multiple current types. A set of long pulses revealed biphasic inactivation kinetics (Fig. 2), further suggesting the possibility of two or more current types. The inactivating currents were plotted on a semi-logarithmic scale to better highlight the slow and fast exponential decays. Fig. 3 depicts such a plot for a step from the holding potential of-80 mV to +20 mV. The main plot shows the two components and a fit to the slow component of inactivation; the inset outlines the residual fast component. Current separation The relative time scales of the fast and slow inactivating components (henceforth referred to as Igand Iks) allowed the possibility that a pre-pulse protocol might serve two separate the two current types by selectively inactivating Ig without affecting Ig, to an appreciable extent. A series of twin pulses were run with different recovery intervals anat several voltages to test this possibility, and to describe recovery from inactivation for Ifast. Fig. 4A through C depict the current measurements for a series of twin-pulse experiments in which all interval durations were varied identically at each of three recovery voltages (see pulse pattern in Fig 4D). These data describe the time course of recovery from inactivation and its voltage dependence, showing total inactivation in some of the traces. Examination of the dependence of total inactivation revealed that pulses to -25 mV were sufficient to totally inactivate Igg (data not shown). The information outlined above was used to construct a pre-pulse procedure that would selectively remove the fast inactivating current before test pulses. This technique yielded current traces like those shown in Fig. 5, which depicts the currents from two stimulation regimes, one with a pre pulse to-25 mV and one without. Test pulses with a pre pulse contain only Igs, allowing Igrto be isolated by subtracting these traces t from their respective non-pre-pulse counterparts (which contain both currents). Igris thus pre- pulse sensitive and Ig, is pre-pulse resistant. The technique was applied to a series of such traces at varying stimulus amplitudes to yield the individual currents plotted in Fig. 6. Fig. 7 shows current-voltage plots for these currents. Conversion to conductance and correction for series resistance To complete Hodgkin-Huxley-like characterization of the two currents displayed in Fig. 6 these traces must be converted into units of conductance (siemens) and normalized by the size of the cell. Furthermore, the sensitivity of the fitting procedure necessitates compensation any significant residual series resistance of the electrode tip. Series resistance was calculated from the capacitance transients of a-10 mV step made with fast (electrode) capacitance balanced, slow-capacitance compensation turned off, and series resistance compensation set to the same level used to record currents. A sample pulse is shown in Fig. 8. The exponential decay is fit, giving a time constant equal to Rsenes times the total membrane capacitance of the cell. Cell capacitance is calculated from the time-integral of the transient current, and the time constant (tap) is divided by this number to give the estimate of Rses. The membrane capacitance is also used to approximate the surface area of the cell, under the assumption that the membrane capacitance per area is near 1 uF per cm (Hille 1992). Values of these parameters for each of the three cells studies are listed in Table 1 The combination of these parameters shown in Eq. 1 yields the conductance per square centimeter of the cell, including compensation for series resistance: 8 1 Pns Rne where A is the surface area of the cell. The Variing term in this equation refers to the driving force, which is equal to the command voltage minus the reversal potential, which was estimated using tail current readings (data not shown). Eq. 1 was applied to each of the traces for lgand Iks. While this equation gives more accurate traces than compensation accomplished through offsetting the membrane command voltage by Ipeak Rseres for each step, the fraction within the denominator tends to amplify noise for small currents and small driving voltages. Therefore the traces from stimuli that elicited no discernable increases in conductance were removed from the each series. The calculations were also made without series resistance compensation to assess its significance. As shown in Fig. 9, for large depolarizations there is a significant offset due to series resistance. Fig. 10 shows the final, compensated conductance traces as calculated for both Igr and Iks. Characterization of Ixt The procedure used for estimating the various parameters utilizes two simultaneous approaches to the same values. First, approximations of the values are made by fitting single exponentials to specific regions of curves for large depolarizations, (to be explained in more detail later). These values are then used as the initial guesses for parameterized functions that are then fit iteratively using a computer program (NEURON, developed by Michael Hines and John W. Moore at the Department of Neurobiology, Duke University). Because the empirical approximations are only reasonably valid for large depolarizations, they are only used as initial guesses to fit the largest command step, after which the fit values for the previous step are used as guesses for the next smaller step. The equation used for fitting the conductance traces is a variation on the equations used in Hodgkin and Huxley's original 1952 paper, the primary difference being that the potassium model has an inactivating particle similar to sodium channels. The equations for the Igemodel are thus: gy -nhg, (2) n-n.-(n.-n)exp(-1/.), (3) h-h.-(h.-ho)exp(-1/7,), (4) where 1o 9. 19. ho G,+B. and r. 9. 18. (6) 8. 19. The traces in Fig. 1OB show near complete inactivation for all pulse amplitudes, thus he can be neglected. Furthermore, pulses from holding potentials more negative than -80 mV (data not shown) did not produce an appreciable decrease in peak current amplitude for large depolarizations, indicating that ho is at a steady level, and can be assumed to equal 1. The gxV curve for Ig(Fig. 11) shows no conductance at the holding potential, so no equals 0. These simplifications lead the following relation when Eq.'s (3) and (4) are substituted back into the equation for fast conductance (gr, Eq. (2): 8y - 8 y 1- exp(-1/1,) exp(1/1.) (7) where (8) 8y -8n This equation was used in the computational fitting of the conductance curves. and the parameters g'kt tn, and ti were approximated and used for the initial guess; N was determined to be 11, the lowest number for which the lag in the activation kinetics was fit to a reasonable level (by eye only). The large value for N is necessitated by the long lag after stimulus initiation before the beginning of the rising activation phase. As all known voltage-gated potassium channels are probably tetramers, the Hodgkin-Huxley model begins to show its weaknesses here. However, as it is applied in this paper, it serves to dissect and characterize many parameters of a given current type, making it an excellent tool for fingerprinting and comparing different currents, regardless of the underlying structural accuracy of the model. g'ky and t, were estimated by fitting the top third of the activation curve (to peak) with a single exponential of the form: KO +Klexp(-K2*1) (9) KO was used as the estimate for g'gyand 1/K2 was used to approximate ty. The same exponential was used to fit the inactivation phase of the conductance curve, using 1/K2 to estimate ti. Once initial guesses were made, each subsequent step used the previous steps fit values for its guesses, and the final set of curves thus fit are depicted in Fig. 12. The values for the parameters and the calculated rate constants are listed in Table 2. Because the gkV curve (Fig. 11) indicates that the saturation point was never reached, an arbitrary adjustment of 20% was added to the highest conductance to approximate gre, (a conservative approximation that will have to stand until future experiments can assess its validity). As would be expected by the complete lack of inactivation recovery in Fig. 4C, ap is 0 for all pulses analyzed. Characterization of Iks The model for Ig, currents is similar to Igg however it varies slightly in applicable assumptions. The most important of these pertains to ha. For Ig, it was assumed that inactivation was complete for all command voltages studied, yet this assumption is not necessarily correct for Iks. The long pulses depicted in Fig. 2 show incomplete inactivation for large depolarizations, a fact that not only invalidates the total inactivation assumption, but creates other complications as well. It is possible that Ig, is comprised of two currents, one that inactivates slowly and one that does not inactivate at all. Even assuming that it is only one current type, the long pulses in Fig. 2 are taken from a different cell than the properly isolated currents in Fig. 10, and it may be that incomplete inactivation is not a constant characteristic in all the cells studied. While every cell studied displayed the complicated activation and inactivation kinetics identifying two current types, the specific questions raised here require further research to be appropriately addressed. In light of these confounding factors, a study and analysis of the inactivation characteristics of Ig, will be reserved until such a time that proper experiments have been completed. However, the activation parameters did not show a high degree of sensitivity to the assumptions necessitated by insufficient inactivation data. Fig. 13 shows curves for three large depolarizations, fit by the following equation: (10) 8k - 8'k 1-exp(-1/1.)I.-(h. -D)exp(1/1,) where (11) 8'k.- 8.7 The curves calculated assuming total inactivation (ha equals 0), and those calculated assuming approximately two-thirds inactivation (an approximation obtained from KO/(KO +Kl) of an exponential fit of the inactivating current, Eq. (9)), are superimposed over the original traces. The fit (in terms of mean-squared-error) is slightly better for the total-inactivation assumption than for partial, but this difference is minimal and barely discernable by eye. The difference between these curves is outlined in Table 3, in which the values for t, are shown to vary significantly between the two assumptions. This sensitivity increases with smaller pulses to the point that a variation of twenty percent in estimates of ha can cause an order of magnitude difference in ty. In contrast, the variation of the activation parameters (g'k and t,) is minimal. This insensitivity allowed analysis of the activation kinetics of Ik, for comparison with those of Igr The difficulty in approximating a changing ho necessitated the use the full-inactivation assumption (ha equals 0). The same procedure described for Igrwas used to fit and calculate the parameters for Ix, the one difference being the use of Eq.'s (10) and (11) in place of (7) and (8). The results of these calculations are summarized in Table 4. (For completeness, the table includes values for the inactivation kinetics as calculated under the assumption of full- inactivation.) DISCUSSION The grV curves (Fig. 11) for the two currents characterized in this study are consistent with two distinct potassium channel types; they display different steepness as depolarization levels increase. This voltage-dependence provides a means for information integration, one of many that these two currents provide. Any relative variation in a parameter between Igrand Igs, with respect to voltage or time, is a possible mechanism for information integration. The relation between the t,'s shown in Fig. 14B outlines the possible mechanism by which the A-type Igcurrent can modulate frequency response for graded stimuli. The time constant of activation for Igris much faster than that for Ig, at negative voltages, and recovery from inactivation is also dependent on negative voltages (Fig. 15). It is likely that the level of Iginactivation plays a major role in establishing the threshold stimulus required to create an action potential. Activation of Igrduring an action potential would lead to rapid and probably incomplete inactivation, and a string of rapidly occurring action potentials would thus produce a progressive increase in excitability as larger portions of A-current were inactivated. This mechanism is just one example of the sort of information processing available to a neuron that possesses multiple current types, and it is the generally accepted means by which A-current modulates frequency output to graded stimuli intensity. Useful speculation towards the physiological purposes of the many other variations is premature for a paper of this scope and the current level of research. The rate constants for the two currents are plotted in Fig. 16. Both currents show little rise in ß at the most negative voltages studied, which in the case of the Igg is result of the 11" order activating term required in fitting its conductance curves. Whether this is artifact or truth will require further study (similarly for the relatively constant value of no). Future studies will include complete characterization of the steady-state inactivation kinetics for Igs, as well more description of the other parameters for relevant voltage ranges outside those studied thus far. With sufficient data, it should be possible to fit continuous functions the graphs of the rate constants, which can then be used to calculate and predict action potentials across the membrane. These topics all serve as avenues for further study in the description of the Strombus pedal-ganglion currents. ACKNOWLEDGEMENTS Matt Brock, for his technical and academic advice and endless patience for inexperienced inquiry. Joseph Schultz for his constant suggestions—often solving problems before they occur and answering questions before they are asked. Professor Stuart H. Thompson for his many insights and conversations on all subjects of neurobiology. And a special thanks to Professor William F. Gilly, for all of his support and late hours of help and encouragement. His high standards are often a cloaked blessing, but a blessing nonetheless. References Cole, K. S. (1949). Dynamic electrical characteristics of the squid axon membrane. Arch. Sci. Physiol. 3, 253-258. Connor, J. A. and C. F. Stevens (1971a). Inward and delayed outward membrane currents in isolated neural somata under voltage clamp. J. Physiol. 213, 1-19. Connor, J. A. and C. F. Stevens (1971b). Voltage clamp studies of a transient outward current in gastropod neural somata. J. Physiol. 213, 21-30. Connor, J. A. and C. F. Stevens (1971c). Prediction of repetitive firing behaviour from voltage clamp data on an isolated neurone soma.. J. Physiol. 213, 31-53. Gilly, W. F. and T. Brismar (1989). Properties of appropriately and inappropriately expressed sodium channels in squid giant axon and its somata. J. Neurosci. 4. 1362-74. Gilly, W. F., M. T. Lucero, and F. T. Horrigan (1990). Control of the spatial distribution of sodium channels in giant fiber lobe neurons of the squid. Neuron. 5, 663-674. Gilly, W. F., R. Gillette, and M. McFarlane (1997). Fast and slow activation kinetics of voltage-gated sodium channels in molluscan neurons. J. Neurophysiol. 77(5). 2373-84. Hille, B. (1992). Ionic Channels of Excitable Membranes, 2"“ Edition. Sinauer Associates, Inc.. Sunderland, Mass. Hodgkin, A. L. and A. F. Huxley (1952). A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. 117 500-544. Hodgkin, A. L., A. F. Huxley and B. Katz (1949). lonic currents underlying activity in the giant axon of the squid. Arch. Sci. Physiol. 3, 129-150. Hodgkin, A. L., A. F. Huxley and B. Katz (1952). Measurements of current-voltage relations in the membrane of the giant axon of Loligo. J. Physiol. (Lond.). 116, 424-448 Lancaster, B. and P. Pennefather (1986). Potassium current evoked by brief depolarizations in bull-frog sympathetic ganglion cells. J. Physiol. 387, 519-548. Smith, S. J. (1978). The mechanism of bursting pacemaker activity in neurons of the mollusk Tritonia diomedia. Ph.D. Thesis, University of Washington. Table 1: Values calculated using the technique outlined in the text and shown in Fig. 8. These values were used in conversion from current to conductance per area. Cmenbrane (pE) Reris (M9. Cell Teg Cell surface area (emt 48.176 1.307705 4.8179E-5 49.55 0.585267 4.955 E-5 23 25.189 0.913097 2.5189 E-5 - staatata- oooooooo ooooooo saa- . ooo takoa- Table 3: Values comparing the sensitivity of various parameters to the full- and partial-inactivation assumptions explored in analysis of Ilow. VCommand gk. (mSem) In (msec) 1h (msec) (mV) Full Partia Full Partial Full Partial 60 1.8582 1.9327 1.2852 1.3214 0 0.3 1.8167 1.9058 1.5323 1.5796 0.3 40 1.7133 1,8081 1,8212 1.8779 0 0.3 800 O ooooooooo ooooooo . S 360 O — — — — sttkoo- Figure Legends Fig. 1: Current traces showing the complicated activation and inactivation kinetics described within the text. Notice the delayed peaks at intermediate voltages. Fig. 2: 250 ms pulses from the holding potential of-80 mV to +20 mV in 10 mV increments. These traces show biphasic inactivation. Fig. 3: Semi-logarithmic plot of the biphasic inactivating current from a 250 ms pulse from-80 mV to +20 mV. The smooth line is a fit for the slow-phase exponential, and the inset depicts the residual fast-inactivating current. Fig. 4: Twin-pulse traces displaying the voltage and temporal dependencies of recovery from inactivation. Fig. 5: Example current separation using trace-subtraction. The top trace is the current measured from a step from the holding potential of-80 mV to +60 mV. The middle trace is the same test pulse with a pre pulse of 55 mV. Traces showing the currents measured from a sequence of pulses with no pre Fig. 6: pulse (A); with a 55 mV pre pulse (to-25 mV command voltage)(B); and the isolated fast-inactivating current (C), calculated using trace subtraction. Fig. 7: IV curves for the currents shown in Fig. 6. Fig. 8: This figure depicts the initial capacitance transient measured for a-10 mV step. Fig. 9: Two sets of conductance-per-area curves, calculated from the same current measurements of Ik, using Eq. (1), the solid line included the Rseries term, the dotted line did not. This shows the significance of the series resistance. Fig. 10: Calculated conductance curves for Ik. (A) and IkT(B). Fig. 11: Relative grV curves for Iks and Igg each scaled independently. Fig. 12: Changes in conductance for Igrin response to various depolarizations (command voltages are listed by each trace). The smooth curves were calculated from fits using Eq. (7), with the resulting parameters described in Table 2. Fig. 13: Traces showing an enlarged fitting of the conductance changes of Ik, in response to three different depolarizations. The smooth curves represent the fits calculated using the assumption of total and partial inactivation, and show the accuracy with which the fit is achieved regardless of the assumption used, as well as the insensitivity of the activation phase to the assumption choice. Fig. 14: Graphs displaying the voltage dependence of na (A) and z, (B). Fig. 15: Semi-logarithmic plot of relative recovery from inactivation. See Fig. 4D for the stimulus pattern. Fig. 16: Plots of the various rate constants for Ik, and Iyg taken from Tables 2 and 4. (inactivation parameters for Ig, are omitted for reasons outlined in the text.) Figure 1 nsünenmmmnense ensmnim ifponmimamilavnnevmnnnn v n 0 -80 mV — 24 ms 1 nA 5 ms +60 my 2 5 mV Figure 2 e itn npen inei es seee 1 nA 50 ms Figure 3 3X102+ 20 22 24 ms 2x1031 20 30 40 45ms Time (ms) Figure 4 Test Pulse Pre Pulse ga H n -100 mV recovery amplitude k 1 nA 5 ms P -60 mV recovery amplitude D Pre pulse Test pulse +10 mV -60 mV (C) 15 -80 mV H-H ----------------- — ------------------------ -100 mV (A) Recording periods — Figure 5 no pre pulse s) 1 nA slow component (igs) 5 ms ieen isolated fast component Uk) Stimulus test pulse +60 5 pre pulse -80 24 ms 50 ms Figure 6 (Total) 1 nA 5 ms itata aaa- w ti 1) ataaakaa- () Simi +60 my 10 mv — — 5mV 80 mV 24 ms Figure 7 6000 - 5000 - — otal + 4000 - 5 3000- § 2000. 1000 -20 -40 20 40 60 Command voltage (mV) Figure 8 membrane capacitance 2 nA tau = RC 200 us Figure 9 Se — compensated Mr f uncompensated 5 4 pee eeen. 0.4 mSlcm aenenene ee eneten, e enenetenete e e e een 5 ms e enen enen enen eneneten ... . Figure 10 eneit ee 0.5 mSlcm e 5 ms e e e e en 2 . - . taaa- . a- F a- he 4 ta e pe e nen — -40 -20 Figure 11 1.0- 0.8 - 0.6 - 0.4 - Command voltage (mV) 40 60 Figure 12 0.2 mSlcm 5 ms 504 M ereeie eeseseessesesseese.e e. 5 204 - .Sesseeeee e 3 - K . sesssessssesssssssesseesssseesssesssessesssessssssssesaksesssssessössssssssselee aa- -10 . seeeeeseeseedbeedeeee ssssessssssessessesseebeseedelssssemesessbeete . Y. Figure 13 partial inactivation full inactivation 0.2 mSlcm 5 ms -------------.---.-.---................................................................ 1.0- 0.9 - 0.8 0.7 0.6 - 4- 2- -20 -20 Figure 14 40 Command voltage (mV) -O 40 Command voltage (mV) 60 1.0- 0.8 - 06- 04- 0.2 0.0 10 2 3 4567 — — recovery voltage =-100 —— recovery voltage = -80 -O recovry voltage =-60 2 34561 1000 2 3 4 Time (ms) Figure 15 0.6 - —0 alpha, beta, 80.4- 0.2 - 00 -20 0 20 40 60 1.5- 3 1.0- —0- alpha, — — beta, 0.5 - -20 0 20 40 60 0.5 - 0.4 - 0.3- 30.2- 50. 0.0- -20 20 40 60 Command Voltage kks kr Kt