- Email: [email protected]
Unequal branch point filtering action in different types of dorsal root ganglion neurons of frogs
Neuroscience Letters, 59 (1985) 15-20
15
Elsevier Scientific Publishers Ireland Ltd.
NSL 03447
UNEQUAL BRANCH POINT FILTERING ACTION IN DIFFERENT TYPES OF DORSAL ROOT GANGLION NEURONS OF FROGS
S. DAVID STONEY, Jr.
Department of Physiology, Medical College of Georgia, Augusta, GA 30912 (U.S.A.) (Received March 5th, 1985; Revised version received and accepted May 6th, 1985)
Key words: axon branch point - filtering action - dorsal root ganglion neuron - primary afferent fiber -
least conduction interval - safety factor - frog
The influence of the intraganglionic branch point on impulse conduction in single neurons of frog dorsal root ganglia (DRG) has been determined by measuring the least interval at which it will conduct two action potentials into the dorsal root. At 21-23°C, branch points of myelinated fibers had long least conduction intervals and low safety factors for orthodromic impulse conduction compared to nodes of Ranvier in peripheral nerve. DRG neurons with broad somatic spikes with a shoulder on the falling phase and slowly conducting myelinated or non-myelinated axons had the longest least conduction intervals (lowest safery factors). DRG neurons with brief somatic spikes with little or no shoulder on the falling phase had short least conduction intervals (higher safety factor) regardless of their conduction velocity. The results indicate that certain DRG neurons have found a way to minimize branch point filtering action.
Axons often undergo extensive collateral and preterminal branching. Branch points or other regions of geometrical inhomogeneity of non-myelinated nerve processes often have a low safety factor* for impulse conduction [3, 6, 12, 19]. Although no systematic studies of orthodromic impulse conduction at branch points of single myelinated fibers have been reported, some results suggest that such branch points may introduce a filtering action on conducted spike trains [1, 4, 9] (see refs. 10 and 16 for recent reviews). I have studied impulse conduction past the branch point of single primary afferent fibers in frog dorsal root ganglia (DRG). The D R G preparation is an ideal model because the stem process of each afferent neuron bifurcates, within the ganglion, to form a dorsal root fiber and a peripheral nerve fiber. Branch points of myelinated fibers always occur at a node of Ranvier [11]. The results show that intraganglionic axon branch points have a lower safety factor for impulse conduction than ordinary nodes of Ranvier in peripheral nerve and that the branch point safety factor is different for different types of D R G neurons. A preliminary report has been published [15]. *Safety factor is defined as 'a property of each region of an axon or other spike-supporting membrane determined by the local threshold measured in critical depolarization across the membrane and the amplitude achieved by the approaching spike, which normally will provide this depolarization by local circuits'
[21. 0304-3940/85/$ 03.30 © 1985 Elsevier Scientific Publishers Ireland Ltd.
16 Glass microelectrodes filled with 3 M KCI (15-30 Mf~) were used for intracellular recordings from D R G neurons of frogs (Rana pipiens). Ganglia were mounted in a perfusion chamber and superfused with an oxygenated solution containing 120 mM NaC1-3 m M KCI-1 m M MgCI2-2 m M CaCI2-10 m M Tris-HCl-10 m M dextrose and Tris base necessary to adjust the p H to 7.8 (the p H of bullfrog plasma at 21-23'-C [13]). Action potentials were evoked by stimulation of peripheral nerve and dorsal root at 0.5-1 Hz with 0.1-0.2-ms current pulses. Experiments were run at room temperature, 21-23'~C. The influence of the branch point on orthodromic impulse conduction was determined by measuring the neuron's least conduction interval (LCI). This was accomplished by gradually reducing the interval between two closely timed, suprathreshold stimuli to peripheral nerve. LCIs were usually determined with 3 × threshold (T) stimuli and then checked with 5-7 T pulses to ensure that no response occurred. LCI was independent of stimulus strength above 2 T for most neurons. The longest interpulse interval at which the intracellularly recorded response to the second stimulus pulse just began to fail in an all-or-none manner (Fig. IA, B) was taken as the neuron's LCI. The M-spike (Fig. 1B, arrows), which antecedes and underlies generation of the full-blown somatic spike in frog D R G neurons, signals invasion of the myelinated segment of the stem process by the approaching orthodromic impulse [7]. Its failure (Fig. 1B), which also occurs when the whole spike complex fails (Fig. I A), indicates that the approaching impulse did not invade the stem process. It should be noted that the LCI, the shortest interpulse interval at which a second conducted action potential is evoked that is capable o f re-exciting the m e m b r a n e locus under study, is not the same as the absolute refractory period, the shortest interpulse interval at which a given membrane locus can be re-excited by a second stimulus of unlimited strength [18]. The LCI is the physiologically important parameter limiting
A
B
Fig. 1. A: LCI determination for orthodromic impulses elicited by two stimuli to peripheral nerve (P) at progressivelyshorter interval. Four superimposed traces of transmembrane potential; 50 mV and 1ms/div. The 2nd stimulus failed to elicit a response when the interputse interval was reduced from 2.76 ms to 2,62 ms, the LCI for this neuron. B: collision of orthodromic and antidromic impulses during LCI determination; three separate transmembrane potential traces of responses to peripheral nerve (p) and dorsal root (c) stimulation; calibrations as in A. The 2nd P stimulus produced an M-spike 2 out of 3 times (arrows). When the M-spike failed in an all-or-none manner (upper trace), no collision occurred, indicating that, at the LCI, there was no invasion of the branch point, stem process or dorsal root fiber by the second action potential. See text for further discussion.
17 transmission of information by high-frequency trains of action potentials in axonal trees. Collision tests showed that when the second action potential failed to invade the stem process it also failed to conduct into the dorsal root. For these tests, a third, appropriately timed stimulus was applied to the dorsal root, so that if the action potential elicited by the second peripheral nerve stimulus had invaded the branch point and dorsal root, it would collide with and abolish the dorsal-root-evoked action potential. The results were very consistent. In all but 3 of 60 neurons tested, the appearance of the antidromic response exactly coincided with the disappearance of the second orthodromic response (Fig. 1B). Thus, the interpuise interval at which the second response initially failed is a reliable measure of the LCI for the neuron's branch point. At the LCI, the second action potential fails to invade the neuron's branch point, stem process or its branch in the dorsal root. Could the response due to the second stimulus have failed at a node of Ranvier in peripheral nerve? To assess this possibility, LCIs for a subset of rapidly conducting DRG neurons [n=44, mean conduction velocity (CV)= 17.3 m/s] were compared with LCIs for peripheral nerve axons (n=36, mean CV= 13.0 m/s). Axons were recorded extracellularly with glass microelectrodes just distal to the ganglion. Their LCI was taken as the least interval at which a second response was observable. Average LCIs (_S.E.M.) were 1.81 _0.11 ms for DRG neurons and 1.03+__0..04 ms for axons. The latter value is very close to that reported by Tasaki [17] for toad peripheral nerve fibers. A t-test showed that the probability that the two means were the same was P<0.0005. This indicates that it was highly improbable that the action potential was failing in the peripheral nerve itself. Altogether, the results above show that this approach provides an unequivocal measure of the influence of the axon branch point on impulse conduction. The LCI is inversely related to the low-pass filtering action of the axon branch point: LC1-1 (x 1000) provides a conservative estimate of the maximum frequency of action potentials that can be conducted into the spinal cord without the advent of intermittent firing. The LCI also allows an accurate comparison between the safety factor (SF) for peripheral nerve and branch point nodes of Ranvier. Tasaki [18] showed that LCI is inversely proportional to SF and that SF---'5-7 for toad peripheral nerve fiber nodes of Ranvier. Thus, 1/ 1.03 x k = 6, and k = 6.18 for nodes of Ranvier of frog peripheral nerve fibers, where 1.03 is the average LCI for peripheral axons (see above) and SF assumed to be 6. Therefore, (I/LCI)x (6.18)=SF for nodes of Ranvier at frog DRG neuron axon branch points. Average LCI and orthodromic CV, as well as estimates of branch point SF for different types of frog DRG neurons are summarized in Fig. 2. Neurons were categorized on the basis of the shape of their somatic action potentials, particularly the presence or absence of a hump on the falling phase (cf. ref. 20). Neurons with brief, smooth spikes were designated as F neurons; neurons with irregularities on the falling phase of their spikes were designated H neurons, with a number to indicate prominence of the hump. Hi neurons had a hump only visible as an inflection in the dv/dt trace; H2 neurons had a hump clearly visible in the voltage record but not as
18
NEURON H4
H3
TYPE H2
F
H1
SPIKE SHAPE
CONDUCTION VELOCITY m/s
.36-+.05
(9)
7.0+--.4
(14) #
16.1-+.9
(52) 4
19.6-+.8
(73) ~
22.1-+.9
(71)
1.73_+.09
(87)
LEAST CONDUCTION INTERVAL ms
SAFETY FACTOR
9.10
2.04
(7)
4.80±.48
1.3
(13) 4
2.20-+.16
( 4 9 ) 4. 1 . 7 9 _ + . 0 8
2.8
3.5
(67)
3.6
Fig. 2. Typical action potential shape and average values [ + S.E.M. (n)] for conduction velocity, LCI and branch point safety factor for orthodromic impulses in different types of DRG neurons at 21-23°C. Neurons were designated on the basis of the presence or absence of a hump on the falling phase of the action potential. F neurons had fast, smooth spikes• The hump on the falling phase of H~ neurons was only visible as an irregularity in the dv/dt trace (arrow); H2 neurons had a moderate hump plainly visiblein the voltage record (arrow). Each picture shows 3-4 superimposed oscilloscope traces, one with no stimulus. Conduction distance was approximately equal for all pictures. Voltage, dv/dt and time calibrations apply to all pictures except H3 and H4: dv/dt = 100 v/s; H4: t = 10 ms. CV was determined by stimulating the peripheral nerve at two different sites a known distance apart. LCI and SF were determined as described in the text. Symbol (#) indicates that a given value is significantly different (P<0.05) from value to its immediate left as determined by analysis of variance. The only neuron types which did not differ from one another in terms of LCI or CV were F and H~ neurons. These neurons should be considered as a homogeneous class with regard to branch point filtering action.
p r o n o u n c e d as in H3 n e u r o n s . This m e t h o d of categorization successfully identified 3 of 5 types o f n e u r o n s in terms of CV, so n e u r o n types were ordered according to c o n d u c t i o n velocity in Fig. 2. Slowly c o n d u c t i n g n e u r o n s , whose action potentials exhibited a distinct h u m p on the falling phase (H4 a n d H 3 neurons), had the longest LCIs a n d lowest safety factors. N e u r o n s with relatively brief, s m o o t h spikes a n d high average CVs (F a n d H~ n e u r o n s ) had the shortest LCIs. F o r the whole sample of neurons, i g n o r i n g action potential shape, LCIs were negatively correlated with ort h o d r o m i c CV (r = - 0.50, P = 0.0001). The correlation between LCI a n d CV for the total sample is s o m e w h a t misleading. W i t h i n n e u r o n groups, only n e u r o n s with m o d e r a t e h u m p s o n the falling phase of their somatic spikes a n d intermediate CV (H2 n e u r o n s ) exhibited a significant correlation between LCI a n d CV (r = - 0 . 3 7 , P = 0.01). The correlation for the whole sample arises m a i n l y from low CVs a n d long LCIs of n e u r o n s with m a r k e d h u m p s o n their somatic spikes (H4 a n d H3 neurons); n e u r o n s with brief, s m o o t h somatic spikes (F a n d H1 n e u r o n s ) had short LCIs even when their CV was low. Thus, the m a g n i tude of the i n t r a g a n g l i o n i c b r a n c h point filtering action is different for different types
19 of D R G neurons and can be most reliably predicted by taking into account the shape of the neuron's somatic action potential and its orthodromic CV. Neurons with marked h u m p s on their somatic spikes and slowly conducting axons will develop intermittent conduction for orthodromic impulses at a lower frequency (~< 100-200 Hz) than neurons with brief, smooth somatic spikes (~<550 Hz). Since amphibian primary afferents discharge at rates as high as 200-400 Hz to receptor stimulation at r o o m temperature [14], branch point filtering action could be of physiological significance for some D R G neurons. In conclusion, these results show that the branch point o f primary afferent fibers in frog D R G can influence orthodromic impulse conduction and that some have a stronger filtering action than others. Certain slowly conducting D R G neurons m a y be considered as relatively p o o r neural signalers due to limitations imposed on their m a x i m u m frequency o f firing by the low safety factor o f their axon branch points. Conversely, other D R G neurons (with brief, smooth somatic action potentials) have found a way to minimize branch point filtering action, thereby increasing the 'bandwidth' and information content [5] o f their neural signals. Whether such differences can be accounted for by unusually short peribifurcation internodes reported for some D R G neurons of amphibians [8] and cats [11] and/or the relative proximity of the high capacitance/low resistance cell body to the branch point remains to be determined. Special thanks to Margene Attaway for excellent technical assistance. Supported by BRS 2 S07 R R 05365 23.
1 Barton, D.H. and Matthews, B.H.C., Intermittent conduction in the spinal cord, J. Physiol. (Lond.), 85 (1935) 73-103. 2 Bullock, T.H., Orkand, R. and Grinnell, A., Introduction to Nervous Systems, W.H. Freeman, San Francisco, 1977, 559 pp. 3 Cunnane, T.C. and Stjame, L., Frequency dependent intermittency and ionic basis of impulseconduction in postganglionicsympathetic fibers of guinea-pig vas deferens, Neuroseience, 11 (1984) 211-229. 4 Dun, F.T., The delay and blockage of sensory impulsesin the dorsal root ganglion, J. Physiol. (Lond.), 127 (1955) 252-264. 5 Gabor, D., Theory of communication, J. Inst. Electr. Eng. (Lond.), 93 (1946) 429-457. 6 Grossman, Y. and Kendig, J.J., General anesthetic block of a bifurcating axon, Brain Res., 245 (1982) 148-153. 7 Ito, M. and Saiga, M., The mode of impulse transmission through the spinal ganglion, 3ap. J. Physiol., 9 (1959) 33-42. 8 Ito, M. and Takahashi, I., Impulse conduction through the spinal ganglion. In Y. Katsuke (Ed.), Electrical Activity of Single Cells, Ikagu Shoin, Tokyo, 1960, pp. 159-179. 9 Krnjevic, K. and Miledi, R., Presynaptic failure of neuromuscular propagation in rats, J. Physiol. (Lond.), 149 (1959) 1-22. 10 Levy, R.A., Presynaptic control of input in the central nervous system, Canad. J. Physiol. Pharm., 58 (1980) 751-766. 11 Lieberman, A.R., Sensory ganglia. In D.N. Landon (Ed.), The Peripheral Nerve, Halsted, New York, 1976, pp. 188-278. 12 Parnas, I., Spira, M.E., Werman, R. and Bergmann, F., Non-homogeneousconduction in giant axons of the nerve cord of Periplantea americana, J. Exp. Biol., 50 (1969) 635-649.
20 13 Reeves, R.B., Intracellular pH in cold blooded vertebrates as a function of body temperature. Resp. Physiol., 28 (1976) 29-47. 14 Shimada, K, and Yai, H., Some properties of cutaneous mechanosensory units in toads, Jap. J, Physiol., 10 (19601 555-570. 15 Stoney, S.D., Jr. and Hershberger, E., Cooling and antiepilcptic drugs depress impulse conduction at branch points of single myelinated fibers, Soc. Neurosci. Abstr., 10 (1984) 109. 16 Swadlow, H.A., Kocsis, A. and Waxman, S.G., Modulation of impulse conduction along the axonal tree, Annu. Rev. Biophys. Biomed. Eng., 9 (1980) 143-179. 17 Tasaki, I., The excitatory and recovery processes in the nerve fiber as modified by temperature changes, Biochim. Biophys. Acta, 3 (1949)498-5(19. 18 Tasaki, I., Nervous Transmission, Thomas, Springfield, IL, 1953. 164 pp. 19 Tauc, T. and Hughes, G.M., Modes of initiation and propagation of spikes in the branching axons of molluscan central neurons, J. Gen. Physiol., 46 (19631 533-549. 20 Yoshida, S. and Matsuda, Y., Studies on sensory neurons of the mouse with intracellular-recording and horseradish peroxidase-injection techniques, J. Neurophysiol., 42 (19791 1134-1145.
15
Elsevier Scientific Publishers Ireland Ltd.
NSL 03447
UNEQUAL BRANCH POINT FILTERING ACTION IN DIFFERENT TYPES OF DORSAL ROOT GANGLION NEURONS OF FROGS
S. DAVID STONEY, Jr.
Department of Physiology, Medical College of Georgia, Augusta, GA 30912 (U.S.A.) (Received March 5th, 1985; Revised version received and accepted May 6th, 1985)
Key words: axon branch point - filtering action - dorsal root ganglion neuron - primary afferent fiber -
least conduction interval - safety factor - frog
The influence of the intraganglionic branch point on impulse conduction in single neurons of frog dorsal root ganglia (DRG) has been determined by measuring the least interval at which it will conduct two action potentials into the dorsal root. At 21-23°C, branch points of myelinated fibers had long least conduction intervals and low safety factors for orthodromic impulse conduction compared to nodes of Ranvier in peripheral nerve. DRG neurons with broad somatic spikes with a shoulder on the falling phase and slowly conducting myelinated or non-myelinated axons had the longest least conduction intervals (lowest safery factors). DRG neurons with brief somatic spikes with little or no shoulder on the falling phase had short least conduction intervals (higher safety factor) regardless of their conduction velocity. The results indicate that certain DRG neurons have found a way to minimize branch point filtering action.
Axons often undergo extensive collateral and preterminal branching. Branch points or other regions of geometrical inhomogeneity of non-myelinated nerve processes often have a low safety factor* for impulse conduction [3, 6, 12, 19]. Although no systematic studies of orthodromic impulse conduction at branch points of single myelinated fibers have been reported, some results suggest that such branch points may introduce a filtering action on conducted spike trains [1, 4, 9] (see refs. 10 and 16 for recent reviews). I have studied impulse conduction past the branch point of single primary afferent fibers in frog dorsal root ganglia (DRG). The D R G preparation is an ideal model because the stem process of each afferent neuron bifurcates, within the ganglion, to form a dorsal root fiber and a peripheral nerve fiber. Branch points of myelinated fibers always occur at a node of Ranvier [11]. The results show that intraganglionic axon branch points have a lower safety factor for impulse conduction than ordinary nodes of Ranvier in peripheral nerve and that the branch point safety factor is different for different types of D R G neurons. A preliminary report has been published [15]. *Safety factor is defined as 'a property of each region of an axon or other spike-supporting membrane determined by the local threshold measured in critical depolarization across the membrane and the amplitude achieved by the approaching spike, which normally will provide this depolarization by local circuits'
[21. 0304-3940/85/$ 03.30 © 1985 Elsevier Scientific Publishers Ireland Ltd.
16 Glass microelectrodes filled with 3 M KCI (15-30 Mf~) were used for intracellular recordings from D R G neurons of frogs (Rana pipiens). Ganglia were mounted in a perfusion chamber and superfused with an oxygenated solution containing 120 mM NaC1-3 m M KCI-1 m M MgCI2-2 m M CaCI2-10 m M Tris-HCl-10 m M dextrose and Tris base necessary to adjust the p H to 7.8 (the p H of bullfrog plasma at 21-23'-C [13]). Action potentials were evoked by stimulation of peripheral nerve and dorsal root at 0.5-1 Hz with 0.1-0.2-ms current pulses. Experiments were run at room temperature, 21-23'~C. The influence of the branch point on orthodromic impulse conduction was determined by measuring the neuron's least conduction interval (LCI). This was accomplished by gradually reducing the interval between two closely timed, suprathreshold stimuli to peripheral nerve. LCIs were usually determined with 3 × threshold (T) stimuli and then checked with 5-7 T pulses to ensure that no response occurred. LCI was independent of stimulus strength above 2 T for most neurons. The longest interpulse interval at which the intracellularly recorded response to the second stimulus pulse just began to fail in an all-or-none manner (Fig. IA, B) was taken as the neuron's LCI. The M-spike (Fig. 1B, arrows), which antecedes and underlies generation of the full-blown somatic spike in frog D R G neurons, signals invasion of the myelinated segment of the stem process by the approaching orthodromic impulse [7]. Its failure (Fig. 1B), which also occurs when the whole spike complex fails (Fig. I A), indicates that the approaching impulse did not invade the stem process. It should be noted that the LCI, the shortest interpulse interval at which a second conducted action potential is evoked that is capable o f re-exciting the m e m b r a n e locus under study, is not the same as the absolute refractory period, the shortest interpulse interval at which a given membrane locus can be re-excited by a second stimulus of unlimited strength [18]. The LCI is the physiologically important parameter limiting
A
B
Fig. 1. A: LCI determination for orthodromic impulses elicited by two stimuli to peripheral nerve (P) at progressivelyshorter interval. Four superimposed traces of transmembrane potential; 50 mV and 1ms/div. The 2nd stimulus failed to elicit a response when the interputse interval was reduced from 2.76 ms to 2,62 ms, the LCI for this neuron. B: collision of orthodromic and antidromic impulses during LCI determination; three separate transmembrane potential traces of responses to peripheral nerve (p) and dorsal root (c) stimulation; calibrations as in A. The 2nd P stimulus produced an M-spike 2 out of 3 times (arrows). When the M-spike failed in an all-or-none manner (upper trace), no collision occurred, indicating that, at the LCI, there was no invasion of the branch point, stem process or dorsal root fiber by the second action potential. See text for further discussion.
17 transmission of information by high-frequency trains of action potentials in axonal trees. Collision tests showed that when the second action potential failed to invade the stem process it also failed to conduct into the dorsal root. For these tests, a third, appropriately timed stimulus was applied to the dorsal root, so that if the action potential elicited by the second peripheral nerve stimulus had invaded the branch point and dorsal root, it would collide with and abolish the dorsal-root-evoked action potential. The results were very consistent. In all but 3 of 60 neurons tested, the appearance of the antidromic response exactly coincided with the disappearance of the second orthodromic response (Fig. 1B). Thus, the interpuise interval at which the second response initially failed is a reliable measure of the LCI for the neuron's branch point. At the LCI, the second action potential fails to invade the neuron's branch point, stem process or its branch in the dorsal root. Could the response due to the second stimulus have failed at a node of Ranvier in peripheral nerve? To assess this possibility, LCIs for a subset of rapidly conducting DRG neurons [n=44, mean conduction velocity (CV)= 17.3 m/s] were compared with LCIs for peripheral nerve axons (n=36, mean CV= 13.0 m/s). Axons were recorded extracellularly with glass microelectrodes just distal to the ganglion. Their LCI was taken as the least interval at which a second response was observable. Average LCIs (_S.E.M.) were 1.81 _0.11 ms for DRG neurons and 1.03+__0..04 ms for axons. The latter value is very close to that reported by Tasaki [17] for toad peripheral nerve fibers. A t-test showed that the probability that the two means were the same was P<0.0005. This indicates that it was highly improbable that the action potential was failing in the peripheral nerve itself. Altogether, the results above show that this approach provides an unequivocal measure of the influence of the axon branch point on impulse conduction. The LCI is inversely related to the low-pass filtering action of the axon branch point: LC1-1 (x 1000) provides a conservative estimate of the maximum frequency of action potentials that can be conducted into the spinal cord without the advent of intermittent firing. The LCI also allows an accurate comparison between the safety factor (SF) for peripheral nerve and branch point nodes of Ranvier. Tasaki [18] showed that LCI is inversely proportional to SF and that SF---'5-7 for toad peripheral nerve fiber nodes of Ranvier. Thus, 1/ 1.03 x k = 6, and k = 6.18 for nodes of Ranvier of frog peripheral nerve fibers, where 1.03 is the average LCI for peripheral axons (see above) and SF assumed to be 6. Therefore, (I/LCI)x (6.18)=SF for nodes of Ranvier at frog DRG neuron axon branch points. Average LCI and orthodromic CV, as well as estimates of branch point SF for different types of frog DRG neurons are summarized in Fig. 2. Neurons were categorized on the basis of the shape of their somatic action potentials, particularly the presence or absence of a hump on the falling phase (cf. ref. 20). Neurons with brief, smooth spikes were designated as F neurons; neurons with irregularities on the falling phase of their spikes were designated H neurons, with a number to indicate prominence of the hump. Hi neurons had a hump only visible as an inflection in the dv/dt trace; H2 neurons had a hump clearly visible in the voltage record but not as
18
NEURON H4
H3
TYPE H2
F
H1
SPIKE SHAPE
CONDUCTION VELOCITY m/s
.36-+.05
(9)
7.0+--.4
(14) #
16.1-+.9
(52) 4
19.6-+.8
(73) ~
22.1-+.9
(71)
1.73_+.09
(87)
LEAST CONDUCTION INTERVAL ms
SAFETY FACTOR
9.10
2.04
(7)
4.80±.48
1.3
(13) 4
2.20-+.16
( 4 9 ) 4. 1 . 7 9 _ + . 0 8
2.8
3.5
(67)
3.6
Fig. 2. Typical action potential shape and average values [ + S.E.M. (n)] for conduction velocity, LCI and branch point safety factor for orthodromic impulses in different types of DRG neurons at 21-23°C. Neurons were designated on the basis of the presence or absence of a hump on the falling phase of the action potential. F neurons had fast, smooth spikes• The hump on the falling phase of H~ neurons was only visible as an irregularity in the dv/dt trace (arrow); H2 neurons had a moderate hump plainly visiblein the voltage record (arrow). Each picture shows 3-4 superimposed oscilloscope traces, one with no stimulus. Conduction distance was approximately equal for all pictures. Voltage, dv/dt and time calibrations apply to all pictures except H3 and H4: dv/dt = 100 v/s; H4: t = 10 ms. CV was determined by stimulating the peripheral nerve at two different sites a known distance apart. LCI and SF were determined as described in the text. Symbol (#) indicates that a given value is significantly different (P<0.05) from value to its immediate left as determined by analysis of variance. The only neuron types which did not differ from one another in terms of LCI or CV were F and H~ neurons. These neurons should be considered as a homogeneous class with regard to branch point filtering action.
p r o n o u n c e d as in H3 n e u r o n s . This m e t h o d of categorization successfully identified 3 of 5 types o f n e u r o n s in terms of CV, so n e u r o n types were ordered according to c o n d u c t i o n velocity in Fig. 2. Slowly c o n d u c t i n g n e u r o n s , whose action potentials exhibited a distinct h u m p on the falling phase (H4 a n d H 3 neurons), had the longest LCIs a n d lowest safety factors. N e u r o n s with relatively brief, s m o o t h spikes a n d high average CVs (F a n d H~ n e u r o n s ) had the shortest LCIs. F o r the whole sample of neurons, i g n o r i n g action potential shape, LCIs were negatively correlated with ort h o d r o m i c CV (r = - 0.50, P = 0.0001). The correlation between LCI a n d CV for the total sample is s o m e w h a t misleading. W i t h i n n e u r o n groups, only n e u r o n s with m o d e r a t e h u m p s o n the falling phase of their somatic spikes a n d intermediate CV (H2 n e u r o n s ) exhibited a significant correlation between LCI a n d CV (r = - 0 . 3 7 , P = 0.01). The correlation for the whole sample arises m a i n l y from low CVs a n d long LCIs of n e u r o n s with m a r k e d h u m p s o n their somatic spikes (H4 a n d H3 neurons); n e u r o n s with brief, s m o o t h somatic spikes (F a n d H1 n e u r o n s ) had short LCIs even when their CV was low. Thus, the m a g n i tude of the i n t r a g a n g l i o n i c b r a n c h point filtering action is different for different types
19 of D R G neurons and can be most reliably predicted by taking into account the shape of the neuron's somatic action potential and its orthodromic CV. Neurons with marked h u m p s on their somatic spikes and slowly conducting axons will develop intermittent conduction for orthodromic impulses at a lower frequency (~< 100-200 Hz) than neurons with brief, smooth somatic spikes (~<550 Hz). Since amphibian primary afferents discharge at rates as high as 200-400 Hz to receptor stimulation at r o o m temperature [14], branch point filtering action could be of physiological significance for some D R G neurons. In conclusion, these results show that the branch point o f primary afferent fibers in frog D R G can influence orthodromic impulse conduction and that some have a stronger filtering action than others. Certain slowly conducting D R G neurons m a y be considered as relatively p o o r neural signalers due to limitations imposed on their m a x i m u m frequency o f firing by the low safety factor o f their axon branch points. Conversely, other D R G neurons (with brief, smooth somatic action potentials) have found a way to minimize branch point filtering action, thereby increasing the 'bandwidth' and information content [5] o f their neural signals. Whether such differences can be accounted for by unusually short peribifurcation internodes reported for some D R G neurons of amphibians [8] and cats [11] and/or the relative proximity of the high capacitance/low resistance cell body to the branch point remains to be determined. Special thanks to Margene Attaway for excellent technical assistance. Supported by BRS 2 S07 R R 05365 23.
1 Barton, D.H. and Matthews, B.H.C., Intermittent conduction in the spinal cord, J. Physiol. (Lond.), 85 (1935) 73-103. 2 Bullock, T.H., Orkand, R. and Grinnell, A., Introduction to Nervous Systems, W.H. Freeman, San Francisco, 1977, 559 pp. 3 Cunnane, T.C. and Stjame, L., Frequency dependent intermittency and ionic basis of impulseconduction in postganglionicsympathetic fibers of guinea-pig vas deferens, Neuroseience, 11 (1984) 211-229. 4 Dun, F.T., The delay and blockage of sensory impulsesin the dorsal root ganglion, J. Physiol. (Lond.), 127 (1955) 252-264. 5 Gabor, D., Theory of communication, J. Inst. Electr. Eng. (Lond.), 93 (1946) 429-457. 6 Grossman, Y. and Kendig, J.J., General anesthetic block of a bifurcating axon, Brain Res., 245 (1982) 148-153. 7 Ito, M. and Saiga, M., The mode of impulse transmission through the spinal ganglion, 3ap. J. Physiol., 9 (1959) 33-42. 8 Ito, M. and Takahashi, I., Impulse conduction through the spinal ganglion. In Y. Katsuke (Ed.), Electrical Activity of Single Cells, Ikagu Shoin, Tokyo, 1960, pp. 159-179. 9 Krnjevic, K. and Miledi, R., Presynaptic failure of neuromuscular propagation in rats, J. Physiol. (Lond.), 149 (1959) 1-22. 10 Levy, R.A., Presynaptic control of input in the central nervous system, Canad. J. Physiol. Pharm., 58 (1980) 751-766. 11 Lieberman, A.R., Sensory ganglia. In D.N. Landon (Ed.), The Peripheral Nerve, Halsted, New York, 1976, pp. 188-278. 12 Parnas, I., Spira, M.E., Werman, R. and Bergmann, F., Non-homogeneousconduction in giant axons of the nerve cord of Periplantea americana, J. Exp. Biol., 50 (1969) 635-649.
20 13 Reeves, R.B., Intracellular pH in cold blooded vertebrates as a function of body temperature. Resp. Physiol., 28 (1976) 29-47. 14 Shimada, K, and Yai, H., Some properties of cutaneous mechanosensory units in toads, Jap. J, Physiol., 10 (19601 555-570. 15 Stoney, S.D., Jr. and Hershberger, E., Cooling and antiepilcptic drugs depress impulse conduction at branch points of single myelinated fibers, Soc. Neurosci. Abstr., 10 (1984) 109. 16 Swadlow, H.A., Kocsis, A. and Waxman, S.G., Modulation of impulse conduction along the axonal tree, Annu. Rev. Biophys. Biomed. Eng., 9 (1980) 143-179. 17 Tasaki, I., The excitatory and recovery processes in the nerve fiber as modified by temperature changes, Biochim. Biophys. Acta, 3 (1949)498-5(19. 18 Tasaki, I., Nervous Transmission, Thomas, Springfield, IL, 1953. 164 pp. 19 Tauc, T. and Hughes, G.M., Modes of initiation and propagation of spikes in the branching axons of molluscan central neurons, J. Gen. Physiol., 46 (19631 533-549. 20 Yoshida, S. and Matsuda, Y., Studies on sensory neurons of the mouse with intracellular-recording and horseradish peroxidase-injection techniques, J. Neurophysiol., 42 (19791 1134-1145.