- Email: [email protected]
Comparison of spontaneous motor pattern generation in non-hemisected and hemisected mouse spinal cord
116
Neuroscience Letters, 144 (1992) 116-120 Q 1992 Elsevier Scientific Publishers Ireland Ltd. All rights reserved 0304-3940/92/$ 05.00
NSL 08934
Comparison of spontaneous motor pattern generation in non-hemisected and hemisected mouse spinal cord Y. Tao and M.H. Droge Department of Biology, Texas Woman’s University, Demon. TX 76204 (USA J (Received 1 May 1992; Revised version received 8 June 1992; Accepted 9 June 1992) Key words:
Spinal cord; Lumbar; Locomotion; Mammalian; In vitro; Central pattern generation
Spontaneous electromyogram (EMG) patterns in the gastrocnemius (G) and tibialis anterior (TA) muscles of spinal cord-hindlimb explants from neonatal mice were investigated. Compared to non-hemisected explants, neither longitudinal hemisection of the spinal cord nor hemisection plus transection at L, significantly altered the incidence of spontaneous motor rhythm. Therefore, not only does each half of the neonatal spinal cord contain sufficient circuitry to generate motor rhythm, but the more reduced preparations were just as likely to produce such activity. Hemisected preparations, however, exhibited slower rhythm, perhaps due to the loss of excitatory commissural connections. No correlation was found between the number of cycles in a rhythmic sequence and cycle period. In hemisected as well as non-hemisected explants, sequences of spontaneous EMG rhythm occurred in either the G or TA muscle, but not in both muscles simultaneously. Consequently. cycle-to-cycle alternation between rhythmic bursting in the G and TA muscles was not observed. The excitability in such preparations was apparently insufficient for maintained activations of both muscles (either for cycle-to-cycle alternation or for co-contraction).
Many researchers interested in central pattern generation in vertebrates have begun work on in vitro preparations to achieve greater access to and control over the circuitry responsible for rhythmogenesis. While in vitro studies of the lamprey spinal cord have contributed greatly to our understanding, much progress has also been made from in vitro work on chick embryos [l 11, Xenopus embryos [ 121 and rats [13]. Virtually all such studies have focused on locomotor patterns that are induced by pharmacological and/or afferent stimulation. More information is needed from in vitro mammalian preparations, especially regarding spontaneous pattern generating capabilities. Longitudinally hemisected spinal cord-hindlimb explants from newborn mice [3, 81 and rats [lo] as well as non-hemisected spinal cord-brainstem explants from rats [ 131 have been established most recently as in vitro locomotor preparations. Hernandez et al. [8] were the first to raise the issue of spontaneous pattern generating capabilities in in vitro preparations. They reported that spon-
Correspondence; M.H. Droge, Department of Biology. Texas Woman’s University, P.O. Box 23971, Denton. TX 76204, USA.
taneous rhythm was an infrequent occurrence that lacked the cycle-to-cycle alternation between electromyogram (EMG) activity in gastrocnemius (G) and tibialis anterior (TA) muscles. Although Kudo and Yamada [lo] did not study spontaneous pattern generation, they did observe alternation between G and TA muscles in
N-methyl-D-aspartic acid (NMDA)-evoked rhythm. Considering the apparent paucity of spontaneous activity and especially spontaneous motor rhythm, in vitro, it is not surprising that this type of activity has received so little attention. Not only have the in vitro studies involved different species and donor ages, the specific type of preparation has varied. Such differences no doubt contribute to conflicting data. For example, Kudo and Yamada [lo] reported a much slower NMDA-evoked rhythm after hemisecting spinal cord-hindlimb explants from neonatal rats. In contrast, no difference in the frequency of rhythmic ventral root potentials was found in an in vitro study of tadpoles; a system in which alternating rhythm involves both sides of the spinal cord [9]. Whether these findings reflect differences in pattern generating capability between mammals and other species is uncertain. A comparison of spontaneous motor rhythm in hemisected and non-hemisected explants from mice would be
117 useful for several reasons. Comparing spontaneous activity in explants at several levels of surgical reduction could provide a better understanding of the relationship between explant size and baseline pattern generating capability. There is always the problem of trauma and tissue damage from explantation surgery and the potentially adverse effects thereof on the physiological and pharmacological status of the preparation. If locomotorlike rhythm can occur spontaneously in a particular in vitro preparation, then one can have even more confidence in that system as a model for further study of pattern generation. Finally, an analysis of spontaneous in vitro activity in mouse spinal tissue would help researchers interpret other studies of hemisected and non-hemisected spinal preparations from mammals as well as from non-mammalian vertebrates. The objective of the present study was to compare spontaneous pattern generation in hemisected and nonhemisected in vitro preparations from neonatal mice. This work should provide useful data for comparing pharmacological results from similar in vitro preparations. Portions of these data have been reported in abstract form [14]. The experiments were performed on spinal cord-hindlimb explants prepared from Balb/c mice ranging in age from birth to 3 days post-partum. Neonatal tissue was selected because it can exhibit the motor behavior of interest and is less susceptible to the transient hypoxia associated with in vitro preparations. Standard (EMG) techniques were used to record activity in the G and TA muscles as described in Hernandez et al. [8]. Methoxyflurane anesthesia was maintained throughout electrode placements and all surgical procedures. Ice chips were placed on the animal to lower body temperature and to minimize the effects of any hypoxia. Bipolar EMG electrodes were implanted prior to the excision of the spinal cord and hindlimbs. The electrodes were constructed from Teflon coated silver wire with a total diameter of 0.003 in (Medwire # AG3T). The 2 electrode tips were separated by at least 0.25 mm but not over 0.50 mm. Small incisions were made in the skin of one hindlimb so that separate EMG electrodes could be inserted into the TA and G muscles. After a sufficient length of wire was provided for limb movement, each electrode was secured to the animal's ankle with a drop of cyanoacrylic cement. After the EGM electrodes were securely in place, the explantation surgery was begun. This surgery included the excision of the vertebral column from T9-Ti0 caudally, together with the hindlimbs. The tissue was transferred to an isolated support chamber containing 4°C artificial cerebrospinal fluid (ACSF; see below) where the surgery was completed. A dorsal laminectomy was
performed to expose the spinal cord and the dorsal pia mater was opened along the midline to increase the access of bath-applied drugs. The ACSF included (in mM concentration): NaC1 128.0, KCI 3.0, CaC12 1.5, MgSO4 1.0, NaHCO3 21.0, NaH2PO4 0.5, glucose 30.0. This solution was equilibrated with a 95% 02 and 5% CO2 gas mixture. To supplement oxygen further, 0.003% H202 was added to the ACSF. Since Mg 2~ can block NMDA transmission, ACSF containing Mg 2+ was used during surgery and the postoperative stabilization period to reduce the chance of excitatory amino acid (EAA) toxicity (as a result of trauma-related increase in EAA release). During subsequent recording sessions, however, the perfusate was switched to Mg2+-free ACSF to promote NMDA transmission. When Mg 2+ was omitted from the ACSF, the concentration of CaC12 was altered to preserve normal osmolarity. All ACSF was administered via a gravity perfusion system at a flow rate of 5 ml/min, which represented a minimum of one total exchange per minute. In 72 experiments, three variations of the spinal cordhindlimb preparation were tested. They included: (1) 21 spinal explants from Tg-TI0 caudally (the non-hemisected preparations); (2) 23 spinal explants longitudinally hemisected from Tg-T10 caudally (the hemisected preparations); and (3) 28 spinal explants transected at Ll and longitudinally hemisected from L~ caudally (the hemisected-lumbar preparations). When surgery was completed, each explant was allowed to return to room temperature while submerged in oxygenated ACSF. After allowing 60 min for each explant to recover from surgery, spontaneous EMG activity was recorded. EMG signals were led via WPI differential preamplifiers to a storage oscilloscope and data were collected using a Polaroid Oscilloscope Camera and a 4-Channel Gould 3400 Chart Recorder. All recordings were obtained at filter settings of 0.1-10 KHz. Motor rhythm was evaluated by the cycle periods in sequences of EMG bursts. A cycle period is the time from the beginning of one burst of impulses to the beginning of the next subsequent burst (in that muscle). A rhythmic sequence was operationally defined as a series of bursts that created at least 5 consecutive cycle periods and maintained a regularity in timing that was consistent with that observed during locomotion. These criteria were based on EMG data from intact, freely moving mice [8]. It is the duration of cycle periods in rhythmic EMG activity that is used in this paper as an index of the timing of motor rhythm. Measurements of cycle periods were obtained using a Zeiss Videoplan Image Analysis System. The statistical analyses included the Student's t-test, the Student's paired t-test, and the one-way ANOVA. The Scheffe test
118 was used for post hoc comparisons [15]. An c~ level of 0.05 was required for rejection of the null hypothesis. The spontaneous EMG activity in the in vitro mouse spinal cord-hindlimb preparations included phasic spiking and/or bursting. Fig. 1A shows an example of spontaneous EMG activity in non-hemisected spinal cordhindlimb preparations and Fig. 1B,C shows spontaneous activity in hemisected preparations. Prolonged sequences of bursting and spiking (e.g. Fig. 1A,B) as well as shorter episodes of such activity (e.g. Fig. IC) occurred in TA and G muscles. Even though these recordings may appear at first to contain totally random activity, they do include brief occurrences of rhythm (e.g. Fig. I C arrow), according to our operational definition of such activity. Since the majority of the EMG activity in these recordings was non-rhythmic, such traces were not included in our measurements of motor rhythm. Concurrent activations of EMG bursts in G and TA muscles did occur in cases where the activity was predominately non-rhythmic (Fig. 1C). Episodes of rhythmic spontaneous activity that were clear and distinct were observed in 34.5% Of all experiments (25 out of 72). Most often these episodes included tens to hundreds of consecutive cycle periods. While stepping-like movements of one or both hindlimbs were occasionally observed, most EMG recordings were not accompanied by detectable limb movement. The 72 experiments included 3 variations of the spinal cord-hindlimb preparation. Spontaneous rhythm occurred in 33.3% of the non-hemisected explants (7 out of 21 ), 43.5% of the hemisected explants (10 out of 23), and 28.9% of the hemisected-lumbar explants (8 out of 28). A z-square test showed there was no significant difference in the incidence of spontaneous rhythmic EMG activity among these 3 types of preparations (Ze=1.2589: P>-0.092). When the timing of rhythmic patterns was compared within each type of preparation, a significant difference was found. The F values for separate one-way ANOVA tests on non-hemisected, hemisected and hemisectedlumbar preparations were 290.44, 153.88 and 3325.93, respectively (all P<0.001). In fact, the timing of separate rhythmic sequences from the same animal differed significantly. The analysis of timing was based on rhythmic sequences of bursting that contained varying numbers of rhythmic cycles. Since no correlation was found between the number of cycles per rhythmic sequence versus the mean cycle period in any preparation it was valid to compare the timing of rhythmic sequences having different numbers of" cycles. The mean cycle period _+ S.D. for rhythmic EMG activity recorded from these 3 types of in vitro preparations was 90_+25 ms (n--222), 97+27 ms (n=545), and 96-+28 ms
(n=798), respectively. Although these mean values may not appear different, a one-way ANOVA showed there was a significant difference among the 3 types of in vitro preparations (~.15~,~=5.56: P=0.039). The Scheffe post hoc test indicated that the cycle periods differed between non-hemisected versus hemisected explants (~)s~,=5.47: P-<0.001), and non-hemisected versus hemisected-lumbar explants (F~0~=3.78: P=0.006). There was no significant difference between the hemisected and hemisectedlumbar in vitro preparation (Ft34~=0.48: P->0.332). Therefore, the frequency of the motor rhythm was slower after spinal cords were hemisected. The spontaneous rhythmic activity in G and TA muscles did not alternate on a cycle-to-cycle basis. Sequences of rhythmic bursting typically occurred in either the TA or the G, but not in both muscles concurrently. Fig. 2 shows 2 different rates of rhythmic bursting in the TA muscle while the G muscle was relatively quiescent. The mean cycle periods + S.D. and coefficient of variation (V) for Fig. 2A,B were 101_+14 ms (n=81), V=14.1%, and 122_+9 ms (n=102), V-7.6%, respectively. In each case the rhythms were highly regular in timing. Spontaneous EMG activity was typically non-rhythmic in the 3 types of in vitro preparations tested (the spinal cord-hindlimb preparation with or without a longitudinal hemisection, and the hemisected lumbar preparation). Such non-rhythmic phasic activity included episodes of spiking and/or bursting that lasted from a few seconds to many minutes. The occurrence of some spontaneous rhythmic activity in non-hemisected (33.3%), hemisected (43.5%), as well as hemisected-lumbar (28.9%) preparations confirms that each half of the mammalian spinal cord is capable of pattern generation. That spontaneous EMG activity also included rhythmic bursting is significant; it indicates that the preparation was not only viable but had a level of excitability sufficient to generate motor patterns. In fact, crude stepping movements were also observed occasionally. The incidence of spontaneous rhythm observed in this study is higher than has been previously reported for any in vitro mammalian spinal cord preparation. One factor that could have contributed to the increase in spontaneous activity compared to previous work was the use of smaller E M G electrodes, which would reduce tissue damage in the hindlimb. The diameter of the electrodes used in this study was 0.003 in compared to the 0.005 in diameter EMG electrodes used by Hernandez et al. [8] on neonatal mice. Future E M G studies of a more proximal set of limb muscles, such as the gluteus superficialis (hip flexor) and semitendinosus (hip extensor), may reveal an even greater incidence of spontaneous motor rhythms in this in vitro mouse preparation.
119
A
JJiU[
L
A
,
f
lllliUuuiii ,ta ll Itttt L[II L 8 +~t.
I i
+~.l
.ll,lti
......
. i.,ild ~1 a_.i~[ ljlta.lit~.l
Ii1
.Lam.,l
.... lialai[
~+,.i...+:~:~;++
li~ak
:
..+
im~.ik+~
l
=iilii. ldi, ~lil It. llii,J'li'~l.ll-lii~l.<.
ttl ! l.l 'l,l!l r lll I.l llri, lllllll III1 [I llll!nml 'l! I
,
I
rIImmrl!llfrrr!rr rrrrlTi
{
,
r
I q,,
i
Fig. 1. Spontaneous E M G activity in 2 different mouse spinal cordhindlimb preparations. Top trace in each group is gastrocnemius recording. Bottom trace in each group is tibialis anterior recording. A: spontaneous activity in a non-hemisected spinal cord-hindlimb explant. B,C: spontaneous activity in 2 different hemisected explants. A,B: prolonged sequences of bursting and spiking. C: short sequences of bursting and spiking. Arrow indicates an example of transient rhythmic bursting. Bar -- 1 s.
The use of high [Ca 2÷] and no Mg 2+ in the perfusate probably contributed to the spontaneous rhythm observed in this study. Czeh and Somjen [2] reported that ventral root activity could be 'induced' in isolated mouse spinal cords bathed in high Ca 2+ and low Mg 2+. For example, when the [Ca 2÷] was raised from 1.2 to 1.8 mM, ventral root activity occurred in 66.7% of their experiments. Increasing the [Ca2+] to 2.4 mM yielded a 100% response. Spontaneous activity also appeared in most cases when the [Mg2+] was lowered to 0.4 mM [2]. It is not surprising, therefore, that a higher incidence of spontaneous EMG rhythm was observed in this study, since the bath contained 2.5 mM Ca 2+ and no Mg 2÷. The timing of rhythm observed in hemisected mouse spinal cords was slower than that for rhythm in nonhemisected preparations. It is possible that the loss of excitatory commissural connections in hemisected preparations reduced the excitability within pattern generating networks and resulted in a slower rhythm. While this study cannot confirm that this is the case, it raised the question as to whether commissural connections in the mammalian spinal cord have more influence on CPG function than in certain other vertebrate species. The frequency of the rhythmic activity in an in vitro rat prepara-
LuL.I LIt2.h.ti t LitLit k. t I.I.L ..t,t LL.
,, ..
Ir rrrrr rI=rt rrrrrrrr rrrrr .... Fig. 2. Spontaneous rhythmic E M G activity in a hemisected spinal cord-hindlimb preparation. Top trace in each group is gastrocnemius recording. Bottom trace in each group is tibialis anterior recording. A: rhythmic bursting in the TA during relative quiescence in the G muscle. B: recordings from the same TA muscle to show slower rhythmic bursting in the same explant. Bar - 1 s.
tion was also found to be slower after splitting the cord [10]. However, a similar study of tadpoles showed the frequency of rhythmic ventral root potentials to be virtually unchanged by the same operation [9]. The most striking result was the observation that spontaneous motor rhythm did not alternate on a cycleto-cycle basis between the G and TA muscles. Grillner [6] proposed that each spinal generator has paired oscillators for extensor and flexor muscles, and they exert strong inhibitory action on each other. In intact animals, spinal CPGs are believed to be simultaneously activated by higher brain centers, while segmental inhibition creates the cycle-to-cycle alternation between antagonists that characterizes locomotion [5]. The lumbar spinal cord and therefore the circuitry containing the inhibitory interneurons responsible for cycle-to-cycle alternation was included in the preparations tested here. The absence of cycle-to-cycle alternations in rhythmic EMG bursting in the G and TA muscles, in vitro, indicates that the expression of reciprocal inhibitory coupling between the antagonist motor nuclei is altered in such preparations. However, entire sequences of rhythmic bursting activity were found to switch from one muscle to the other, suggesting that some form of reciprocal inhibition is preserved. Glycine has been shown to be required for cycle-to-cycle alternations between antagonist swim muscles in adult lampreys [1]. Both glycine and GABA should be further investigated in future attempts to understand the central generation of alternating rhythmic patterns in mammals.
120
The authors wish to thank Drs. L. Higgins and L. Uphouse for their critical review of the manuscript. Dr. Higgins' work in preparing the figures was also much appreciated. This work was funded by NIH NS25250 and an institutional grant from Texas Woman's University. 1 Alford, S. and Williams, T.L., Endogenous activation of glycine and NMDA receptors in lamprey spinal cord during ficlive locomotion, J. Neurosci., 9 (1989) 2792 28011. 2 Czeh, G. and Somjen, G.G., Spontaneous activity induced in isolated mouse spinal cord by high extracellular calcium and by low extracellular magnesium, Brain Res., 495 (1989) 89 99. 3 Droge, M.H., Gross, G.W., Hightower, M.H. and Czisny, L.E., Multietectrode analysis of coordinated, multisite, rhythmic bursting in cultured CNS monolayer networks. J. Neurosci., 6 (1986) 1583 1592. 4 Garthwaite, J., Garthwaite, G. and Hajos. F., Amino acid neurotoxicity: relationship to neuronal depolarization in rat cerebellar slices, Neuroscience, 18 (1986)449 460. 5 Gelfand, I.M., Orlovsky, G.N. and Shik, M.L., Locomotion and scratching in tetrapods. In Cohen el al. (Eds.), Neural Control of Rhythmic Movements in Vertebrates, Wiley. New York, 1988, pp. 167 201. 6 Grillner, S., Control of locomotion in bipeds, tetrapods, and fish. In V.B. Brooks tEd.), Handbook of Physiology, Section I. The Nervous System, Vol. 2., Motor control. American Physiological Society, Bethesda. 1981, pp. 1179 1236.
7 Hajos, F.. Garthwaite, G. and Garthwaite, J.. Reversible and irrc versible neuronal damage caused by excitatory amino acid analogues in rat cerebella r slices, Neuroscience, 18 i1986)417 436. 8 Hernandez, R, Elbert, K. and Droge, M.H., Spontaneous and NMDA evoked motor rhythms in the neonatal mouse spinal cord: an in vitro study with comparisons to in situ activity. Exp. Brain Res., 85 (1991) 66 74. 9 Kahm J.A. and Roberts, A., Experiments on the central pattern generator for swimming in amphibian embryo, Philos. Trans, R. Soc. London Ser. B, 296 (1982) 229 243. 10 Kudo, N. and Yamada. T., N-Methyl-D,t,-aspartate-induced locomotor activity in a spinal cord-hindlimb preparation of the newborn rat studied in vitro, Neurosci. Lett., 75 (1987) 43 48. 11 O'Donovan. M.J., Developmental approaches to the analysis of vertebrate central pattern generators, J. Neurosci. Methods, 21 11987) 275 286. 12 Roberts, A., Soffe, S,R. and Dale, N., Spinal interneurons and swimming in frog embryos. In S, Grillner et al. (Ed./, Neurobiology of Vertebrate Locomotion, Macmillan, London, 1986, pp. 279-306. 13 Smith, J.C., Feldman, J.L. and Schmidt, B.J., Neural mechanisms generating locomotion studied in mammalian brain stem-spinal cord in vitro, FASEB J,, 2 (19881 2283 2288. 14 Tao. Y. and Droge, M.H., Spontaneous motor rhythm and glycine effects on rhythm generation in mouse spinal cord, in vitro, Soc. Neurosci. Abstr., 1991. 15 Zar. J.H., Biostatistical Analysis, 2nd edn., Prentice-Hall, Englewood Cliffs, New Jersey, 1984.
Neuroscience Letters, 144 (1992) 116-120 Q 1992 Elsevier Scientific Publishers Ireland Ltd. All rights reserved 0304-3940/92/$ 05.00
NSL 08934
Comparison of spontaneous motor pattern generation in non-hemisected and hemisected mouse spinal cord Y. Tao and M.H. Droge Department of Biology, Texas Woman’s University, Demon. TX 76204 (USA J (Received 1 May 1992; Revised version received 8 June 1992; Accepted 9 June 1992) Key words:
Spinal cord; Lumbar; Locomotion; Mammalian; In vitro; Central pattern generation
Spontaneous electromyogram (EMG) patterns in the gastrocnemius (G) and tibialis anterior (TA) muscles of spinal cord-hindlimb explants from neonatal mice were investigated. Compared to non-hemisected explants, neither longitudinal hemisection of the spinal cord nor hemisection plus transection at L, significantly altered the incidence of spontaneous motor rhythm. Therefore, not only does each half of the neonatal spinal cord contain sufficient circuitry to generate motor rhythm, but the more reduced preparations were just as likely to produce such activity. Hemisected preparations, however, exhibited slower rhythm, perhaps due to the loss of excitatory commissural connections. No correlation was found between the number of cycles in a rhythmic sequence and cycle period. In hemisected as well as non-hemisected explants, sequences of spontaneous EMG rhythm occurred in either the G or TA muscle, but not in both muscles simultaneously. Consequently. cycle-to-cycle alternation between rhythmic bursting in the G and TA muscles was not observed. The excitability in such preparations was apparently insufficient for maintained activations of both muscles (either for cycle-to-cycle alternation or for co-contraction).
Many researchers interested in central pattern generation in vertebrates have begun work on in vitro preparations to achieve greater access to and control over the circuitry responsible for rhythmogenesis. While in vitro studies of the lamprey spinal cord have contributed greatly to our understanding, much progress has also been made from in vitro work on chick embryos [l 11, Xenopus embryos [ 121 and rats [13]. Virtually all such studies have focused on locomotor patterns that are induced by pharmacological and/or afferent stimulation. More information is needed from in vitro mammalian preparations, especially regarding spontaneous pattern generating capabilities. Longitudinally hemisected spinal cord-hindlimb explants from newborn mice [3, 81 and rats [lo] as well as non-hemisected spinal cord-brainstem explants from rats [ 131 have been established most recently as in vitro locomotor preparations. Hernandez et al. [8] were the first to raise the issue of spontaneous pattern generating capabilities in in vitro preparations. They reported that spon-
Correspondence; M.H. Droge, Department of Biology. Texas Woman’s University, P.O. Box 23971, Denton. TX 76204, USA.
taneous rhythm was an infrequent occurrence that lacked the cycle-to-cycle alternation between electromyogram (EMG) activity in gastrocnemius (G) and tibialis anterior (TA) muscles. Although Kudo and Yamada [lo] did not study spontaneous pattern generation, they did observe alternation between G and TA muscles in
N-methyl-D-aspartic acid (NMDA)-evoked rhythm. Considering the apparent paucity of spontaneous activity and especially spontaneous motor rhythm, in vitro, it is not surprising that this type of activity has received so little attention. Not only have the in vitro studies involved different species and donor ages, the specific type of preparation has varied. Such differences no doubt contribute to conflicting data. For example, Kudo and Yamada [lo] reported a much slower NMDA-evoked rhythm after hemisecting spinal cord-hindlimb explants from neonatal rats. In contrast, no difference in the frequency of rhythmic ventral root potentials was found in an in vitro study of tadpoles; a system in which alternating rhythm involves both sides of the spinal cord [9]. Whether these findings reflect differences in pattern generating capability between mammals and other species is uncertain. A comparison of spontaneous motor rhythm in hemisected and non-hemisected explants from mice would be
117 useful for several reasons. Comparing spontaneous activity in explants at several levels of surgical reduction could provide a better understanding of the relationship between explant size and baseline pattern generating capability. There is always the problem of trauma and tissue damage from explantation surgery and the potentially adverse effects thereof on the physiological and pharmacological status of the preparation. If locomotorlike rhythm can occur spontaneously in a particular in vitro preparation, then one can have even more confidence in that system as a model for further study of pattern generation. Finally, an analysis of spontaneous in vitro activity in mouse spinal tissue would help researchers interpret other studies of hemisected and non-hemisected spinal preparations from mammals as well as from non-mammalian vertebrates. The objective of the present study was to compare spontaneous pattern generation in hemisected and nonhemisected in vitro preparations from neonatal mice. This work should provide useful data for comparing pharmacological results from similar in vitro preparations. Portions of these data have been reported in abstract form [14]. The experiments were performed on spinal cord-hindlimb explants prepared from Balb/c mice ranging in age from birth to 3 days post-partum. Neonatal tissue was selected because it can exhibit the motor behavior of interest and is less susceptible to the transient hypoxia associated with in vitro preparations. Standard (EMG) techniques were used to record activity in the G and TA muscles as described in Hernandez et al. [8]. Methoxyflurane anesthesia was maintained throughout electrode placements and all surgical procedures. Ice chips were placed on the animal to lower body temperature and to minimize the effects of any hypoxia. Bipolar EMG electrodes were implanted prior to the excision of the spinal cord and hindlimbs. The electrodes were constructed from Teflon coated silver wire with a total diameter of 0.003 in (Medwire # AG3T). The 2 electrode tips were separated by at least 0.25 mm but not over 0.50 mm. Small incisions were made in the skin of one hindlimb so that separate EMG electrodes could be inserted into the TA and G muscles. After a sufficient length of wire was provided for limb movement, each electrode was secured to the animal's ankle with a drop of cyanoacrylic cement. After the EGM electrodes were securely in place, the explantation surgery was begun. This surgery included the excision of the vertebral column from T9-Ti0 caudally, together with the hindlimbs. The tissue was transferred to an isolated support chamber containing 4°C artificial cerebrospinal fluid (ACSF; see below) where the surgery was completed. A dorsal laminectomy was
performed to expose the spinal cord and the dorsal pia mater was opened along the midline to increase the access of bath-applied drugs. The ACSF included (in mM concentration): NaC1 128.0, KCI 3.0, CaC12 1.5, MgSO4 1.0, NaHCO3 21.0, NaH2PO4 0.5, glucose 30.0. This solution was equilibrated with a 95% 02 and 5% CO2 gas mixture. To supplement oxygen further, 0.003% H202 was added to the ACSF. Since Mg 2~ can block NMDA transmission, ACSF containing Mg 2+ was used during surgery and the postoperative stabilization period to reduce the chance of excitatory amino acid (EAA) toxicity (as a result of trauma-related increase in EAA release). During subsequent recording sessions, however, the perfusate was switched to Mg2+-free ACSF to promote NMDA transmission. When Mg 2+ was omitted from the ACSF, the concentration of CaC12 was altered to preserve normal osmolarity. All ACSF was administered via a gravity perfusion system at a flow rate of 5 ml/min, which represented a minimum of one total exchange per minute. In 72 experiments, three variations of the spinal cordhindlimb preparation were tested. They included: (1) 21 spinal explants from Tg-TI0 caudally (the non-hemisected preparations); (2) 23 spinal explants longitudinally hemisected from Tg-T10 caudally (the hemisected preparations); and (3) 28 spinal explants transected at Ll and longitudinally hemisected from L~ caudally (the hemisected-lumbar preparations). When surgery was completed, each explant was allowed to return to room temperature while submerged in oxygenated ACSF. After allowing 60 min for each explant to recover from surgery, spontaneous EMG activity was recorded. EMG signals were led via WPI differential preamplifiers to a storage oscilloscope and data were collected using a Polaroid Oscilloscope Camera and a 4-Channel Gould 3400 Chart Recorder. All recordings were obtained at filter settings of 0.1-10 KHz. Motor rhythm was evaluated by the cycle periods in sequences of EMG bursts. A cycle period is the time from the beginning of one burst of impulses to the beginning of the next subsequent burst (in that muscle). A rhythmic sequence was operationally defined as a series of bursts that created at least 5 consecutive cycle periods and maintained a regularity in timing that was consistent with that observed during locomotion. These criteria were based on EMG data from intact, freely moving mice [8]. It is the duration of cycle periods in rhythmic EMG activity that is used in this paper as an index of the timing of motor rhythm. Measurements of cycle periods were obtained using a Zeiss Videoplan Image Analysis System. The statistical analyses included the Student's t-test, the Student's paired t-test, and the one-way ANOVA. The Scheffe test
118 was used for post hoc comparisons [15]. An c~ level of 0.05 was required for rejection of the null hypothesis. The spontaneous EMG activity in the in vitro mouse spinal cord-hindlimb preparations included phasic spiking and/or bursting. Fig. 1A shows an example of spontaneous EMG activity in non-hemisected spinal cordhindlimb preparations and Fig. 1B,C shows spontaneous activity in hemisected preparations. Prolonged sequences of bursting and spiking (e.g. Fig. 1A,B) as well as shorter episodes of such activity (e.g. Fig. IC) occurred in TA and G muscles. Even though these recordings may appear at first to contain totally random activity, they do include brief occurrences of rhythm (e.g. Fig. I C arrow), according to our operational definition of such activity. Since the majority of the EMG activity in these recordings was non-rhythmic, such traces were not included in our measurements of motor rhythm. Concurrent activations of EMG bursts in G and TA muscles did occur in cases where the activity was predominately non-rhythmic (Fig. 1C). Episodes of rhythmic spontaneous activity that were clear and distinct were observed in 34.5% Of all experiments (25 out of 72). Most often these episodes included tens to hundreds of consecutive cycle periods. While stepping-like movements of one or both hindlimbs were occasionally observed, most EMG recordings were not accompanied by detectable limb movement. The 72 experiments included 3 variations of the spinal cord-hindlimb preparation. Spontaneous rhythm occurred in 33.3% of the non-hemisected explants (7 out of 21 ), 43.5% of the hemisected explants (10 out of 23), and 28.9% of the hemisected-lumbar explants (8 out of 28). A z-square test showed there was no significant difference in the incidence of spontaneous rhythmic EMG activity among these 3 types of preparations (Ze=1.2589: P>-0.092). When the timing of rhythmic patterns was compared within each type of preparation, a significant difference was found. The F values for separate one-way ANOVA tests on non-hemisected, hemisected and hemisectedlumbar preparations were 290.44, 153.88 and 3325.93, respectively (all P<0.001). In fact, the timing of separate rhythmic sequences from the same animal differed significantly. The analysis of timing was based on rhythmic sequences of bursting that contained varying numbers of rhythmic cycles. Since no correlation was found between the number of cycles per rhythmic sequence versus the mean cycle period in any preparation it was valid to compare the timing of rhythmic sequences having different numbers of" cycles. The mean cycle period _+ S.D. for rhythmic EMG activity recorded from these 3 types of in vitro preparations was 90_+25 ms (n--222), 97+27 ms (n=545), and 96-+28 ms
(n=798), respectively. Although these mean values may not appear different, a one-way ANOVA showed there was a significant difference among the 3 types of in vitro preparations (~.15~,~=5.56: P=0.039). The Scheffe post hoc test indicated that the cycle periods differed between non-hemisected versus hemisected explants (~)s~,=5.47: P-<0.001), and non-hemisected versus hemisected-lumbar explants (F~0~=3.78: P=0.006). There was no significant difference between the hemisected and hemisectedlumbar in vitro preparation (Ft34~=0.48: P->0.332). Therefore, the frequency of the motor rhythm was slower after spinal cords were hemisected. The spontaneous rhythmic activity in G and TA muscles did not alternate on a cycle-to-cycle basis. Sequences of rhythmic bursting typically occurred in either the TA or the G, but not in both muscles concurrently. Fig. 2 shows 2 different rates of rhythmic bursting in the TA muscle while the G muscle was relatively quiescent. The mean cycle periods + S.D. and coefficient of variation (V) for Fig. 2A,B were 101_+14 ms (n=81), V=14.1%, and 122_+9 ms (n=102), V-7.6%, respectively. In each case the rhythms were highly regular in timing. Spontaneous EMG activity was typically non-rhythmic in the 3 types of in vitro preparations tested (the spinal cord-hindlimb preparation with or without a longitudinal hemisection, and the hemisected lumbar preparation). Such non-rhythmic phasic activity included episodes of spiking and/or bursting that lasted from a few seconds to many minutes. The occurrence of some spontaneous rhythmic activity in non-hemisected (33.3%), hemisected (43.5%), as well as hemisected-lumbar (28.9%) preparations confirms that each half of the mammalian spinal cord is capable of pattern generation. That spontaneous EMG activity also included rhythmic bursting is significant; it indicates that the preparation was not only viable but had a level of excitability sufficient to generate motor patterns. In fact, crude stepping movements were also observed occasionally. The incidence of spontaneous rhythm observed in this study is higher than has been previously reported for any in vitro mammalian spinal cord preparation. One factor that could have contributed to the increase in spontaneous activity compared to previous work was the use of smaller E M G electrodes, which would reduce tissue damage in the hindlimb. The diameter of the electrodes used in this study was 0.003 in compared to the 0.005 in diameter EMG electrodes used by Hernandez et al. [8] on neonatal mice. Future E M G studies of a more proximal set of limb muscles, such as the gluteus superficialis (hip flexor) and semitendinosus (hip extensor), may reveal an even greater incidence of spontaneous motor rhythms in this in vitro mouse preparation.
119
A
JJiU[
L
A
,
f
lllliUuuiii ,ta ll Itttt L[II L 8 +~t.
I i
+~.l
.ll,lti
......
. i.,ild ~1 a_.i~[ ljlta.lit~.l
Ii1
.Lam.,l
.... lialai[
~+,.i...+:~:~;++
li~ak
:
..+
im~.ik+~
l
=iilii. ldi, ~lil It. llii,J'li'~l.ll-lii~l.<.
ttl ! l.l 'l,l!l r lll I.l llri, lllllll III1 [I llll!nml 'l! I
,
I
rIImmrl!llfrrr!rr rrrrlTi
{
,
r
I q,,
i
Fig. 1. Spontaneous E M G activity in 2 different mouse spinal cordhindlimb preparations. Top trace in each group is gastrocnemius recording. Bottom trace in each group is tibialis anterior recording. A: spontaneous activity in a non-hemisected spinal cord-hindlimb explant. B,C: spontaneous activity in 2 different hemisected explants. A,B: prolonged sequences of bursting and spiking. C: short sequences of bursting and spiking. Arrow indicates an example of transient rhythmic bursting. Bar -- 1 s.
The use of high [Ca 2÷] and no Mg 2+ in the perfusate probably contributed to the spontaneous rhythm observed in this study. Czeh and Somjen [2] reported that ventral root activity could be 'induced' in isolated mouse spinal cords bathed in high Ca 2+ and low Mg 2+. For example, when the [Ca 2÷] was raised from 1.2 to 1.8 mM, ventral root activity occurred in 66.7% of their experiments. Increasing the [Ca2+] to 2.4 mM yielded a 100% response. Spontaneous activity also appeared in most cases when the [Mg2+] was lowered to 0.4 mM [2]. It is not surprising, therefore, that a higher incidence of spontaneous EMG rhythm was observed in this study, since the bath contained 2.5 mM Ca 2+ and no Mg 2÷. The timing of rhythm observed in hemisected mouse spinal cords was slower than that for rhythm in nonhemisected preparations. It is possible that the loss of excitatory commissural connections in hemisected preparations reduced the excitability within pattern generating networks and resulted in a slower rhythm. While this study cannot confirm that this is the case, it raised the question as to whether commissural connections in the mammalian spinal cord have more influence on CPG function than in certain other vertebrate species. The frequency of the rhythmic activity in an in vitro rat prepara-
LuL.I LIt2.h.ti t LitLit k. t I.I.L ..t,t LL.
,, ..
Ir rrrrr rI=rt rrrrrrrr rrrrr .... Fig. 2. Spontaneous rhythmic E M G activity in a hemisected spinal cord-hindlimb preparation. Top trace in each group is gastrocnemius recording. Bottom trace in each group is tibialis anterior recording. A: rhythmic bursting in the TA during relative quiescence in the G muscle. B: recordings from the same TA muscle to show slower rhythmic bursting in the same explant. Bar - 1 s.
tion was also found to be slower after splitting the cord [10]. However, a similar study of tadpoles showed the frequency of rhythmic ventral root potentials to be virtually unchanged by the same operation [9]. The most striking result was the observation that spontaneous motor rhythm did not alternate on a cycleto-cycle basis between the G and TA muscles. Grillner [6] proposed that each spinal generator has paired oscillators for extensor and flexor muscles, and they exert strong inhibitory action on each other. In intact animals, spinal CPGs are believed to be simultaneously activated by higher brain centers, while segmental inhibition creates the cycle-to-cycle alternation between antagonists that characterizes locomotion [5]. The lumbar spinal cord and therefore the circuitry containing the inhibitory interneurons responsible for cycle-to-cycle alternation was included in the preparations tested here. The absence of cycle-to-cycle alternations in rhythmic EMG bursting in the G and TA muscles, in vitro, indicates that the expression of reciprocal inhibitory coupling between the antagonist motor nuclei is altered in such preparations. However, entire sequences of rhythmic bursting activity were found to switch from one muscle to the other, suggesting that some form of reciprocal inhibition is preserved. Glycine has been shown to be required for cycle-to-cycle alternations between antagonist swim muscles in adult lampreys [1]. Both glycine and GABA should be further investigated in future attempts to understand the central generation of alternating rhythmic patterns in mammals.
120
The authors wish to thank Drs. L. Higgins and L. Uphouse for their critical review of the manuscript. Dr. Higgins' work in preparing the figures was also much appreciated. This work was funded by NIH NS25250 and an institutional grant from Texas Woman's University. 1 Alford, S. and Williams, T.L., Endogenous activation of glycine and NMDA receptors in lamprey spinal cord during ficlive locomotion, J. Neurosci., 9 (1989) 2792 28011. 2 Czeh, G. and Somjen, G.G., Spontaneous activity induced in isolated mouse spinal cord by high extracellular calcium and by low extracellular magnesium, Brain Res., 495 (1989) 89 99. 3 Droge, M.H., Gross, G.W., Hightower, M.H. and Czisny, L.E., Multietectrode analysis of coordinated, multisite, rhythmic bursting in cultured CNS monolayer networks. J. Neurosci., 6 (1986) 1583 1592. 4 Garthwaite, J., Garthwaite, G. and Hajos. F., Amino acid neurotoxicity: relationship to neuronal depolarization in rat cerebellar slices, Neuroscience, 18 (1986)449 460. 5 Gelfand, I.M., Orlovsky, G.N. and Shik, M.L., Locomotion and scratching in tetrapods. In Cohen el al. (Eds.), Neural Control of Rhythmic Movements in Vertebrates, Wiley. New York, 1988, pp. 167 201. 6 Grillner, S., Control of locomotion in bipeds, tetrapods, and fish. In V.B. Brooks tEd.), Handbook of Physiology, Section I. The Nervous System, Vol. 2., Motor control. American Physiological Society, Bethesda. 1981, pp. 1179 1236.
7 Hajos, F.. Garthwaite, G. and Garthwaite, J.. Reversible and irrc versible neuronal damage caused by excitatory amino acid analogues in rat cerebella r slices, Neuroscience, 18 i1986)417 436. 8 Hernandez, R, Elbert, K. and Droge, M.H., Spontaneous and NMDA evoked motor rhythms in the neonatal mouse spinal cord: an in vitro study with comparisons to in situ activity. Exp. Brain Res., 85 (1991) 66 74. 9 Kahm J.A. and Roberts, A., Experiments on the central pattern generator for swimming in amphibian embryo, Philos. Trans, R. Soc. London Ser. B, 296 (1982) 229 243. 10 Kudo, N. and Yamada. T., N-Methyl-D,t,-aspartate-induced locomotor activity in a spinal cord-hindlimb preparation of the newborn rat studied in vitro, Neurosci. Lett., 75 (1987) 43 48. 11 O'Donovan. M.J., Developmental approaches to the analysis of vertebrate central pattern generators, J. Neurosci. Methods, 21 11987) 275 286. 12 Roberts, A., Soffe, S,R. and Dale, N., Spinal interneurons and swimming in frog embryos. In S, Grillner et al. (Ed./, Neurobiology of Vertebrate Locomotion, Macmillan, London, 1986, pp. 279-306. 13 Smith, J.C., Feldman, J.L. and Schmidt, B.J., Neural mechanisms generating locomotion studied in mammalian brain stem-spinal cord in vitro, FASEB J,, 2 (19881 2283 2288. 14 Tao. Y. and Droge, M.H., Spontaneous motor rhythm and glycine effects on rhythm generation in mouse spinal cord, in vitro, Soc. Neurosci. Abstr., 1991. 15 Zar. J.H., Biostatistical Analysis, 2nd edn., Prentice-Hall, Englewood Cliffs, New Jersey, 1984.