- Email: [email protected]
The neural control of cyclic motor activity in the fetal rat (Rattus norvegicus)
Physiology&Behavior,Vol. 47, pp. 121-126. ©PergamonPress plc, 1990. Printedin the U.S.A.
0031-9384/90$3.00 + .00
The Neural Control of Cyclic Motor Activity in the Fetal Rat (Rattus norvegicus) STEVENS. ROBERTSON
Department of Human Development and Family Studies, Cornell University, Ithaca, NY 14853 AND W I L L I A M P. S M O T H E R M A N
Center for Developmental Psychobiology, Department of Psychology State University of New York at Binghamton, Binghamton, NY 13905 Received 18 A u g u s t 1989
ROBERTSON, S. S., AND W. P. SMOTHERMAN. The neuralcontrol of cyclic motor activity in the fetal rat (Rattus norvegicus). PHYSIOL BEHAV 47(1) 121-126, 1990.--The spontaneous behavior of rat fetuses (in a saline bath with fetal-placental-uterine connections intac0 was observed directly for 30 minutes on Day 20 of gestation. Rearleg and nonrearleg movements from fetuses with a mid-thoracic spinal cord transection or sham operation were analyzed for cyclic organization. Oscillations in rearlegactivity occurred at the same frequency in fetuses with spinal cord transections (0.74 cycle/rain) and sham-operated fetuses (0.72 cycle/rain). However, oscillations in nonrearleg activity were much slower in the fetuses with spinal cord transections (0.30 vs. 0.77 cycle/rain). Other characteristics of the cyclic patterns in motor activity were unaffected. The findings demonstrate 1) the caudal half of the spinal cord can generate cyclic output in the absence of descending input from the brain, 2) there is no single timing center, and 3) rostral sources are slower. Fetus
Rat
Motor activity
Cyclic
Spinal cord
OSCILLATION is a pervasive property of living systems which must be explained by any theory of spontaneous activity. In the human fetus and neonate, there are cyclic fluctuations in spontaneous motor activity comprised of general movements of the limbs, trunk, and head, and more isolated or stereotyped movements, with a cycle time of 1-5 minutes (10,17). The oscillations are apparent by midgestation (11), and qualitative observations using ultrasound suggest they may characterize fetal activity in the first trimester as well (3). Basic properties of the oscillations, such as their frequency and strength, remain virtually unchanged throughout the second half of gestation in normal fetuses (11), during a time when other aspects of neurobehavioral organization (e.g., behavioral state) change dramatically. The properties of fetal cyclic motor activity (CM) persist relatively unchanged after birth during active sleep (12). Although newborn CM is stronger and more regular in the nonsleep states compared to active sleep, its cyclic properties appear to be independent of the amount of movement (12). Fetal CM is disrupted by maternal diabetes, but returns to normal by the end of gestation (16), and remains so after birth even in newborns with evidence of increased glucose supply in utero (13). Thus, oscillations in spontaneous motor activity of the fetus and the neonate appear to be stable and robust. Their persistence with little if any quantitative change across birth
is consistent with other evidence of neurobehavioral continuity from prenatal to postnatal life, such as sleep state organization and isolated movement patterns (7). The oscillations may play a role in early neuromuscular development and regulate interaction with the environment (14), although these possibilities have not been tested. A full understanding of this behavioral pattern will require knowledge of its underlying mechanism. Two questions of considerable importance arise. First, can the spinal motor system generate the characteristic oscillations in general activity without descending neural input from the brain? If so, then a class of simple top-down models of CM, in which, for example, supraspinal fluctuations in general arousal might be responsible for the oscillations in motor activity, could be ruled out. In birds, evidence for the spinal origin of CM comes from a classic series of studies designed to investigate the role of sensory and supraspinal input on the motility oftbe chick embryo [e.g., (4, 6, 8, 9)]. In the human, qualitative descriptions of the exaggerated burst-pause patterns in the movements of anencephalic fetuses are also consistent with a spinal origin for CM (24). A second important question about the mechanism underlying CM is whether there is more than one source of oscillation? If so, then models which are based on a single, localized timing center 121
122
ROBERTSON AND SMOTHERMAN
would be ruled out and attention could be focused on characterizing the multiple sources, the process by which they generate cyclic output, and the dynamics of their interactions. The data on spontaneous motility and its neurophysiological correlates in the chick embryo cited above suggest that isolated sections of the spinal cord are capable of generating cyclic output, which is consistent with a multisource model. In the human, the relatively common finding of double peaks in the movement spectrum [e.g., (17)] is consistent with the existence of more than one source of oscillation which under some circumstances become uncoupled. The present experiment was designed to address these questions directly in a mammalian model of human CM. Recent work suggests that the fetal rat may be useful animal model in which to explore the neural substrate of CM (22). In that study, methods for the direct observation of fetal rat behavior (19), combined with analytic techniques for detecting and quantifying cyclicity in spontaneous motor activity (11), revealed oscillations in the spontaneous activity of fetuses delivered into a warm saline bath with fetal-placental-uterine connections intact. Quantitative differences were found between fetal rat and human CM, but they were not large. In the present experiment, the same observational and analytic techniques were used to study the spontaneous activity of fetal rats following transection of the spinal cord at a midthoracic level (or a sham operation). Rearleg and nonrearleg activity were analyzed to determine the presence and quantitative properties of any cyclic patterns generated above and below the transection. METHOD
Subjects Subject fetuses were the offspring of female Sprague-Dawley rats (Simonsen Laboratories, Gilroy, CA) time-mated to LongEvans males. Vaginal smears were taken daily to identify the day of conception (first detectable sperm=Day 0 of gestation). Females were housed in groups of three in polycarbonate cages (33 × 38 × 10 cm) until testing on Day 20 of gestation. Cages were kept in a temperature- and humidity-controlled colony room under 12:12 hr light/dark cycle (lights on at 0700); testing took place between 1300 and 1800. Females were provided with ad lib food and water and at all times were maintained in accordance with guidelines for animal care established by the National Institutes of Health, the International Society for Developmental Psychobiology and the Animal Behavior Society.
Surgical Preparation Direct observation of fetal behavior requires preparation of the pregnant female and manipulation of the immediate environment of the fetus. On Day 20 of gestation, pregnant rats were prepared to eliminate sensation from the lower body while avoiding the suppressive effects of general anesthesia on fetal behavior. Under ether anesthesia, females were administered a chemomyelotomy which involved injection of 100 t~l of 100% ethanol into the spinal column between the first and second lumbar vertebrae. This procedure produces an irreversible spinal anesthesia posterior to the site of injection (19). The female was then placed in a Plexiglas holding apparatus, her uterus was externalized through a midventral incision, and the uterus and lower body were immersed in a bath containing a buffered saline solution maintained at 37.5°C. The mother and fetuses were allowed to recover from the ether anesthesia and acclimate to the bath for 20 min before the onset of behavioral observation.
Preparation of the Fetal Microenviroment After the female was placed in the bath, a single subject fetus
was selected from the ovarian end of the uterine horn for manipulation and observation. A small incision was made in the side of the uterus and the fetus was delivered through the opening into the saline bath. The embryonic membranes were removed from the subject fetus while preserving its umbilical circulation and placental attachment.
Spinal Cord Transection A wire knife (0.22 mm diameter) was used to sever the spinal cord of 16 fetuses in a midthoracic location. Ten control fetuses were not transected, but received a sham treatment consisting of insertion and withdrawal of the knife from the back. The general condition of sham-operated and transected fetuses was good and virtually indistinguishable from unmanipulated fetuses studied previously. However, data from 2 sham-operated fetuses were not used because rearleg activity was completely absent for 5-6 min following the operation. At the conclusion of the experiment all transected fetuses were preserved in buffered formalin solution, spinal cords sectioned in a sagittal plane and examined under a binocular dissecting microscope to confirm the site and completeness of transection. In all fetuses the spinal cord was completely severed. Transections were localized to the midthoracic region between T2 and T5.
Behavioral Observations Observation of spontaneous fetal behavior began 5 min after spinal transection or the sham operation and continued for a period of 30 min. During the behavior observations, each instance of fetal movement was entered into a microcomputer serving as an event recorder which stored the time of occurrence of each movement. This procedure has been shown to be a highly reliable method for recording the occurrence of fetal rat movements (20). Individual movements of foreleg, rearleg, head, mouth, and body curl were distinguished (18,21). The sum of the individual movements of different body regions provided a measure of overall fetal activity (component activity). In addition, these individual movements could occur singly or in any combination (complex activity).
Analysis of Cyclic Patterns For each fetus, two separate movement time series were constructed (Fig. 1). One was formed by counting the number of rearleg movements in each successive 5-see interval throughout the 30-min period of observation. The other time series was formed by counting all other movements excluding movements of the rearlegs. Since we did not isolate the neural control of the various components of nonrearleg activity, they were not distinguished in the analysis of cyclic patterns. Each time series (rearleg, nonrearleg) was analyzed separately for cyclic patterns using methods described in detail in previous reports (11,22), and summarized briefly below. In the first stage of analysis, each time series was subjected to a Fourier analysis and the resulting cumulative variance distribution (in the frequency domain) was compared to the distribution expected from a random process using a Kolmogorov-Smirnov test (5). If this analysis revealed that the fluctuations in activity were not random (p<0.05), the time series was spectral analyzed to identify specific cyclic patterns and quantify their basic properties. A Tukey lag window was used which was equivalent to a spectral window with a bandwidth of 0.32 cycle/min (5). Linear trends, which accounted for 0 to 8.5 percent of the movement variance, were removed before the time series were spectral analyzed. A peak in movement spectrum was considered to reflect a cyclic
NEURAL CONTROL OF CYCLIC MOTOR ACTIVITY
123
800"
A
Sham I"1 Sham • Transect
600'
tO t-
"--
400"
3
E
Rearleg
200'
>
o
0
0
5
10
15
20
,
,
25
30
o
CA
F
Time (min)
H
M
C
N, CX
Categories of M o v e m e n t
Transect
B
R
FIG. 2. Mean number (___SEM) of movements per 30 min in five behavioral categories: F (foreleg), R (rearleg), H (head), M (mouth), C (body curl) as well as CA (component activity) and CX (complex movements).
10 Non-reorleg o
to tQ)
significant oscillations in both readeg and nonrearleg activity in fetuses with spinal cord transections as well as in sham-operated fetuses. In three of the fetuses with spinal cord transections there was insufficient activity of the readegs (4-32 movements in 30 min) to justify the analysis of cyclic patterns. In 8 of the remaining
3
E
Reorleg
> o ~E
^A 0
5
10
15 Time
2"0
25
,)
3'0
(min)
t-
.9 FIG. 1. Movement time series for (A) a sham-operated fetus and (B) a fetus with a transected spinal cord. The time series in these illustrations were smoothed with a five-term moving average filter to reduce the visual effects of high-frequency variation. pattern in motor activity if it exceeded the 95% confidence limits of the spectral estimates of random data.
"5 t/)
t-I-
13.7
E t~
t03
(~
65.6
RESULTS
Overall Activity Overall activity, the five individual movement categories, and complex activity, were compared for transected and sham-operated fetuses (Fig. 2). Transection of the fetal spinal cord in the midthoracic region resulted in a 36% reduction in overall fetal activity, t(19)= 2.61, p<0.05. The reduction was due mainly to decreased movements of the rearlegs, t(19)= 7.43, p<0.01, and body cuds, t(19)=4.58, p<0.01. Some rearleg movements are normally driven by activity above the level of the transection, and body curls involve the coordination of rostral and caudal activity. Furthermore, transected fetuses performed fewer complex movements, t(19)=5.02, p<0.01. Forelegs and rearlegs were rarely active at the same time in the fetuses with transected spinal cords, while this pattern was common in the sham-operated animals (Fig. 3). Thus, there was strong behavioral evidence, as well as histological verification, of the isolation of the caudal half of the spinal cord.
Cyclic Patterns Analysis of the movement time series provided evidence for
5.1
°
tO _
"5 o~ t-
@ FIG. 3. Diagrams depicting linkage relations between pairs of body regions. Lines connect body regions that move in combination. Values list actual mean frequency of combined movement.
124
A
c
ROBERTSON AND SMOTHERMAN
that the surgical procedure (not the spinal cord transection) may in some cases abolish the temporal organization of rearleg activity, perhaps transiently. The complete, but transient, cessation of movement in the rearlegs of 2 additional sham-operated fetuses (see the Methods section) may represent more extreme effects of the same type. Subsequent spectral analysis of the time series from fetuses with nonrandom fluctuations in both rearleg and nonrearleg activity revealed a dominant cyclic pattern in all of them (both rearleg and nonrearleg), indicated by the presence of a significant peak in each of the movement spectra. Three characteristics of the peaks were quantified in order to summarize basic properties of the corresponding oscillations in motor activity: 1) The frequency at which the peak occurred was used as a measure of the dominant rate of oscillation in motor activity. 2) The magnitude or height of the peak (maximum spectral density) was used to measure the relative strength of the oscillation in motor activity. 3) The width of the peak at its half-maximum point was used to measure the dispersion of movement variance in the frequency domain around the dominant cycle. If a local maximum occurred before the half-maximum point, straight line extrapolation from the peak through the nearest local minimum was used, and if the halfmaximum point was not reached before zero frequency, the width was taken to be twice the measurable half-width. The group means for these measures, based on fetuses with statistical evidence of cyclic patterns in both rearleg and nonrearleg activity, are shown in Fig. 4. Rearleg vs. nonrearleg differences are shown in Table 1. Figure 4 and Table 1 show that the frequency of oscillation in nonrearleg activity was dramatically affected by the spinal cord transection. For sham-operated fetuses, the rate of oscillation was nearly identical in nonrearleg and rearleg movement. For the cord-transected fetuses, however, the rate of oscillation in nonrearleg activity was decreased to less than half the rate of oscillation in rearleg activity. There was no effect of spinal cord transection on either the strength of the cyclic patterns or the dispersion of movement variance around the dominant cycle. In both the cord-transected and sham-operated fetuses there was a tendency for oscillations to be slightly stronger in nonrearleg activity, although the effect in both groups was marginal.
1.0-
0.8"
~>,
0.6"
O ¢'-
0.4"
-I-
(7"
m
0.2-
0.0"
B
1.0-
0.8" e-
"~
0.6"
e-
0.4"
0.2" 0.0"
C
1.0
0.8 ¢-
0.6
g e"O
Unoperated Fetuses 0.4
0.2
0.0
Transect
Sham Transect
Control
FIG. 4. (A) Frequency, (B) magnitude, and (C) width (mean---SEM) of the peak in the movement spectrum corresponding to the dominant oscillation in motor activity in each time series, for cord-transected, sham-operated, and unoperated control fetuses.
13 fetuses, the cumulative variance distributions derived from the Fourier analysis of both the rearleg and nonrearleg time series all differed (p<0.05) from the expected distributions of white noise, indicating that the fluctuations in movement were not random. Similarly, in 6 of the 8 sham-operated fetuses the fluctuations in motor activity in both the rearleg and nonrearleg time series were not random. The other 5 fetuses with spinal cord transections and 2 sham-operated fetuses had rearleg activity (but not nonrearleg activity) which appeared to be random. This raises the possibility
To determine whether the sham operation might have influenced cyclic motor activity, the measures obtained from the sham-operated fetuses in this experiment were compared to those obtained from unoperated fetuses studied previously (22) at the same gestational age and using the same methods. Movement time series for nonrearleg and rearleg activity were constructed for 10 unoperated fetuses. In 9 of the I0 unoperated fetuses there was statistical evidence for cyclic patterns in both nonrearleg and rearleg activity. The results are shown in Fig 4. The nonrearleg vs. rearleg differences were similar in the sham-operated and unoperated fetuses for the frequency of oscillation, the strength of oscillation, and the dispersion of movement variance around the dominant cycle (all ps>0.20). DISCUSSION
The results demonstrate that the caudal half of the spinal cord in the fetal rat is capable of generating cyclic motor output to the rearlegs in the absence of descending neural input from the brain. The results also demonstrate that there is no single timing center or source of oscillation underlying cyclic motor activity in the fetal rat, since cyclic patterns were generated above and below the spinal cord transection. Furthermore, the oscillations generated by
NEURAL CONTROL OF CYCLIC MOTOR ACTIVITY
125
TABLE 1 PARAMETERSOF MOVEMENTSPECTRA:NONREARLEGVS. REARLEGACTIVITY Frequency (cycles/rain) Transect (n=8) Sham (n=6)
Magnitude (min)
Width (cycles/min)
• I - 0 . 4 5 +-- 0.075" 0.14 --- 0.08:~ 0.01 -+ 0.03 [
0.05 + 0.23
0.11 -- 0.05~
0.14 -- 0.14
Mean difference -+ SEM in the frequency, magnitude, and width of the peaks in the movement spectra corresponding to the dominant oscillations in nonrearleg and rearleg activity. *p<0.05. tp<0.001. :~0.05
the rostral half of the spinal cord (including the brain) are dramatically slower than those generated by the caudal half of the cord, which in turn are similar to those generated by the intact motor system. These findings rule out a simple top-down model of cyclic motor activity, such as one based on fluctuations in general arousal generated by supraspinal sources. They do not, however, eliminate the possibility that the brain participates in the generation of cyclic motor activity, or influences the characteristic properties of the oscillations such as their frequency, amplitude, or regularity. In fact, data from the chick embryo indicate that by the middle of incubation supraspinal structures influence the temporal patterning of spinal motor output in that species (6). In the present experiment, the tendency for oscillations in nonrearleg activity to be slightly stronger may be due to supraspinal influences. The findings are consistent with a 2-source model, with one source above the midthoracic region of the spinal cord (which includes the brain), and one below. If there are two sources, the data from the sham-operated fetuses indicate that the sources are normally coupled in such a way as to oscillate at nearly the same frequency. The results from the fetuses with spinal cord transections, however, provide convincing evidence that the preferred frequencies are very different, with rostral sources slower. The results of a recent study of human neonates (15) would be predicted by such a model in which rostral sources of oscillation are slower. In that study a sound pulse induced a brief (less than 10 sec) burst of movement and a relative slowing of oscillations over the next 10-15 min. The slowing would be expected if supraspinal activation of the motor system results in a lasting influence of rostral sources with a lower preferred frequency. In the present experiment, the fact that the preferred frequency of the caudal portion of the spinal cord is the same as that of the intact
motor system suggests that the caudal cord may entrain the rest of the system under normal circumstances. That is, the data are consistent with a bottom-up model (in the caudal-rostral, not hierarchical, sense). Without data from fetuses with both cervical and midthoracic transections, we cannot say whether supraspinal structures are responsible for the slower oscillations in nonrearleg activity. However, it is unlikely that the effects on the frequency of oscillation are due to load changes. That is, the mechanical load of the motor system above and below the transection was reduced, which would not be expected to lead to slower oscillations in either nonrearleg or rearleg activity, and certainly not in just nonrearleg activity. In addition to the different preferred frequency of oscillations generated in the caudal and rostral portions of the motor system, a basic characteristic of cyclic motor activity which must be accounted for by any model is the irregular nature of the fluctuations. Cycle time is not strongly regulated. The fluctuations in spontaneous activity appear to be noisy. The variability may reflect instability in the sources of oscillation. If so, the instability is not due to descending input from the brain because the oscillations in rearleg activity in cord-transected fetuses are also variable: the width of the spectral peaks for nonrearleg and rearleg activity are nearly identical. Perhaps afferent input to the cord destabilizes the mechanism responsible for the oscillations in motor output. An alternative approach is to consider the variability as an intrinsic property of the mechanism responsible for the oscillations. The somewhat erratic quality of the fluctuations in motor activity may emerge directly from the dynamics of the underlying mechanism. If there are just 2 localized sources of oscillation, each may have this property. On the other hand, there may be no sources of oscillation. Rather, interactions among more distributed active elements of the spinal motor system may result in erratic fluctuations in pooled motor output which are nevertheless constrained to a limited portion of the frequency domain (reflected in the characteristic width of the peak in the movement spectrum). From this point of view, a dynamical model with a chaotic attractor might be useful in trying to explain the fluctuations in spontaneous motor activity (1, 23, 25). The results of this experiment put constraints on how we model the fluctuations in spontaneous motor activity. In the fetal rat, and perhaps in other mammals including the human, cyclic motor activity arises from more than one source in the motor system (perhaps no sources), and the brain appears to be unnecessary, although important contributions of supraspinal sources cannot be ruled out. ACKNOWLEDGEMENTS This research is supported by NIH grants HD 23814-02 and HD 11089-11 to S. S. Robertson, and HD 16102-08 and Research Career Development Award HD 00719-04 to W. P. Smotherman.
REFERENCES 1. Abraham, R. H.; Shaw, C. D. Dynamics: The geometry of behavior, part 2: Chaotic behavior. Santa Cruz, CA: Aerial Press; 1983. 2. Aschoff, J, ed. Handbook of behavioral neurobiology, vol. 4: Biological rhythms. New York: Plenum Press; 1981. 3. de Vries, J. I. P.; Visser, G. H. A.; Prechtl, H. F. R. The emergence of fetal behavior. I. Qualitative aspects. Early Hum. Dev. 7:301-322; 1982. 4. Hamburger, V.; Wenger, E.; Oppenheim, R. Motility in the chick embryo in the absence of sensory input. J. Exp. Zool. 162:133-160; 1966. 5. Jenkins, G. M.; Watts, D. G. Spectral analysis and its applications. San Francisco: Holden-Day; 1968. 6. Oppenheim, R. W. The role of supraspinal input in embryonic
7.
8. 9. 10.
motility: A reexamination in the chick. J. Comp. Neurol. 160:37-50; 1975. Prechtl, H. F. R. Continuity and change in early neural development. In: Prechtl, H. R. F., ed. Continuity of neural functions from prenatal to postnatal life. (Clinics in Developmental Medicine No. 94.) Philadelphia: Lippincott; 1984:1-15. Provine, R. R.; Rogers, L. Development of spinal cord bioelectric activity in spinal chick embryos and its behavioral implications. J. Neurobiol. 8:217-228; 1977. Ripley, K. L.; Provine, R. R. Neural correlates of embryonic motility in the chick. Brain Res. 46:127-134; 1972. Robertson, S. S. Intrinsic temporal patterns in the spontaneous movement of awake neonates. Child Dev. 53:1016-1021; 1982.
126
11. Robertson, S. S. Cyclic motor activity in the human fetus after midgestation. Dev. Psychobiol. 18:411-419; 1985. 12. Robertson, S. S. Human cyclic motility: Fetal-newborn continuities and newborn state differences. Dev. Psychobiol 20:425--442; 1987. 13. Robertson, S. S. Infants of diabetic mothers: Late normalization of fetal cyclic motility persists after birth. Dev. Psychobiol. 21:477-490; 1988. 14. Robertson, S. S. Mechanism and function of cyclicity in spontaneous movement. In Smotherman, W. P.; Robinson, S. R., eds. Behavior of the fetus. Caldwell, NJ: Telford Press; 1989:77-94. 15. Robertson, S. S. The dynamics of newborn cyclic motor activity. Paper presented to the Society for Research in Child Development. Kansas City, MO; 1989. 16. Robertson, S. S.; Dierker, L. J. The development of cyclic motility in fetuses of diabetic mothers. Dev. Psychobiol. 19:223-234; 1986. 17. Robertson, S. S.; Dierker, L. J.; Sorokin, Y.; Rosen, M. G. Human fetal movement: spontaneous oscillations near one cycle per minute. Science 218:1327-1330; 1982. 18. Robinson, S. R.; Smotherman, W. P. Environmental determinants of
ROBERTSON AND SMOTHERMAN
19. 20. 21. 22. 23. 24. 25.
behaviour in the rat fetus, II. The emergence of synchronous movement. Anita. Behav. 35:1652-1662; 1987. Smotherman, W. P.; Richards, L. S.; Robinson, S. R. Techniques for observing fetal behavior in utero: A comparison of chemomyelotomy and spinal transection. Dev. Psychobiol. 17:661-674; 1984. Smotherman, W. P.; Robinson, S. R. The rat fetus in its environment: behavioral adjustments to novel, familiar, aversive, and conditioned stimuli presented in utero. Behav. Neurosci. 99:521-530; 1985. Smotherman, W. P.; Robinson, S. R. Environmental determinants of behaviour in the rat fetus. Anim. Behav. 34:1859-1873; 1986. Smotherman, W. P.; Robinson, S. R.; Robertson, S. S. Cyclic motor activity in the rat fetus. J. Comp. Psychol. 102:78-82; 1988. Swinney, H. L. Observations of order and chaos in nonlinear systems. Physica 7D:3-15; 1983. Visser, G. H. A.; Laurini, R. N.; de Vries, J. I. P.; Bekedam, D. J.; Prechtl, H. F. R. Abnormal motor behavior in anencephalic fetuses. Early Hum. Dev. 12:173-182; 1985. Winfree, A. T. The geometry of biological time. New York: SpringerVerlag; 1980.
0031-9384/90$3.00 + .00
The Neural Control of Cyclic Motor Activity in the Fetal Rat (Rattus norvegicus) STEVENS. ROBERTSON
Department of Human Development and Family Studies, Cornell University, Ithaca, NY 14853 AND W I L L I A M P. S M O T H E R M A N
Center for Developmental Psychobiology, Department of Psychology State University of New York at Binghamton, Binghamton, NY 13905 Received 18 A u g u s t 1989
ROBERTSON, S. S., AND W. P. SMOTHERMAN. The neuralcontrol of cyclic motor activity in the fetal rat (Rattus norvegicus). PHYSIOL BEHAV 47(1) 121-126, 1990.--The spontaneous behavior of rat fetuses (in a saline bath with fetal-placental-uterine connections intac0 was observed directly for 30 minutes on Day 20 of gestation. Rearleg and nonrearleg movements from fetuses with a mid-thoracic spinal cord transection or sham operation were analyzed for cyclic organization. Oscillations in rearlegactivity occurred at the same frequency in fetuses with spinal cord transections (0.74 cycle/rain) and sham-operated fetuses (0.72 cycle/rain). However, oscillations in nonrearleg activity were much slower in the fetuses with spinal cord transections (0.30 vs. 0.77 cycle/rain). Other characteristics of the cyclic patterns in motor activity were unaffected. The findings demonstrate 1) the caudal half of the spinal cord can generate cyclic output in the absence of descending input from the brain, 2) there is no single timing center, and 3) rostral sources are slower. Fetus
Rat
Motor activity
Cyclic
Spinal cord
OSCILLATION is a pervasive property of living systems which must be explained by any theory of spontaneous activity. In the human fetus and neonate, there are cyclic fluctuations in spontaneous motor activity comprised of general movements of the limbs, trunk, and head, and more isolated or stereotyped movements, with a cycle time of 1-5 minutes (10,17). The oscillations are apparent by midgestation (11), and qualitative observations using ultrasound suggest they may characterize fetal activity in the first trimester as well (3). Basic properties of the oscillations, such as their frequency and strength, remain virtually unchanged throughout the second half of gestation in normal fetuses (11), during a time when other aspects of neurobehavioral organization (e.g., behavioral state) change dramatically. The properties of fetal cyclic motor activity (CM) persist relatively unchanged after birth during active sleep (12). Although newborn CM is stronger and more regular in the nonsleep states compared to active sleep, its cyclic properties appear to be independent of the amount of movement (12). Fetal CM is disrupted by maternal diabetes, but returns to normal by the end of gestation (16), and remains so after birth even in newborns with evidence of increased glucose supply in utero (13). Thus, oscillations in spontaneous motor activity of the fetus and the neonate appear to be stable and robust. Their persistence with little if any quantitative change across birth
is consistent with other evidence of neurobehavioral continuity from prenatal to postnatal life, such as sleep state organization and isolated movement patterns (7). The oscillations may play a role in early neuromuscular development and regulate interaction with the environment (14), although these possibilities have not been tested. A full understanding of this behavioral pattern will require knowledge of its underlying mechanism. Two questions of considerable importance arise. First, can the spinal motor system generate the characteristic oscillations in general activity without descending neural input from the brain? If so, then a class of simple top-down models of CM, in which, for example, supraspinal fluctuations in general arousal might be responsible for the oscillations in motor activity, could be ruled out. In birds, evidence for the spinal origin of CM comes from a classic series of studies designed to investigate the role of sensory and supraspinal input on the motility oftbe chick embryo [e.g., (4, 6, 8, 9)]. In the human, qualitative descriptions of the exaggerated burst-pause patterns in the movements of anencephalic fetuses are also consistent with a spinal origin for CM (24). A second important question about the mechanism underlying CM is whether there is more than one source of oscillation? If so, then models which are based on a single, localized timing center 121
122
ROBERTSON AND SMOTHERMAN
would be ruled out and attention could be focused on characterizing the multiple sources, the process by which they generate cyclic output, and the dynamics of their interactions. The data on spontaneous motility and its neurophysiological correlates in the chick embryo cited above suggest that isolated sections of the spinal cord are capable of generating cyclic output, which is consistent with a multisource model. In the human, the relatively common finding of double peaks in the movement spectrum [e.g., (17)] is consistent with the existence of more than one source of oscillation which under some circumstances become uncoupled. The present experiment was designed to address these questions directly in a mammalian model of human CM. Recent work suggests that the fetal rat may be useful animal model in which to explore the neural substrate of CM (22). In that study, methods for the direct observation of fetal rat behavior (19), combined with analytic techniques for detecting and quantifying cyclicity in spontaneous motor activity (11), revealed oscillations in the spontaneous activity of fetuses delivered into a warm saline bath with fetal-placental-uterine connections intact. Quantitative differences were found between fetal rat and human CM, but they were not large. In the present experiment, the same observational and analytic techniques were used to study the spontaneous activity of fetal rats following transection of the spinal cord at a midthoracic level (or a sham operation). Rearleg and nonrearleg activity were analyzed to determine the presence and quantitative properties of any cyclic patterns generated above and below the transection. METHOD
Subjects Subject fetuses were the offspring of female Sprague-Dawley rats (Simonsen Laboratories, Gilroy, CA) time-mated to LongEvans males. Vaginal smears were taken daily to identify the day of conception (first detectable sperm=Day 0 of gestation). Females were housed in groups of three in polycarbonate cages (33 × 38 × 10 cm) until testing on Day 20 of gestation. Cages were kept in a temperature- and humidity-controlled colony room under 12:12 hr light/dark cycle (lights on at 0700); testing took place between 1300 and 1800. Females were provided with ad lib food and water and at all times were maintained in accordance with guidelines for animal care established by the National Institutes of Health, the International Society for Developmental Psychobiology and the Animal Behavior Society.
Surgical Preparation Direct observation of fetal behavior requires preparation of the pregnant female and manipulation of the immediate environment of the fetus. On Day 20 of gestation, pregnant rats were prepared to eliminate sensation from the lower body while avoiding the suppressive effects of general anesthesia on fetal behavior. Under ether anesthesia, females were administered a chemomyelotomy which involved injection of 100 t~l of 100% ethanol into the spinal column between the first and second lumbar vertebrae. This procedure produces an irreversible spinal anesthesia posterior to the site of injection (19). The female was then placed in a Plexiglas holding apparatus, her uterus was externalized through a midventral incision, and the uterus and lower body were immersed in a bath containing a buffered saline solution maintained at 37.5°C. The mother and fetuses were allowed to recover from the ether anesthesia and acclimate to the bath for 20 min before the onset of behavioral observation.
Preparation of the Fetal Microenviroment After the female was placed in the bath, a single subject fetus
was selected from the ovarian end of the uterine horn for manipulation and observation. A small incision was made in the side of the uterus and the fetus was delivered through the opening into the saline bath. The embryonic membranes were removed from the subject fetus while preserving its umbilical circulation and placental attachment.
Spinal Cord Transection A wire knife (0.22 mm diameter) was used to sever the spinal cord of 16 fetuses in a midthoracic location. Ten control fetuses were not transected, but received a sham treatment consisting of insertion and withdrawal of the knife from the back. The general condition of sham-operated and transected fetuses was good and virtually indistinguishable from unmanipulated fetuses studied previously. However, data from 2 sham-operated fetuses were not used because rearleg activity was completely absent for 5-6 min following the operation. At the conclusion of the experiment all transected fetuses were preserved in buffered formalin solution, spinal cords sectioned in a sagittal plane and examined under a binocular dissecting microscope to confirm the site and completeness of transection. In all fetuses the spinal cord was completely severed. Transections were localized to the midthoracic region between T2 and T5.
Behavioral Observations Observation of spontaneous fetal behavior began 5 min after spinal transection or the sham operation and continued for a period of 30 min. During the behavior observations, each instance of fetal movement was entered into a microcomputer serving as an event recorder which stored the time of occurrence of each movement. This procedure has been shown to be a highly reliable method for recording the occurrence of fetal rat movements (20). Individual movements of foreleg, rearleg, head, mouth, and body curl were distinguished (18,21). The sum of the individual movements of different body regions provided a measure of overall fetal activity (component activity). In addition, these individual movements could occur singly or in any combination (complex activity).
Analysis of Cyclic Patterns For each fetus, two separate movement time series were constructed (Fig. 1). One was formed by counting the number of rearleg movements in each successive 5-see interval throughout the 30-min period of observation. The other time series was formed by counting all other movements excluding movements of the rearlegs. Since we did not isolate the neural control of the various components of nonrearleg activity, they were not distinguished in the analysis of cyclic patterns. Each time series (rearleg, nonrearleg) was analyzed separately for cyclic patterns using methods described in detail in previous reports (11,22), and summarized briefly below. In the first stage of analysis, each time series was subjected to a Fourier analysis and the resulting cumulative variance distribution (in the frequency domain) was compared to the distribution expected from a random process using a Kolmogorov-Smirnov test (5). If this analysis revealed that the fluctuations in activity were not random (p<0.05), the time series was spectral analyzed to identify specific cyclic patterns and quantify their basic properties. A Tukey lag window was used which was equivalent to a spectral window with a bandwidth of 0.32 cycle/min (5). Linear trends, which accounted for 0 to 8.5 percent of the movement variance, were removed before the time series were spectral analyzed. A peak in movement spectrum was considered to reflect a cyclic
NEURAL CONTROL OF CYCLIC MOTOR ACTIVITY
123
800"
A
Sham I"1 Sham • Transect
600'
tO t-
"--
400"
3
E
Rearleg
200'
>
o
0
0
5
10
15
20
,
,
25
30
o
CA
F
Time (min)
H
M
C
N, CX
Categories of M o v e m e n t
Transect
B
R
FIG. 2. Mean number (___SEM) of movements per 30 min in five behavioral categories: F (foreleg), R (rearleg), H (head), M (mouth), C (body curl) as well as CA (component activity) and CX (complex movements).
10 Non-reorleg o
to tQ)
significant oscillations in both readeg and nonrearleg activity in fetuses with spinal cord transections as well as in sham-operated fetuses. In three of the fetuses with spinal cord transections there was insufficient activity of the readegs (4-32 movements in 30 min) to justify the analysis of cyclic patterns. In 8 of the remaining
3
E
Reorleg
> o ~E
^A 0
5
10
15 Time
2"0
25
,)
3'0
(min)
t-
.9 FIG. 1. Movement time series for (A) a sham-operated fetus and (B) a fetus with a transected spinal cord. The time series in these illustrations were smoothed with a five-term moving average filter to reduce the visual effects of high-frequency variation. pattern in motor activity if it exceeded the 95% confidence limits of the spectral estimates of random data.
"5 t/)
t-I-
13.7
E t~
t03
(~
65.6
RESULTS
Overall Activity Overall activity, the five individual movement categories, and complex activity, were compared for transected and sham-operated fetuses (Fig. 2). Transection of the fetal spinal cord in the midthoracic region resulted in a 36% reduction in overall fetal activity, t(19)= 2.61, p<0.05. The reduction was due mainly to decreased movements of the rearlegs, t(19)= 7.43, p<0.01, and body cuds, t(19)=4.58, p<0.01. Some rearleg movements are normally driven by activity above the level of the transection, and body curls involve the coordination of rostral and caudal activity. Furthermore, transected fetuses performed fewer complex movements, t(19)=5.02, p<0.01. Forelegs and rearlegs were rarely active at the same time in the fetuses with transected spinal cords, while this pattern was common in the sham-operated animals (Fig. 3). Thus, there was strong behavioral evidence, as well as histological verification, of the isolation of the caudal half of the spinal cord.
Cyclic Patterns Analysis of the movement time series provided evidence for
5.1
°
tO _
"5 o~ t-
@ FIG. 3. Diagrams depicting linkage relations between pairs of body regions. Lines connect body regions that move in combination. Values list actual mean frequency of combined movement.
124
A
c
ROBERTSON AND SMOTHERMAN
that the surgical procedure (not the spinal cord transection) may in some cases abolish the temporal organization of rearleg activity, perhaps transiently. The complete, but transient, cessation of movement in the rearlegs of 2 additional sham-operated fetuses (see the Methods section) may represent more extreme effects of the same type. Subsequent spectral analysis of the time series from fetuses with nonrandom fluctuations in both rearleg and nonrearleg activity revealed a dominant cyclic pattern in all of them (both rearleg and nonrearleg), indicated by the presence of a significant peak in each of the movement spectra. Three characteristics of the peaks were quantified in order to summarize basic properties of the corresponding oscillations in motor activity: 1) The frequency at which the peak occurred was used as a measure of the dominant rate of oscillation in motor activity. 2) The magnitude or height of the peak (maximum spectral density) was used to measure the relative strength of the oscillation in motor activity. 3) The width of the peak at its half-maximum point was used to measure the dispersion of movement variance in the frequency domain around the dominant cycle. If a local maximum occurred before the half-maximum point, straight line extrapolation from the peak through the nearest local minimum was used, and if the halfmaximum point was not reached before zero frequency, the width was taken to be twice the measurable half-width. The group means for these measures, based on fetuses with statistical evidence of cyclic patterns in both rearleg and nonrearleg activity, are shown in Fig. 4. Rearleg vs. nonrearleg differences are shown in Table 1. Figure 4 and Table 1 show that the frequency of oscillation in nonrearleg activity was dramatically affected by the spinal cord transection. For sham-operated fetuses, the rate of oscillation was nearly identical in nonrearleg and rearleg movement. For the cord-transected fetuses, however, the rate of oscillation in nonrearleg activity was decreased to less than half the rate of oscillation in rearleg activity. There was no effect of spinal cord transection on either the strength of the cyclic patterns or the dispersion of movement variance around the dominant cycle. In both the cord-transected and sham-operated fetuses there was a tendency for oscillations to be slightly stronger in nonrearleg activity, although the effect in both groups was marginal.
1.0-
0.8"
~>,
0.6"
O ¢'-
0.4"
-I-
(7"
m
0.2-
0.0"
B
1.0-
0.8" e-
"~
0.6"
e-
0.4"
0.2" 0.0"
C
1.0
0.8 ¢-
0.6
g e"O
Unoperated Fetuses 0.4
0.2
0.0
Transect
Sham Transect
Control
FIG. 4. (A) Frequency, (B) magnitude, and (C) width (mean---SEM) of the peak in the movement spectrum corresponding to the dominant oscillation in motor activity in each time series, for cord-transected, sham-operated, and unoperated control fetuses.
13 fetuses, the cumulative variance distributions derived from the Fourier analysis of both the rearleg and nonrearleg time series all differed (p<0.05) from the expected distributions of white noise, indicating that the fluctuations in movement were not random. Similarly, in 6 of the 8 sham-operated fetuses the fluctuations in motor activity in both the rearleg and nonrearleg time series were not random. The other 5 fetuses with spinal cord transections and 2 sham-operated fetuses had rearleg activity (but not nonrearleg activity) which appeared to be random. This raises the possibility
To determine whether the sham operation might have influenced cyclic motor activity, the measures obtained from the sham-operated fetuses in this experiment were compared to those obtained from unoperated fetuses studied previously (22) at the same gestational age and using the same methods. Movement time series for nonrearleg and rearleg activity were constructed for 10 unoperated fetuses. In 9 of the I0 unoperated fetuses there was statistical evidence for cyclic patterns in both nonrearleg and rearleg activity. The results are shown in Fig 4. The nonrearleg vs. rearleg differences were similar in the sham-operated and unoperated fetuses for the frequency of oscillation, the strength of oscillation, and the dispersion of movement variance around the dominant cycle (all ps>0.20). DISCUSSION
The results demonstrate that the caudal half of the spinal cord in the fetal rat is capable of generating cyclic motor output to the rearlegs in the absence of descending neural input from the brain. The results also demonstrate that there is no single timing center or source of oscillation underlying cyclic motor activity in the fetal rat, since cyclic patterns were generated above and below the spinal cord transection. Furthermore, the oscillations generated by
NEURAL CONTROL OF CYCLIC MOTOR ACTIVITY
125
TABLE 1 PARAMETERSOF MOVEMENTSPECTRA:NONREARLEGVS. REARLEGACTIVITY Frequency (cycles/rain) Transect (n=8) Sham (n=6)
Magnitude (min)
Width (cycles/min)
• I - 0 . 4 5 +-- 0.075" 0.14 --- 0.08:~ 0.01 -+ 0.03 [
0.05 + 0.23
0.11 -- 0.05~
0.14 -- 0.14
Mean difference -+ SEM in the frequency, magnitude, and width of the peaks in the movement spectra corresponding to the dominant oscillations in nonrearleg and rearleg activity. *p<0.05. tp<0.001. :~0.05
the rostral half of the spinal cord (including the brain) are dramatically slower than those generated by the caudal half of the cord, which in turn are similar to those generated by the intact motor system. These findings rule out a simple top-down model of cyclic motor activity, such as one based on fluctuations in general arousal generated by supraspinal sources. They do not, however, eliminate the possibility that the brain participates in the generation of cyclic motor activity, or influences the characteristic properties of the oscillations such as their frequency, amplitude, or regularity. In fact, data from the chick embryo indicate that by the middle of incubation supraspinal structures influence the temporal patterning of spinal motor output in that species (6). In the present experiment, the tendency for oscillations in nonrearleg activity to be slightly stronger may be due to supraspinal influences. The findings are consistent with a 2-source model, with one source above the midthoracic region of the spinal cord (which includes the brain), and one below. If there are two sources, the data from the sham-operated fetuses indicate that the sources are normally coupled in such a way as to oscillate at nearly the same frequency. The results from the fetuses with spinal cord transections, however, provide convincing evidence that the preferred frequencies are very different, with rostral sources slower. The results of a recent study of human neonates (15) would be predicted by such a model in which rostral sources of oscillation are slower. In that study a sound pulse induced a brief (less than 10 sec) burst of movement and a relative slowing of oscillations over the next 10-15 min. The slowing would be expected if supraspinal activation of the motor system results in a lasting influence of rostral sources with a lower preferred frequency. In the present experiment, the fact that the preferred frequency of the caudal portion of the spinal cord is the same as that of the intact
motor system suggests that the caudal cord may entrain the rest of the system under normal circumstances. That is, the data are consistent with a bottom-up model (in the caudal-rostral, not hierarchical, sense). Without data from fetuses with both cervical and midthoracic transections, we cannot say whether supraspinal structures are responsible for the slower oscillations in nonrearleg activity. However, it is unlikely that the effects on the frequency of oscillation are due to load changes. That is, the mechanical load of the motor system above and below the transection was reduced, which would not be expected to lead to slower oscillations in either nonrearleg or rearleg activity, and certainly not in just nonrearleg activity. In addition to the different preferred frequency of oscillations generated in the caudal and rostral portions of the motor system, a basic characteristic of cyclic motor activity which must be accounted for by any model is the irregular nature of the fluctuations. Cycle time is not strongly regulated. The fluctuations in spontaneous activity appear to be noisy. The variability may reflect instability in the sources of oscillation. If so, the instability is not due to descending input from the brain because the oscillations in rearleg activity in cord-transected fetuses are also variable: the width of the spectral peaks for nonrearleg and rearleg activity are nearly identical. Perhaps afferent input to the cord destabilizes the mechanism responsible for the oscillations in motor output. An alternative approach is to consider the variability as an intrinsic property of the mechanism responsible for the oscillations. The somewhat erratic quality of the fluctuations in motor activity may emerge directly from the dynamics of the underlying mechanism. If there are just 2 localized sources of oscillation, each may have this property. On the other hand, there may be no sources of oscillation. Rather, interactions among more distributed active elements of the spinal motor system may result in erratic fluctuations in pooled motor output which are nevertheless constrained to a limited portion of the frequency domain (reflected in the characteristic width of the peak in the movement spectrum). From this point of view, a dynamical model with a chaotic attractor might be useful in trying to explain the fluctuations in spontaneous motor activity (1, 23, 25). The results of this experiment put constraints on how we model the fluctuations in spontaneous motor activity. In the fetal rat, and perhaps in other mammals including the human, cyclic motor activity arises from more than one source in the motor system (perhaps no sources), and the brain appears to be unnecessary, although important contributions of supraspinal sources cannot be ruled out. ACKNOWLEDGEMENTS This research is supported by NIH grants HD 23814-02 and HD 11089-11 to S. S. Robertson, and HD 16102-08 and Research Career Development Award HD 00719-04 to W. P. Smotherman.
REFERENCES 1. Abraham, R. H.; Shaw, C. D. Dynamics: The geometry of behavior, part 2: Chaotic behavior. Santa Cruz, CA: Aerial Press; 1983. 2. Aschoff, J, ed. Handbook of behavioral neurobiology, vol. 4: Biological rhythms. New York: Plenum Press; 1981. 3. de Vries, J. I. P.; Visser, G. H. A.; Prechtl, H. F. R. The emergence of fetal behavior. I. Qualitative aspects. Early Hum. Dev. 7:301-322; 1982. 4. Hamburger, V.; Wenger, E.; Oppenheim, R. Motility in the chick embryo in the absence of sensory input. J. Exp. Zool. 162:133-160; 1966. 5. Jenkins, G. M.; Watts, D. G. Spectral analysis and its applications. San Francisco: Holden-Day; 1968. 6. Oppenheim, R. W. The role of supraspinal input in embryonic
7.
8. 9. 10.
motility: A reexamination in the chick. J. Comp. Neurol. 160:37-50; 1975. Prechtl, H. F. R. Continuity and change in early neural development. In: Prechtl, H. R. F., ed. Continuity of neural functions from prenatal to postnatal life. (Clinics in Developmental Medicine No. 94.) Philadelphia: Lippincott; 1984:1-15. Provine, R. R.; Rogers, L. Development of spinal cord bioelectric activity in spinal chick embryos and its behavioral implications. J. Neurobiol. 8:217-228; 1977. Ripley, K. L.; Provine, R. R. Neural correlates of embryonic motility in the chick. Brain Res. 46:127-134; 1972. Robertson, S. S. Intrinsic temporal patterns in the spontaneous movement of awake neonates. Child Dev. 53:1016-1021; 1982.
126
11. Robertson, S. S. Cyclic motor activity in the human fetus after midgestation. Dev. Psychobiol. 18:411-419; 1985. 12. Robertson, S. S. Human cyclic motility: Fetal-newborn continuities and newborn state differences. Dev. Psychobiol 20:425--442; 1987. 13. Robertson, S. S. Infants of diabetic mothers: Late normalization of fetal cyclic motility persists after birth. Dev. Psychobiol. 21:477-490; 1988. 14. Robertson, S. S. Mechanism and function of cyclicity in spontaneous movement. In Smotherman, W. P.; Robinson, S. R., eds. Behavior of the fetus. Caldwell, NJ: Telford Press; 1989:77-94. 15. Robertson, S. S. The dynamics of newborn cyclic motor activity. Paper presented to the Society for Research in Child Development. Kansas City, MO; 1989. 16. Robertson, S. S.; Dierker, L. J. The development of cyclic motility in fetuses of diabetic mothers. Dev. Psychobiol. 19:223-234; 1986. 17. Robertson, S. S.; Dierker, L. J.; Sorokin, Y.; Rosen, M. G. Human fetal movement: spontaneous oscillations near one cycle per minute. Science 218:1327-1330; 1982. 18. Robinson, S. R.; Smotherman, W. P. Environmental determinants of
ROBERTSON AND SMOTHERMAN
19. 20. 21. 22. 23. 24. 25.
behaviour in the rat fetus, II. The emergence of synchronous movement. Anita. Behav. 35:1652-1662; 1987. Smotherman, W. P.; Richards, L. S.; Robinson, S. R. Techniques for observing fetal behavior in utero: A comparison of chemomyelotomy and spinal transection. Dev. Psychobiol. 17:661-674; 1984. Smotherman, W. P.; Robinson, S. R. The rat fetus in its environment: behavioral adjustments to novel, familiar, aversive, and conditioned stimuli presented in utero. Behav. Neurosci. 99:521-530; 1985. Smotherman, W. P.; Robinson, S. R. Environmental determinants of behaviour in the rat fetus. Anim. Behav. 34:1859-1873; 1986. Smotherman, W. P.; Robinson, S. R.; Robertson, S. S. Cyclic motor activity in the rat fetus. J. Comp. Psychol. 102:78-82; 1988. Swinney, H. L. Observations of order and chaos in nonlinear systems. Physica 7D:3-15; 1983. Visser, G. H. A.; Laurini, R. N.; de Vries, J. I. P.; Bekedam, D. J.; Prechtl, H. F. R. Abnormal motor behavior in anencephalic fetuses. Early Hum. Dev. 12:173-182; 1985. Winfree, A. T. The geometry of biological time. New York: SpringerVerlag; 1980.