Recovery of Oscillator Function Following Spinal Regeneration in the Sea Lamprey

Joseph Ayers
Department of Biology and
Marine Science Center
Northeastern University
East Point, Nahant, MA 01908

Ayers, J. (1989) Recovery of oscillator function following spinal regeneration in the sea lamprey. In: Cellular and Neuronal Oscillators. J. Jacklet, [ed]. Marcel Dekker, New York, Pp. 349-383.

Table of Contents


Neuronal oscillators are subject to ongoing modulation by intersegmental and segmental command, coordinating and sensory inputs (Pinsker and Ayers, 1983). In many cases, neuronal oscillators can recover function following regeneration from both central and peripheral lesions which eliminate these sources of extrinsic moduation (Rovainen, 1976; Selzer, 1978; Freed, de Medinaceli and Wyatt, 1985). In general regeneration between the central nervous system and peripheral sense and effector organs occurs in most animal groups, but the capability for regeneration between parts of the CNS is lost in mammals (Eidelberg, 1981; Nichols, 1982; Seil, Herbert and Carlson, 1987). This loss is not due to an intrinsic inability of mammalian axons to regenerate (Aguayo, David and Bray 1981), but may be due to target activation of an intrinsic "stop" mechanism in the regenerating neurons similar to that which terminates axonal elongation during development (Liuzzi and Lasek, 1987). In invertebrates and lower vertebrates axonal regeneration in the CNS is more common and may, in some cases, lead to functional recovery. Recent studies in several lower vertebrates have indicated that recovery of some behaviors, such as undulatory (Bernstein and Gelderd, 1970; Rovainen, 1976; Selzer, 1978; Borgens, Roederer and Cohen, 1981, Currie and Ayers, 1983; Sharma et al. 1986) and even ambulatory (Duffy, Davis and Simpson, 1987) locomotion, can occur following spinal transection.

What has not been shown in any model of spinal cord regeneration is how the regeneration of descending neurons leads to behavioral recovery. Behavioral recovery following CNS lesions requires at least two prerequisite processes: (1) regeneration of ascending and/or descending intersegmental interneurons across the lesion and (2) recovery of interneuronal function resulting from synapse formation by the regenerating neurons (Guth et al., 1980).

Students of regeneration have traditionally relied on anatomical techniques (Cajal, 1928, Seil, Herbert and Carlson, 1987). Diagnosis of behavioral recovery, however, requires kinematic analysis in the time frame over which regeneration and functional recovery occur. Elucidation of how regeneration leads to behavioral recovery requires an assay which is capable of diagnosis of the normalcy of the recovered behavior. An assay for recovery of locomotor function following peripheral sciatic nerve regeneration (Sciatic Functional Index) has been developed in the rat (de Medinaceli, Freed and Wyatt, 1982). This assay is based on stride parameters estimated from footfalls, and has made it possible to quantify recovery from peripheral lesions as well as assay the effectiveness of nerve growth promoting agents (de Medinaceli et al. 1986). Anyone who has attempted to catch a fleeing small mammal will quickly recognize the limitations of this assay in the diagnosis of anything other than forward locomotion.

Our goal has been to develop an assay which is capable of distinguishing closely related axial undulatory behaviors as well as serving as a quantitative assay of behavioral recovery following spinal transection in a simple vertebrate model, the sea lamprey. We have relied on an analysis of motion pictures of freely behaving animals to characterize the different forms of undulatory behavior. Using these normal behaviors as a standard we then compare them with the behaviors that occur in regenerated specimens following recovery from complete spinal cord transection.

Lampreys exhibit recovery of several behaviors following spinal transection (Ayers et al., 1980, 1981) and offer the opportunity to address the underlying mechanisms of recovery in a quantitative fashion in terms of the function of identified neurons (Rovainen, 1976; Wood, and Cohen, 1979; Yin and Selzer, 1983; 1984; Mackler and Selzer, 1985; Currie and Ayers, 1987). The lamprey spinal cord provides an example of a distributed oscillatory system which operates under both complex descending intersegmental and local segmental reflex modulation (Rovainen, 1979; Grillner et al., 1986). Thus it has been possible to evaluate recovery in the context of the command neuron (Currie and Ayers, 1983), coordinating neuron (Cohen, Mackler and Selzer, 1986), central pattern generator (Grillner et al., 1986) model of the organization of motor systems (Stein, 1978).


A. Analysis of Propagating Flexion Waves.

Lampreys exhibit a variety of behaviors, all of which consist of lateral axial undulations. In order to characterize undulatory behavior we developed an assay which enabled us to quantify their movements both in terms of their timing as well as the amplitude of the lateral undulations (Ayers et al. 1983). The assay relies on frame by frame numerical analysis of motion pictures of spontaneous behavior. Motion pictures are projected on to a computer digitizing tablet and the shape of the animal is digitized on a frame by frame basis as shown in Fig. 1A.

Figure 1. Summary of analysis of undulatory behavior during the initiation of a swimming sequence. Fig 1a. Normalized images of a freely swimming ammocoete larvae. The number to the left of each image is the motion picture frame number. Fig. 1b. Parameters determined from the analysis of curvature: Locus - position from nose to tail expressed as a % of body length; Curvature - 1/radius of best fit circle at curvature maximum. Fig. 1c. Graph of the locus of flexions from Fig 1a. as a function of time. Closed circles indicate flexions to the left while open circles indicate flexions to the right.. Fig. 1d. Regression analysis of flexion waves illustrating the three parameters of timing diagrams. Fig. 1e. Graph of the curvature of flexions as a function of the locus where they are determined. Flexions are sorted on the basis of whether they occur in the head, mid-body (hatched area) or tail region. The numbers to the right of this diagram indicate the mean curvature of flexions observed in the mid-body region.

Quantification of undulatory behavior is based on a analysis of body curvature (Ayers et al. 1983). Each set of points representing the shape of the lamprey in a given frame of the movie is fit with a fifth order polynomial and the polynomial is analyzed for curvature. We define a flexion as a region of curvature maxima and we assign to each flexion a locus which is its position as a percent of body length and curvature which is the inverse of the radius of the best-fit circle at the curvature maxima (Fig. 1B). When we graph the locus of flexions as a function of time (Fig. 1C), we find that the flexions to each side propagate down the body from frame to frame and by regression analysis can be fit with a straight line (Fig. 1D). We define a flexion wave as an ordered set of flexions which propagate along one side and consider these the functional unit of all undulatory behavior.

In a graph of locus as a function of time (Fig 1D), we can estimate several dynamic parameters of the flexion waves. The first parameter, period, the inverse of undulation frequency, is the time between the nose intercept of successive flexion waves to the same side. Propagation time is the time it takes a flexion wave to progress from the nose to the tail. During normal swimming propagation time is roughly equivalent to the period of swimming. By determination of the allometric constants of percent of body length per segment, we can estimate the intersegmental delay during swimming which is roughly 1 percent of the total cycle time and from intersegmental delay and period, the intersegmental mechanical phase lag.

Our measurement of the amplitude of swimming movements is based on a graph of the curvature of flexion waves as a function of time or of flexion locus. When plotted in this fashion, one observes a time-independent envelope of flexion curvatures which is characteristic of each lamprey behavior (Fig. 1E).

Table 1. Computed Parameters Of the Ammocoete Sequence Indicated in Figure 1.

wave 1 2 3 4 5 6 7 Means
side l r l r l r l
period 0.841 0.548 0.547 0.543 0.600 0.616
prop 0.903 0.620 0.603 0.595 0.606 0.679 0.628 0.665
prat 1.074 1.131 1.101 1.095 1.010 1.082
delay 7.43 5.10 4.96 4.90 4.99 5.59 5.17 5.45
phase 0.0088 0.0093 0.0091 0.0090 0.0083 0.0089

Left Right Bias
Head 0.322 -.513 0.546 -.512 0.506 -.497 -.508 0.473 1.075L
Body -.266 0.450 -.429 0.498 -.417 0.455 -.432 -.379 0.466 1.230R
Tail -.611 0.516 -.563 0.560 -.539 0.540 -.571 0.536 1.065L

The analysis results in a table of swimming parameters where each column of the table corresponds to a flexion wave (Table I). The analysis estimates the period, the propagation time, the intersegmental phase lag and the ratio of propagation time to period. The first row of Table I (wave) indicates the wave number from left to right. The second row (side) indicates whether the flexion wave is to the left (L) or to the right (R). The third row (period) is the computed cycle period (sec.) of the wave estimated from the start times of flexion waves to the same side. The fourth row (prop) is the propagation time. The fifth row (prat) is the ratio of propagation time over period. The sixth row (delay) is the intersegmental mechanical delay estimated from propagation time and allometric constants. The seventh row (phase) is the intersegmental mechanical phase lag (intersegmental delay / period).

We find it useful to divide the flexion waves into those curvatures which occur in three separate regions of the body. The eighth through tenth row of Table I is the average curvature of the flexion wave in the Head (0-25%), Body (25-75%) and Tail (75-100%) regions. In this table, the average curvature was computed by taking the average value of the curvature of all the flexions detected in the wave in the indicated region. We can also estimate the bilateral asymmetry of curvature which we express as the ratio of curvature on the side which is flexing the most over the side which is flexing the least. In other words, a bilateral asymmetry of 1.5 to the left would indicate that the animal is flexing its body one and a half times more to the left than to the right. It is the parameters of these tables which form the basis for quantification of the development of swimming, distinguishing different undulatory behaviors and diagnosis of behavioral recovery following spinal transection..

Figure 2. Analysis of swimming behavior in an ammocoete larvae (Fig. 2a-b) and a recently transformed adult (Fig. 2c-d). Figures 2a and 2c are graphs of the locus of flexion waves as a function of time as in fig. 1d. Figures 2b and 2d are graphs of the curvature of flexion waves as a function of locus as in Fig. 1e.

2. Development of Swimming Behavior.

Lampreys are anadromous species and develop as aquatic ammocoete larvae living in mud tubes in fresh water streams. At 5-7 years of age, these filter-feeding larvae metamorphose into parasitic feeding phase adults between July and September. They both develop new anatomical features such as eyes and the sucker mouth, and reorganize much of their central nervous system. Larval and adult lampreys have quite different life styles. Larval lamprey are tube-dwelling filter feeders whereas the adults are free living parasites which must be capable of pursuing and capturing prey. One would expect, therefore that their swimming behavior might be quite different and that is indeed what we found on the basis of our analysis. Examples of swimming in larval and adult lamprey are found in Fig. 2. Larval lamprey tend to swim at a relatively constant velocity with both constant period and curvature along the length of the body. Adult swimming on the other hand is much more variable in velocity. Adult lamprey tend to swim in spurts where they speed up and slow down. Despite their speed variations the period of swimming stays relatively constant and more variability is observed in curvature.

Figure 3. Developmental changes in swimming parameters during metamorphosis. Fig 3a. Histograms of the period of swimming undulations in ammocoetes (upper panel: n = 84; mean = 0.453; s.d. = 0.151) and transformers (lower panel: n = 102; mean = 0.278; s.d. = 0.125). The periods are significantly different (p<0.001). Fig 3b. Histograms of the propagation time of flexion waves in ammocoetes (upper panel: n = 120; mean = 0.544; s.d. = 0.220) and transformers (lower panel: n = 144; mean = 0.344; s.d. = 0.180). The propagation times are significantly different (p<0.001). Fig 3c. Histograms of the intersegmental phase lag (phase) of flexion waves in ammocoetes (upper panel: n = 86; mean = 0.011; s.d. = 0.012) and transformers (lower panel: n = 106; mean = 0.013; s.d. = 0.003). The phase lags are not significantly different (p>0.05). Fig 3d. Histograms of the mid-body curvature of flexions in ammocoetes (upper panel: n = 120; mean = 0.0410; s.d. = 0.127) and transformers (lower panel: n = 147; mean = 0.335; s.d. = 0.116). The curvatures are not significantly different (p>0.05).

When pooled data of swimming in ammocoetes and transformers are compared, significant differences are observed in the overall timing although the relative timing is similar. The period of ammocoete swimming is significantly different from that of transformer swimming (Fig. 3A). as is the propagation time (Fig. 3B). In contrast, the relative timing (Fig. 3C) and mid-body curvature (Fig. 3D) are not significantly different. In other words, the spinal clocks are phase constant, but they keep different time. This similarity in relative timing and curvature accounts for the apparent similarity in larval and adult swimming.

3. Other Forms of Swimming.

We distinguish normal swimming from escape swimming. One of the distinguishing features of escape swimming in both larvae and adults is a profound decrease in period and large parametric variations in curvature. In other words swimming parameters are quite stereotyped during normal swimming and become quite variable during escape swimming. Lamprey will also attempt to swim when taken out of water. This terrestrial swimming is quite similar to escape swimming with short periods and high curvature although the propagation of the waves in the tail region is damped. In other words the waves propagate from the gill region to the cloaca and frequently, the caudal body curvature simply decreases at the cloaca rather than the flexion propagating to the tail (J. Kinch, unpublished).

D. Other Types of Undulatory Behavior

1. CAUDALLY PROPAGATING FLEXION WAVES. Lamprey exhibit at least two other undulatory behaviors which result from undulations which propagate from nose to tail. The first of these is crawling behavior. Crawling behavior can occur under water, but is most typically exhibited when the animals are placed on a smooth surface out of water. The crawling movements tend to be much slower than swimming movements and exhibit a much higher curvature. In fact we often see curvatures which are up to four times as great as those which characterize swimming movements (Fig. 4D).

The third behavior which results from front to rear propagating flexion waves is burrowing behavior. Burrowing consists of three components, the first of which is simply swimming into the substrate. Once the animals have gained purchase in the substrate, they decrease the frequency of the swimming movements (Fig. 4A) and increase the curvature of flexion waves, especially in the head region (Fig. 4B). Specimens observed during this phase of burrowing frequently assume C or L shapes, in contrast to the S shapes which characterize swimming. During these movements, the posterior body and tail appear to shove the head region into the substrate. During the third phase of burrowing, lateral undulations cease and the specimen appears to pull itself into the substrate with head flexions (G. Eaholtz, unpublished).

2. ROSTRALLY PROPAGATING FLEXION WAVES. A second class of undulatory behaviors occurs when the flexion waves propagate rostrally from tail to nose. The behavior which we categorize as backward crawling is observed most commonly when specimens are struggling or crawling out of water. Backward crawling is always a slow behavior with highest curvatures typically occurring in the head region.

Figure 4. Other forms of undulatory behavior. Fig 4a-b Analysis of the shoving phase of burrowing in a transformer. Fig. 4c-d. Analysis of forward crawling in a transformer which was placed on a wet tray out of water. Note the higher range of curvatures exhibited during this behavior. Fig. 4e-f. Analysis of backward crawling in a transformer. Fig. 4g-h. Analysis of aversive withdrawl in response to a lateral tap to the gill region in an ammocoete larvae. The lateral tap was delivered around frames 3-4. The dashed line indicates the reversing flexion wave.

3. REVERSING FLEXION WAVES. The third major category of undulatory behaviors consists of an aversive withdrawal behavior which occurs in response to noxious stimuli to the head region. Aversive withdrawal is characterized by a flexion wave which first propagates from tail to near mid-body, then reverses to propagate from mid body to tail and transform into a bout of escape swimming. Aversive withdrawal is most readily elicited by a lateral tap to the gill region and is rarely observed when specimens collide directly with an aquarium wall. It tends to make the specimen back up, turn and swim away.

Figure 5. Stages of ammocoete recovery of swimming behavior. The specimens examined in this figure recovered at room temperature (ca. 21šC). The lesion locus is indicate by the dark stipple around locus = 25% Fig 5a-b Analysis of head-wagging behavior in a stage 2 specimen. The head wags occur predominately to the left and do not propagate beyond locus of ~35%. Note the propagating caudal flexion wave to the left which occurs in frames 19-24 indicative of achievement of stage 2. Fig. 5c-d. Analysis of attempted swimming in a stage 3 specimen. Note the lack of coordinated propagation of flexions across the lesion site. Fig. 5e-f. Analysis of coordinated swimming in a stage 4 specimen. Fig. 5g-h. Analysis of coordinated swimming in a stage 5 specimen


We have analyzed regeneration and its relationship to behavioral recovery using behavioral, anatomical and electrophysiological techniques. In most of our experiments, specimens were subjected to a complete spinal cord transection at 25% of body length from the oral hood.

A. Sequential Events in the Recovery of Swimming

The behavioral recovery of swimming which occurs following spinal transection evolves slowly and can be resolved in several distinct stages based on functional capabilities of the transectees (Rovainen, 1976; Ayers et al., 1981; Wilbur, Margolin and Ayers, 1987) . Spontaneously, acutely transected specimens are capable only of rapid head wagging movements which occur at a frequency which can be higher than 20 hz (J. Kinch, unpublished). The caudal extent of the body typically remains limp although isolated flexion waves and struggling can in some cases be evoked by tail pinch. Stage 1 specimens can propel themselves by head wagging and frequently bump into the walls of the aquarium in which they are held. When this occurs in acutely transected specimens, they tend to continue to headwag into the aquarium wall. The initial sign of recovery occurs when specimens are capable of eliciting one caudally propagating flexion wave which turns the head away from the wall (Scott Currie, unpublished). We consider such specimens to have achieved stage 2 of recovery and propose that this capability indicates the initial stage of recovery of the descending command for swimming. Stage 2 specimens rapidly develop the capability of eliciting multiple caudally propagating flexions waves to achieve Stage 3. In Stage 3 specimens, the caudal body can generate swimming undulations but the anterior body often continues to head wag (Fig. 5C-D)

In the fourth stage of recovery, the head movements begin to become coordinated with the tail movements as illustrated in Fig. 5E-F. Here flexions propagate continuously from head to tail. This intersegmental coordination can be shown to rely on neuronal coordinating elements for coordinated swimming persists when the spinal cord is isolated by removal of musculature and gut in the region of the transection (Eaholtz, 1985) and coordinated bursting can be observed in isolated neuraxis preparations (Cohen, Mackler and Selzer, 1986). We propose, therefore, that achievement of stage 4 represents the recovery of function of intersegmental coordinating systems (Stein, 1978; Pinsker and Ayers, 1983).

B. Recovery of Righting During Swimming.

Stage 4 specimens typically exhibit a bilateral asymmetry of curvature to the two sides of the body (Table II). In other words, they produce larger amplitude flexions to one side of the body than to the other. Perhaps consequently, they exhibit an important functional deficit in that they are incapable of maintaining their primary orientation relative to gravity. Stage 4 specimens will often roll while swimming and are incapable of staying dorsal side up. In other words, stage 4 specimens exhibit the lack of control of equilibrium characteristic of chronically spinalized higher vertebrates (Barbeau and Rossingol, 1987).

Figure 6. Recovery of righting behavior during swimming. Righting was diagnosed by determining the percentage of time specimens maintained a dorsal-side -up orientation relative to gravity during swimming during a 2 minute epoch of swimming. In this graph, this proportion was determined for three ammocoetes recovering at 13šC at four different times following transection. Notice that the ability to right during swimming recovers gradually, but eventually totally.

The last major stage of recovery occurs, therefore, when specimens regain the ability to right while swimming. Recovery of righting behavior occurs gradually. We monitor their ability to right by determining the proportion of time which they spend dorsal side up while swimming. Stage 4 specimens are typically capable of staying dorsal side up only 50% of the time. In other words their orientation is random. If one graphs the proportion of time, specimens stay dorsal side up as specimens progress from stage 4 to stage 5, one observes that they evolve from random orientation to being able to stay dorsal side up almost 100% of the time (Fig. 6). This recovery of the ability to right is often accompanied by an elimination of bilateral asymmetrys of curvature, although specimens with curvature asymmetrys can often right quite effectively (C. Wilbur, unpublished).

C. Effect of Temperature on Recovery

Quite by accident, we found that the timing of the recovered swimming behavior in stage 5 specimens is contingent on the temperature which the specimens are held during the recovery process (Ayers and Wilbur, 1986). We developed a holding facility in a World War II bunker which maintains a year round temperature of ~13šC due to its thermal mass. Specimens held at room temperature (~21šC) exhibit a consistent deficit as shown in Fig. 5G-H. Although their swimming movements are coordinated and they are capable of staying dorsal side up while swimming, they make extremely rapid swimming movements when compared to a normal specimen. If specimens are allowed to recover in the bunker at lower holding temperatures (e.g. 13š C), we observe that specimens recover so completely after recovery times on the order of 4 to 6 months, that they become indistingishable from normal animals (compare. Fig. 7A-B with Fig. 2A-B).

Figure 7. Effect of recovery temperature on swimming parameters. Fig. 7a-b Analysis of recovered swimming in a specimen which recovered at reduced temperature (ca. 13šC). Compare with the swimming movements of a normal ammocoete (Fig. 2a-b) and a specimen which recovered at room temperature (Fig. 5g-h). Fig. 7c. Relationship of the period of swimming to specimen length for normal ammocoetes (closed circles). specimens which recovered at room temperature (21šC, closed squares) and specimens which recovered at reduced temperature (13šC, open circles). Notice that the room temperature transectees exhibit a different relationship from the reduced temperature transectees which are indistinguishable from normal ammocoetes. Fig. 7d. Relationship of the propagation time of flexion waves to specimen length for the same three classes of specimens as shown in Fig. 7c. Fig. 7e. Relationship of the mid-body curvature of flexion waves to specimen length for the same three classes of specimens as shown in Fig. 7c., Fig. 7f. Relationship of the intersegmental mechanical phase lag of flexion waves to specimen length for the same three classes of specimens as shown in Fig. 7c. Note that the effect of temperature on recovery is specific to the timing of swimming movements.

It is important to point out that many of the parameters are directly related to the size of specimens so any quantitative assay of recovery must take into account this length dependence. The deficit which is observed in specimens which recover at room temperature is highly specific to the timing of swimming. Specimens which recover at 21š differ over the entire range of body lengths in the frequency of their swimming movements (Fig. 7C), as well as the propagation time of the flexion waves (Fig. 7D). The animals are quite indistinguishable from normal specimens and specimens which recover at reduced temperature, however, in terms of the body curvature they achieve during swimming (Fig. 7E) and intersegmental mechanical phase lag (Fig. 7F).

D. Diagnosis of Behavioral Recovery.

Regression analysis allows us to estimate the expected "normal" value of any parameter for a lesioned specimen of a given length (Figs 7C-F). We can therefore express the average values of a recovered specimen as a percent of the "normal" value. Table II represents the average parameters extracted from the wave table (as in Table 1) for the stage 4 partially recovered ammocoete indicated in Fig. 5E-F. The first column indicates the measured parameter. The second column is the mean value for the complete sequence of recovered swimming. The third column is the mean value expressed as percentage of the expected value for specimen of the same length (fourth column), determined from the regression coefficients indicated in Fig. 7C-F. The mean mid-body flexion curvatures are indicated and the lateral curvature bias indicated.

Table II. Functional Recovery Diagnosis

Parameter Recovered % of Control Normal
Period 0.264 55.4% 0.477
Propagation Time 0.362 71.1% 0.509
Propagation Time/Period 1.370 125.7% 1.090
Mechanical Intersegmental Delay 2.78 67.2% 4.14
Mechanical Phase Lag 0.0106 121.9% 0.0087
Left Body Curvature -0.203 47.1% -0.431
Right Body Curvature 0.309 71.7% 0.431
Lateral Bias 1.52R

Note that this specimen is characterized by a period which is half that of normal specimens, but by a lesser reduction of propagation time. This results in a greater intersegmental mechanical phase lag than that expected for normal specimens. The mean body curvature is reduced on both sides but much more on the left than the right leading to highly assymetric flexion wave amplitudes.

The scheme described above applies to the majority of stage 4 specimens which recover at room temperature. There is considerable variability observed, however, and not all specimens exhibit complete recovery, regardless of the holding temperature. The major source of variability has to do with bilateral asymmetrys and the extent of propagation. These asymmetrys can be so extreme as to involve the recovery only of unilateral swimming movements. An example of this is shown in Fig. 8A-B. Note that this specimen is able to generate repetitive flexion waves to the right side of the body, however there is no alternation to the left side. Note also that the caudal waves to the right result from an initial flexion to the left in the head region.

Figure 8. Do bilateral curvature asymmetrys result from bilaterally asymmetric regeneration? Fig. 8a. Graph of the locus of flexions vs. time for an ammocoete which was capable of generating flexion waves only on one side of the body. Fig. 8b. Graph of curvature of flexions as a function of flexion locus. Fig. 8c-d, corresponding analysis of the swimming movements acute to a hemisection of the left side of the spinal cord at 25% of body length. Notice the bilateral flexion waves.

We assumed that this indicated that such recovery might result from asymmetric recovery of the descending motor command function. To further examine this proposal, we analyzed the swimming movements of specimens which had only an acute hemisection of the spinal cord. Hemisected specimens exhibit a bilateral asymmetry (Fig. 8D), but do show alternating propagating flexion waves (Fig. 8C). Their timing is within the range characteristic of normal specimens (Fig. 8C).This finding suggests that the descending normal command at the level of the hemisection contains information adequate to initiate propagating flexion waves on both sides of the body, but that regeneration can include subsets of the systems which are adequate to initiate flexion waves on one side only.


One issue which is unresolved by experiments on larval lamprey is the relationship of regenerative processes to the processes which occur during normal development. We have addressed this issue by comparing the recovered behavior which occurs in larval lamprey to the recovered behavior which occurs in specimens which were transected both prior to and after the adult metamorphosis (Ayers et al., 1982; Margolin and Ayers, 1987). As illustrated in Fig. 9A, recovered transformed adults can result from two independent pathways. Normal ammocoetes can be lesioned to recover as ammocoetes and then transform into adults. In contrast, ammocoetes can first transform into adults and then be lesioned and recover as transformers.

Figure 9. Swimming movements in recovered transformed lamprey. Fig. 9a Recovered transformers can result from regeneration preceeding transformation when the specimens are transected as ammocoetes (Figs. 9b-e). or from transformation preceeding regeneration when the specimens are transected as transformers (Fig. 9f-g). The specimen in figure 9b-c was transected as an ammocoete but recovered at room temperature (21šC). Compare the swimming movements with those of the recovered 21š ammocoete in Fig. 5g-h. In contrast, the specimen in figure 9d-e was transected as an ammocoete but recovered at reduced temperature (21šC). Compare the swimming movements with those of the recovered 13š ammocoete in Fig. 6a-b. The specimen in figure 9f-g was transected over 30 days subsequent to the achievement of stage 6 of metamorphosis and recovered at reduced temperature (13šC). Compare the swimming movements with those of the normal transformer in Fig. 2c-d

Recovering adult specimens exhibit a different set of phenomena than larvae. Acutely transected adults exhibit spontaneous spinal undulations when attached to the aquarium by their oral sucker (Ayers et al.,1982). These spinal undulations tend to come and go but may be observed in specimens which have achieved stage 4 of free recovered swimming. Spinal undulations are much slower and lower amplitude than normal transformer swimming and are more characteristic of the rhythm evoked by bath applied d-Glutamic acid (Ayers et al., 1983).

Figure 10. Recovery of behaviors other than swimming. Fig. 10a-b example of the recovered shoving phase of burrowing in a stage 5 room temperature ammocoete. Fig. 10c-d Example of recovered backward crawling in a stage 5 room temperature ammocoete. Fig. 10e-f Example of aversive withdrawal in a stage 5 room temperature transformer.

Transformed lamprey can exhibit swimming movements following recovery from transection and, in fact, their recovery is as robust as that observed in larvae (Margolin and Ayers, 1987). Fig 9B-C illustrates the recovered swimming movements of a specimen which was transected prior to metamorphosis and recovered at room temperature. Notice that it exhibits the rapid swimming movements which are characteristic of larval specimens which recover at room temperature. In other words its recovered behavior is similar to that of an ammocoete which recovers at room temperature. Specimens which are transected prior to metamorphosis but allowed to recover at reduced temperature exhibit slower swimming movements more characteristic of normal ammocoetes. If a lamprey is transected after the adult metamorphosis it then can recover almost normal swimming movements as shown in Fig. 9F-G. Thus behavioral recovery can occur, even if the transection occurs after metamorphosis. This finding differs from the observations in anurans where behavioral recovery is contingent on metamorphosis (Forehand and Farel, 1982).


In addition to recovering swimming movements, lamprey can also recover other undulatory behaviors. Fig. 10A-B illustrates an example of recovered burrowing behavior in an ammocoete. Notice that the specimen is attempting to shove into the substrate as observed during normal burrowing and exhibits a rather low frequency component at the start of the behavior. We have also demonstrated the recovery of backward crawling as shown in Fig. 10C-D. Notice here that the flexion waves propagate from tail to nose and have the low frequency characteristic of normal backward crawling.

Transformers can also recover these other undulatory behaviors. In fact they are capable of recovering even aversive withdrawal behavior as illustrated in Fig. 10E-F. Aversive withdrawal is one of the rarest recovered behaviors observed in ammocoetes and its occurrence in transected adults indicates just how robust the recovery is of the normal undulatory behavioral repertoire.


We have examined the regeneration of reticulospinal systems in both larval and adult lamprey by retrograde transport of Horseradish peroxidase (Mesulam, 1982). Normal specimens exhibit a considerable difference in the overall morphology of reticulospinal system between larval (Fig. 11A), and transformed (Fig. 11C) specimens. Counts of absolute cell number, however, indicate that few new neurons are added to the reticulospinal population during transformation (Swain and Ayers, 1986).

Following spinal transection, specimens are allowed to recover to behavioral criteria and re-transected at a distance of about 5 mm distal to the original transection. HRP injected into such a second transection fills only cells which have regenerated across the original lesion (J. Snedekor and M. Selzer, personal communication). Although considerable variability is observed between recovered specimens, we generally find that neurons in all of the four major reticulospinal cell groups (mesencephalic, isthmic, bulbar and vagal) fill in recovered specimens (Fig. 11B, D). In general, we have observed a gradual increase in number of regenerated neurons as the recovery time increases in parallel with progress to behavioral normalcy. We conclude from these experiments that the regeneration of the reticulospinal system following spinal transection is profound and relatively non-specific. In other words most cell groups regenerate as opposed to only a few groups regenerating. There is, however, a greater tendency for neurons with smaller cell bodies to regenerate to 5mm than giant reticulospinal neurons (G. Swain, unpublished).

Figure 11. Regeneration of reticulospinal neurons. The two left diagrams are camera lucida tracings of Horseradish peroxidase fills of the descending population of reticulospinal neurons which project to 25% of body length in a normal ammocoete larvae (Fig. 11a) and a normal transformed adult (Fig. 11c). The two right diagrams are corresponding maps of the regenerated neurons which project at least 5 mm distal to a regenerated transection at 25% of body length in an ammocoete which had recovered to stage 5 (Fig. 11b), and a transformed adult transected as a transformer and recovered to stage 5 (Fig. 11d). Notice that there are filled cells in all 4 major cell groups in the recovered specimens. Gary P. Swain, unpublished


What integrative processes underlie this recovered behavior? To address this issue, we have examined the physiological effects of electrical stimulation of neuron populations and identified neurons in the brainstem. Like most vertebrates (Forssberg, 1982; Mori, 1987), lamprey exhibit a descending command system which can elicit bursting discharge in isolated transformer neuraxes (McClellan, 1984; McClellan and Grillner, 1983).

Figure 12. Behavioral effects of brainstem microstimulation. Fig. 12a. In situ brain stimulation preparation. The specimen is pinned down by the gill arches and the brain surgically exposed. Fig. 12b. Electrical microstimulation (7-46µa) were delivered through double-barreled glass microelectrodes into the lateral brainstem. (stippled region). Fig. 12c. Response to subthreshold stimulation. Fig. 12d. Response to a threshold stimulus. Fig. 12e. Response to stimulation at 2x threshold. Notice that the period and propagation time of the flexion waves decreases with increasing current. Fig. 12f. .Response to stimulus at 4x threshold at which the swimming movements become uncoordinated due to the extreme prolongation of propagation times

A. Normal Function Of Descending Command Systems.

We have determined the behavior elicited by this descending system by analysis of the movements evoked by electrical microstimulation of normal ammocoetes and transformers in situ (Margolin,Kaufman, and Ayers, 1985). For these experiments we use an exposed brain preparation which is pinned down by its gill arches so that the caudal body is free to undulate (Fig. 12A). An example of the undulatory movements elicited by electrical microstimulation in the lateral isthmic region (Fig. 12B) in a normal ammocoete is shown in Fig. 12. The undulations evoked by microstimulation depend profoundly on the intensity of stimulus current. Subthreshold stimulation produces no undulation (Fig. 12A) while achievement of a discrete threshold produces coordinated flexion waves which propagate from head to tail. The swimming command can be enhanced by recruitment, for further increases in stimulus current above threshold can increase the frequency of the evoked movements. Further increases in the stimulus current tends to greatly increase the propagation time of the movements until coordinated swimming movements are abolished (Fig. 12F). There is little effect of increased stimulus frequency when the stimulus current is held constant in normal specimens.

Figure 13. Brainstem microstimulation of a command system can initiate swimming behavior in ammocoete larvae. The upper two panels (Fig. 13a-b), are an analysis of swimming behavior in a normal 14.5 cm ammocoete larvae. The lower two panels (Fig. 13c-d) are a corresponding analysis of the undulations evoked by brainstem microstimulation on the right side in a normal ammocoete of the same length.

When the undulations evoked by brainstem microstimulation are compared to the normal swimming movements of specimens of comparable size, they are quite similar both in timing and amplitude (Fig. 13). We have had similar results in transformed adults. These responses differ from the undulations evoked by pharmacological (D-Glutamate) stimulation in ammocoetes and transformers which, though phase constant are much slower and of greatly reduced amplitude (Ayers et al. 1983; Wallen and Williams, 1984). In other words, the D-Glutamate activated spinal clock generates undulations, but it keeps different time than the brainstem activated clock.

B. Function Of The Undulatory Command In Recovered Specimens.

The swimming command system recovers function following spinal transection (Currie and Ayers, 1983). Specimens in early stages of recovery show no propagating flexion waves in response to command stimulation. Whole body flexions to one side are most common in these earlier stages of recovery (Kaufman, Margolin and Ayers, 1985). In contrast, recovered descending control can be readily demonstrated in specimens which have achieved stage 5 of recovery. Figure 14A-B indicates the recovered behavior of a specimen which regenerated at 13š. When this same specimen was prepared for brainstem microstimulation, coordinated propagating undulations could be evoked with similar periods and propagation times to the recovered behavior (Fig. 14C). The brainstem evoked undulations had considerably reduced amplitude in this recovered specimen when compared to a similarly prepared normal specimen (compare Figs. 14D and 13D). Presumably the recovered behavior resulted from the additive effects of additional elements (Fig. 11B) than those elicited by the microstimulus.

Figure 14. Swimming command system stimulation can initiate swimming in recovered ammocoete transectees. The upper two panels (Fig. 14a-b), are an analysis of swimming behavior in a stage 5 recovered ammocoete larvae.The lower two panels (Figs. 14c-d) are a corresponding analysis of the undulations evoked in the same specimen following surgical reduction and brainstem microstimulation on the left side. Notice the similarity in timing but a marked reduction in the amplitude of the microstimulation evoked swimming.

As in normal ammocoetes, recovered ammocoetes demonstrate increases in the frequency and amplitude of swimming undulations in response to increases in stimulus current and stimuli in great excess of threshold current tended to eliminate the coordination of the swimming movements (D. Kaufman, unpublished). Ammocoetes which have recovered from spinal transection may demonstrate errors in command system function. For example we found specimens which exhibited different propagation times on the two sides or strong effects of increases in stimulus frequency.

In contrast to the considerable recovery of command system function in ammocoetes, we have found a lack of functional recovery of the command system in adult specimens. In recovered adults, the behavior evoked by brainstem microstimulation rarely approximates the recovered behavior (L. Margolin, unpublished). Brainstem microstimulation in recovered adults may initiate lateral undulations immediately caudal to the lesion, but in contrast to the corresponding responses in recovered ammocoetes these rarely propagate to the tail. These findings indicate that the recovered behavior observed in transformed adults probably does not result from the parametric effects of a single descending system (Pinsker and Ayers, 1983), but may require the summated effects of several descending systems (Fig. 11D), acting in concert.

Figure 15. Plasticity of fin postural command system function following spinal transection. Fig 15a In situ intracellular stimulation preparation for fin posture analysis. Fig 15b. Action potentials were evoked in the Muller neuron I1 by intracellular pulses which were repeated in trains. Fig. 15c. Lateral fin movement evoked by trains of I1 action potentials. Fig 15d . Plasticity in the efferent effects of I1 at various times after transection between the first and second dorsal fins.The graph indicates the stimulus frequency necessary for threshold and maximal contractions of the dorsal fin. The response threshold initially increases over the first two hours after which no response can be evoked. Responses of the anterior find did not return until the specimens had achieved stage 5 of swimming recovery. Adapted from Currie and Ayers, 1987

C. Recovery Of Fin Postural Command Function.

In one case, we have been able to extend our observations on recovery on interneuronal function to an identified reticulospinal interneuron, I1 (Currie and Ayers, 1982; 1987). Intracellular stimulation of I1 controls the posture of both dorsal fins (Fig. 15). Transection of the spinal cord between the dorsal fins results not only in a loss of posterior dorsal fin posture, but a rapid loss in the ability to control anterior dorsal fin posture as well (Fig. 15D). This deficit in the ability to control anterior dorsal fin posture recovers by the time the ability to initiate caudal swimming movements recovers. Loss of function rostral to a spinal lesion parallels the time course of the formation of retraction bulbs and their transformation to growth cones (Hall and Cohen, 1983). This finding indicates that this neuron experiences a retrograde loss of function which gradually recovers anterior to the lesion. We have never observed I1 to recover the ability to control posterior dorsal fin posture caudal to a transection, even in specimens which have achieved stage 5 of recovery (Currie and Ayers, 1983).


Recovery from spinal cord transection in the lamprey constitutes an excellent model for establishment of the mechanisms of regeneration and functional recovery of descending systems following spinal cord trauma. The lamprey model exhibits all of the criteria necessary for a model of spinal regeneration leading to functional recovery and in addition affords the opportunity to compare developmental processes with regenerative processes. In addition, the temperature at which recovery occurs affords a probe with which regeneration and recovery can be experimentally manipulated.

Recovery of the ability to initiate swimming following spinal transection in ammocoete larvae appears to result from regeneration and recovery of function in the descending reticulospinal system which mediates the swim command in normal specimens. At this time we have not established the mechanism of recovery in adult specimens, although their behavioral recovery is every bit as robust as that observed in ammocoetes.

Acknowledgments: I thank my colleagues Gail A. Carpenter, Scott N. Currie, Galen Eaholtz, Dean Kaufman, James F. Kinch, Lee Margolin, Gary P. Swain and Cricket Corwin Wilbur for their considerable contributions to this project.


  • Aguayo, AJ, David, S and Bray, G M (1981) Influences of the glial environment on the elongation of axons after injury: Transplantation studies in adult Rrodents. J Exp Biol 95: 221-240.
  • Ayers, J, and Wilbur, C (1986) Temperature determines the quality of recovered swimming in larval sea lamprey. Soc Neurosci Abstr 12: 1574.
  • Ayers, J, Carpenter, G, Currie, S and Kinch, J (1980) Quantitative analysis of normal and regenerating behaviors in the sea lamprey. Soc Neurosci Abstr 6: 431
  • Ayers, J, Carpenter, G, Currie, S and Kinch, J (1981) Behavioral analysis of spinal cord regeneration in the Sea Lamprey. Soc Neurosci Abstr 7: 681.
  • Ayers, J, Carpenter, G, Currie, S and Kinch, J (1983) Which behavior does the lamprey central motor program mediate?, Science 221: 1312-1314.
  • Ayers, J, Currie, S, Kinch, J and Pereira, W (1982) Adult lamprey can recover from complete spinal cord transection. Soc Neurosci Abstr 8: 868.
  • Barbeau, H and Rossignol, S (1987) Recovery of locomotion after chronic spinalization in the adult cat. Brain Res 412: 84-95.
  • Bernstein, J and Gelderd, J (1970) Regeneration of the long-spinal tracts in the goldfish. Brain Res 20: 33-38.
  • Borgens, R, Roederer, E and Cohen, MJ (1981) Enhanced spinal cord regeneration in lamprey by applied electric fields. Science 213: 611-617.
  • Cajal, SR (1928) Degeneration and Regeneration of the Nervous System. Oxford Univ Press.
  • Cohen, AH, Mackler, SA and Selzer, ME (1986) Functional regeneration following spinal transection demonstrated in the isolated spinal cord of the larval sea lamprey. Proc Natl Acad Sci USA 83: 2763-2766.
  • Currie, S and Ayers, J (1982) Gradual loss and partial recovery of a proximal motor reflex following spinal transection in lamprey (Petromyzon marinus). Soc Neurosci Abstr 8: 870.
  • Currie, S, and Ayers, J (1983) Regeneration of locomotor command systems in the lamprey. Brain Res 279: 238-240.
  • Currie, S, and Ayers, J (1987) Plasticity of fin command system function following spinal cord transection in larval sea lamprey, Petromyzon marinus, Brain Research 415: 337-341.
  • de Medinacelli, L, Freed, WJ and Wyatt, RJ (1982) An index of the functional condition of rate sciatic nerve based on measurements made from walking tracks. Exp Neurol 77: 634-643.
  • de Medinacelli, L, Quach, T , Duchemin, A and Wyatt, RJ (1986) Is vigor of regeneration a key factor in recovery from peripheral nerve injuries? Exp Neurol 94: 788-790.
  • Duffy, MT, Davis, BM and Simpson, SB (1987) Bulbospinal and propriospinal connections in the salamander regenerated spinal cord. Soc Neurosci Abstr 13: 749.
  • Eaholtz, G (1985) Behavioral recovery results from spinal cord regeneration in larval sea lamprey. Soc Neurosci Abstr 11: 589.
  • Eidelberg, E (1981) Consequences of spinal cord lesions upon motor function, with special reference to locomotor activity. Prog Neurobiol 17: 185-202.
  • Forssberg, H (1982) Spinal locomotor functions and descending control. In: Brain Stem Control of Spinal Mechanisms. B Sjolund and A Bjorklund, [eds] Elsevier Biomedical Press, Amsterdam. Pp. 253-271.
  • Forehand, C and Farel, P (1982) Anatomical and behavioral recovery from the effects of spinal cord transection: dependence on metamorphosis in anuran larvae. J Neurosci 5: 654-662.
  • Freed, WJD de Medinaceli, L and Wyatt, RJ (1985) Promoting functional plasticity in the damaged nervous system. Science 227: 1544-1552.
  • Grillner, S, Brodin, L Sigvardt, K and Dale, N (1986) On the spinal network generating locomotion in lamprey: transmitters, membrane properties and circuitry. In: Neurobiology of Vertebrate Locomotion. S Grillner, P Stein, D Stuart, H Forssberg and R Herman [eds] Macmillan Press, Hampshire, England, Pp. 335-352.
  • Guth, L, Brewer, CR, Collins, WF, Goldberger, ME and Perl, ER (1980) Criteria for evaluating spinal cord regeneration experiments. Exp Neurol 69: 1-3.
  • Hall, G and Cohen, MJ (1983) Extensive dendritic sprouting induced by close axotomy of central neurons in the lamprey. Science 222: 518-521.
  • Kaufman, D, Margolin, L and Ayers, J (1985) Behavioral analysis of the recovery of command system function following spinal transection in larval and adult sea lamprey. Soc Neurosci Abstr 11: 589.
  • Liuzzi, FJ and Lasek, RJ (1987) Astrocytes block axonal regeneration in mammals by activating the physiological stop pathway. Science 337: 642-647
  • Mackler, SA and ME Selzer (1985) Regeneration of functional synapses between individual recognizable neurons in the lamprey spinal cord. Science 229: 774-776.
  • Margolin, L, and Ayers, J (1987) The effect of temperature on the quality of recovered swimming following spinal transection in metamorphosing sea lamprey. Soc Neurosci Abstr 13: 1287.
  • Margolin, L, Kaufman, D and Ayers, J (1985) Development of command system function during transformation in the Sea Lamprey. Soc Neurosci Abstr 11: 1287.
  • McClellan, AD (1984) Descending control and sensory gating of 'fictive' swimming and turning responses elicited in an in vitro preparation of the lamprey brainstem/spinal cord. Brain Res 302: 151-162.
  • McClellan, AD and Grillner, S (1983) Activation of 'fictive swimming' by electrical microstimulation of brainstem locomotor regions in an in vitro preparation of the lamprey central nervous system. Brain Res 300: 357-361.
  • Mesulam, MM (1982) Tracing neural connections with horseradish peroxidase IBRO handbook series: Methods in the neurosciences. New York: John Wiley and Sons.
  • Mori, S (1987) Integration of posture and locomotion in acute decerebrate cats and in awake freely moving cats. Prog Neurobiol 28: 161-195.
  • Nichols, JG (1982) [ed] Repair and Regeneration of the Nervous System. Dahlem Konferenzen, Springer-Verlag, Berlin.
  • Pinsker, HM and Ayers, J (1983) Neuronal Oscillators. Chapter 9 In: The Clinical Neurosciences. Section Five, Neurobiology. W D Willis [ed] Churchill Livingstone Publishers. Pp. 203-266.
  • Rovainen, C M (1976) Regeneration of Müller and Mauthner Axons after Spinal Transection in Larval Lampreys. J Comp Neurol 168: 545-554.
  • Rovainen, CM (1979) Neurobiology of Lampreys. Physiol Rev 59: 1007-1077.
  • Seil, F, Herbert, E, Carlson, B [eds] (1987) Neural Regeneration, Elsevier, New York.
  • Selzer, ME (1978) Mechanisms of functional recovery and regeneration after spinal cord transection in larval lampreys. J Physiol (London) 277: 395-408.
  • Sharma, SC, Breckwoldt, R Astartita, V and Berthoud, M (1986) Regeneration of descending axons in goldfish. Soc Neurosci Abstr 12: 625
  • Stein, PSG (1978) Motor systems with specific reference to the control of locomotion. Ann Rev Neurosci 1: 61-81.
  • Swain, GP and Ayers, J (1986) Development of descending reticulospinal systems during transformation in the Sea Lamprey. Soc Neurosci Abstr 12: 318
  • Wallen, P and Williams, T (1984) Fictive locomotion in the lamprey spinal cord in vitro compared with swimming in the intact and spinal animal. J Physiol 347: 225-239.
  • Wilbur, C, Margolin, L, and Ayers, J (1987) The effect of temperature on the chronology of behavioral recovery following spinal transection in metamorphosing Sea Lamprey. Soc Neurosci Abstr 13: 1287
  • Wood, M and Cohen, M (1979) Synaptic regeneration in identified neurons of the lamprey spinal cord. Science 206: 344-347.
  • Yin, HS and Selzer, ME (1983) Axonal regeneration in lamprey spinal cord. J Neurosci 3: 1135-1144.
  • Yin, HS and Selzer, ME (1984) Electrophysiological evidence of regeneration of lamprey spinal neurons. Exp Neurol 83: 618-628.

  • .