period differences between segmental oscillators produce

Period Differences Between Segmental Oscillators ProduceIntersegmental Phase Differences in the Leech HeartbeatTiming Network

MARK A. MASINO AND RONALD L. CALABRESEBiology Department, Emory University, Atlanta, Georgia 30322

Received 25 April 2001; accepted in final form 25 October 2001

Masino, Mark A. and Ronald L. Calabrese. Period differencesbetween segmental oscillators produce intersegmental phase differ-ences in the leech heartbeat timing network. J Neurophysiol 87:1603–1615, 2002; 10.1152/jn.00338.2001. Considerable experimentaland theoretical effort has been exerted to understand how constantintersegmental phase relationships are produced between oscillators insegmentally organized pattern generating networks. The phase rela-tionship between the segmental oscillators in the isolated timingnetwork of the leech heartbeat central pattern generator is quiteregular within individual preparations. However, it varies consider-ably among different preparations. Our goal is to determine how thephase relationships in this network are established. Here we assesswhether inherent period differences, as suggested by the excitability-gradient hypothesis, play a role in establishing the phase relationshipsbetween the two coupled segmental oscillators of the leech heartbeattiming network. To do this we developed methods for reversiblyuncoupling the segmental oscillators (sucrose knife) and pharmaco-logical manipulation of the individual oscillators (split bath). Differ-ences in inherent cycle periods between the third and fourth segmentaloscillators (G3 and G4) were present in most (20 of 26) preparations.These period differences correlated with the phase differences ob-served between the segmental oscillators in the recoupled timingnetwork, such that the oscillator with the faster cycle period, regard-less of the segment in which it was located, led in phase in proportionto its period difference with the other oscillator. The phase differencesbetween the original (coupled) and recoupled states of individualpreparations were similar. Thus application and removal of the su-crose knife did not alter the period difference between the segmentaloscillators in the timing network. Pharmacological manipulation ofthe uncoupled oscillators to alter the period difference between theoscillators led to similar correlated phase differences in the recoupledtiming network. Across all experiments the uncoupled segmentaloscillator with the faster cycle period established the cycle period ofthe timing network when recoupled. In conclusion, our findings indi-cate that an excitability-gradient plays a role in establishing the phaserelationship between the segmental oscillators of the leech heartbeatcentral pattern generator since inherent period differences presentbetween the oscillators are correlated to the phase relationships of thecoupled/recoupled timing network.

I N T R O D U C T I O N

Segmental oscillators are coordinated to produce constantphase relationships that are independent of cycle period be-tween segments in isolated nerve cord preparations in many

segmentally distributed motor pattern generating networks,including lamprey swimming (Cohen 1987a; Grillner et al.1991, 1995) and crayfish swimmeret (Murchison et al. 1993)networks. Considerable experimental and theoretical effort hasbeen exerted to understand how these constant phase relation-ships are produced (crayfish, Braun and Mulloney 1995; Mul-loney 1997; Skinner and Mulloney 1998; Stein 1971; lamprey,Cohen 1987a; Grillner et al. 1993; Kotaleski et al. 1999a,b;Sigvardt 1993; Wadden et al. 1997; Wallen et al. 1992). Anemerging consensus is that asymmetries in synaptic couplingbetween oscillators give rise to the phase differences betweensegments (Skinner and Mulloney 1998; Wadden et al. 1997).An alternative hypothesis (Excitability-Gradient), which statesthat differences in inherent cycle periods between symmetri-cally coupled segmental oscillators produce the observed phasedifferences, received some early support (Grillner et al. 1993).However, this hypothesis is now widely discounted in severalsystems (crayfish, Mulloney 1997; Skinner and Mulloney1998; Skinner et al. 1997; lamprey, Cohen 1987a; Sigvardt1993; Sigvardt and Williams 1996).

In contrast to these systems, the first paper of this series(Masino and Calabrese 2002) described the wide range ofphase relationships observed in isolated ganglionic chains [G3to G4 (G3-G4) or headbrain to G4 (HB-G4)] between the twosegmental oscillators that form the timing network of the leechheartbeat central pattern generator. We also showed that, al-though the connections between the coordinating heart inter-neurons [in G1 and G2 (G1,2)] and the oscillator heart inter-neurons (in G3 and G4) are complex and anatomicallyasymmetric, the system functions mainly in a symmetric mode(Masino and Calabrese 2002). Most of the action potentials inthe coordinating interneurons arise at an initiation site in G4that is inhibited by the oscillator interneurons located in bothG3 and G4. The modeling studies of the previous paper (Hill etal. 2002) predict that when the system functions in this sym-metric mode the phase differences arise from differences in theinherent periods of the half-center oscillators that constitute thecore of the two segmental oscillators. Moreover, these studiespredict that the period of the timing network should be theperiod of the faster of the two independent segmental oscilla-tors.

Address for reprint requests: R. L. Calabrese, Biology Dept., Emory Uni-versity, 1510 Clifton Rd., Atlanta, GA 30322 (E-mail: [email protected]).

The costs of publication of this article were defrayed in part by the paymentof page charges. The article must therefore be hereby marked ‘‘advertisement’’in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

J Neurophysiol87: 1603–1615, 2002; 10.1152/jn.00338.2001.

16030022-3077/02 $5.00 Copyright © 2002 The American Physiological Societywww.jn.org

In this study, we developed methods for reversibly uncou-pling the segmental oscillators (sucrose knife) and pharmaco-logical manipulation of the individual oscillators (split bath) totest these predictions experimentally. In these experiments, thesegmental oscillators were uncoupled, their independent peri-ods assessed and in some cases altered with pharmacologicalmanipulations, and then the timing network reconstituted. Thusthe flexibility in phase relationship between segmental oscilla-tors in the leech heartbeat timing network affords an opportu-nity to explore the adaptability of networks that produce co-ordinated segmental motor output and to uncover potentiallynovel mechanisms for intersegmental phase regulation.

M E T H O D S

The methods for dissection, extracellular recording, data acquisi-tion, and data analysis are as described in Masino and Calabrese(2002). All split-bath experiments were carried out on isolated G3–G4chains, while concentration response experiments were carried out onisolated, single G3 and G4 ganglia.

Sucrose knife technique

We developed a method (sucrose knife technique) to reversiblyuncouple segmental ganglia by blocking the propagation of actionpotentials in the interganglionic connective with a sucrose solution(260 mM) similar in osmotic concentration to normal leech saline.Small diameter tubing (1/16-in. ID, 1/8-in. OD) with a notch cut in itsmiddle was placed across the center of the silicone elastomer (Syl-gard)–lined preparation dish. The G3 to G4 (G3/G4) interganglionicconnective was placed in the notch so that it traversed the tube (Fig.1A), and petroleum jelly (Vaseline) was applied with a syringe to sealthe notch in the tubing. Two different solutions could be gravity fedthrough the sucrose knife tubing. Normal leech saline was fed into thetubing under control conditions (coupled or recoupled states), whilethe sucrose solution was fed into the tubing to uncouple the ganglia.Reversible uncoupling and recoupling of the ganglia were used toassess period differences between segmental oscillators of the heart-beat timing network and their effects on the recoupled phase relation-ships.

Pot electrodes were used for extracellular recording and stimulationof the connective nerves in tests of the efficacy of the sucrose knife onG3–G4 chains. A Vaseline pot was formed around the cut end of the

FIG. 1. A: the split bath–sucrose knifetechnique. The sucrose knife tubing separatedthe preparation dish into 2 separate compart-ments (anterior and posterior baths). TheG3–G4 chain preparation is shown pinned(ventral surface up) in the dish with theG3–G4 interganglionic connective placed inthe notch so that it traverses the tube. See textfor details of operation. Under normal condi-tions, the ganglion in each bath was super-fused with normal saline. To modify the cycleperiod of one oscillator, however, saline con-taining either myomodulin or Cs� was appliedto its bath. The electrical activity of the oscil-lator heart interneurons was recorded with ex-tracellular suction electrodes. B: the sucroseknife technique reversibly blocked the propa-gation of action potentials in the G3–G4 inter-ganglionic connective. Pot electrodes wereformed around the cut ends of the connectiveanterior to G3 (stimulation) and of the connec-tive posterior to G4 (recording). Coupled:prior to applying sucrose solution to theG3/G4 connective, a complex compound ac-tion potential generated by an extracellularelectrical stimulus applied to the connectiveanterior to G3 was observed in an extracellularrecording of the connective posterior to G4.Uncoupled: when the sucrose knife was ap-plied to the G3/G4 connective (indicated bylarge X), however, activity generated by thestimulus did not cross the “knife.” Recoupled:when the sucrose knife was removed by re-placing sucrose solution in the knife tubingwith normal saline, the activity generated bythe stimulus was once again observed in theconnective posterior to G4. The stimulus arti-fact (SA) was clipped in each panel. The ex-tracellular records were scaled to the same sizebefore and after the sucrose knife.

1604 M. A. MASINO AND R. L. CALABRESE

J Neurophysiol • VOL 87 • MARCH 2002 • www.jn.org

connective, and recording/stimulating silver wire was inserted into theSylgard inside the pot. Recordings and stimuli were then referenced toa second silver wire in the general bath, and standard extracellularstimulating and recording techniques were applied. Electrical stimuli(1 ms) applied to the connective at one end of the coupled chain (withnormal saline in the “knife”) evoked a complex compound actionpotential recorded at the other end (Fig. 1B, Coupled). This responsewas blocked when the ganglia were uncoupled (with sucrose solutionin the knife; Fig. 1B, Uncoupled) by the sucrose knife and returnedwhen the ganglia were recoupled (with normal saline in the knife; Fig.1B, Recoupled).

Concentration response curves

Concentration response curves for leech myomodulin (provided byC. Sahley, Purdue University), molluscan myomodulin (PeninsulaLabs, San Carlos, CA), and Cs� (Cesium Chloride, Sigma, St. Louis,MO) on cycle period were measured in isolated G3 and G4 segmentaloscillators and in the intact heartbeat timing network (HB–G4). Myo-modulin and Cs� solutions were made fresh in normal leech salinebefore each experiment. Increasing concentrations of myomodulin(10�9 to 5 � 10�9 M) or Cs� (10�4 to 5 � 10�3 M) solutions wereapplied to the bath with 3- to 5-min normal saline washes betweenapplications. The mean oscillator cycle period at each concentrationwas measured from at least 12 consecutive bursts and normalized tothe cycle period of the timing network at the beginning of theexperiment. These normalized periods were then averaged at eachconcentration across preparations.

Split bath experiments

Split bath preparations were created by forming a Vaseline partitionbetween the G3 and G4 segmental ganglia across the G3/G4 connec-tive, such that G3 was located in the anterior bath and G4 was locatedin the posterior bath (Fig. 1A). The segmental oscillators remainedcoupled for the duration of these experiments since the sucrose knifetechnique was not used. We applied saline containing myomodulin(10�6 M) to one bath, either anterior or posterior, while normal salinewas applied to the other. The effects on network cycle period and theG3 to G4 phase relationship were assessed.

Split bath–sucrose knife experiments

The sucrose knife tubing separated the preparation dish into twocompartments (split bath), such that G3 was in the anterior bath andG4 in the posterior bath (Fig. 1A). In the split bath–sucrose knifeexperiments, we created large period differences or reversed theperiod differences between segmental oscillators, which had beenuncoupled, by applying saline containing myomodulin or Cs� to onebath while normal saline was applied to the other. The effects of theseperiod differences on the phase relationships of the segmental oscil-lators could then be assessed by recoupling the ganglia.

R E S U L T S

Do inherent cycle period differences between segmentaloscillators determine their phase relationships?

We attempted to determine whether the different inherentcycle periods (T) of the G3 and G4 segmental oscillatorsdetermined their phase (�) relationships, as predicted by theprevious modeling studies (Hill et al. 2002), by reversiblyuncoupling these oscillators and correlating their different in-herent periods with their coupled and recoupled phase differ-ences.

The sucrose knife, which blocked the conduction of action

potentials in the G3/G4 connective (METHODS), was used toreversibly uncouple the G3 and G4 segmental oscillators, andtheir inherent periods were measured. An example demonstrat-ing the efficacy of the sucrose knife technique is shown in Fig.2. As expected, the period of the timing network was regularcycle-by-cycle, and the burst activities in ipsilateral G3 and G4oscillator interneurons were phase locked in the coupled state(Fig. 2, Coupled). When sucrose solution was fed into thesucrose knife tubing, there was a brief interval during whichnetwork behavior was somewhat chaotic (Fig. 3). This intervallasted �3 min, after which the G3 and G4 oscillator interneu-rons began to burst independently (Fig. 2, Uncoupled). In thiscase, the G3 oscillator interneuron slowed, while the G4 os-cillator interneuron sped with respect to the coupled period. In

FIG. 2. Reversibly uncoupling the G3 and G4 oscillators with the sucroseknife allowed them to express their inherent cycle periods. Coupled: ipsilateraloscillator interneurons in G3 and G4 were phase locked prior to applying thesucrose knife. Notice that the cycle period was regular and that HN(R,3) cellled HN(R,4) in phase (indicated by lines connecting symbols in each of thepaired bursts). Uncoupled: when the sucrose knife was applied to the connec-tive, the G3 and G4 segmental oscillators cycled independently. The oscillatorinterneurons in G3 and G4 had different cycle periods when uncoupled, suchthat the cycle period of the G3 oscillator interneuron [HN(R,3) cell] was fasterand the cycle period of the G4 oscillator interneuron [HN(R,4) cell] was slowerthan the original cycle period of the coupled network. Recoupled: the segmen-tal oscillators assumed the same period and became phase locked (indicated bylines connecting symbols in each of the paired bursts) when the sucrose knifewas removed. The original (coupled) G3 to G4 phase relationship was pre-served in the recoupled state. Also, the recoupled and the faster uncoupledoscillator’s (G4) cycle periods were comparable. A raster point above eachspike indicated a detected spike event. The median spike in each burst wasindicated by a symbol above the burst; circles represent G3 oscillator inter-neurons and asterisks represent G4 oscillator interneurons. This symbol rep-resentation is preserved in Figs. 3 and 6–9.

1605PERIOD DIFFERENCES PRODUCE INTERSEGMENTAL PHASE LAGS


many preparations, the period of the G3 oscillator interneuronwas more regular than that of the G4 oscillator interneuronwhen uncoupled. When saline was fed back into the sucroseknife tubing (Recoupled), the oscillator interneurons quickly(�3 min) attained the same cycle period and became phaselocked close to their original G3 to G4 phase relationship (Fig.2, Recoupled). In some preparations, during the transitionperiod from the coupled to uncoupled or the uncoupled torecoupled states, regular bursting broke down and was replacedby either intense firing [Fig. 3, HN(L,3) in Transition toRecoupled State] or by sporadic bursting for a brief period(�30 s). This disorganized behavior was expressed by theoscillator interneuron in G3 or G4, or by both concurrently, andwas followed by a normal bursting pattern.

Similar results were obtained in all 26 preparations uncou-pled with the sucrose knife in this phase of the study. Therecoupled period and the coupled period were strongly corre-lated (r � 0.9, P � 0.001) in these preparations, but there wassome tendency for preparations with long periods to haveshorter periods when recoupled (Fig. 4A). Moreover, the re-coupled G3 to G4 phase relationship was strongly correlated tothe coupled G3 to G4 phase relationship (Fig. 4B; r � 0.9, P �0.001). The coefficient of variation (CV) was used to comparethe variability of the uncoupled G3 and G4 cycle periods. Wecalculated CV for the mean cycle period of the uncoupledoscillators in each preparation (n � 26). The mean CV for eachgroup (CVG3 or CVG4) was calculated from the individual CVsacross all preparations in that group. The periods of the un-coupled G3 oscillators were more regular (indicated by a lowCVG3 � 4.5 � 1.7%, mean � SD) than were the periods of theG4 oscillators (indicated by a high mean CVG4 � 10.1 � 5.3%)across preparations. Variability of cycle period in the G3

oscillators was significantly lower than in the G4 oscillators(paired t � �5.7, P � 0.001, n � 26). Similar results wereobtained with isolated ganglia used in the concentration responseanalyses of Fig. 5, B and C (mean CVG3 � 4.0 � 1.8% andmean CVG4 � 9.0 � 6.0%; paired t � �2.8, P � 0.03, n � 6).

These results indicate that the sucrose knife can be used torapidly and reversibly uncouple the G3 and G4 segmentaloscillators without fundamentally altering their original periodand phase when recoupled. The uncoupled segmental oscilla-tors cycle independently and appear to express their own inherentperiods; however, the G3 oscillator is more regular than the G4oscillator (Fig. 2). The bases of this difference are not knownbut may reflect the greater complexity of the G4 oscillator. Theaxonal processes of the G3 oscillator interneurons remainactive in isolated ganglia, but their activity is not well coordi-nated with the other elements of the network (Peterson 1983a).They thus represent a potential source of sporadic input to thecoordinating interneurons in isolated or uncoupled G4.

The G3 and G4 segmental oscillators expressed differentinherent periods in the majority (20 of 26) of the preparations,when uncoupled with the sucrose knife. There were a numberof preparations in which the G4 oscillator was faster (9 of 26)and a number in which it was slower (11 of 26). In thosepreparations where their periods were similar, the oscillatorsnevertheless cycled independently as judged by the variabilityin their phase relationships. We correlated both the coupledand the recoupled G3 to G4 phase difference (�3 – �4) with theinherent period difference (T3 – T4) of the segmental oscillators(Fig. 4C) and found a strong and significant correlation foreach (coupled, r � 0.7, P � 0.001; recoupled, r � 0.7, P �0.001). These results indicate that the period difference be-tween the uncoupled oscillators is a good predictor of the

FIG. 3. Brief transition states occurred when thesucrose knife was applied (top) to and removed (bot-tom) from the G3/G4 interganglionic connective.Transition to Uncoupled State: the G3 and G4 oscil-lator interneurons were phase locked (indicated bylines connecting symbols in each of the 1st 3 pairedbursts) prior to the start (arrow) of the sucrose knife.The oscillator interneurons experienced a brief tran-sition period after applying the sucrose knife. Fol-lowing the transition period the G3 and G4 oscillatorinterneurons cycled independently (uncoupled).Transition to Reoupled State: the G3 and G4 oscil-lator interneurons cycled independently prior to re-moving the sucrose knife (arrow), experienced a brieftransition period once the sucrose knife was re-moved, and became phase locked (indicated by linesconnecting symbols in each of the last 3 pairedbursts) following the transition. In some prepara-tions, a period of chaotic spiking, similar to thatobserved in the HN(L,3) cell, occurred during thetransitions.



recoupled phase relationship of the G3 and G4 oscillators,regardless of which oscillator is faster or which is slower.

How is the cycle period of the timing network determined?

The modeling studies of the previous paper (Hill et al. 2002)predict that the period of the coupled timing network should be

the period of the faster segmental oscillator. To determinewhether the cycle period of the recoupled timing network wasestablished by one or the other, or both of the segmentaloscillators, the recoupled period was plotted against the periodof the faster (F) and of the slower (E) of the two uncoupledsegmental oscillators for each preparation (Fig. 4D). The plot-ted points for the faster oscillators fell along and on either

FIG. 4. Analysis of the sucrose knife experiments. A: plot of the recoupled timing network cycle periods vs. the coupled timingnetwork cycle periods. The cycle period was similar in the recoupled and the coupled states; however, there was some tendencyfor preparations with long periods to have shorter periods when recoupled. B: plot of the recoupled vs. the coupled G3 to G4 phaserelationships. The phase relationships were similar in the coupled and recoupled states, which indicated that application andremoval of the sucrose knife did not alter the period difference between the segmental oscillators in the timing network. C: plotof the recoupled G3 to G4 phase relationships vs. the differences in inherent cycle period between the G3 and G4 segmentaloscillators. Inherent period differences between the uncoupled segmental oscillators predict the coupled and recoupled G3 to G4phase differences. The inherently faster uncoupled oscillator led in phase and established the cycle period in the recoupled timingnetwork. Phase (�3 – �4) values indicated whether the G4 (positive phase) or G3 (negative phase) oscillator led in phase whenrecoupled, while period differences (T3 – T4) indicated whether the G3 (negative change in period) or the G4 (positive change inperiod) oscillator had the inherently faster cycle period prior to recoupling (in the uncoupled state). D: plot of the recoupled timingnetwork cycle periods vs. the inherent cycle periods of the uncoupled segmental oscillators. For each preparation the faster andslower of the 2 oscillators was determined, and they were plotted separately. Notice that the cycle period of the faster oscillators(●) fell more closely to the unity line than did the cycle period of the slower oscillators (E). The recoupled cycle period of the timingnetwork was established by the cycle period of the inherently faster uncoupled oscillator.



side of the unity line, while all but one of the points for theslower oscillators fell below the unity line. Thus the inherentlyfaster uncoupled segmental oscillator, regardless of its gan-glion of origin, established the cycle period of the recouplednetwork.

Testing the period difference hypothesis

Both the sucrose knife experiments described above and themodeling studies described in the previous paper (Hill et al.2002) lead to the hypothesis that inherent period differencesbetween segmental oscillators produced the phase relation ofthe heartbeat timing network and that the faster of the twosegmental oscillators determined its period. To further test thishypothesis, we markedly altered the period of one of theoscillators in the uncoupled state and assessed the effects onthe recoupled phase difference and period.

First, we sought exogenous substances that could alter theperiod of an isolated oscillator in a concentration-dependentfashion. The neuropeptide myomodulin has been shown tomodulate ion channels and motor activity in a variety ofinvertebrate systems including Aplysia (Cropper et al. 1987a,b)and Lymnaea (Santama et al. 1994a,b). Recently, a novelmyomodulin-like peptide was identified and characterized inthe CNS of the leech (Wang et al. 1998). Synthetic leech andmolluscan (Aplysia) myomodulins showed identical neuronalmodulatory effects on the Retzius cell in the leech (Wang et al.1998), thus we assessed whether leech and molluscan myo-modulins affected the cycle period of the heartbeat segmentaloscillators in isolated ganglia. Leech myomodulin in normalsaline produced a concentration-dependent decrease in cycleperiod for the isolated G3 segmental oscillator (Fig. 5A). Themechanism by which myomodulin decreased cycle period isnot known, but it is likely that it may modulate the intrinsicmembrane properties of the heart interneurons, as does thepeptide FMRFamide (Nadim and Calabrese 1997). We usedmolluscan myomodulin instead of the leech form for the fol-lowing experiments, however, because it produced similareffects on cycle period (Fig. 5B) as did the leech form andbecause it was more readily available.

The mean cycle period of the oscillators at each myomodulinconcentration was normalized to the mean cycle period prior toany myomodulin application (control). At low concentrations(�10�8 M) the cycle period of the segmental oscillators re-mained close to the control period. The cycle period of thesegmental oscillators, however, decreased when 5 � 10�8 Mmyomodulin was present in the normal saline and continued todecrease as the myomodulin concentrations increased. At aconcentration of 10�5 M myomodulin, the G3 and G4 oscil-lator cycle periods were nearly twice as fast (�0.5; normal-ized) as their original periods. We did not measure responses atconcentrations �10�5 M because the robust response at con-centrations between 10�7 and 10�5 M was sufficient for ourneeds. In subsequent split bath and split bath–sucrose knife

FIG. 5. Concentration response curves for the effect of myomodulin andCs� on the cycle period of the oscillators in the heartbeat timing network. A:plot of the normalized cycle period for isolated G3 segmental oscillators ateach concentration of leech myomodulin. Leech myomodulin had a similareffect to molluscan myomodulin on cycle period in the isolated G3 oscillator(compare plots A and B). B: plot of the normalized cycle period for isolated G3and G4 segmental oscillators at each concentration of myomodulin. Molluscanmyomodulin produced a similar concentration-dependent decrease in cycleperiod for the 2 segmental oscillators. C: plot of normalized cycle period forisolated G3 and G4 segmental oscillators at each concentration of Cs�. Cs�

produced a concentration-dependent change in cycle period that was similarfor the 2 segmental oscillators. For each concentration response curve, the datawere pooled from 6 preparations. Error bars indicate SD.



experiments, we typically used 10�6 M molluscan myomodu-lin to speed one segmental oscillator.

In addition, we assessed the concentration response of myo-modulin in cycle period for G3–G4 chains over a limited range(5 � 10�8 to 5 � 10�6 M) of the most effective concentra-tions. The chain preparations produced a concentration re-sponse relationship that was similar to the relationship ob-served for the isolated ganglia, however, as the cycle perioddecreased with increased concentrations of myomodulin, theG3 to G4 phase relationship usually shifted toward zero (datanot shown).

Low concentrations of Cs� in normal saline produced aconcentration-dependent increase in cycle period that was sim-ilar for the isolated G3 and G4 segmental oscillators (Fig. 5C),presumably by blocking the hyperpolarization-activated in-ward current (Ih) (Angstadt and Calabrese 1989). At higherconcentrations, however, this effect was reversed. The meancycle period of the oscillators at each Cs� concentration (10�4

to 5 � 10�3 M) was normalized to the mean cycle period priorto any Cs� application (control). At the lowest (10�4 M) con-centration, the normalized cycle periods of the oscillators weresimilar to the control period. The period of the segmentaloscillators increased, however, as the Cs� concentrations(10�4 � 2 � 10�3 M) increased. The period peaked at 2 �10�3 M, and at higher (3 � 10�3 M to 5 � 10�3 M) concen-trations the period of the segmental oscillators decreased to-ward the control period (Fig. 5C). At concentrations above 2 �10�3 M, it is possible that nonspecific effects masked theeffects of Ih blockade. In subsequent split bath–sucrose knifeexperiments, we typically used 10�3 to 2 � 10�3 M Cs� toslow one segmental oscillator.

In split bath preparations, the cycle period of the intacttiming network (no sucrose knife) decreased when myomodu-lin (10�6 M) was applied to one of the baths, regardless ofwhether it was applied to the G3 (anterior bath) or G4 (poste-rior bath) segmental oscillator. The cycle period of the modu-lated oscillator sped up (data not shown) in a manner consistentwith the myomodulin concentration response curve (Fig. 5, Aand B). During modulation, the unmodulated oscillator as-sumed the same cycle period as the modulated oscillator, andthe two oscillators remained phase locked. Once myomodulinwas washed out, the timing network returned to a period closeto the original cycle period. In five of six preparations, themodulated oscillator led in phase, regardless of its originalphase relationship (leading or lagging). In one preparation,where the original G3 to G4 phase difference was moderatelylarge (�9%), the modulated G3 oscillator did not assume aphase lead, but instead lagged the unmodulated G4 oscillator inphase. However, the original phase difference was reducedfrom �9% to a modulated phase of �4.

Combining the split bath–sucrose knife technique with theability to alter cycle period with myomodulin and Cs� allowedus to assess how artificially large and/or reversed period dif-ferences between the uncoupled segmental oscillators shapedthe phase relationship and period in the recoupled timingnetwork. The four examples illustrated in Figs. 6–9 are typicalresults from various experiments in which we altered theperiod of one oscillator, when G3 and G4 were uncoupled, withthe application of molluscan myomodulin or Cs�. The basicparadigm was to uncouple the G3 and G4 segmental oscillators

and either speed the slower one or slow the faster one, asillustrated in Figs. 6–9. In some preparations, however, thefaster oscillator was sped or the slower one slowed. Regardlessof which oscillator was altered or whether it was sped orslowed, when recoupled the faster oscillator led in phase andthe period of the coupled system was close to that of the fasteruncoupled oscillator. In 23 of 24 such split bath–sucrose knifeexperiments, the faster oscillator led in phase when the oscil-lators were recoupled. In some cases, these manipulationscaused an oscillator that had led in the coupled state to lag inthe recoupled state, and an oscillator that had lagged in thecoupled state to lead in the recoupled state (Figs. 6, 7, and 9).

We correlated the recoupled G3 to G4 phase difference (�3 –�4) with the inherent period difference (T3 – T4) of the uncou-pled segmental oscillators (Fig. 10A) for both the split bath–sucrose knife experiments by themselves (open and gray sym-bols) and in conjunction with the sucrose knife experiments ofFig. 4C (control; indicated by black triangles in Fig. 10A).There was a strong and significant correlation in both cases(r � 0.8, P � 0.001 and r � 0.8, P � 0.001, respectively).Period differences, both naturally occurring and altered bymyomodulin or Cs�, between the segmental oscillators deter-mined the recoupled G3 to G4 phase difference.

We also lumped the data from these split bath–sucrose knifeand control sucrose knife (Fig. 4C, control) experiments toassess the relation of the recoupled period to the period of theuncoupled segmental oscillators. As in Fig. 4D, the recoupledperiod was plotted against the period of the faster (F) and ofthe slower (E) of the two uncoupled segmental oscillators foreach preparation (Fig. 10B). The plotted points for the fasteroscillators fell along and nearly evenly on either side of theunity line, while all but one of the points for the sloweroscillators fell below the unity line. Thus the inherently orartificially faster uncoupled segmental oscillator establishedthe cycle period of the recoupled network.

D I S C U S S I O N

The leech heartbeat pattern generating network is an attrac-tive model to study the intersegmental coordination of coupledsegmental oscillators because of its relatively simple designand well-described circuitry (Calabrese and Peterson 1983;Calabrese et al. 1995; Peterson 1983a,b; Schmidt and Cala-brese 1992). In the first paper of this series (Masino andCalabrese 2002), we showed that the coupled segmental oscil-lators in the heartbeat pattern generating network are organizeddifferently from coupled oscillators in other rhythmic motorprogram pattern generating networks such as the crayfishswimmeret system (Mulloney et al. 1993, 1998; Murchison etal. 1993; Stein 1971) or the swimming networks in leech(Friesen and Pearce 1993; Hocker et al. 2000) and lamprey(Cohen 1987a; Cohen et al. 1982; Grillner et al. 1995). In theseother systems, oscillators in adjacent segments produce a con-stant intersegmental phase lag that is independent of cycleperiod in both the intact animal and in the isolated nerve cord(either intact or reduced) (crayfish swimmeret, Braun andMulloney 1993; lamprey swimming, Cohen and Wallen 1980;Wallen and Williams 1984; leech swimming, Friesen andPearce 1993; Kristan et al. 1974;). In contrast, the phase lagbetween coupled oscillators in the isolated leech heartbeat



timing network, although quite regular within individual prep-arations, varies considerably among different preparations. Ineffect, there is not a constant phase relationship between thesegmental oscillators in the isolated timing network.

To address what mechanisms might establish the interseg-mental phase differences observed in the isolated timing net-work, we used the split bath–sucrose knife technique as a toolto reversibly uncouple the segmental oscillators, and measureand manipulate their inherent cycle periods. We then recoupledthe timing network and related period differences to the phaserelationships in the network. Differences in the inherent cycleperiods between oscillators correlated to the coupled and re-coupled phase relationships in the timing network (Fig. 4C).The same relationship between cycle period difference andphase was observed when larger than normal period differ-ences were induced or when period differences were reversedbetween the segmental oscillators by markedly altering the

period of one of the oscillators in the uncoupled state (Figs.6–9 and 10A). These experiments indicated that the uncoupledperiod difference (inherent and altered) between the segmentaloscillators was a good predictor of the recoupled phase rela-tionship (Fig. 10A). Regardless of which oscillator was fasteror which was slower, the magnitude of the phase was directlyrelated to the magnitude of the period difference (Fig. 10A),and the cycle period of the recoupled timing network wasestablished by the inherently faster segmental oscillator (Fig.10B). These data are consistent with the excitability-gradienthypothesis, which states that differences in the inherent cycleperiods between symmetrically coupled segmental oscillatorsgenerate the intersegmental phase differences in rhythmic ac-tivity among oscillators (Grillner et al. 1993; Matsushima andGrillner 1990, 1992). Moreover, modeling studies of the leechheartbeat timing network presented in the second paper of thisseries (Hill et al. 2002) indicate that inherent period differences

FIG. 6. Speeding of the G3 segmentaloscillator caused it to lead in phase. In Figs.6–9 the layout is as follows: the left columnshows extracellular recordings from a singlepreparation before (Coupled), during (Un-coupled; 1 oscillator sped/slowed with myo-modulin/Cs�), and after (Recoupled) thesegmental oscillators are uncoupled with thesucrose knife. Actograms in the middle col-umn demonstrate the timing relationships ofthe G3 and G4 oscillator interneurons. Ineach panel, the actogram’s reference cycle isdefined by the mean cycle period (TC 'coupled period) of the phase marker cellfrom the coupled state. This convention waschosen so that the actograms would indicatethe period differences, both natural and al-tered, from the original coupled state as wellas the phase relation slope in the recoupledstate. The cartoons in the right column illus-trate the state of coupling in the G3–G4chain and to which bath (anterior or poste-rior) myomodulin (MM) or Cs� was ap-plied. Coupled State: the cycle period of thetiming network (TC � 13.6 s) and the phaserelationship between the ipsilateral oscilla-tor interneurons were regular when the os-cillators were coupled. Notice the G4 oscil-lator interneuron had a slight phase lead.Uncoupled State: the segmental oscillatorswere uncoupled by applying the sucrose knifeto the G3/G4 interganglionic connective.Once uncoupled, applying 10�6 M myo-modulin (MM) to the anterior bath sped theG3 segmental oscillator. The electrical tracesand actogram show that the oscillators cycledindependently at different periods. Notice theindependent cycle period of the unmodu-lated G4 oscillator was similar to the origi-nal (coupled) cycle period. The modulatedcycle period of the G3 oscillator, however,was considerably faster than the cycle pe-riod of both the originally coupled timingnetwork and the uncoupled G4 segmentaloscillator; the G3 symbols drift sharply tothe left. Recoupled State: the segmental os-cillators were recoupled by removing thesucrose knife. Myomodulin was still presentin the anterior bath. The faster oscillator(G3) led in phase and established the cycleperiod of the recoupled network; both theG3 and G4 symbols drift sharply to the left.



between the segmental oscillators determine the phase relation-ships in both the symmetric and asymmetric versions of thenetwork. Asymmetries in synaptic connectivity incorporated inthese models are insufficient to generate phase differences,which are produced only when period differences between thesegmental oscillators are present.

Excitability-gradient hypothesis and phase generation

The excitability-gradient hypothesis has been tested andrejected in a variety of different pattern generating systems(crayfish, Braun and Mulloney 1995; Mulloney 1997; lampreyswimming, Cohen 1987a; Sigvardt 1993; Sigvardt and Wil-liams 1996; leech swimming, Pearce and Friesen 1985). Forexample, since the crayfish swimmeret rhythmic motor activityoriginates in the most posterior abdominal ganglion, the excit-ability-gradient hypothesis predicts that this oscillator shouldcycle faster than the oscillators in more anterior ganglia (Mul-

loney 1997). No evidence for an excitability-gradient wasfound along the nerve cord, however. When anterior and pos-terior segments of the cord were isolated by blocking informa-tion flow in the connective between the third and fourth seg-mental ganglia by applying TTX or sucrose solution to theconnective or by severing the connective, Mulloney and co-workers found that the anterior ganglia cycled faster than theposterior ganglia (Braun and Mulloney 1995; Mulloney 1997).Further, they showed that the rear-to-front phase relationshipof the system was not reversed when a gradient of excitationalong the nerve cord (anterior ganglia were more stronglyexcited with modulators than the posterior ganglia), whichshould oppose this phase relationship, was created. To assesswhether asymmetry in the network connectivity might generatephase differences, Skinner and Mulloney (1998) modeled sev-eral alternative swimmeret circuits, each with a different con-nectivity between the oscillator modules in the four adjacentsegmental ganglia, and examined their activity. One of these

FIG. 7. Speeding the G4 segmental oscil-lator caused it to lead in phase. Figure layoutas in Fig. 6. Coupled State: the cycle periodof the timing network (TC � 8.1 s) and thephase relationship between the ipsilateral os-cillator interneurons were regular when theoscillators were coupled. Notice that the G3oscillator led in phase. Uncoupled State: thesegmental oscillators were uncoupled by ap-plying the sucrose knife to the G3/G4 inter-ganglionic connective. Once uncoupled, ap-plication of 5 � 10�6 M myomodulin (MM)to the posterior bath sped the G4 oscillator.The electrical traces and actogram show thatthe segmental oscillators had different peri-ods and thus cycled independently. Noticethe independent cycle period of the unmodu-lated G3 oscillator was slower than the orig-inal (coupled) cycle period. The modulatedcycle period of the G4 oscillator, however,was considerably faster than the cycle periodof both the originally coupled timing net-work and the uncoupled G3 segmental os-cillator. Recoupled State: the segmental os-cillators were recoupled by removing thesucrose knife. Myomodulin was still presentin the posterior bath. The faster oscillator(G4) led in phase and established the cycleperiod of the recoupled network.



circuits effectively reproduced the invariant intersegmentalphase relationships and relative durations of activity that char-acterize the crayfish swimmeret network in the absence of anexcitability gradient. These modeling studies support the alter-native hypothesis that asymmetry in the pattern of interseg-mental synaptic connections, not an excitability gradient alongthe nerve cord, may account for the generation of the observedintersegmental phase relationships (Skinner and Mulloney1998).

The excitability-gradient hypothesis has also been tested inthe lamprey swim pattern generating network, which producesa front-to-rear progression of motor activity. Swimming inlamprey, which is characterized by a lateral undulation of thebody, is governed by a spinal locomotor pattern generatingnetwork that consists of �100 coupled oscillators distributedsegmentally along the length of the spinal cord (Grillner et al.1983). The phase lag between adjacent segments is typically�1% and is independent of cycle period, which ensures the

production of a constant wavelength of activity along the bodyaxis (Cohen 1987a; Grillner et al. 1993; Sigvardt 1993).

To ascertain whether different regions of the lamprey nervecord produced a gradient of swim cycle periods, Cohen(1987b) cut the spinal cord into numerous small pieces andmeasured their inherent cycle periods. No evidence was foundfor any consistent differences in cycle period among the seg-ments. However, indications of asymmetry in the intersegmen-tal coordinating system of the swim network were seen inforcing experiments where rhythmic bending applied to oneend of the nerve cord entrained the activity of the entirenetwork (Williams et al. 1990). Activity was entrained over abroader range when the cord was forced from the caudal endthan when it was forced from the rostral end, indicating thatascending coupling dominates in the network.

In another series of experiments, the split-bath techniquewas used to examine intersegmental phase lags produced whenone-half of the spinal cord was more excited than the other half

FIG. 8. Slowing the G3 segmental oscil-lator caused it to lag in phase. Figure layoutas in Fig. 6. Coupled State: the cycle periodof the timing network (TC � 9.6 s) and thephase relationship between the ipsilateral os-cillator interneurons were regular when theoscillators were coupled. Notice that therewas no phase difference between the G3 andG4 segmental oscillators. Uncoupled State:the segmental oscillators were uncoupled byapplying the sucrose knife to the G3/G4interganglionic connective. Once uncoupled,application of Cs� (10�3 M) to the anteriorbath slowed the G3 oscillator. The electricaltraces and actogram show that the segmentaloscillators cycled independently at differentcycle periods. Notice the independent cycleperiod of the unmodulated G4 oscillator wasslower than the original (coupled) cycle pe-riod. The altered cycle period of the G3oscillator, however, was considerably slowerthan the cycle period of both the originallycoupled timing network and the uncoupledG4 segmental oscillator. Recoupled State:the segmental oscillators were recoupled byremoval of the sucrose knife. Note that Cs�

was still present in the anterior bath. Theslower oscillator (G3) lagged in phase, whilethe faster oscillator (G4) established the cy-cle period of the recoupled timing network.



(Sigvardt and Williams 1996). Theoretical analyses of splitbath experiments indicate that, in an asymmetrically couplednetwork, the dominant coupling determines the intersegmentalphase lag in that half of the spinal cord from which thedominant coupling pathway originates and forces the phase lagin the other half to change (Kopell and Ermentrout 1990). Forexample, if ascending coupling dominates in the lamprey swimnetwork then bathing the two halves of the spinal cord indifferent concentrations of excitatory amino acid would causethe phase lag in the rostral half of the cord to change, while thephase lag in the caudal half would remain unchanged (changesin phase were determined relative to the control phase lags).Anterior and posterior sections of the spinal cord were differ-entially excited by applying different [high (0.8 mM) or low(0.2 mM)] concentrations of excitatory amino acid to each bathof the split bath preparation (Sigvardt and Williams 1996).Regardless of which concentration (high or low) of excitatoryamino acid was applied to each bath, phase in the posterior half

of the cord did not change, while phase in the anterior half ofthe cord did change (phase decreased when the concentrationwas low and increased when the concentration was high).These experiments corroborate the result of the forcing exper-iments and are consistent with the suggestion that an asymme-try in the network, rather than an excitability gradient, gener-ates the phase relationship between the segmental oscillators.

In a continuous network computer model of the lampreyswim system the cells of the rhythmic generating network aredispersed along the length of the nerve cord such that there areno segmental boundaries (Wadden et al. 1997). This network iscomposed of a column of 420 excitatory interneurons and 300inhibitory interneurons on each half of the nerve cord. Theexcitatory interneurons project equal distances in the rostraland caudal directions, while the inhibitory interneurons havelonger caudal than rostral projections, thereby introducingasymmetry to the network. This model produces stable, front-to-rear phase relationships (�0.5–2.5% per segment) that in-

FIG. 9. Slowing the G4 segmental os-cillator caused it to lag in phase. Figurelayout as in Fig. 6. Coupled State: the cycleperiod of the timing network (TC � 9.8 s)and the phase relationship between the ip-silateral oscillator interneurons were regu-lar when the oscillators were coupled. No-tice that the G4 segmental oscillator led inphase. Uncoupled State: the segmental os-cillators were uncoupled by applying thesucrose knife to the G3/G4 interganglionicconnective. Once uncoupled, applicationof Cs� (10�3 M) to the posterior bathslowed the G4 oscillator. The electricaltraces and actogram show that the segmen-tal oscillators cycled independently at dif-ferent cycle periods. Notice the indepen-dent cycle period of the unmodulated G3oscillator was faster than the original (cou-pled) cycle period. The altered cycle periodof the G4 oscillator, however, was consid-erably slower than the cycle period of boththe originally coupled timing network andthe uncoupled G3 segmental oscillator. Re-coupled State: the segmental oscillatorswere recoupled by removal of the sucroseknife. Cs� was still present in the posteriorbath. The slower oscillator (G4) lagged inphase, while the faster oscillator (G3) es-tablished the cycle period of the recouplednetwork.



crease slightly as the cycle period decreases. Nonetheless,these phase relationships are similar to those observed in bothisolated spinal cord and in intact preparations (Wallen andWilliams 1984). Thus Wadden et al. (1997) suggest that theasymmetric coupling present in the network generates theintersegmental phase lags. This model abandons the traditionalthinking about the lamprey swim system as a series of coupledsegmental oscillators and views the swim central pattern gen-

erator as a distributed network of concatenated interneuronsthat produce a temporal wave of activity and drives segmentalmotor outflow.

What of asymmetries in the leech heartbeat timing network?

Asymmetry in the leech heartbeat timing network is evidentin the connections of the timing network; the G3 oscillatorinterneurons inhibit the coordinating interneurons in both G3and G4, whereas the G4 oscillator interneurons only inhibit thecoordinating interneurons in G4 (Fig. 1A, Masino and Cala-brese 2002). Asymmetry is also indicated by the observationthat the phase of the coordinating interneurons is more tightlytied to the G3 than the G4 oscillator interneurons (Masino andCalabrese 2002, Fig. 11C). Presently, the functional signifi-cance of this apparent asymmetry in the timing network is notunderstood. To examine the mechanisms underlying interseg-mental phase generation in the heartbeat timing network, Hillet al. (2002) created two versions (symmetric or asymmetriccoupling) of a simplified computer model. In the symmetricmodel where G3 and G4 oscillator interneurons both inhibit theG1,2 coordinating interneurons, when two segmental oscilla-tors with different cycle periods are coupled, the faster oneleads in phase and determines the period of the coupled system(see Hill et al. 2002, Figs. 3 and 4). In contrast, in the asym-metric model where only G3 oscillator interneurons inhibit theG1,2 coordinating interneurons, when two segmental oscilla-tors with different cycle periods are coupled, the G3 oscillatorinterneurons, but not the G4 oscillator interneurons, can lead inphase and can control the period of the coupled system (seeHill et al. 2002, Fig. 10). Clearly this later condition does notpertain to the isolated heartbeat timing network.

In the split bath–sucrose knife experiments (G3–G4 prepa-rations) described here, where each segmental oscillator en-trains the other over a similar range of period and phaserelationships (Fig. 10A), the isolated timing network appears tofunction in a symmetric mode when tested under closed loopconditions. Under these experimental conditions (mutual en-tertainment between the 2 oscillators in the recoupled condi-tion) there is a closed feedback loop between the two oscilla-tors. A functional asymmetry may be expressed when thenetwork is tested under open loop conditions, for example, in“driving” experiments like those described in the precedingpaper (Hill et al. 2002, Figs. 8, 9, and 11), where one of thesegmental oscillators is forced to assume a cycle period estab-lished by injecting current pulses into one of the paired oscil-lator interneurons in that ganglion. In such driving (open loop)experiments, feedback from the nondriven (“follower”) oscil-lator onto the driven oscillator will be substantially reducedand thus the follower oscillator should lose its ability to influ-ence activity in the timing network. If so, we predict that theG3 oscillator will be more effective than the G4 oscillator atdriving and entraining the timing network to various cycleperiods because the G3 oscillator interneurons inhibit the co-ordinating interneurons at both their G3 and G4 spike initiationsites, while the G4 oscillator interneurons inhibit the coordi-nating interneurons only at their G4 spike initiation site. Futureexperimental and modeling studies will focus on exploringsuch potential asymmetries and their functional consequences.

We thank Dr. Andrew A. V. Hill for providing the Matlab scripts used fordata analysis and J. S. Fromowitz for performing the leech myomodulin

FIG. 10. Analysis of the split bath–sucrose knife experiments. A: plot of therecoupled G3 to G4 phase relationships vs. differences in cycle period betweenthe G3 and G4 segmental oscillators. Period differences, both naturally occur-ring (control; Œ) and altered by myomodulin (MM) or Cs� (open and graysymbols), between the segmental oscillators predict the recoupled G3 to G4phase difference. The faster (inherent or altered) uncoupled oscillator led inphase. B: plot of the recoupled timing network cycle periods vs. the inherentcycle periods of the uncoupled segmental oscillators. For each preparation thefaster and slower of the 2 segmental oscillators was determined, and they wereplotted separately. Notice that the cycle period of the faster oscillators (●) fellmore closely to the unity line than did the cycle period of the slower oscillators(E). The recoupled cycle period of the timing network was established by thecycle period of the faster (inherent or altered) uncoupled oscillator.



concentration response experiments. In addition, we thank both Dr. GennadyS. Cymbalyuk and A. E. Tobin for critical evaluations of the manuscript.

This work was supported by National Institute of Neurological Disordersand Stroke Grant NS-24072.

REFERENCES

ANGSTADT JD AND CALABRESE RL. A hyperpolarization-activated inward cur-rent in heart interneurons of the medicinal leech. J Neurosci 9: 2846–2857,1989.

BRAUN G AND MULLONEY B. Cholinergic modulation of the swimmeret motorsystem in crayfish. J Neurophysiol 70: 2391–2398, 1993.

BRAUN G AND MULLONEY B. Coordination in the crayfish swimmeret system:differential excitation causes changes in intersegmental phase. J Neuro-physiol 73: 880–885, 1995.

CALABRESE RL, NADIM F, AND OLSEN ØH. Heartbeat control in the medicinalleech: a model system for understanding the origin, coordination, andmodulation of rhythmic motor patterns. J Neurobiol 27: 390–402, 1995.

CALABRESE RL AND PETERSON E. Neural control of heartbeat in the leechHirudo medicinalis. Symp Soc Exp Biol 37: 195–221, 1983.

COHEN AH. Intersegmental coordinating system of the lamprey central patterngenerator for locomotion. J Comp Physiol 160: 181–193, 1987a.

COHEN AH. Effects of oscillator frequency on phase-locking in the lampreycentral pattern generator. J Neurosci Methods 21: 113–125, 1987b.

COHEN AH, HOLMES PJ, AND RAND RH. The nature of the coupling betweensegmental oscillators of the lamprey spinal generator for locomotion: amathematical model. J Math Biol 13: 345–369, 1982.

COHEN AH AND WALLEN P. The neuronal correlate of locomotion in fish‘fictive swimming’ induced in an in vitro preparation of the lamprey spinalcord. Exp Brain Res 41: 11–18, 1980.

CROPPER EC, LLOYD PE, REED W, TENENBAUM R, KUPFERMANN I, AND WEISS

KR. Multiple neuropeptides in cholinergic motor neurons of Aplysia: evi-dence for modulation intrinsic to the motor circuit. Proc Natl Acad Sci USA84: 3486–3490, 1987a.

CROPPER EC, TENENBAUM R, KOLKS MA, KUPFERMANN I, AND WEISS KR.Myomodulin: a bioactive neuropeptide present in an identified cholinergicbuccal motor neuron of Aplysia. Proc Natl Acad Sci USA 84: 5483–5486,1987b.

FRIESEN WO AND PEARCE RA. Mechanisms of intersegmental coordination inleech locomotion. Semin Neurosci 5: 41–47, 1993.

GRILLNER S, DELIAGINA T, EKEBERG O, EL MANIRA A, HILL RH, LANSNER A,ORLOVSKY GN, AND WALLEN P. Neural networks that co-ordinate locomo-tion and body orientation in lamprey. Trends Neurosci 18: 270–279, 1995.

GRILLNER S, MATSUSHIMA T, WADDEN T, TEGNER J, EL MANIRA A, AND

WALLEN P. The neurophysiological bases of undulatory locomotion invertebrates. Semin Neurosci 5: 17–27, 1993.

GRILLNER S, WALLEN P, BRODIN L, AND LANSNER A. Neuronal networkgenerating locomotor behavior in lamprey: circuitry, transmitters, mem-brane properties, and simulation. Annu Rev Neurosci 14: 169–199, 1991.

GRILLNER S, WALLEN P, MCCLELLAN A, SIGVARDT K, WILLIAMS T, AND

FELDMAN J. The neural generation of locomotion in the lamprey: an incom-plete account. In: Neural Origin of Rhythmic Movements, edited by RobertsA and Roberts BL. Cambridge, MA: Cambridge Univ. Press, 1983, p.285–303.

HILL AAV, MASINO MA, AND CALABRESE RL. Model of intersegmental coor-dination in the leech heartbeat neuronal network. J Neurophysiol 87: 1586–1602, 2002.

HOCKER CG, YU X, AND FRIESEN WO. Functionally heterogeneous segmentaloscillators generate swimming in the medical leech. J Comp Physiol [A]186: 871–883, 2000.

KOPELL N AND ERMENTROUT GB. Phase transitions and other phenomena inchains of coupled oscillators. SIAM J Appl Math 50: 1014–1052, 1990.

KOTALESKI JH, GRILLNER S, AND LANSNER A. Neural mechanisms potentiallycontributing to the intersegmental phase lag in lamprey. I. Segmental oscil-lations dependent on reciprocal inhibition. Biol Cybern 81: 317–330, 1999a.

KOTALESKI JH, LANSNER A, AND GRILLNER S. Neural mechanisms potentiallycontributing to the intersegmental phase lag in lamprey. II. Hemisegmentaloscillations produced by mutually coupled excitatory neurons. Biol Cybern81: 299–315, 1999b.

KRISTAN WB, STENT GS, AND ORT CA. Neuronal control of swimming in themedicinal leech. J Comp Physiol 94: 97–119, 1974.

MASINO MA AND CALABRESE RL. Phase relationships between segmentallyorganized oscillators in the leech heartbeat pattern generating network.J Neurophysiol 87: 1572–1585, 2002.

MATSUSHIMA T AND GRILLNER S. Intersegmental co-ordination of undulatorymovements—a “trailing oscillator” hypothesis. Neuroreport 1: 97–100,1990.

MATSUSHIMA T AND GRILLNER S. Neural mechanisms of intersegmental coor-dination in lamprey: local excitability changes modify the phase couplingalong the spinal cord. J Neurophysiol 67: 373–388, 1992.

MULLONEY B. A test of the excitability-gradient hypothesis in the swimmeretsystem of crayfish. J Neurosci 17: 1860–1868, 1997.

MULLONEY B, MURCHISON D, AND CHRACHRI A. Modular organization of thepattern-generating circuits in a segmental motor system: the swimmerets ofthe crayfish. Semin Neurosci 5: 49–57, 1993.

MULLONEY B, SKINNER FK, NAMBA H, AND HALL WM. Intersegmental coor-dination of swimmeret movements: mathematical models and neural cir-cuits. Ann NY Acad Sci 860: 266–280, 1998.

MURCHISON D, CHRACHRI A, AND MULLONEY B. A separate local pattern-generating circuit controls the movements of each swimmeret in crayfish.J Neurophysiol 70: 2620–2631, 1993.

NADIM F AND CALABRESE RL. A slow outward current activated by FMRF-amide in heart interneurons of the medicinal leech. J Neurosci 17: 4461–4472, 1997.

PEARCE RA AND FRIESEN WO. Intersegmental coordination of the leech swim-ming rhythm. I. Roles of cycle period gradient and coupling strength.J Neurophysiol 54: 1444–1459, 1985.

PETERSON EL. Generation and coordination of heartbeat timing oscillation inthe medicinal leech. I. Oscillation in isolated ganglia. J Neurophysiol 49:611–626, 1983a.

PETERSON EL. Generation and coordination of heartbeat timing oscillation inthe medicinal leech. II. Intersegmental coordination. J Neurophysiol 49:627–638, 1983b.

SANTAMA N, BRIERLEY M, BURKE JF, AND BENJAMIN PR. Neural networkcontrolling feeding in Lymnaea stagnalis: immunocytochemical localizationof myomodulin, small cardioactive peptide, buccalin, and FMRFamide-related peptides. J Comp Neurol 342: 352–365, 1994a.

SANTAMA N, WHEELER CH, BURKE JF, AND BENJAMIN PR. Neuropeptidesmyomodulin, small cardioactive peptide, and buccalin in the central nervoussystem of Lymnaea stagnalis: purification, immunoreactivity, and artifacts.J Comp Neurol 342: 335–351, 1994b.

SCHMIDT J AND CALABRESE RL. Evidence that acetylcholine is an inhibitorytransmitter of heart interneurons in the leech. J Exp Biol 171: 329–347,1992.

SIGVARDT KA. Intersegmental coordination in the lamprey central patterngenerator for locomotion. Semin Neurosci 5: 3–15, 1993.

SIGVARDT KA AND WILLIAMS TL. Effects of local oscillator frequency onintersegmental coordination in the lamprey locomotor CPG: theory andexperiment. J Neurophysiol 76: 4094–4103, 1996.

SKINNER FK, KOPELL N, AND MULLONEY B. How does the crayfish swimmeretsystem work? Insights from nearest-neighbor coupled oscillator models.J Comput Neurosci 4: 151–160, 1997.

SKINNER FK AND MULLONEY B. Intersegmental coordination of limb move-ments during locomotion: mathematical models predict circuits that driveswimmeret beating. J Neurosci 18: 3831–3842, 1998.

STEIN PS. Intersegmental coordination of swimmeret motoneuron activity incrayfish. J Neurophysiol 34: 310–318, 1971.

WADDEN T, HELLGREN J, LANSNER A, AND GRILLNER S. Intersegmental coor-dination in the lamprey: simulations using a network model without seg-mental boundaries. Biol Cybern 76: 1–9, 1997.

WALLEN P, EKEBERG O, LANSNER A, BRODIN L, TRAVEN H, AND GRILLNER S.A computer-based model for realistic simulations of neural networks. II. Thesegmental network generating locomotor rhythmicity in the lamprey. J Neu-rophysiol 68: 1939–1950, 1992.

WALLEN P AND WILLIAMS TL. Fictive locomotion in the lamprey spinal cord invitro compared with swimming in the intact and spinal animal. J Physiol(Lond) 347: 225–239, 1984.

WANG Y, PRICE DA, AND SAHLEY CL. Identification and characterization of amyomodulin-like peptide in the leech. Peptides 19: 487–493, 1998.

WILLIAMS TL, SIGVARDT KA, KOPELL N, ERMENTROUT GB, AND REMLER MP.Forcing of coupled nonlinear oscillators: studies of intersegmental coordi-nation in the lamprey locomotor central pattern generator. J Neurophysiol64: 862–871, 1990.



period differences between segmental oscillators produce

Documents