http://jbr.sagepub.com/Journal of Biological Rhythms

http://jbr.sagepub.com/content/21/1/45The online version of this article can be found at:

 DOI: 10.1177/0748730405283654

2006 21: 45J Biol RhythmsJuan J. Chiesa, Montserrat Anglès-Pujolràs, Antoni Díez-Noguera and Trinitat Cambras

auratus)History-Dependent Changes in Entrainment of the Activity Rhythm in the Syrian Hamster (Mesocricetus

  

Published by:

http://www.sagepublications.com

On behalf of: 

  Society for Research on Biological Rhythms

can be found at:Journal of Biological RhythmsAdditional services and information for    

  http://jbr.sagepub.com/cgi/alertsEmail Alerts:

 

http://jbr.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://jbr.sagepub.com/content/21/1/45.refs.htmlCitations:  

What is This? 

- Jan 31, 2006Version of Record >>

by guest on June 5, 2013jbr.sagepub.comDownloaded from

45

History-Dependent Changes in Entrainment of theActivity Rhythm in the Syrian Hamster

(Mesocricetus auratus)

Juan J. Chiesa,1 Montserrat Anglès-Pujolràs, Antoni Díez-Noguera, and Trinitat CambrasDepartament de Fisiologia, Facultat de Farmacia, Universitat de Barcelona, Barcelona, Spain

Abstract The authors have studied the activity rhythm of Syrian hamstersexposed to square LD cycles with a 22-h period (T22) with the aim of testing theeffects of the previous history on the rhythmic pattern. To do so, sequentialchanges of different lighting environments were established, followed by thesame LD condition. Also, the protocol included T22 cycles with varying light-ing contrasts to test the extent to which a computational model predicts exper-imental outcomes. At the beginning of the experiment, exposure to T22 with300 lux and dim red light occurring respectively at photophase and scotophase(LD300/dim red) mainly generated relative coordination. Subsequent transfer tocycles with ∼0.1-lux dim light during the scotophase (LD300/0.1) promotedentrainment to T22. However, a further reduction in light intensity to 10 luxduring the photophase (LD10/0.1) generated weak and unstable T22 rhythms.When, after that, animals were transferred again to the initial LD300/dim red cycles,the amplitude of the rhythm still remained very low, and the phases were veryunstable. Exposure to constant darkness partially restored the activity rhythm,and when, afterwards, the animals were submitted again to LD300/dim red cycles,a robust T22 rhythm appeared. The results demonstrate history-dependentchanges in the hamster circadian system because the locomotor activity patternunder the same T22 cycle can show relative coordination or unstable or robustentrainment depending on the prior lighting condition. This suggests that thecircadian system responds to environmental stimuli depending on its previoushistory. Moreover, computer simulations allow the authors to predict entrain-ment under LD300/0.1 cycles and indicate that most of the patterns observed inthe animals due to the light in the scotophase can be explained by differentdegrees of coupling among the oscillators of the circadian system.

Key words entrainment, hamster, T cycle, coupling, multioscillatory model

1. To whom all correspondence should be addressed: Juan J. Chiesa, Departament de Fisiologia, Facultat de Farmacia,Universitat de Barcelona, Av Joan XXIII s/n, 08028 Barcelona, Spain; e-mail: jchiesa@ub.edu, e-mail 2: cambras@ub.edu.

JOURNAL OF BIOLOGICAL RHYTHMS, Vol. 21 No. 1, February 2006 45-57DOI: 10.1177/0748730405283654© 2006 Sage Publications

The temporal adaptation of organisms to theirenvironment depends on the ability of the circadianpacemaker to entrain, using daily environmentalinformation provided primarily by the LD transitionsimposed by the solar day. Photic information is

processed by the pacemaker, which is located inthe SCN of the hypothalamus. Entrainment involvesadjustments in the phase and period of pacemakeractivity, and hence their rhythmic outputs synchronizewith the environmental cycle (Daan and Aschoff, 2001).

by guest on June 5, 2013jbr.sagepub.comDownloaded from

These fine-tuning entrainment mechanisms can bemodeled within the laboratory, using either continu-ous or discrete light action. The continuous model isbased on the assumption that the free-running period(τ) of the pacemaker continuously decreases orincreases by daily changes in light intensity (Aschoff,1981). In the discrete model, τ is considered a constantof the pacemaker, and its adjustment to the zeitgeberperiod (T) occurs solely via phase shifts elicited by dis-crete LD (or DL) transitions (Pittendrigh, 1981a). Thedirection and magnitude of the phase shift dependon the PRC, which has been regarded as a functionalcharacteristic of the pacemaker that rigidly defines theT cycle range into which it will be entrained, as T = τ ±maximum phase shift at PRC (Pittendrigh, 1981a).

The hamster circadian system appears to be morecomplex than a single pacemaker that invariablyresponds with the same PRC to environmental stim-uli. Of course, a flexible phase angle becomes impor-tant for the hamster, a photoperiodic species that isseasonally entrained to a varying day length. Thephotoperiod modulates entrainment, generatingbehavioral and physiological rhythms appropriate foreach phase of the year (Pittendrigh and Daan, 1976b;Elliott and Tamarkin, 1994; Schwartz et al., 2001).Even under constant darkness, the circadian pace-maker retains the influence of the previous entrainingphotoperiod, as reflected in the rhythm’s τ, wave-form, and photic PRC (Pittendrigh and Daan, 1976a;Pittendrigh et al., 1984), and this influence is alsoreflected in the SCN activity (Mrugala et al., 2000;Schaap et al., 2003).

Pittendrigh and Daan (1976b) considered that thehamster circadian clock is made up of a complex pace-maker consisting of 2 mutually coupled oscillators:morning (M), accelerated by light and synchronizedto dawn, and evening (E), decelerated by light andsynchronized to dusk. These oscillators may adopt arange of variable phase angles to regulate the overtrhythm, the α phase duration reflecting the phaserelationship between them (ΨEM). The period of thecoupled system depends on the interaction of these2 components, while the strength of this interactiondepends on ΨEM. Distinct phase relationships for bothE and M oscillators with respect to the zeitgeber havebeen considered to explain the effects of light inten-sity changes on motor activity (Gorman et al., 1997)and pineal melatonin rhythms (Elliott and Tamarkin,1994). However, the historical dependence of the pace-maker state, reflected in the form of aftereffects on bothτ and α (Aschoff, 1981; Pittendrigh and Daan, 1976a),

suggests that the actual state is not solely dependenton a given lighting condition. The ΨEM is also establishedhistorically by the previous environmental conditions,namely, hysteresis phenomena, initially described byHoffmann (1971).

It has been demonstrated that dual circadian oscil-lation is present in the SCN tissue in vitro, whichcorresponds to M and E oscillators underlying photo-periodic phenomena (Jagota et al., 2000). However, itis not known whether they are a property of individualSCN cells or instead emerge from an intercellular net-work interaction. The principal functional bases ofmammalian pacemakers were also modeled from amultioscillatory standpoint (Winfree, 1967; Díez-Noguera, 1994; Achermann and Kunz, 1999). Thesemodels take the central assumption that many pace-maker characteristics may arise at the level of intercel-lular interactions. It is well known that single circadianoscillators are arranged in individual cultured neu-ronal clocks (Welsh et al., 1995) with spontaneouslydispersed periods (Herzog et al., 1998; Honma et al.,2004). Neuronal clocks synchronize among themwithin the integrated SCN tissue by their networkinteractions, attaining stable phase relationships andgenerating the circadian output (Yamaguchi et al.,2003). Indeed, the circadian period of both pacemakerand behavioral rhythms may be determined by themean period arising from such interactions betweenclock cells (Liu et al., 1997; Low-Zeddies and Takahashi,2001). Neural communication via both electrical (Longet al., 2005), Na+-dependent action potential (Honmaet al., 2000; Yamaguchi et al., 2003) and chemicalsynapses (Aton et al., 2005) participates in both syn-chrony and rhythmicity maintenance among SCNcells. Remarkably, Long et al. (2005) demonstratedthat impaired coupling at the SCN produced a lesscoherent behavioral output with reduced precision.

Rats show arrhythmic or ultradian behavioral pat-terns in prolonged exposure to LL (Eastman andRechtschaffen, 1983; Depres-Brummer et al., 1995).Furthermore, it was suggested that light could promptsuch behavioral reorganization by inhibiting internalcoupling at the pacemaker (Vilaplana et al., 1997).Importantly, recent data have shown that LL gener-ated desynchronization of SCN cultured neurons dis-sected from behaviorally arrhythmic mice (Ohta et al.,2005). Also, some evidence indicated that entrainmentof the SCN pacemaker to LD cycles (Quintero et al.,2003) or to light pulses (Kuhlman et al., 2003) involvedchanges affecting SCN internal coupling and that lightintensity may participate in such changes.

46 JOURNAL OF BIOLOGICAL RHYTHMS / February 2006

by guest on June 5, 2013jbr.sagepub.comDownloaded from

In our own laboratory, we employed a computationalmodel to simulate the functioning of the circadiansystem (Díez-Noguera, 1994). Built on possible physi-ological correlates, it can be used as an inductive tool,potentially identifying unknown elements of the sys-tem (Helfrich-Förster and Díez-Noguera, 1993). Themodel’s 2 main assumptions are as follows: 1) that amultioscillatory structure is at work and 2) that itsfunctioning, and thus its output pattern, depends onthe coupling strength between oscillators. The purposeof the present experiment was to determine whetherchanges in entrainment of activity rhythm elicited bymodifying the illuminance contrast of the LD cyclecan be explained by coupling changes in our modelsimulations. Moreover, we tested the effect of priorlighting conditions on entrainment to the same LDcycle to study a possible plasticity in the response ofthe system. A comparison of the results obtained withbehavioral and simulation experiments would allowfurther insight into the mechanisms underlying thecircadian system.

MATERIALS AND METHODS

Animals and Housing

Twenty male golden hamsters (Mesocricetus auratus)arrived at the laboratory directly from the provider(Harlan France SARL). These 2-month-old animalswere housed individually in transparent metacrylatecages measuring 22 × 22 × 15 cm (Panlab, Spain) andcovered with a stainless steel grid, with wood shavingbedding. Cages were maintained in a soundproof,temperature-controlled chamber (18-22 ºC, humidity40%-80%) with time-adjustable illumination cycles.Throughout the entire experiment, the animalsreceived food (Harlan Teklad 2040 global diets®) andtap water ad libitum. Cage cleaning, along with foodand water replenishment, was done once a week atirregular time intervals, taking care to avoid per-forming maintenance tasks on the days close to stagetransitions.

Procedures

Once they had arrived at our laboratory, the animalswere submitted to light-dark cycles with a 22-h period(T22). T cycles initially comprised an 11-h photophaseof 300 lux and an 11-h scotophase of 0.1-lux dim redlight (LD300/dim red, LD1, stage duration: 19 cycles). From

the initial group of 20 animals, 8 clearly exhibiting arelative coordination pattern throughout this stage wereselected for the present experiment (see the processof selection in the Results section). Subsequently, T22cycles that differed in their illuminance contrast wereused to verify 2 predictions generated by model simu-lations (see Simulations below). Hence, a 2nd stage ofT22 cycles with a 300-lux photophase and a 0.1-luxdim illuminance at scotophase was imposed (LD300/0.1,stage duration: 77 cycles). As rhythms became unsta-ble, this stage was prolonged for more than a monthand then for an additional month to achieve stableentrainment. Then, a next stage of T22 cycles reducingilluminance to 10 lux at the photophase and maintainingthe same previous 0.1-lux dim light at scotophase wasimposed (LD10/0.1, stage duration: 27 cycles). LD300/dim red

condition was reinstated to test the different entrain-ability of rhythm (LD2, stage duration: 31 cycles).Afterwards, the animals were transferred to constantdarkness, maintaining lights-off from the last scotophase(DD, stage duration: 25 cycles). Finally, LD300/dim red cycleswere reinstated to thereby compare the resultingrhythm with previous LD1 and LD2 (LD3, stage duration:55 cycles). A scheme in Figure 1 outlines the entireexperimental paradigm, and light sources had thefollowing characteristics:

• For LD300/dim red: two 36-W Mazdafluor fluorescent tubesyielded ∼300 lux of indirect lighting at photophase. Thespectral composition featured 2 major irradiance peaks atthe 440- and 540-nm wavelength, as well as a bandwidthmeasuring between 550 and 600 nm. Lighting at scotophasewas delivered by indirect dim red light of ∼0.1 lux, with amajor band from 650 nm extending to the infrared region.

• For LD300/0.1: lighting at photophase was the same asfor LD300/dim red, and scotophase consisted of illuminancelevel supplied by light-emitting diodes (LEDs) produc-ing ∼0.1 lux of indirect dim light. LEDs provided a majorirradiance bandwidth of 450 to 470 nm, as well as a sec-ondary one measuring 530 to 620 nm.

• For LD10/0.1: lighting at photophase was delivered by two18-W Philips fluorescent tubes laterally covered byopaque wooden badges, producing ∼10 lux of reflectedlight with a spectra similar to 300-lux lighting. Scotophaseused the same LEDs as LD300/0.1.

Illuminance and irradiance were measured at cagelevel with a Gossen Metrawatt Camille Bauer, Mavolux5032B digital luxmeter (GMC-Instruments Schweiz SA,Zurich, Switzerland) and with a Handheld FieldSpec®

spectroradiometer (Analytical Spectral Devices, Boulder,CO), respectively.

Chiesa et al. / HISTORY-DEPENDENT CHANGES 47

by guest on June 5, 2013jbr.sagepub.comDownloaded from

Data Collection and Time-Series Analysis

Animal motor activity was measured by means ofan activity meter consisting of 2 perpendicularinfrared light beams crossing the cage at a height of7 cm above the floor, coupled to a suitable electronicdevice. Each beam interruption due to animal move-ment represented an activity count that was recordedand compiled at 15-min intervals. Individual datasamples were acquired in parallel channels andstored in a computer for further analysis. Lightingcycles were recorded simultaneously with activity ina separate channel connected to a photocell pulsegenerator.

Time-series analysis was carried out separately fordata series corresponding to each stage. Time serieswere smoothed by a 30-min step moving average toenhance circadian tendency viewing, and data seg-ments of 19 cycles (LD1 stage length) were averaged toobtain the mean daily activity profile at each stageat T22. LD300/0.1 profiles comprised 3 consecutive seg-ments of 19 cycles, the first beginning at the 20th cycleof the stage when the rhythm phase appeared gener-ally stabilized on the actograms. For the remainingstages, data from the first 19 cycles of each one wereused. χ2 periodograms (Sokolove and Bushell, 1978)were calculated by taking raw data corresponding toeach of the previously defined 19-cycle segments,together with all of the data at LD1 and the first 19cycles at DD. The highest significant peak (p < 0.05),scanned within a range from 20 to 26 h, was consideredfor period estimation, hence obtaining the percentageof variance explained by the dominant rhythm. Whena nonsignificant peak occurred on the periodogram,we still used the value of percentage of varianceexplained by the expected periodicity at 22 h, as it waspreviously described (Cambras et al., 2004).

The rhythm phase control exerted by the previousLD cycle was examined at the DD stage. First, an eye-fitted line was drawn on the actograms by taking thetendency of the rhythm’s onset during the first 10 daysin DD. This line was projected backwards to the lastday of the LD2 stage. Then the projected activity onsetsof the individuals were studied with the Rayleigh z test(Batschelet, 1981), which calculates the temporal distri-bution of individual phases across the T cycle. Hence,a sidedness distribution suggests that, due to entrain-ment, the rhythm free-runs from the phase held by thezeitgeber when it was present.

Graphs and calculations were made using theintegrated package for analysis in chronobiology,“El Temps” (Antoni Díez-Noguera, University of

Barcelona, Spain, 1998-2005, http://www.ub.es/dpfisiv/soft/ElTemps/).

Computer Simulations

Model description. The model we used for simulationsassumes that, functionally, the pacemaker consists ofa population of coupled limit-cycle oscillators witha distribution of angular velocities (Díez-Noguera,1994). Each individual oscillator is mathematicallydescribed by 2 differential equations for 2 state vari-ables. In the model, the state variables of a singleoscillator are continuously modified by the effect ofthe state variables of the other oscillators. These inter-actions act in such a way that oscillators tend toapproach the average value of the whole system andlinearly depend on a global coupling factor. In such amodel, coupling changes generate phase shifts of therhythmic output of the system.

Simulations were done just by varying couplingstrength, which was cyclically imposed as a squarewave, by alternating a part of high coupling values(Hc) with another of lower ones (Lc), emulating an LDcycle with abrupt transitions. To verify model predic-tions in the experiment, the scotophase of the LD cyclewas assumed to be analogous to Hc in model simula-tions, whereas photophase corresponded to Lc.

Although different pairs of equations that fulfill theformal requirements needed for the model may beused, for this study, we used Selkov (1968) equations.This oscillator had produced good simulations inprevious experiments (Díez-Noguera, 1994; Díez-Noguera et al., 2003). Importantly, the Selkov modelprovides a PRC to coupling interactions that has ashape similar to the photic PRC of rodents and is ade-quate for present simulation aims. In previous studiesconducted in our laboratory (Díez-Noguera et al.,2003), we demonstrated that the PRC amplitude is notgreatly affected by the number of oscillators (between2 and 64). Although a more realistic model shouldinclude the ∼10.000 neurons that make up the SCNtissue, a small percentage of these cells may participatethrough coupling interactions in the rhythmic outputof the pacemaker (Long et al., 2005; Aton et al., 2005).Also, as in other assayed models of the circadian system(Oda et al., 2000), the single mathematical oscillatormay represent a subset of pacemaker cells. Thus, forthis study, we used 32 oscillators, enough to ensure ahigh number of stable and short-termed simulations.

Simulations. 1) First of all, we carried out a set ofsimulations conducted to gauge a range of couplingstrength values for both Hc and Lc parts. To do so, the

48 JOURNAL OF BIOLOGICAL RHYTHMS / February 2006

by guest on June 5, 2013jbr.sagepub.comDownloaded from

coupling cycle was composed of 88 calculation steps(44 Hc/44 Lc) in such a way that it paralleled a 22-hcycle, with intervals of 15 min defined for motor activ-ity data collection (15 min × 88 = 1320 min = 22 h). Onehundred cycles were defined for all simulationlengths. The mean velocity of the system was fine-tuned to obtain a stable free-running period near 24 h,when Hc = Lc. Subsequently, the Hc/Lc ratio was var-ied in a systematic study encompassing a wide rangeof values. Thus, a parametric region where outputpatterns changed from stable entrainment to relativecoordination was selected (Hc range = 0.5-0.6; Lcrange = 0.1-0.2). Then, both Hc and Lc were varied ona logarithmic scale to improve graphic representation.

2) The 2nd set of simulations was carried out tomodel the transitions between the first 2 stages of theparadigm used for animal experiments. LD1 andLD300/0.1 were simulated by dividing 100 coupling cyclesinto 2 parts: in the 1st part, Hc and Lc were definedaccording to 1), in such a way that relative coordinationwould appear; in the 2nd part, the value of Lc wasfixed but that of Hc was reduced in a linear manner tosimulate the light increase during the scotophase.

RESULTS

Animal Experiments

In the first T22 stage (LD1) when LD300/dim red cycleswere imposed, the initial group of 20 hamsters showeddifferent activity patterns. Since one of the purposes ofthe experiment was to verify model simulations (seeSimulations below), by exposing the animals to T22cycles that differed in their illuminance contrast, weselected a group with a homogeneous activity pattern.Because relative coordination was the most commonpattern of overt rhythm at LD1 stage (n = 11), weselected 8 of them for the experiment (8 hamsters wasthe optimal number allowed by animal facilities atthat time). The rest of the animals were used for otherexperiments. Relative coordination was establishedby the observation of the actograms on the basis of thefollowing criteria (Aschoff, 1981): 1) a free-runningrhythm whose phase changes periodically with thezeitgeber, 2) the rhythm has a different period depend-ing on whether it coincides with light or with darkness,and 3) there is an α phase modulation, with compres-sion during light and decompression during darkness.

An actogram in a 22-h double-plotted format withdata on a representative hamster throughout the entireexperimental period is shown in Figure 1. The 22-hactivity profile obtained by averaging data of the 8

individuals is also displayed. During LD1, the relativecoordination pattern of the selected animals featureda period value between the endogenous τ and exter-nal T (mean ± SD: 23.5 ± 0.06 h). The subsequent expo-sure of hamsters to cycles with 0.1-lux dim light at thescotophase during the LD300/0.1 stage gradually pro-moted the period adjustment of the rhythm to T22.However, different rhythmic patterns were observedduring the initial days of this stage (Fig. 2). Two ani-mals still tended to free-run, and this continued forseveral days (panels B and C). In 2 other hamsters, agreat phase jump was observed during the 1st dayunder LD300/0.1, which was indistinguishable in termsof advance or delay (panels D and E). In the remaining4 animals, the α phase was rapidly confined to the sco-tophase (panels A, F, G, and H) but presented someonset fluctuations. Generally, after major phase stabi-lization occurred, activity rhythm synchronized toT22, acquiring a stable phase relationship during thefollowing days (8 of 8 animals with significant 22-hrhythm in both the 1st and middle 19-cycle segments).Afterwards, the α duration gradually increased, withthe rhythm displaying a dampened amplitude towardthe last part of the stage (5 of 8 animals with significant22-h rhythm in the last 19 cycles). These changes can beobserved from the mean daily activity profile obtainedafter major period stabilization at T22 (Fig. 1). Duringthe 1st and intermediate days at LD300/0.1, it can be seenthat a clear bimodal pattern remained confined to thescotophase. After that, the rhythm amplitude appeareddampened, and the α phase tended to be enlargedduring the last days, with certain loss of bimodality.

When the animals were transferred to the LD10/0.1

stage, by changing photophase illuminance from 300 luxto 10 lux, an unstable pattern was still observed in mosthamsters (4 of 8 animals with significant rhythm). Also,the mean pattern of the 22-h profile revealed a very lowamplitude of rhythm with an almost indiscernibleα phase (Fig. 1). During further reexposure to LD300/dim red

cycles for a month (LD2 stage), the activity rhythmdid not recover, and a flattened low-amplitude rhythmremained (5 of 8 animals with significant rhythm).Despite this unstable pattern, the control of therhythm phase by the zeitgeber was verified by signifi-cant clustering of the activity onsets (Rayleigh z test:r = 0.98, p < 0.01).

During continuous exposure for nearly a monthto constant darkness (DD stage), the χ2 periodogramdetected 5 of 8 animals with significant free-runningrhythms (mean ± SD, τ = 23.4 ± 0.1). After DD, rein-stating of LD cycles (LD3 stage) was accompaniedby a stable circadian rhythm synchronized to T22

Chiesa et al. / HISTORY-DEPENDENT CHANGES 49

by guest on June 5, 2013jbr.sagepub.comDownloaded from

50 JOURNAL OF BIOLOGICAL RHYTHMS / February 2006

Figure 1. (A) Double-plotted actogram of a 22-h module for a representative hamster’s motor activity. Single-plotted black rectanglerepresents scotophase timing of the T22 cycle. Subdivisions of the bar on the right indicate the corresponding stages with the variouslighting regimes used (see text). (B) Panels on the right are the mean rhythm profiles double-plotted at T22, obtained from the 8 indi-viduals at the corresponding stages indicated on the actogram. Data at the LD300/0.1 stage were subdivided into 3 consecutive segments,the first beginning at the 40th cycle, when stable T22 rhythm appeared. Shaded area indicates the scotophase of the LD cycle. The y-axisrepresents activity counts, and the horizontal line depicts the daily mean.

(8 of 8 animals with significant rhythms). A clear 22-hrhythm can be observed in the daily activity profile,including a major onset peak (Fig. 1).

In addition, the percentage of variance explained bythe rhythm quantifies the entrainability of the activitypattern to the different T22 and DD stages (Fig. 3).Starting from stable circadian rhythmicity with highvalues at baseline at LD1, a decrease elicited by LD300/0.1

cycles was sustained throughout this stage, maintaining

minimal values at LD10/0.1, LD2, and DD. An increase ofthe percentage of variance was registered when rein-stating the LD300/dim red condition during the last LD3

stage, after exposure to DD.Thus, 3 different patterns appeared under the same

LD illuminance (LD1, LD2, and LD3), although withdifferent prior conditions: 1) direct transfer of naiveanimals to T22 (LD1 stage) produced a large homoge-neous group showing relative coordination, 2) a

by guest on June 5, 2013jbr.sagepub.comDownloaded from

rhythm with a dampened amplitude and unstabledaily pattern revealed by a low percentage of variancewas obtained during LD2, and 3) a stable rhythm witha great percentage of variance synchronized better toT22 during LD3 following previous exposure to DD.

Model Simulations

Model simulations were conducted to predict theresponse of the hamster circadian system, observedunder different illuminance contrasts of T22 cycles.The functioning of the system for different degrees ofcoupling strength for both Hc and Lc parts is shown inFigure 4A. One can see that the system output changesfrom free running, relative coordination, or entrain-ment depending on the ratio between Hc and Lc.Moreover, a reduction in the coupling strength in bothHc and Lc parts drives the system period towardentrainment. Thus, from these simulations, 2 predic-tions can be obtained: 1) a reduction in the coupling

strength during the Hc part (analogous to the LD1 toLD300/0.1 change in the animal experiment) tendstoward entrainment, and 2) an increment in the cou-pling strength during the Lc part (analogous to theLD300/0.1 to LD10/0.1 change in the animal experiment)tends toward relative coordination.

Figure 4B displays the simulations that successfullypredict the experimental results obtained after transferfrom LD1 to LD300/0.1. An initial part with a fixed Hc/Lcratio (selected from Fig. 4A) corresponds to LD1 and isfollowed by another part with varying values for Hc,which corresponds to LD300/0.1. A linear augmentationof the coupling strength in the Hc part is shown ineach of the 10 successive actograms displayed in the2 rows. One can see that depending on the value of Hcin the 2nd part of the simulation, intermediate tran-sitions between relative coordination and entrainmentare obtained. Some of these patterns resemble those ofthe animals during the initial part of the LD300/0.1 stage(see Fig. 2).

Chiesa et al. / HISTORY-DEPENDENT CHANGES 51

Figure 2. Double-plotted actograms in a 22-h module for each individual during LD1 and the following 20 cycles in LD300/0.1. The blackrectangle within each panel indicates the LD300/0.1 scotophase. Differences between individual responses during the initial cycles inLD300/0.1 are shown (see text).

by guest on June 5, 2013jbr.sagepub.comDownloaded from

The 2nd model prediction failed because instead ofdetecting relative coordination, as the model predictsfor an increment of Lc values, the activity pattern ofanimals in the transfer from LD300/0.1 to LD10/0.1 tendedto a decrease in the rhythm manifestation.

DISCUSSION

This experiment provides evidence about 2 points:1) that changes in the motor activity rhythm due tolight intensity may be parallel to those produced inthe output of a multioscillatory system, by changes ofcoupling among the oscillators, and, more important,2) that the overt rhythm can be very changeabledepending on the previous lighting conditions.

The starting point of our work was the selection ofhamsters for this experiment. The simulations withthe computer system revealed transitions betweenstable entrainment, relative coordination, and freerunning, induced by very small changes in internalcoupling. These changes only affect the ratio betweenHc and Lc, which can be defined as the degree of cou-pling of the whole circadian system. When hamsterswere exposed to initial T22 cycles, several patternswere also observed, although the most common wasrelative coordination. According to the model, these

patterns could be interpreted to mean that eachhamster may also have a different coupling degree inits circadian system (compare Fig. 2A vs. 4B). At pre-sent, it is not known what coupling is, but it is proba-bly a multifactorial outcome (see Miche and Colwell,2001, for review). In any case, any of these factors canaccept different degrees, which could explain the indi-vidual differences under a single lighting condition.This suggests the importance of studying couplingfrom both experimental and simulation points of view.With this idea, we selected 8 hamsters with a similaractivity pattern for the experiments, assuming thatthey corresponded to the same internal coupling degreeas the model, and thus a more homogeneous responsewas to be expected when light conditions were changed.Actually, the general response to light was similar forthe selected hamsters under all the lighting conditions.

The model simulations indicate that, from a specificcoupling degree of the system that produces relativecoordination in its output, a decrease of the couplingdegree in the scotophase induces entrainment. This iscorroborated in vivo because hamsters submitted toT22 cycles (LD1 stage) changed from an initial relativecoordination to entrainment when dim nocturnal illu-mination was applied (LD300/0.1 stage). We cannot defin-itively confirm that entrainment occurs during LD300/0.1

because both the adoption of a unique and stable phase

52 JOURNAL OF BIOLOGICAL RHYTHMS / February 2006

1 2 3 0

10

20

30

% o

f va

rian

ce

LD1 LD300 / 0.1 LD10 / 0.1 LD2 DD LD3

Figure 3. Mean and standard deviation of the percentage of variance calculated using the χχ2 periodogram on equal 19-cycle raw datasegments corresponding to the stages of the experiment. Values were obtained from relative coordination patterns at LD1, from free-running periodicities at DD, and from the 22-h rhythm for the rest of the stages. Data corresponding to the LD300/0.1 stage were subdi-vided into 3 sequential segments. The dotted line indicates the percentage of variance value for the defined significance level, p == 0.05.

by guest on June 5, 2013jbr.sagepub.comDownloaded from

relationship, as well as the free running of rhythmunder constant conditions from the phase predictedby the zeitgeber, must be demonstrated (Daan andAschoff, 2001). We do believe, however, that activechanges in pacemaker activity certainly did occur atthis stage. Activity onset gradually advanced, chieflyduring the initial part of the LD300/0.1 stage with a conse-quent α expansion, which implies active modifications

in the circadian system (i.e., Figs. 1 and 2). Evidencefrom various experiments employing dim nocturnallight that induced splitting (Gorman et al., 2003), shortphotoperiod reentrainment (Gorman and Elliott, 2004),and entrainment to lengthening T cycles (Gorman et al.,2005) suggests that scotopic light may alter the phaserelationship between oscillators. Because we do nothave more conclusive data, we are still puzzled as to

Chiesa et al. / HISTORY-DEPENDENT CHANGES 53

Figure 4. (A) Matrix of double-plotted actograms for the model’s periodic outputs from 100-cycle simulation for a different high-coupling/low-coupling (Hc/Lc) ratio. The black rectangle indicates the selected transition from relative coordination to entrainment,for simulations detailed in B. (B) Model prediction for the experimental outcome. Selected values from the Hc/Lc ratio in A were fixedduring the first 50 cycles in the simulation (LD1). In the second 50 cycles (LD300/0.1), Hc was variable by linear increments of 0.05, whichcorrespond to double-plotted actograms numbered from 1 to 10 (Hc range == 0.25-0.7). Hc in the model simulation was assumed tobe analogous to the scotophase in the animal experiment, whereas Lc represented the photophase of an LD cycle.

by guest on June 5, 2013jbr.sagepub.comDownloaded from

why entrainment occurred with such an illuminancechange. A possible explanation for the observed changefrom relative coordination to entrainment could also bearrived at by considering the efficiency of light relatedto its action spectra. Although a 2-h pulse of red lightcan modify the phase of the free-running rhythm in rats(McCormack and Sontag, 1980), scotopic light used inthe present work featured an adequate spectral com-position for irradiance detection in the circadiansystem (Berson et al., 2002; Foster, 2005). Consideringthe Aschoff rules (Aschoff, 1981), increasing lightintensity produces τ lengthening; thus, entrainment toa 22-h T cycle should be prevented. However, this isnot the case in our hamsters but fits with other exper-iments involving rats in which we found that entrain-ment to T22 cycles improved when the photoperiodwas augmented (Cambras et al., 2004).

Model simulations predicted a change from relativecoordination to entrainment if light intensity increasedat scotophase. This was based on the central assump-tion that the coupling strength would diminish propor-tionally with light intensity. Light has been proposed toact on the circadian system by inhibiting the pacemakercoupling in several rodent species (Vilaplana et al.,1997; Honma and Honma, 1999; Ohta et al., 2005). It isknown that prolonged constant light (LL) disruptsbehavioral rhythms by disrupting the cellular synchro-nization within the SCN pacemaker, without stoppingthe core molecular clocks of individual cells (Ohta et al.,2005). Arrhythmicity induced by LL in hamsters wasalso explained as the uncoupling of multiple oscillators(Pittendrigh, 1981b). However, the LL effects appearedto be more complex than arrhythmicity alone. Indeed,some animals show splitting of activity rhythms into 2components that remained coupled with a stable phaserelationship of 180º or an unstable one (Pittendrigh andDaan, 1976b; Pittendrigh, 1981b). Moreover, based onthe E- and M-composed pacemaker, simulations with adual oscillator model explain experimental data in thehamster with the assumption that prolonged exposureto constant darkness gradually reduces the couplingstrength (Oda et al., 2000). Models for circadian pace-makers have to contribute to understanding their mech-anism and/or function (i.e., Oda et al., 2000), as well aspredict the outcome of new experiments. In this sense,there are no “correct” models to represent the reality ofthe circadian pacemaker, but there are some that can beuseful for achieving these aims (Beersma, 2005). Ourmodel seems useful for predicting changes in the activ-ity pattern by assuming that coupling strength dimin-ishes with light intensity, when the daily action of lightmay affect the coupling mechanisms.

The model also predicted that if there wereentrainment, then incremental changes in the cou-pling strength during the Lc part would drive the sys-tem from entrainment to relative coordination. Toverify this, our experiment involved a further reduc-tion in illuminance intensity from 300 to 10 lux duringthe photophase. Contrary to this prediction, not onlywas relative coordination not recorded as expectedfrom simulations, but a very low-amplitude circadianrhythm was generated. Although we cannot explainthese changes with the data available in the presentwork, the possible saturation of the entrainment path-ways (Nelson and Takahashi, 1999) could be consid-ered in terms of the incremental instability of rhythmthroughout LD300/0.1. Thus, despite the fact that 10-luxilluminance is above the response threshold for SCNelectrical activity (Meijer et al., 1986), a virtual “con-stant light” setting during the LD10/0.1 stage rapidlydrove the system toward instability. LD10/0.1 conditionsmay also provide weak photic input to visual detectionsystems, which may merge with the circadian systemat the receptor level (Berson et al., 2002; Hattar et al.,2003), disrupting the output pathways due to possibleambiguities in the light information between the pho-tophase and scotophase. In addition, overt rhythmmanifestation is due to the functioning of the circa-dian system together with other structures mediatingpossible masking processes (Mrosovsky, 1999) thatcould be modulated by lower light contrast, henceproducing a lower amplitude rhythm. However, wemust take into account that under the previous stage,LD300/0.1, the system was continually changing, asobserved in the evolution of the rhythm profile in thisstage. Thus, one can suppose that under LD10/0.1, thecontrast between the 2 light phases is not strongenough to stabilize the rhythm, and hence it main-tains with the same dynamics as in the previous stage.

While our model successfully simulates functionalfeatures of the circadian pacemaker (i.e., entrainment,PRC, dependence of period on light intensity, split-ting) by controlling the coupling strength betweenoscillatory elements (Helfrich-Förster and Díez-Noguera, 1993; Díez-Noguera, 1994; Díez-Nogueraet al., 2003), it is unable to generate dynamic changes.Because the functional state of a given oscillator isdirectly modified by those of the rest, and no mnesicimprints are acquired, none of the effects from the pre-vious history is exerted. Therefore, the model shouldbe more effective in predicting rhythm changes overthe short term, as occurs from LD1 to LD300/0.1. Thissuggests that models cannot be thought of as systemsthat produce a specific response (circadian pattern) for

54 JOURNAL OF BIOLOGICAL RHYTHMS / February 2006

by guest on June 5, 2013jbr.sagepub.comDownloaded from

a specific stimulus (a particular LD cycle) because theinternal structure of the system can change through-out the time. This was even more evident when weobserved that the rhythm was strongly dependent onthe activity rhythm pattern under previous conditions.Thus, a “memory component” of the system can behypothesized, in which the pacemaker’s response is acontextual interaction between its previous historyand present conditions.

The system’s progress throughout the differentstages of the experiment was primarily influenced byprevious experience, producing a hysteresis in the cir-cadian system response (Hoffmann, 1971). Notably, wefound important differences in rhythm manifestationamong 3 identical LD conditions that had been pre-ceded by different illuminance environments (see Fig. 1for LD1, LD2, and LD3). First, relative coordinationduring LD1 was found in naive animals moved directlyto a 22-h T cycle. Second, after the LD300/0.1 to LD10/0.1

stage sequence, a condition that generated instabilityof rhythm, it could not be reestablished under LD2.And finally, a stable rhythmic pattern during LD3

occurred following DD conditions that synchronizedbetter to the T cycle. Similar effects under prior con-stant darkness were observed in the Siberian hamster,facilitating reentrainment to a phase-delayed LD cycle(Ruby et al., 2002). Therefore, if the circadian systemproperties are also modified, such plasticity mayinvolve history-dependent changes in the couplingstrength between pacemaker oscillators, as may happenduring classic aftereffects.

Although on the basis of observed longitudinalchanges, our experimental protocol revealed novelhistory-dependent changes on entrainment of activityrhythm, the possible effects of aging should be takeninto account. Age-related changes in circadian systemproperties are not yet well established because evidenceboth for (Pittendrigh and Daan, 1974; Morin, 1988)and against (Davis and Viswanathan, 1998) τ changeshas been published, and the same has occurred withregard to entrainment (Morin, 1988 vs. Aton et al.,2004). In any case, aging is associated with an overallincrease in the lability of circadian phase and a reduc-tion in the amplitude of endocrine, metabolic, andbehavioral rhythms in various animal species (seeTurek et al., 1995 for review). Specifically for theSyrian hamster, it was shown that animals 12.25 monthsold, submitted to LD 14:10, exhibited a more greatlyreduced amplitude than 5.5-month-old ones (Labyaket al., 1998). When comparing present changes inamplitude, our hamsters were ∼6 months old at thebeginning of LD2, whereas ∼8-month-old animals at

LD3 exhibited changes in amplitude opposite to thoseassociated with aging. Also, old hamsters reentrainbetter to phase-advanced LD 14:10 cycles (Zee et al.,1992). This was attributed to the shortening of activityrhythm τ with age. Although this could account forthe better entrainment observed in LD3 when com-pared with LD2, hamsters show no important differ-ences in age between these 2 stages. Thus, we assumethat the effect of aging did not account for theobserved differences reported here. Also, the repro-ductive state, which we did not test, could havechanged because of illuminance modifications (Ferraroand McCormack, 1985) and affected the entrainmentpattern. Because effects of entrainment history mustbe evidenced by the animal experience itself, thedefinition of present sequential design excludes theuse of a control group. However, additional experi-ments with the same light-dark schedules in a differentsequence could definitely demonstrate whether suchobserved longitudinal changes in entrainment dependsolely on previous history or other factors are involved.

In summary, 2 main conclusions can be drawn fromthe present study: 1) the assumption of both a coupledmultioscillatory system and that light may act byinhibiting coupling explains entrainment under a T22cycle, and 2) importantly, the actual state of the circa-dian system is established by its prior light (entrain-ment) history, perhaps by a memory component.Although the principle of coupling was conceivednearly 4 decades ago (Winfree, 1967), its physiologicalcorrelates remain elusive. Meanwhile, the presentwork offers a novel combination of 2 experimentalstrategies that allow us to observe intriguing changesin the output of the circadian multioscillator system,employing animal experiments in tandem with com-putational model simulations. They have permitted usto conclude that changes in hamster activity patterns,including entrainment to T22 cycles, are partially relatedto modulations exerted by light intensity on the cou-pling strength between oscillators. Unfortunately, thereis still a lack of physiological evidence for coupling inthe circadian pacemaker, and plasticity is lacking inmost of the computational models of the pacemaker.

ACKNOWLEDGMENTS

This work was funded by the Ministerio de Cienciay Tecnología, Spain, BFI 2003-03489. J. J. Chiesa hasa Programa Nacional de Formación de ProfesoradoUniversitario predoctoral fellowship from theMinisterio de Educación y Ciencia, Spain.

Chiesa et al. / HISTORY-DEPENDENT CHANGES 55

by guest on June 5, 2013jbr.sagepub.comDownloaded from

REFERENCES

Achermann P and Kunz H (1999) Modeling circadianrhythm generation in the suprachiasmatic nucleus withlocally coupled self-sustained oscillators: Phase shiftsand phase response curves. J Biol Rhythms 14:460-468.

Aschoff J (1981) Freerunning and entrained circadianrhythms. In Handbook of Behavioural Neurobiology: BiologicalRhythms, Aschoff J, ed, pp 81-93, New York, Plenum.

Aton SJ, Block GD, Tei H, Yamazaki S, and Herzog ED(2004) Plasticity of circadian behavior and the suprachia-matic nucleus following exposure to non-24-hour lightcycles. J Biol Rhythms 19:198-207.

Aton SJ, Colwell CS, Harmar AJ, Waschek J, and Herzog ED(2005) Vasoactive intestinal polypeptide mediates circa-dian rhythmicity and synchrony in mammalian clockneurons. Nat Neurosci 8:476-483.

Batschelet E (1981) Circular Statistics in Biology. London:Academic Press.

Beersma DG (2005) Why and how do we model circadianrhythms? J Biol Rhythms 20:304-313.

Berson DM, Dunn FA, and Takao M (2002) Phototransductionby retinal ganglion cells that set the circadian clock.Science 295:1070-1073.

Cambras T, Chiesa J, Araujo J, and Díez-Noguera A (2004)Effects of photoperiod on rat motor activity rhythm atthe lower limit of entrainment. J Biol Rhythms 19:216-225.

Daan S and Aschoff J (2001) The entrainment of circadiansystems. In Handbook of Behavioural Neurobiology: CircadianClocks, Takahashi JS, Turek FW, and Moore RY, eds, pp 7-43,New York, Kluwer/Plenum.

Davis FC and Viswanathan N (1998) Stability of circadiantiming with age in Syrian hamsters. Am J Physiol RegulIntegr Comp Physiol 275(Pt 2):R960-R968.

Depres-Brummer P, Levi F, Metzger G, and Touitou Y (1995)Light-induced suppression of the rat circadian system.Am J Physiol Regul Integr Comp Physiol 268:R1111-R1116.

Díez-Noguera A (1994) A functional model of the circadiansystem based on the degree of intercommunicationin a complex system. Am J Physiol Regul Integr CompPhysiol 267:R1118-R1135.

Díez-Noguera A, Cambras T, and Acosta O (2003) Propertiesof circadian system: Are they just common properties ofmultioscillatory systems? Paper presented at the VII Latin-American Symposium on Chronobiology, November,Tlaxcala, México.

Eastman C and Rechtschaffen A (1983) Circadian tempera-ture and wake rhythms of rats exposed to prolongedcontinuous illumination. Physiol Behav 31:417-427.

Elliott JA and Tamarkin L (1994) Complex circadian regula-tion of pineal melatonin and wheel-running in Syrianhamsters. J Comp Physiol 174:469-484.

Ferraro JS and McCormack CE (1985) Gonadal regressiondespite light pulses coincident with locomotor activity inthe Syrian hamster. Biol Reprod 33:93-102.

Foster RG (2005) Neurobiology: Bright blue times. Nature433:698-699.

Gorman MR and Elliott JA (2004) Dim nocturnal illumina-tion alters coupling of circadian pacemakers in Siberianhamsters, Phodopus sungorus. J Comp Physiol A NeuroetholSens Neural Behav Physiol 190:631-639.

Gorman MR, Elliott JA, and Evans JA (2003) Plasticity ofhamster circadian entrainment patterns depends on lightintensity. Chronobiol Int 20:233-248.

Gorman MR, Freeman DA, and Zucker I (1997)Photoperiodism in hamsters: Abrupt versus gradualchanges in day length differentially entrain morningand evening circadian oscillators. J Biol Rhythms12:122-135.

Gorman MR, Kendall M, and Elliott JA (2005) Scotopic illu-mination enhances entrainment of circadian rhythms tolengthening light:dark cycles. J Biol Rhythms 20:38-48.

Hattar S, Lucas RJ, Mrosovsky N, Thompson S, Douglas RH,Hankins MW, Lem J, Biel M, Hofmann F, Foster RG, et al.(2003) Melanopsin and rod-cone photoreceptive systemsaccount for all major accessory visual functions in mice.Nature 424:76-81.

Helfrich-Förster C and Díez-Noguera A (1993) Use of a mul-tioscillatory system to simulate experimental resultsobtained for the period-mutants of Drosophila melanogaster.Proceedings of the 9th meeting ESC. J Interdiscipl CycleRes 24:225-231.

Herzog ED, Takahashi JS, and Block GD (1998) Clock controlscircadian period in isolated suprachiasmatic nucleus neu-rons. Nat Neurosci 1:708-713.

Hoffmann K (1971) Splitting of the circadian rhythm as afunction of light intensity. In Biochronometry, Menaker M,ed, pp 134-150, Washington, DC, National Academy ofScience.

Honma S and Honma K (1999) Light-induced uncoupling ofmultioscillatory circadian system in a diurnal rodent,Asian chipmunk. Am J Physiol Regul Integr Comp Physiol276:R1390-R1396.

Honma S, Nakamura W, Shirakawa T, and Honma K(2004) Diversity in the circadian periods of singleneurons of the rat suprachiasmatic nucleus depends onnuclear structure and intrinsic period. Neurosci Lett358:173-176.

Honma S, Shirakawa T, Nakamura W, and Honma K (2000)Synaptic communication of cellular oscillations in the ratsuprachiasmatic neurons. Neurosci Lett 294:113-116.

Jagota A, de la Iglesia HO, and Schwartz WJ (2000) Morningand evening circadian oscillations in the suprachiasmaticnucleus in vitro. Nat Neurosci 3:372-376.

Kuhlman S, Silver R, Le Sauter J, Bult-Ito A, and McMahon D(2003) Phase resetting light pulses induce Per1 and per-sistent spike activity in a subpopulation of biologicalclock neurons. J Neurosci 23:1441-1450.

Labyak SE, Turek FW, Wallen EP, and Zee PC (1998) Effectsof bright light on age-related changes in the locomotoractivity of Syrian hamsters. Am J Physiol Regul IntegrComp Physiol 274:R830-R839.

Liu C, Weaver DR, Strogatz SH, and Reppert SM (1997)Cellular construction of a circadian clock: Period deter-mination in the suprachiasmatic nuclei. Cell 91:855-860.

Long MA, Jutras MJ, Connors BW, and Burwell RD (2005)Electrical synapses coordinate activity in the suprachias-matic nucleus. Nat Neurosci 8:61-66.

Low-Zeddies SS and Takahashi JS (2001) Chimera analysisof the Clock mutation in mice shows that complexcellular integration determines circadian behavior. Cell105:25-42.

56 JOURNAL OF BIOLOGICAL RHYTHMS / February 2006

by guest on June 5, 2013jbr.sagepub.comDownloaded from

McCormack CE and Sontag CR (1980) Entrainment by redlight of running activity and ovulation rhythms of rats.Am J Physiol Regul Integr Comp Physiol 239:R450-R453.

Meijer JH, Groos GA, and Rusak B (1986) Luminance codingin a circadian pacemaker: The suprachiasmatic nucleusof the rat and the hamster. Brain Res 382:109-118.

Miche S and Colwell CS (2001) Cellular communication andcoupling within the suprachiasmatic nucleus. ChronobiolInt 18:579-600.

Morin LP (1988) Age-related changes in hamster circadianperiod, entrainment, and rhythm splitting. J BiolRhythms 3:237-248.

Mrosovsky N (1999) Masking: History, definitions, andmeasurement. Chronobiol Int 16:415-429.

Mrugala M, Zlomanczuk P, Jagota A, and Schwartz WJ (2000)Rhythmic multiunit neural activity in slices of hamstersuprachiasmatic nucleus reflect prior photoperiod. AmJ Physiol Regul Integr Comp Physiol 278:987-994.

Nelson DE and Takahashi JS (1999) Integration and satura-tion within the circadian photic entrainment pathway ofhamsters. Am J Physiol Regul Integr Comp Physiol277:R1351-R1361.

Oda GA, Menaker M, and Friesen WO (2000) Modeling thedual pacemaker system of the tau mutant hamster. J BiolRhythms 15:246-264.

Ohta H, Yamazaki S, and McMahon DG (2005) Constantlight desynchronizes mammalian clock neurons. NatNeurosci 8:267-269.

Pittendrigh CS (1981a) Circadian systems: Entrainment. InHandbook of Behavioural Neurobiology: Biological Rhythms,Aschoff J, ed, pp 95-124, New York, Plenum.

Pittendrigh CS (1981b) Circadian systems: General perspec-tive. In Handbook of Behavioural Neurobiology: BiologicalRhythms, Aschoff J, ed, pp 57-80, New York, Plenum.

Pittendrigh CS and Daan S (1974) Circadian oscillations inrodents: A systematic increase of their frequency with age.Science 186:548-550.

Pittendrigh CS and Daan S (1976a) A functional analysis of cir-cadian pacemakers in nocturnal rodents. IV. Entrainment:Pacemaker as clock. J Comp Physiol 106:291-351.

Pittendrigh CS and Daan S (1976b) A functional analysis ofcircadian pacemakers in nocturnal rodents. V. Pacemakerstructure: A clock for all seasons. J Comp Physiol106:333-355.

Pittendrigh CS, Elliott JA, and Takamura T (1984) The circa-dian component in photoperiodic induction. Ciba FoundSymp 104:26-47.

Quintero J, Kuhlman SJ, and McMahon DG (2003) The bio-logical clock nucleus: A multiphasic oscillator networkregulated by light. J Neurosci 23:8070-8076.

Ruby NF, Joshi N, and Heller HC (2002) Constant darknessrestores entrainment to phase-delayed Siberian hamsters.Am J Physiol Regul Integr Comp Physiol 283:R1314-R1320.

Schaap J, Albus H, Tjebbe VH, Eilers PH, Detari L, andMeijer JH (2003) Heterogeneity of rhythmic suprachiasmaticnucleus neurons: Implications for circadian waveformand photoperiodic encoding. Proc Natl Acad Sci USA100:15994-15999.

Schwartz WJ, de la Iglesia HO, Zlomanczuk P, andIllnerova H (2001) Encoding le quattro stagioni within themammalian brain: Photoperiodic orchestration throughthe suprachiasmatic nucleus. J Biol Rhythms 16:302-311.

Selkov EE (1968) Self oscillations in glycolysis. Eur J Biochem4:79-86.

Sokolove PG and Bushell WN (1978) A chi square peri-odogram: Its utility for the analysis of circadian rhythms.J Theor Biol 72:131-160.

Turek FW, Penev P, Zhang Y, Van Reeth O, and Zee PC(1995) Effects of age on the circadian system. NeurosciBiobehav Rev 19:53-58.

Vilaplana J, Cambras T, and Díez-Noguera A (1997)Dissociation of motor activity circadian rhythm in ratsafter exposure to LD cycles of 4-h period. Am J PhysiolRegul Integr Comp Physiol 272:R95-R102.

Yamaguchi S, Isejima H, Matsuo T, Okura R, Yagita K,Kobayashi M, and Okamura H (2003) Synchronization ofcellular clocks in the suprachiasmatic nucleus. Science302:1408-1412.

Welsh DK, Logothetis DE, Meister M, and Reppert SM(1995) Individual neurons dissociated from rat suprachi-asmatic nucleus express independently phased circadianfiring rhythms. Neuron 14:697-706.

Winfree A (1967) Biological rhythms and the behavior ofpopulations of coupled oscillators. J Theor Biol 16:15-42.

Zee PC, Rosenberg RS, and Turek FW (1992) Effects of agingon entrainment and rate of resynchronization of circa-dian locomotor activity. Am J Physiol Regul Integr CompPhysiol 263:R1099-R1103.

Chiesa et al. / HISTORY-DEPENDENT CHANGES 57

by guest on June 5, 2013jbr.sagepub.comDownloaded from