Physiology&Behavior,Vol.52, pp. 231-235, 1992
0031-9384/92 $5.00 + .00 Copyright© 1992PergamonPressLtd.
Printed in the USA.
Activity Rhythms in Mice III: Stability and Plasticity of Rhythm Characteristics in Experimental and Environmental Conditions J A C Q U E S BEAU
GOn~tique, Neurogdn~tique el Comportement, Universit~ Ren~ Descartes U.F.R. BiomOdicale, U.R.A. 1294 C.N.R.S., 45 rue des Saints POres 75270 Paris Cedex 06 France Received 14 J u n e 1991 BEAU, J. Activity rhythms in mice Ili: Stability and plasticity ~["rhythm characteristics in experimental and environmental conditions. PHYSIOL BEHAV 52(2) 231-235, 1992.--The expression of a biologicalrhythm as measured by a given descriptor results from a set of components: the subject, the measuring device, and the experimental conditions. Rhythm of activity under four experimental conditions was observed in two strains of mice: BALB/c and C57BL/6. Condition 1, the optimal normal situation, was used as a reference; condition 2 was a retest of the animals used in condition 1 twelve days later; condition 3 tested animals for a period of 3 days in a food-deprivation situation; in condition 4, animals were isolated for 2 weeks prior to testing. Data analysis of rhythm of activity in the synchronic mode is based on a series of tests in the frequency and temporal domains. Analysisof condition-linkedvariationin the findingsindicatesthat these parametersare only slightlyaffected by the four experimental conditions. The results, however, can be hierarchizedaccordingto condition and accordingto the kind of parameter. The temporal parameters are more sensitivethan the frequencyparameters, the least sensitivebeing the temporo-frequential measures. Isolation produces the greatest variation in parameter magnitude. The results show that rhythm of activity is fairly stable across a number of experimental conditions, a finding that could considerably simplify experimental protocols. Chronobiology
Circadian rhythm
Inbred m i c e
Activity
THE persistence of biological rhythms in the absence of a luminous synchronizer is one indication that they are endogenous and suggests the involvement of an internal clock(s). However, the question still remains as to whether the features of biological rhythms are stable or whether they exhibit a certain degree of plasticity in particular under varying experimental conditions. A partial answer to this question can be found in the numerous works conducted on free run activity: Aschoff's rule [so termed by Pittendrigh (16)] accounts for the effect of the luminous env i r o n m e n t - t h e fundamental synchronizer--on the characteristics of a biological rhythm. It yields a value expressing the relationship between increase in the period when free run is in LL and decrease when free run is in DD for nocturnal animals; this rule can be reversed for diurnal animals. Standard practice in chronobiology is to express the temporal schedule of the synchronizer in terms of duration of the light (L) and dark (D) phases; for example LD 12:12 indicates 12 h of light followed by 12 h of darkness; by extension, LL signifies continuous light and DD continuous darkness. In identical situations, however (DD or LL), the value of period r is stable from one day to the next (17), although mice exhibit long-term variations in r with age and as a function of prior experience (the after-effect phenomenon). In contrast to the effect of light (Aschoff's Rule), these same authors (8) report that administration of D20 produces longer periods in diurnal as well as in nocturnal animals, thus demonstrating biochemical responsiveness. Variations in
Environment
Experimentalconditions
r in the hamster as a function of the physical environment have been observed by Pratt (19). The r period in DD varies as a function of whether activity is measured when the animal is simply outside its burrow or in a running wheel. Similarly, activity is affected by the presence or absence of a running wheel in the rearing cages. These differences should be placed in the same category as variations caused by differences in measuring devices (14). Studies on the stability of biological rhythms in the synchronized mode provide consistent evidence for stability in different environments. The environmental conditions can be classified into two groups whose distinguishing feature is when--prior to or during the assessment--modification in the environment took place. Research falling into group 1 (prior modification) indicate that there is no change in rhythm as a function of 1) exposure to a previous experiment, submersion, electric shock, restriction of activity when a pup (21), on the rat or 2) an LD schedule (15), on the mouse. Isolation prior to testing, however, has been reported to affect the amount of activity and the ultradian rhythm (10). In group 2, in studies which focus on environmental modifications, LePape and Lasaile (12) report that the D/L ratio (see the parameters defining a biological rhythm) varies as a function of whether the test was conducted in the rearing cages or in a seminatural environment. My contention here is in this type of finding, as the free-running period, it is the apparatus which is important. In fact, the rhythm is not affected by a different en-
231
232
BEAt
vironment, hut rather another aspect of the rhythm is being assessed by a different measuring apparatus. Surprisingly, in this respect, extensive studies by Reinberg et al. (20) indicate that biological rhythms are unaffected in both humans and animals in situations of food deprivation. This finding was originally obtained by Chossat (7) in an ad mortem experiment on the pigeon. The studies in both groups, however, have only dealt with elementary features of rhythm such as the D/L ratio or the presence or absence of rhythm. What is lacking is further research on the impact of a variety of experimental situations on a wide range of parameters in both the temporal and frequency domains, thus accounting for all the facets of biological rhythm. This is the aim of the present study, which serves a dual purpose: 1. to control the stability of rhythm of activity in highly inbred strains of mice in terms of the component structures of rhythm as function of environmental variations: 2. to show that certain standardised but not extremely stringent experimental conditions are suitable for assessing the fine structure of activity rhythms in the synchronic mode. METHOD
Measuring Device A detailed description of the data collection and recording device can be found in a preceding article (1). The device is composed of 16 actographs driven by a microcomputer. Each actograph consists ofatoric cage having a 4.5 X 4 cm rectangular section, and a mean radius of 15 cm. The animal's movements inside the cage are detected by an optoelectronic system which yields a measurement of locomotor activity. Amplitude of totalised locomotor activity between two samplings is automatically sampled by the microcomputer every 8.44 minutes. Food and water are provided automatically ad lib for the entire duration of the experiment.
E.\7~erimental Protocol The protocol is identical to those used previously for data collection with the same type of measuring device (2,5). Measurement lasts 3 days. A 72-h observation window is suitable for the study of synchronized rhythm when the structure of circadian rhythm is under study, because in this case it is the frequency of period longer than 24 h which is interesting and predominate in most of the parameters. The animal is placed in the actograph at 0730. Hours are expressed in universal time (UT) to avoid ambiguities created by daylight savings in France, and differences in international legislation. Recording of actograph data starts at 1330, i.e., the middle of the diurnal phase which is thought to be the period of least activity in the mouse. Recording stops at 1330, 72 h later. Four types of measurement were used. The first was identical to a previous protocol and was conducted on 16 strain C and strain B6 mice. This constitutes the normal condition (group N). Five of the subjects in these groups were retested in condition 2--the retest condition (group R). The second experiment took place 12 days after the first. Experiment 3 consisted of testing the animals for the 3-day period in a situation of food deprivation (group D). Experiment 4 was conducted after the animals had been isolated in individual cages for 2 weeks prior to the experiment (group I).
Sllbjecls The subjects used in the experiments were all 8- to 10-weekold female mice from inbred strains. Since these experiments
are the complement to a genetic analysis on parental lines BABL cBy, abbreviated C and C57BL/6By abbreviated B6. their Fls and their seven recombinant inbreds (5), the two parentals were used in the present experiment. All the animals were born and reared in the laboratory. l h e animals were weaned at 4 weeks of age and placed in a preexperimental room with standardized conditions. The animals were housed in Makrolon cages measuring 42 x 21 , 18 cm containing dust-free sawdust. Nontreated water and Sourrifarat (IFFA-CREDO) food were provided ad lib. Temperature was maintained at 23 ± 1°. Alternation of light and dark and level of illumination were identical to experimental conditions. 3his gave the animals at least a l-month period of highly standardized conditioning to prepare them for the experimenl.
Parameter.~ De[ining Rhythm Raw activity data in the temporal and frequency domains (Fig. l a,b) were examined for each experiment. Data were characterized by the seven parameters which provide the most complete description of activity rhythm: Rms~, Rmsr, D/L, EBI, EB3, 7~, and T~. A brief definition of these parameters is given in the following paragraphs. A detailed description, rationale, and complete properties can be fbund elsewhere (3,4). Two parameters are calculated in the temporal domain (Fig. l a): Rms~ and D/L. Rmst is obtained by calculating the root mean square of the temporal function characterizing activity. The formula for Rmst, calculated on the temporal expression of the activity R(t), is as follows: Rmst :
/ j,, 7 ~",~,,;"L~,, R2(t) dt,
7"is the duration of the measure. This value measures the amount of activity and takes both the mean level of activity and local bursts of activity into account. D/L is the ratio of the amount of activity recorded during the dark part of the cycle to the amount of activity recorded during the light part. If animals are perfectly adapted to the day/ night synchronizer, they are only active at night and D/L tends towards infinity, if, in contrast, animals are as active during the night as the day, D/L equals 1. Thus, D/L measures the fitting of the rhythm to the natural synchronizer. It is sensitive to the difference in phase between the period of intense activity and the onset of the dark phase and is, thus, not perfect. Three parameters are calculated, in the frequency domain (Fig, l b), on the activity signal spectrum S(f) obtained by a FFT (Fast Fourier Transform): Rmsf, EBI, and EB3. Rmsr, filtered root mean square data, like Rms~, is also linked to amount of activity. It is defined as the energy produced by the temporal signal R(t) whose ideal filter would eliminate the ultradian components with a period T = I/F less than 1.125 h. This parameter usually exhibits less dispersion than its counterpart Rms,, which can be attributed to the filtering. For the parameters described below, the spectrum is partitioned into four frequency bands. The division by logarithmic increment, which is standard practice in frequency studies, was necessary because the harmonics of these frequencies were related. The energy in each band is indicative of a specific characteristic of rhythm. Only two bands were used in the present study. EB 1 is the proportion of activity signal energy located in the band period oc > T > 18 h as compared to the total energy of the entire spectrum and is calculated as follows:
ACTIVITY RHYTHMS Ill
233
A
"
I
"
I
18H30
a)
-
I
6H30
I
18H30
6H30
I
18H30
I~
6H30
b)
". UT TXME
FREQUENCY
DONAIN
Pulsatile activity
ATic Spectral Center/
24H : 12H l B1
EB3
DOMAIN
I
SPECTRUM OF ACTIVITY FOR C57/B6-11
Circadian level
EBI
TINE
&H
,
3H
B2
90"
B3
PERIODS w
B4
FIG. 1. (a) Example of activity rhythm evolution in time domain in arbitrary units. The white and black bars on the temporal axis (marked in universal time) represents the light and dark phases. (b) Spectral analysis of this activity rhythm. S(.D represents the module of the Fourier transform. The zones BI, B2, and B3 represent the components of activity situated respectively in the circadian zone and the close and middle ultradian zones. Tc and Tr arrows indicate the position of spectral center and spectral range.
, $2(./3 d f EBI S2(,JO
as a bandwidth. Its value in ultradian periods decreases as the spectral range increases with considerable energy toward the high frequencies.
d.l"
RESULTS Given the type of frequency band, EBI is connected to the circadian level (related to a 24-h period). The higher the EBI value, the better the fit of the temporal curve of activity rhythm to a pure sinusoidal wave with a 24-h period. EB3 is calculated in the same way as EB 1 and is the proportion of activity signal energy located in the period band 4, 5 h > T > 1.25 h; it is related to the pulsatile nature of activity and reflects a tendency for production of short bursts of activity outside the range of circadian components. T~ and Tr are temporo-frequency descriptors (4) which can represent the major portion of the information of the spectrum by using only these two measures. This procedure, used in EEG analysis (11), is designed to obtain as simple a characterization as possible of what is an already complex spectrum (2 indices). These two indices correspond to periods and are measured in hours and calculated as follows:
where F is the maximum frequency in the spectrum. T¢, the spectral center is visible as the mean period of the spectrum. The fewer the high frequencies in the spectrum (corresponding to bursts of pulsatile or erratic activity) the greater the value of Tc approaching 24 h. Tr, the spectral range, appears
The means for the raw data appear in Table 1 that also presents the outcome of a 2-way analysis of variance where Exp covers the four conditions--normal N, repeated experiment R, food deprivation condition D, and isolation condition I, and Strain defines the strains: C and B6. The ANOVA was calculated either on the raw data or on data which had been functionally transformed for homoscedasticity with power and logarithmic functions. These techniques are customary in statistics and applied frequently in genetics (13). The results show a significant Strain effect for all parameters except EB3; this was expected since these parameters were chosen because they are informative for genetic analysis. No Exp effect was observed for Rmsf, EBI, or Tr. There is a main Exp effect for Rmst, D/L, and EB3, and a Strain. Exp interaction for T¢. This significant interaction effect for Tc calls for a separate strain-by-strain analysis. The results appear in Table 2. No significant effects for strain were observed. Inspection of the raw data for the three parameters exhibiting an Exp effect (Rmst, D/L, and EB3) suggests a main effect for isolation which is partially confirmed by the statistical analysis. First, an ANOVA excluding I (Table 3) indicates that Rmst and EB3 form a homogeneous group for conditions N, R, and D; second, partial t-test comparisons show that only group I differs significantly from group N [respectively Rmst, t (42) = 2.09, p < 0.05; EB3, t (42) = 2.11, p < 0.05]. Even without group I,
234
BF,\I
RAW D A I A
Raw Data Strain group C N R D 1 B6 N R D 1 Statistical analysis Factor df Exp 3 Strain 1 Exp • Strain 3
1 ABI.E 1 AND STATISTIC&I AN&LYSIS
D/I.
Rms~
Rms~
EBI
EB3
1,-
1.88_+0.15 1.93 _+ 0.26 1.26 _+ 0.24 1.34+0,26
197.97_+ 1 8 . 3 253.32 _+ 32.1 212.40 _+ 28.4 238.74_+26.0
111.72_+ 11.1 141.97 -+ 17.4 119.70 _+ 20.8 133.71 + 1 4 . 7
24.21 _+2.26 21.65 _+ 2.83 33.34 _-+6.97 23.30+3.13
37.31 _+2.17 40.79 _+ 3.5 28.45 +_ 4.6 28.05+ 1.93
3.27+0.12 3.31 _+ 0.09 3.91 _+ 0.30 3.73 +(t.15
1.78_+0.03 1.80 -+ 0.05 1.78 _+_0.02 1.82 ±(I.05
3.86_+0,26 2.81_+0,19 2.93_+0,21 2.91 -+0.21
119.92_+ 122.42_+ 135.72_+ 176.22_+
69.77_+ 7.2 68.29_+ 2.0 77.73_+ 4.8 96.04_+ 9.4
18.87+ 1 . 7 2 14.09_+2.16 21.78+3.04 17.57 + 1 . 1 5
35.74_+ 1 . 9 9 39.24+4.1 40.66_+2.3 33.63_+2.97
2.93_+ 1.0 2.75_+0.16 2.53_+(I.13 3.07_+0.15
1.72_+(I.02 1.74+0.07 1.56.0.04 1.71 _+0.05
2.25 8.88 $ .22
2.97 * 2.73 2.22
1.83 38.05 t 3.32*
2.4 I 14.61 $ 1.49
2.92" 52.38 # 1.21
10.3 4.3 6.2 2.3
3.22" 33.53 t 1.31
2.17 26.03 t .93
The raw data values of rhythm parameters indicated in the table are the means values with standard error. C designates strain BALB/c, B6 designates strain C57BL/6. N designates the normal condition tested on 16 subjects. R designates the repeated condition where five subjects were retested. D designates the food-deprivation condition tested on six subjects. I designates the isolation condition where six subjects were isolated prior to testing. Statistical analysis is a 2-way ANOVA on parameters defining rhythm of activity. Exp is the experiment factor covering groups N, R, D, and I. Strain covers strains C and B6. Exp. Strain represents the interaction, df indicates the degrees of freedom. F the Fisher test and the associated level of significance is: Significant at p < *0.05:-t0.0001: :~0.01.
D / L still has a significant effect that could be predicted from the values in Table 1 since the normal situation tends to dichotomize. DISCUSSION The study of parameters characterizing r h y t h m of activity in strains C a n d B6 indicates that the parameters measured here are relatively insensitive to experimentally induced environmental variations but nevertheless form a hierarchy (see Table 4). However, these environmental variations always remain lower than genetic variations (see "Fables 1 a n d 4). As shown in Table 4, the temporo-frequency descriptors Tc and T r are totally unresponsive to experimental conditions and can, thus, be considered to be robust with respect to experimental protocols. This can be attributed to their m e t h o d of calculation, since they are global measures of the features defining the spectrum. It should be pointed out, however, that the Strain. Exp interaction indicates that Tc undergoes a differential effect according to the strains. The parameters in the frequency d o m a i n are highly insensitive to experimental conditions in contrast to the parameters in the temporal domain. C o m p a r i n g Rmsf a n d Rms~, which b o t h represent a value linked to a m o u n t of activity, indicates that they
TABLE 2 ANALYSIS BY STRAIN FOR T~ Strain
df
F
C B6
3 3
2.1 2.48
Effect of Exp on parameter T~ for conditions N, R, D, and I on strains C and B6; no tests are significant.
differ in sensitivity as a function of the m a n n e r in which the value is calculated: when Rmsf does not include the high frequencies there is less trend towards dispersion. W h e n a m u c h more classic variable associated with a m o u n t of a c t i v i t y - - t h e m e a n v a l u e - - w a s calculated (data not given here), the results are comparable to those obtained for Rmst. EB3, which is a frequency measure located in the upper part of the spectrum, shows a sensitivity to experimental conditions as did Rmst. This sensitivity-high frequency characteristic is an additional indication of the d i s c r i m i n a n t value of the u n t r a d i a n zones. It drops out, however, if the isolation condition is excluded from analysis. Only D / L exhibits an overall sensitivity to experimental conditions.
TABLE 3 EFFECT OF ISOLATION Parameter/Factor Rmst Exp Strain Exp. Strain Exp D/L Strain Exp. Strain EB3 Exp Strain Exp. Strain
d/
F
I
31.30 *
2 2
0.72 3.33t
I
36.6l* .72
1.35
2 2 1
2
.48 .33 3.06
Parameters exhibiting an Exp effect when I (isolation condition) is excluded: *t Significant at *p < 0.001- tp < 0.05.
ACTIVITY R H Y T H M S llI
235
TABLE 4 STABILITY OF RHYTHM PARAMETERS AS A FUNCTION OF DOMAIN (TIME OR FREQUENCY) AND EXPERIMENTAL CONDITION Time Parameters Overall sensitivity to experimental conditions Sensibility to experimental conditions excluding 1
Frequency
Time/ Frequency
D/L
Rms~ Rmsr EB1
EB3
T~
T,
Yes
Yes
No
No
Yes
No
No
Yes
No
No
No
No
No
No
D/L, a highly classical temporal parameter in chronobiological assessment in the synchronic mode, thus appears to be sensitive to the experimental situation and should be considered to be less robust in particular in genetic analysis. In our opinion, this lack of robustness according to environmental conditions is due to phase sensibility of this parameter which was criticized in Beau (6) because of its inefficiency, in particular, to show maternal effect. The insensitivity of the parameters measured here implies that the rhythm characteristics of each strain are stable. First of all, the value of a parameter for a given strain and a given experimental situation is only rarely comparable to the value obtained for the other strain in another situation. Secondly, the strain hierarchy remains constant under the different experi-
mental situations. Each component of rhythm thus remains characteristic of the strain despite modifications of the experimental environment. This is true for even the most arduous situation tested here: 3 days of food deprivation. The results concerning repetition of experiments are useful in cases where repeated measures are needed on rare or hardto-find animals. It is also a good indicator of the absence of a sequential effect for experiments in sequence (an effect which can only make experimental protocols more cumbersome). This latter point is taken up by Possidente and Hegmann (18) in a study on seven inbred strains of mice including C and Bl0 for parameters in free run (period) and in synchro/free run (variation in period). The effect of isolation--which is particularly apparent for the temporal measures and those affecting features linked to a m o u n t of activity--has the greatest impact on the parameters which disrupt rhythm of activity. This point has been made by D'Amato (9) for augmentation of activity on Swiss mice. In summary, the components of activity rhythm in the mouse exhibit variable stability: 1. component values vary with the type of parameter used (or rather, its domain time or frequency); 2. component values vary across experimental conditions; 3. component values exhibit different trends as a function of strain. This study has demonstrated high stability for the set of parameters associated with rhythm of activity in situations where the experimental environment is modified. Repeated testing, and a more stressful condition, food deprivation, were examined here. Rhythm components, and, in particular, parameters derived from measures in the temporal domain, were also shown to be sensitive to modifications which touch on the social environment, such as prior isolation of animals.
REFERENCES 1. Beau, J. Microcomputer-driven digital data processing ofmouse actograph activity. Physiol. Behav. 3:435-441 ; 1986. 2. Beau, J. Components comparison of activity rhythm in three inbred mice strains: In: Medioni, J.; Vaysse, G., eds. Genetic approaches to behaviour. Toulouse: Privat IEC; 1986:19-27. 3. Beau, J. Spline interpolation of the Fast Fourier transform of a biological rhythm: Determination of the spectral components. J. Interdiscipl. Cycle Res. 19:129-140; 1988. 4. Beau, J. Calculation and properties of two temporal and frequency descriptors: The spectral center and the spectral range. J. Interdisciplin. Cycle Res. 19:141-152; 1988. 5. Beau, J. Activity rhythms in inbred mice I: Genetic analysis with recombinant inbred strains. Behav. Genet. 21 (2): 117-129; 1991. 6. Beau, J. Fitting a biological rhythm to a synchronizer: Applied illustration of mouse activity rhythm in genetic analysis. J. lnterdiscipl. Cycle Res. 22(3):271-280; 1991. 7. Chossat, C. Recherche experimentale sur l'inanition. Mem. Acad. R. Sci. Inst. Fr.; 1843:430-640. 8. Daan, S.; Pittendrigh, C. S. A functional analysis of circadian pacemakers in nocturnal rodents: II1 Heavy water and constant light: Homeostasis of frequency? J. Comp. Physiol. 106:267-290; 1976. 9. D'Amato, F. R. Time budgets and behavioural synchronization in aggregated and isolated male and female mice. Behav. Processes 13: 385-397; 1986, 10. Del Pozo, F.; DeFeudis, F. V.; Jimenez, J, M. Motilities of isolated
and aggregated mice: A difference in ultradian rhythmicity. Experientia 34( 10):1302-1304:1978. 11. Hjorth, B. EEG analysis based on time domain properties. Electroencephalogr. Clin. Neurophysiol. 29:306-310; 1970. 12. Le Pape, G.: Lassalle, J. M. Activitt~ locomotrice de souris isoll~es, de deux lignFzesconsanguines, dans un environnement semi-naturel ou en cages d¥1evage. Behav. Processes 4:221-230; 1979. 13. Mather, K.; Jinks, J. Biometrical genetics. London: Chapman; 1971. 14. Mather, J. G. Wheel-runningactivity: A new interpretation. Mammal Res. 11:41-51: 1981. 15. Oliverio, A.: Malorni, W. Wheel running and sleep in two strains of mice: Plasticity and rigidity in the expression of circadian rhythmicity. Brain Res. 163:121-133; 1979. 16. Pittendrigh, C. S. Circadian rhythms and the circadian organization of living systems. Cold Spring Harbor Syrup. Q. Biol. 25:159-184; 1960. 17. Pittendrigh, C. S.; Dann, S. A functional analysis of circadian pacemakers in nocturnal rodents: ! The stability and lability of spontaneous frequency. J. Comp. Physiol. 106:223-252; 1976. 18. Possidente, B.; Hegmann, J. Gene differences modify AschoWs rule in mice. Physiol. Behav. 28(1):199-200; 1982. 19. Pratt, B. L.; Goldman, B. D. Environmental influence upon circadian periodicity of Syrian hamsters. Physiol. Behav. 36:91-95; 1986. 20. Reinberg, A.; Levi, F.; Debry, G. Chronobiologie et nutrition. Encycl. Med. Chir. 10390 AI0:I-10: 1984. 21, Webb, W. B.; Friedman, J. Attempts to modify the sleep pattern of the rat. Physiol. Behav. 6(4):459-460; 1971.
0031-9384/92 $5.00 + .00 Copyright© 1992PergamonPressLtd.
Printed in the USA.
Activity Rhythms in Mice III: Stability and Plasticity of Rhythm Characteristics in Experimental and Environmental Conditions J A C Q U E S BEAU
GOn~tique, Neurogdn~tique el Comportement, Universit~ Ren~ Descartes U.F.R. BiomOdicale, U.R.A. 1294 C.N.R.S., 45 rue des Saints POres 75270 Paris Cedex 06 France Received 14 J u n e 1991 BEAU, J. Activity rhythms in mice Ili: Stability and plasticity ~["rhythm characteristics in experimental and environmental conditions. PHYSIOL BEHAV 52(2) 231-235, 1992.--The expression of a biologicalrhythm as measured by a given descriptor results from a set of components: the subject, the measuring device, and the experimental conditions. Rhythm of activity under four experimental conditions was observed in two strains of mice: BALB/c and C57BL/6. Condition 1, the optimal normal situation, was used as a reference; condition 2 was a retest of the animals used in condition 1 twelve days later; condition 3 tested animals for a period of 3 days in a food-deprivation situation; in condition 4, animals were isolated for 2 weeks prior to testing. Data analysis of rhythm of activity in the synchronic mode is based on a series of tests in the frequency and temporal domains. Analysisof condition-linkedvariationin the findingsindicatesthat these parametersare only slightlyaffected by the four experimental conditions. The results, however, can be hierarchizedaccordingto condition and accordingto the kind of parameter. The temporal parameters are more sensitivethan the frequencyparameters, the least sensitivebeing the temporo-frequential measures. Isolation produces the greatest variation in parameter magnitude. The results show that rhythm of activity is fairly stable across a number of experimental conditions, a finding that could considerably simplify experimental protocols. Chronobiology
Circadian rhythm
Inbred m i c e
Activity
THE persistence of biological rhythms in the absence of a luminous synchronizer is one indication that they are endogenous and suggests the involvement of an internal clock(s). However, the question still remains as to whether the features of biological rhythms are stable or whether they exhibit a certain degree of plasticity in particular under varying experimental conditions. A partial answer to this question can be found in the numerous works conducted on free run activity: Aschoff's rule [so termed by Pittendrigh (16)] accounts for the effect of the luminous env i r o n m e n t - t h e fundamental synchronizer--on the characteristics of a biological rhythm. It yields a value expressing the relationship between increase in the period when free run is in LL and decrease when free run is in DD for nocturnal animals; this rule can be reversed for diurnal animals. Standard practice in chronobiology is to express the temporal schedule of the synchronizer in terms of duration of the light (L) and dark (D) phases; for example LD 12:12 indicates 12 h of light followed by 12 h of darkness; by extension, LL signifies continuous light and DD continuous darkness. In identical situations, however (DD or LL), the value of period r is stable from one day to the next (17), although mice exhibit long-term variations in r with age and as a function of prior experience (the after-effect phenomenon). In contrast to the effect of light (Aschoff's Rule), these same authors (8) report that administration of D20 produces longer periods in diurnal as well as in nocturnal animals, thus demonstrating biochemical responsiveness. Variations in
Environment
Experimentalconditions
r in the hamster as a function of the physical environment have been observed by Pratt (19). The r period in DD varies as a function of whether activity is measured when the animal is simply outside its burrow or in a running wheel. Similarly, activity is affected by the presence or absence of a running wheel in the rearing cages. These differences should be placed in the same category as variations caused by differences in measuring devices (14). Studies on the stability of biological rhythms in the synchronized mode provide consistent evidence for stability in different environments. The environmental conditions can be classified into two groups whose distinguishing feature is when--prior to or during the assessment--modification in the environment took place. Research falling into group 1 (prior modification) indicate that there is no change in rhythm as a function of 1) exposure to a previous experiment, submersion, electric shock, restriction of activity when a pup (21), on the rat or 2) an LD schedule (15), on the mouse. Isolation prior to testing, however, has been reported to affect the amount of activity and the ultradian rhythm (10). In group 2, in studies which focus on environmental modifications, LePape and Lasaile (12) report that the D/L ratio (see the parameters defining a biological rhythm) varies as a function of whether the test was conducted in the rearing cages or in a seminatural environment. My contention here is in this type of finding, as the free-running period, it is the apparatus which is important. In fact, the rhythm is not affected by a different en-
231
232
BEAt
vironment, hut rather another aspect of the rhythm is being assessed by a different measuring apparatus. Surprisingly, in this respect, extensive studies by Reinberg et al. (20) indicate that biological rhythms are unaffected in both humans and animals in situations of food deprivation. This finding was originally obtained by Chossat (7) in an ad mortem experiment on the pigeon. The studies in both groups, however, have only dealt with elementary features of rhythm such as the D/L ratio or the presence or absence of rhythm. What is lacking is further research on the impact of a variety of experimental situations on a wide range of parameters in both the temporal and frequency domains, thus accounting for all the facets of biological rhythm. This is the aim of the present study, which serves a dual purpose: 1. to control the stability of rhythm of activity in highly inbred strains of mice in terms of the component structures of rhythm as function of environmental variations: 2. to show that certain standardised but not extremely stringent experimental conditions are suitable for assessing the fine structure of activity rhythms in the synchronic mode. METHOD
Measuring Device A detailed description of the data collection and recording device can be found in a preceding article (1). The device is composed of 16 actographs driven by a microcomputer. Each actograph consists ofatoric cage having a 4.5 X 4 cm rectangular section, and a mean radius of 15 cm. The animal's movements inside the cage are detected by an optoelectronic system which yields a measurement of locomotor activity. Amplitude of totalised locomotor activity between two samplings is automatically sampled by the microcomputer every 8.44 minutes. Food and water are provided automatically ad lib for the entire duration of the experiment.
E.\7~erimental Protocol The protocol is identical to those used previously for data collection with the same type of measuring device (2,5). Measurement lasts 3 days. A 72-h observation window is suitable for the study of synchronized rhythm when the structure of circadian rhythm is under study, because in this case it is the frequency of period longer than 24 h which is interesting and predominate in most of the parameters. The animal is placed in the actograph at 0730. Hours are expressed in universal time (UT) to avoid ambiguities created by daylight savings in France, and differences in international legislation. Recording of actograph data starts at 1330, i.e., the middle of the diurnal phase which is thought to be the period of least activity in the mouse. Recording stops at 1330, 72 h later. Four types of measurement were used. The first was identical to a previous protocol and was conducted on 16 strain C and strain B6 mice. This constitutes the normal condition (group N). Five of the subjects in these groups were retested in condition 2--the retest condition (group R). The second experiment took place 12 days after the first. Experiment 3 consisted of testing the animals for the 3-day period in a situation of food deprivation (group D). Experiment 4 was conducted after the animals had been isolated in individual cages for 2 weeks prior to the experiment (group I).
Sllbjecls The subjects used in the experiments were all 8- to 10-weekold female mice from inbred strains. Since these experiments
are the complement to a genetic analysis on parental lines BABL cBy, abbreviated C and C57BL/6By abbreviated B6. their Fls and their seven recombinant inbreds (5), the two parentals were used in the present experiment. All the animals were born and reared in the laboratory. l h e animals were weaned at 4 weeks of age and placed in a preexperimental room with standardized conditions. The animals were housed in Makrolon cages measuring 42 x 21 , 18 cm containing dust-free sawdust. Nontreated water and Sourrifarat (IFFA-CREDO) food were provided ad lib. Temperature was maintained at 23 ± 1°. Alternation of light and dark and level of illumination were identical to experimental conditions. 3his gave the animals at least a l-month period of highly standardized conditioning to prepare them for the experimenl.
Parameter.~ De[ining Rhythm Raw activity data in the temporal and frequency domains (Fig. l a,b) were examined for each experiment. Data were characterized by the seven parameters which provide the most complete description of activity rhythm: Rms~, Rmsr, D/L, EBI, EB3, 7~, and T~. A brief definition of these parameters is given in the following paragraphs. A detailed description, rationale, and complete properties can be fbund elsewhere (3,4). Two parameters are calculated in the temporal domain (Fig. l a): Rms~ and D/L. Rmst is obtained by calculating the root mean square of the temporal function characterizing activity. The formula for Rmst, calculated on the temporal expression of the activity R(t), is as follows: Rmst :
/ j,, 7 ~",~,,;"L~,, R2(t) dt,
7"is the duration of the measure. This value measures the amount of activity and takes both the mean level of activity and local bursts of activity into account. D/L is the ratio of the amount of activity recorded during the dark part of the cycle to the amount of activity recorded during the light part. If animals are perfectly adapted to the day/ night synchronizer, they are only active at night and D/L tends towards infinity, if, in contrast, animals are as active during the night as the day, D/L equals 1. Thus, D/L measures the fitting of the rhythm to the natural synchronizer. It is sensitive to the difference in phase between the period of intense activity and the onset of the dark phase and is, thus, not perfect. Three parameters are calculated, in the frequency domain (Fig, l b), on the activity signal spectrum S(f) obtained by a FFT (Fast Fourier Transform): Rmsf, EBI, and EB3. Rmsr, filtered root mean square data, like Rms~, is also linked to amount of activity. It is defined as the energy produced by the temporal signal R(t) whose ideal filter would eliminate the ultradian components with a period T = I/F less than 1.125 h. This parameter usually exhibits less dispersion than its counterpart Rms,, which can be attributed to the filtering. For the parameters described below, the spectrum is partitioned into four frequency bands. The division by logarithmic increment, which is standard practice in frequency studies, was necessary because the harmonics of these frequencies were related. The energy in each band is indicative of a specific characteristic of rhythm. Only two bands were used in the present study. EB 1 is the proportion of activity signal energy located in the band period oc > T > 18 h as compared to the total energy of the entire spectrum and is calculated as follows:
ACTIVITY RHYTHMS Ill
233
A
"
I
"
I
18H30
a)
-
I
6H30
I
18H30
6H30
I
18H30
I~
6H30
b)
". UT TXME
FREQUENCY
DONAIN
Pulsatile activity
ATic Spectral Center/
24H : 12H l B1
EB3
DOMAIN
I
SPECTRUM OF ACTIVITY FOR C57/B6-11
Circadian level
EBI
TINE
&H
,
3H
B2
90"
B3
PERIODS w
B4
FIG. 1. (a) Example of activity rhythm evolution in time domain in arbitrary units. The white and black bars on the temporal axis (marked in universal time) represents the light and dark phases. (b) Spectral analysis of this activity rhythm. S(.D represents the module of the Fourier transform. The zones BI, B2, and B3 represent the components of activity situated respectively in the circadian zone and the close and middle ultradian zones. Tc and Tr arrows indicate the position of spectral center and spectral range.
, $2(./3 d f EBI S2(,JO
as a bandwidth. Its value in ultradian periods decreases as the spectral range increases with considerable energy toward the high frequencies.
d.l"
RESULTS Given the type of frequency band, EBI is connected to the circadian level (related to a 24-h period). The higher the EBI value, the better the fit of the temporal curve of activity rhythm to a pure sinusoidal wave with a 24-h period. EB3 is calculated in the same way as EB 1 and is the proportion of activity signal energy located in the period band 4, 5 h > T > 1.25 h; it is related to the pulsatile nature of activity and reflects a tendency for production of short bursts of activity outside the range of circadian components. T~ and Tr are temporo-frequency descriptors (4) which can represent the major portion of the information of the spectrum by using only these two measures. This procedure, used in EEG analysis (11), is designed to obtain as simple a characterization as possible of what is an already complex spectrum (2 indices). These two indices correspond to periods and are measured in hours and calculated as follows:
where F is the maximum frequency in the spectrum. T¢, the spectral center is visible as the mean period of the spectrum. The fewer the high frequencies in the spectrum (corresponding to bursts of pulsatile or erratic activity) the greater the value of Tc approaching 24 h. Tr, the spectral range, appears
The means for the raw data appear in Table 1 that also presents the outcome of a 2-way analysis of variance where Exp covers the four conditions--normal N, repeated experiment R, food deprivation condition D, and isolation condition I, and Strain defines the strains: C and B6. The ANOVA was calculated either on the raw data or on data which had been functionally transformed for homoscedasticity with power and logarithmic functions. These techniques are customary in statistics and applied frequently in genetics (13). The results show a significant Strain effect for all parameters except EB3; this was expected since these parameters were chosen because they are informative for genetic analysis. No Exp effect was observed for Rmsf, EBI, or Tr. There is a main Exp effect for Rmst, D/L, and EB3, and a Strain. Exp interaction for T¢. This significant interaction effect for Tc calls for a separate strain-by-strain analysis. The results appear in Table 2. No significant effects for strain were observed. Inspection of the raw data for the three parameters exhibiting an Exp effect (Rmst, D/L, and EB3) suggests a main effect for isolation which is partially confirmed by the statistical analysis. First, an ANOVA excluding I (Table 3) indicates that Rmst and EB3 form a homogeneous group for conditions N, R, and D; second, partial t-test comparisons show that only group I differs significantly from group N [respectively Rmst, t (42) = 2.09, p < 0.05; EB3, t (42) = 2.11, p < 0.05]. Even without group I,
234
BF,\I
RAW D A I A
Raw Data Strain group C N R D 1 B6 N R D 1 Statistical analysis Factor df Exp 3 Strain 1 Exp • Strain 3
1 ABI.E 1 AND STATISTIC&I AN&LYSIS
D/I.
Rms~
Rms~
EBI
EB3
1,-
1.88_+0.15 1.93 _+ 0.26 1.26 _+ 0.24 1.34+0,26
197.97_+ 1 8 . 3 253.32 _+ 32.1 212.40 _+ 28.4 238.74_+26.0
111.72_+ 11.1 141.97 -+ 17.4 119.70 _+ 20.8 133.71 + 1 4 . 7
24.21 _+2.26 21.65 _+ 2.83 33.34 _-+6.97 23.30+3.13
37.31 _+2.17 40.79 _+ 3.5 28.45 +_ 4.6 28.05+ 1.93
3.27+0.12 3.31 _+ 0.09 3.91 _+ 0.30 3.73 +(t.15
1.78_+0.03 1.80 -+ 0.05 1.78 _+_0.02 1.82 ±(I.05
3.86_+0,26 2.81_+0,19 2.93_+0,21 2.91 -+0.21
119.92_+ 122.42_+ 135.72_+ 176.22_+
69.77_+ 7.2 68.29_+ 2.0 77.73_+ 4.8 96.04_+ 9.4
18.87+ 1 . 7 2 14.09_+2.16 21.78+3.04 17.57 + 1 . 1 5
35.74_+ 1 . 9 9 39.24+4.1 40.66_+2.3 33.63_+2.97
2.93_+ 1.0 2.75_+0.16 2.53_+(I.13 3.07_+0.15
1.72_+(I.02 1.74+0.07 1.56.0.04 1.71 _+0.05
2.25 8.88 $ .22
2.97 * 2.73 2.22
1.83 38.05 t 3.32*
2.4 I 14.61 $ 1.49
2.92" 52.38 # 1.21
10.3 4.3 6.2 2.3
3.22" 33.53 t 1.31
2.17 26.03 t .93
The raw data values of rhythm parameters indicated in the table are the means values with standard error. C designates strain BALB/c, B6 designates strain C57BL/6. N designates the normal condition tested on 16 subjects. R designates the repeated condition where five subjects were retested. D designates the food-deprivation condition tested on six subjects. I designates the isolation condition where six subjects were isolated prior to testing. Statistical analysis is a 2-way ANOVA on parameters defining rhythm of activity. Exp is the experiment factor covering groups N, R, D, and I. Strain covers strains C and B6. Exp. Strain represents the interaction, df indicates the degrees of freedom. F the Fisher test and the associated level of significance is: Significant at p < *0.05:-t0.0001: :~0.01.
D / L still has a significant effect that could be predicted from the values in Table 1 since the normal situation tends to dichotomize. DISCUSSION The study of parameters characterizing r h y t h m of activity in strains C a n d B6 indicates that the parameters measured here are relatively insensitive to experimentally induced environmental variations but nevertheless form a hierarchy (see Table 4). However, these environmental variations always remain lower than genetic variations (see "Fables 1 a n d 4). As shown in Table 4, the temporo-frequency descriptors Tc and T r are totally unresponsive to experimental conditions and can, thus, be considered to be robust with respect to experimental protocols. This can be attributed to their m e t h o d of calculation, since they are global measures of the features defining the spectrum. It should be pointed out, however, that the Strain. Exp interaction indicates that Tc undergoes a differential effect according to the strains. The parameters in the frequency d o m a i n are highly insensitive to experimental conditions in contrast to the parameters in the temporal domain. C o m p a r i n g Rmsf a n d Rms~, which b o t h represent a value linked to a m o u n t of activity, indicates that they
TABLE 2 ANALYSIS BY STRAIN FOR T~ Strain
df
F
C B6
3 3
2.1 2.48
Effect of Exp on parameter T~ for conditions N, R, D, and I on strains C and B6; no tests are significant.
differ in sensitivity as a function of the m a n n e r in which the value is calculated: when Rmsf does not include the high frequencies there is less trend towards dispersion. W h e n a m u c h more classic variable associated with a m o u n t of a c t i v i t y - - t h e m e a n v a l u e - - w a s calculated (data not given here), the results are comparable to those obtained for Rmst. EB3, which is a frequency measure located in the upper part of the spectrum, shows a sensitivity to experimental conditions as did Rmst. This sensitivity-high frequency characteristic is an additional indication of the d i s c r i m i n a n t value of the u n t r a d i a n zones. It drops out, however, if the isolation condition is excluded from analysis. Only D / L exhibits an overall sensitivity to experimental conditions.
TABLE 3 EFFECT OF ISOLATION Parameter/Factor Rmst Exp Strain Exp. Strain Exp D/L Strain Exp. Strain EB3 Exp Strain Exp. Strain
d/
F
I
31.30 *
2 2
0.72 3.33t
I
36.6l* .72
1.35
2 2 1
2
.48 .33 3.06
Parameters exhibiting an Exp effect when I (isolation condition) is excluded: *t Significant at *p < 0.001- tp < 0.05.
ACTIVITY R H Y T H M S llI
235
TABLE 4 STABILITY OF RHYTHM PARAMETERS AS A FUNCTION OF DOMAIN (TIME OR FREQUENCY) AND EXPERIMENTAL CONDITION Time Parameters Overall sensitivity to experimental conditions Sensibility to experimental conditions excluding 1
Frequency
Time/ Frequency
D/L
Rms~ Rmsr EB1
EB3
T~
T,
Yes
Yes
No
No
Yes
No
No
Yes
No
No
No
No
No
No
D/L, a highly classical temporal parameter in chronobiological assessment in the synchronic mode, thus appears to be sensitive to the experimental situation and should be considered to be less robust in particular in genetic analysis. In our opinion, this lack of robustness according to environmental conditions is due to phase sensibility of this parameter which was criticized in Beau (6) because of its inefficiency, in particular, to show maternal effect. The insensitivity of the parameters measured here implies that the rhythm characteristics of each strain are stable. First of all, the value of a parameter for a given strain and a given experimental situation is only rarely comparable to the value obtained for the other strain in another situation. Secondly, the strain hierarchy remains constant under the different experi-
mental situations. Each component of rhythm thus remains characteristic of the strain despite modifications of the experimental environment. This is true for even the most arduous situation tested here: 3 days of food deprivation. The results concerning repetition of experiments are useful in cases where repeated measures are needed on rare or hardto-find animals. It is also a good indicator of the absence of a sequential effect for experiments in sequence (an effect which can only make experimental protocols more cumbersome). This latter point is taken up by Possidente and Hegmann (18) in a study on seven inbred strains of mice including C and Bl0 for parameters in free run (period) and in synchro/free run (variation in period). The effect of isolation--which is particularly apparent for the temporal measures and those affecting features linked to a m o u n t of activity--has the greatest impact on the parameters which disrupt rhythm of activity. This point has been made by D'Amato (9) for augmentation of activity on Swiss mice. In summary, the components of activity rhythm in the mouse exhibit variable stability: 1. component values vary with the type of parameter used (or rather, its domain time or frequency); 2. component values vary across experimental conditions; 3. component values exhibit different trends as a function of strain. This study has demonstrated high stability for the set of parameters associated with rhythm of activity in situations where the experimental environment is modified. Repeated testing, and a more stressful condition, food deprivation, were examined here. Rhythm components, and, in particular, parameters derived from measures in the temporal domain, were also shown to be sensitive to modifications which touch on the social environment, such as prior isolation of animals.
REFERENCES 1. Beau, J. Microcomputer-driven digital data processing ofmouse actograph activity. Physiol. Behav. 3:435-441 ; 1986. 2. Beau, J. Components comparison of activity rhythm in three inbred mice strains: In: Medioni, J.; Vaysse, G., eds. Genetic approaches to behaviour. Toulouse: Privat IEC; 1986:19-27. 3. Beau, J. Spline interpolation of the Fast Fourier transform of a biological rhythm: Determination of the spectral components. J. Interdiscipl. Cycle Res. 19:129-140; 1988. 4. Beau, J. Calculation and properties of two temporal and frequency descriptors: The spectral center and the spectral range. J. Interdisciplin. Cycle Res. 19:141-152; 1988. 5. Beau, J. Activity rhythms in inbred mice I: Genetic analysis with recombinant inbred strains. Behav. Genet. 21 (2): 117-129; 1991. 6. Beau, J. Fitting a biological rhythm to a synchronizer: Applied illustration of mouse activity rhythm in genetic analysis. J. lnterdiscipl. Cycle Res. 22(3):271-280; 1991. 7. Chossat, C. Recherche experimentale sur l'inanition. Mem. Acad. R. Sci. Inst. Fr.; 1843:430-640. 8. Daan, S.; Pittendrigh, C. S. A functional analysis of circadian pacemakers in nocturnal rodents: II1 Heavy water and constant light: Homeostasis of frequency? J. Comp. Physiol. 106:267-290; 1976. 9. D'Amato, F. R. Time budgets and behavioural synchronization in aggregated and isolated male and female mice. Behav. Processes 13: 385-397; 1986, 10. Del Pozo, F.; DeFeudis, F. V.; Jimenez, J, M. Motilities of isolated
and aggregated mice: A difference in ultradian rhythmicity. Experientia 34( 10):1302-1304:1978. 11. Hjorth, B. EEG analysis based on time domain properties. Electroencephalogr. Clin. Neurophysiol. 29:306-310; 1970. 12. Le Pape, G.: Lassalle, J. M. Activitt~ locomotrice de souris isoll~es, de deux lignFzesconsanguines, dans un environnement semi-naturel ou en cages d¥1evage. Behav. Processes 4:221-230; 1979. 13. Mather, K.; Jinks, J. Biometrical genetics. London: Chapman; 1971. 14. Mather, J. G. Wheel-runningactivity: A new interpretation. Mammal Res. 11:41-51: 1981. 15. Oliverio, A.: Malorni, W. Wheel running and sleep in two strains of mice: Plasticity and rigidity in the expression of circadian rhythmicity. Brain Res. 163:121-133; 1979. 16. Pittendrigh, C. S. Circadian rhythms and the circadian organization of living systems. Cold Spring Harbor Syrup. Q. Biol. 25:159-184; 1960. 17. Pittendrigh, C. S.; Dann, S. A functional analysis of circadian pacemakers in nocturnal rodents: ! The stability and lability of spontaneous frequency. J. Comp. Physiol. 106:223-252; 1976. 18. Possidente, B.; Hegmann, J. Gene differences modify AschoWs rule in mice. Physiol. Behav. 28(1):199-200; 1982. 19. Pratt, B. L.; Goldman, B. D. Environmental influence upon circadian periodicity of Syrian hamsters. Physiol. Behav. 36:91-95; 1986. 20. Reinberg, A.; Levi, F.; Debry, G. Chronobiologie et nutrition. Encycl. Med. Chir. 10390 AI0:I-10: 1984. 21, Webb, W. B.; Friedman, J. Attempts to modify the sleep pattern of the rat. Physiol. Behav. 6(4):459-460; 1971.
