- Email: [email protected]
Effects of meal timing on tumor progression in mice
Life Sciences 75 (2004) 1181 – 1193 www.elsevier.com/locate/lifescie
Effects of meal timing on tumor progression in mice M.W. Wu a,b, X.M. Li a, L.J. Xian b, F. Le´vi a,* a
INSERM E 0354 ‘‘Chronothe´rapeutique des cancers’’, Hoˆpital Paul Brousse and Universite´ Paris XI, 94807 Villejuif Cedex, France b Cancer Center, Sun Yat-sen University, Guangzhou, China Received 2 December 2003; accepted 9 February 2004
Abstract Meal timing can reset circadian clocks in peripheral tissues. We investigated the effects of such non– photic entrainment on tumor growth rate. Two experiments involved a total of 61 male B6D2F1 mice synchronized with an alternation of 12 h of light (L) and 12 h of darkness (D) (LD12:12). Mice were randomly allocated to have access to food ad libitum, or restricted to 4 or 6 h during L or D. Rest-activity and body temperature, two circadian outputs, were monitored with an intra-peritoneal sensor. Glasgow osteosarcoma was inoculated into both flanks of each mouse ten days after meal timing onset. Before tumor inoculation, meal timing during D amplified the 24-h rhythms in rest-activity and body temperature with minimal phase alteration as compared to ad libitum feeding. Conversely, meal timing during L induced dominant 12-h or 8-h rhythmic components in activity, nearly doubled the 24-h amplitude of body temperature and shifted its acrophase (time of maximum) from f mid-D to f mid-L. Thirteen days after tumor inoculation, mean tumor weight ( F SEM, mg) was 1503 F 150 in ad libitum mice, 1077 F 157 in mice fed during D and 577 F 139 in mice fed during L (ANOVA, p < 0.0001). Overall survival was prolonged in the mice fed during L (median, 17.5 days, d) as compared with those fed during D (14.5 d) or ad libitum (14 d) (Log Rank, p = 0.0035). The internal desynchronization produced by meal timing during L slowed down tumor progression, an effect possibly resulting from improved host-mediated tumor control and/or altered tumor circadian clocks. D 2004 Elsevier Inc. All rights reserved. Keywords: Circadian rhythms; Tumor progression; Feeding schedule; Entrainment; Mouse
Introduction Tumor growth was reported to be hampered by underfeeding in mice with transplanted tumors nearly 100 years ago (reviewed in Kritchevsky, 2001). Thirty years later, food restriction was * Corresponding author. INSERM E 0354, Hoˆpital. Paul Brousse, 14 avenue P.V. Couturier, 94807 Villejuif Cedex, France. Tel.: +33-1-45-59-38-55; fax: +33-1-45-59-36-02. E-mail address: [email protected] (F. Le´vi). 0024-3205/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.lfs.2004.02.014
1182
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
proposed to lower the incidence of spontaneous tumors in rats (McCay et al., 1935, 1939), a finding subsequently confirmed by other investigators (Nelson and Halberg, 1986) and extended to chemically-induced tumors as well (Pollard et al., 1984; Roebuck et al., 1993). Food restriction was also shown to slow down the growth of transplanted tumors (Siegel et al., 1988; Hursting et al., 1994), with low calorie intake rather than vitamin or mineral deficiencies appearing as a critical factor (Pariza, 1986). Several mechanisms have been proposed as accounting for the tumor growth inhibition produced by calorie restriction, a dietary regimen low in calories yet not causing undernutrition (Koubova and Guarente, 2003). These include alterations of the secretory patterns of glucocorticoids, mammotrophic hormones or insulin, as well as reduced growth factor responsiveness of malignant cells or decreased malignant cell proliferation (Lakatua et al., 1983; Scheving et al., 1983; Hodgson et al., 1997). Nevertheless, most of these studies have not carefully attempted to differentiate the effect of calorie restriction from that of circadian entrainment by meal timing, as restricted feeding usually consisted of a limitation of the access to food to a few hours during the light span (Scheving et al., 1983; Hotz et al., 1987; Duffy et al., 1990). Mammalian circadian rhythms result from interlocked molecular loops involving specific genes, which have been found in almost all cells. These cellular clocks are coordinated at the organism’s level by the suprachiasmatic nuclei (SCN), a central pacemaker located in the hypothalamus. The SCN further helps the adjustment of the circadian time structure to the regular alternation of light and darkness over 24 hours, the main synchronizer of the endogenous circadian rhythms (Reppert and Weaver, 2002; Schibler and Sassone-Corsi, 2002). Nevertheless, the restriction of food availability to 4 hours during the light span was shown to entrain the rhythm in clock gene mPer1 expression in liver kidney, heart, pancreas, skeletal muscle and lung, while no such effect was found in the SCN (Damiola et al., 2000; Hara et al., 2001; Stokkan et al., 2001; Rutter et al., 2002). Thus, whereas peripheral clocks were synchronized with restricted feeding during light, a time when the animals normally rest, the central pacemaker remained entrained to the light-dark schedule. This resulted in internal desynchronization, as the physiological relation between several endogenous rhythms were lost (Hara et al., 2001). Recently, our team reported that malignant growth was accelerated by the disruption of the circadian coordination, which resulted from SCN destruction or repeat alterations of the light-dark schedule (Filipski et al., 2002, 2003). These findings have indicated that the circadian clock was a control point in tumor growth. They have called for investigating whether other modifications of circadian entrainment could rather slow down cancer progression. Here, we sought whether restricted feeding could alter the growth rate of a transplanted tumor in mice synchronized with an alternation of 12 h of light and 12 h of darkness, as a function of meal timing.
Materials and methods Animals and synchronization Two experiments (Exp.) involved male B6D2F1 mice (male DBA/2 female C57BL/6, Charles river, L’Arbresle, France). All the mice were singly housed and kept in an autonomous chronobi-
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
1183
ologic facility which is equipped with temperature control (23 F 1 jC) and comprises 6 compartments, each one being provided with filtrated air (100 F 10 l/min) and separate lighting regimens (Jouan, Saint-Herblain, France). The mice were synchronized with an alternation of 12 h of light (L) and 12 h of darkness (D) (LD 12:12) with water ad libitum. In Exp. 1, 28 male mice, aged 6 weeks, were randomly allocated to one of three groups, group A: food ad libitum (n = 8) with L, 9:00–21:00; group B: meal timing during L (MTL, n = 9) with L, 9:00–21:00; group C: meal timing during D (MTD, n = 11) with L, 21:00–9:00. In Exp. 2, 33 mice aged 4 to 5 weeks were randomly allocated to one of four groups, A: food ad libitum (n = 10) with L, 9:00–21:00; B: MTL (n = 7) with L, 9:00–21:00; C: MTD (n = 8) with L, 21:00–9:00; D: food ad libitum (n = 8) with L, 21:00–9:00 (Table 1). Two weeks following LD synchronization onset, food (A04–10, UAR, Villemoisson-Sur-Orge, France) was made available at specific daily times for both meal-fed groups. Feeding time was from 2 to 6 Hours After Light Onset - HALO - for group B (MTL) or from 14 to 18 HALO for group C (MTD) in exp. 1. For Exp. 2, feeding times were from 2 to 8 HALO for group B (MTL) or from 14 to 20 HALO for group C (MTD). Body temperature and locomotor activity monitoring In Exp. 2, an intraperitoneal sensor was implanted into 17 mice for continuous monitoring of body temperature and rest-activity. The procedure was performed 7 days before starting food restriction. The animals were anesthetized for 30 min with intraperitoneal injection of tribromoethanol (Fluka, Saint Quentin Fallavier Fallavier, France: 10 mg per mouse). A 2 cm-incision was made in the skin and the peritoneum, the telemetry sensor (PhysiolTel, TA10TA-F20, Data Sciences Inc., MN, USA) was implanted into the abdominal cavity and the abdominal wall sutured. After implantation, calibration values of each transmitter were entered into the DataquestR A.R.T. data acquisition system. Signals emanation from each transmitter were received by an antenna mounted on a receiver board (model RLA 1000, Data Sciences International) placed below the animal’s cage. Intraperitoneal temperature and locomotor activity of each animal were automatically monitored every 10 min during the entire study. Data were processed by a PC equipped with a dedicated recording and analysis system (DataquestR A.R.T. version 2.3, Data Sciences International). Table 1 Synchronization with feeding and photoperiodic schedules Experiment 1
2
Group A B C A B C D
Feeding schedule Ad libitum MTD MTL Ad libitum MTD MTL Ad libitum
Meal Timing during Darkness (MTD) or during Light (MTL). N.A., not applicable.
Feeding times (HALO)
Lighting times (clock hours)
Start
Onset
Offset
09:00 09:00 21:00 09:00 09:00 21:00 21:00
21:00 21:00 09:00 21:00 21:00 09:00 09:00
N.A. 02:00 14:00 N.A. 02:00 14:00 N.A.
Finish 06:00 18:00 08:00 20:00
1184
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
Body weight In Exp 1, body weight was measured both at 2 and 6 HALO for ad libitum mice and for MTL mice and both at 14 and 18 HALO for ad libitum mice and for MTD mice. In Exp. 2, body weight was measured at 2 and 8 HALO in ad libitum mice (group A) and in MTL mice and at 14 and 20 HALO in MTD mice (group C) and in mice fed ad libitum (group D). Tumor Glasgow osteosarcoma was obtained from Aventis Pharma S.A and maintained in C57BL/6 mice then transplanted into male B6D2F1 mice for the purpose of the experiment. The tumor was dissected from a single donor male B6D2F1 mouse for each experiment, and 3 3 3 mm3 tumor fragments were prepared and suspended in Hank’s buffer at 4 jC. A tumor fragment was implanted subcutaneously into both flanks of each mouse using a 12-gauge trocar within the 2 h following sampling. The tumor was inoculated to all the mice 10 days after food restriction (day 0). Two perpendicular diameters (mm) of each tumor were measured daily with a caliper. Tumor weight (mg) was computed as: tumor weight = (length width2)/2. Each mouse with tumor weight reaching 10% of the mouse body weight ( f 2500 mg) along the course of the study was sacrificed for ethical reasons and was considered as dead from tumor progression on this date (Granda et al., 2002). Tumor weight and mortality were recorded daily until the end of each experiment. Statistical analyses Temperature (jC) and activity data (arbitrary units) of each mouse were converted into ASCII format. Data were analyzed with cosinor analysis for periods of 24, 12 or 8 h (Nelson et al., 1979; Filipski et al., 2002). Mean and SEM were calculated for body weight and tumor weight and differences between groups or sampling days were analyzed by ANOVA. Survival data were analyzed with Log rank using the SPSS 11.5 software.
Results Body temperature and locomotor activity rhythms Both variables displayed similar circadian rhythms prior to the onset of food restriction in all three groups. A striking decrease in nadir body temperature was found over the initial 5 days of meal timing in the mice fed during L, which was not the case for mice fed during D. Activity and temperature data of each mouse were double plotted throughout the entire experimental span (Fig. 1) and averaged over the 3-day span preceding tumor inoculation (Fig. 2). These graphic displays clearly indicated that both activity and temperature patterns were altered with meal timing. Both MTD and MTL produced a marked activity bout near feeding onset, short after the beginning of darkness or light, respectively. Yet, a lower activity bout occurred after the end of the feeding span in the MTD mice, whereas some level of activity persisted after feeding and during D in the
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
1185
Fig. 1. Double-plotted representation of locomotor activity (left panel) and body temperature (right panel) in a mouse fed ad libitum, a mouse meal-fed during darkness (MTD) and a mouse meal-fed during light (MTL) in experiment 2. Two consecutive days are shown on the x-axis to visualize the regular alternation of high and low values of each variable from one day to the next. Open and black boxes in horizontal axis respectively represent 12 h of light and 12 h of darkness for photoperiodic synchronization. Time is expressed in hours after light onset (HALO). Shaded boxes represent times of food availability. The feeding condition is shown on the x-axis as follow: ad libitum food (top panel); MTD = meal timing during darkness (from 14 to 20 HALO, middle panel); MTL = meal timing during light (from 2 to 8 HALO, bottom panel).
1186
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
Fig. 2. Time course of locomotor activity (left panel) and body temperature (right panel) in ad libitum fed mice (top panel), in MTL mice (middle panel) or in MTD animals (bottom panel). Figure displays mean circadian pattern of each variable over 3 days before tumor inoculation. Each point represents mean F (SEM) of 3-5 animals. Shaded boxes represent time of food availability.
MTL animals. The body temperature pattern was reinforced by MTD, yet its phase was similar to that observed in ad libitum condition. Conversely, high temperature values were observed during L rather than during D, in MTL mice.
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
1187
Cosinor analysis of these data revealed that the acrophase of both variables were similar in ad libitum and in MTD mice, yet the circadian amplitude of both rhythms were significantly increased with meal timing during D as compared to ad libitum. Conversely MTL altered the 24-h period of the rest-activity cycle resulting in dominant 12-h or 8-h periodic components in activity (splitting). In these animals, the amplitude of the body temperature rhythm was nearly doubled and its acrophase was shifted from f 17 HALO to f 8 HALO. Quite interestingly, the amplitude increase was greater with MTL as compared to MTD (Table 2). Body weight change In both experiments, the body weight of the ad libitum mice remained stable over the time span preceding tumor inoculation, reaching an average value ( F SEM) of 27.2 F 0.2 g in Exp. 1 and to 26.6 F 0.4 g in Exp. 2. Conversely, all the meal-fed mice lost weight during the initial 5 days of meal timing. The mice subsequently increased the daily amount of ingested food up to 6 g, so that they maintained a stable or gradually increasing body weight. Thus, over the 4-day span preceding tumor inoculation, the average body weight gain during meal feeding was similar in MTD and in MTL mice (4.1 F 0.1% vs 4.4 F 0.2%, p from ANOVA = 0.20). On the day preceding tumor inoculation, mean body weight before feeding was 23.2 F 0.3 g for MTD mice and 21.3 F 0.4 g for MTL animals (p from ANOVA < 0.001). Thus, as compared to ad libitum fed mice, the MTD animals lost on average 10.7 to 16.6% of their body weight over 10 days, pending on experiment. The mean relative body weight loss of the MTL mice ranged from 19.4 to 21.3% (Fig. 3). Tumor weight In Exp. 1, tumor became palpable in 7 of 8 mice fed ad libitum eight days after tumor inoculation (87.5%). This was the case for 8 of 11 mice fed during D (72.7%) and for 4 of 9 mice fed during L (44.4%). Six days later (day 14), two ad libitum mice and one MTD mouse died from tumor progression Table 2 Mean circadian parameters F SEM or ultradian parameters of body temperature and locomotor activity in Exp. 2 Variable
Group
Period (H in h)
Amplitude F SEM (jC)
Acrophase (HALO) F SEM (min)
p
Body temperature
Ad libitum MTD MTL Ad libitum MTD MTL # 1 #2 #3
24 24 24 24 24 12 12 8
0.8 1.6 2.1 2.6 3.9 3.4 4.3 3.3
17:20 F 10 18:26 F 8 7:54 F 24 17:16 F 8 15:39 F 31 2:24 and 14:24 2:08 and 14:08 2:05, 10:05 and 18:05
< 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 0.003 < 0.0001 < 0.0001
Activity
F F F F F
0.1 0.2 0.12 0.2 1.4
Individual parameters were obtained with single cosinor analysis of data collected over the 3-day span preceding tumor inoculation. #1, 2 or 3 mean different individuals, with corresponding dominant periods.
1188
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
Fig. 3. Body weight ( F SEM) on the day preceding tumor inoculation, as a function of feeding schedule. Results from Exp. 1 (upper panel) and Exp. 2 (lower panel). Body weight measured at 2 HALO in ad libitum fed mice or before feeding in the MTL group (left panels) and at 14 HALO in ad libitum fed mice or before feeding in the MTD group (right panels).
as tumor weight reached 2500 mg. On this day, mean tumor weight ( F SEM) was 1432 F 421 mg in controls, 1140 F 224 mg in MTD mice and 280 F 96 mg in MTL animals (ANOVA, p < 0.0001). Thus tumor grew more slowly in the meal-fed mice as compared to the ad libitum ones (ANOVA, p < 0.0001), and even more so in the MTL animals as compared to the MTD ones (ANOVA, p < 0.0001). In Exp. 2, tumor growth was similar in both ad libitum groups (A and D) despite light onset occurred 12 h apart (ANOVA, p = 0.7). Thus, the data of both groups were pooled and considered as a single
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
1189
control group. Thirteen days after inoculation, the tumor weight of three ad libitum fed mice reached 2500 mg. On this day, mean tumor weight ( F SEM) was 1795 F 660 mg in controls, 1616 F 528 mg in MTD mice and 1114 F 368 mg in MTL animals (ANOVA, p < 0.0001). The inspection of the tumor growth curves of pooled experiments indicated that meal timing slowed down tumor growth, and even more so if it occurred during L. Thirteen days after tumor inoculation, a time when the first mouse died from tumor progression, mean tumor weight ( F SEM) was 1503 F 150 mg in controls, 1077 F 157 mg in MTD and 577 F 139 mg in MTL (2-way ANOVA: Feeding schedule, p < 0.001; experiment, p < 0.001; interaction, p = 0.99). The overall differences between the tumor growth curves were statistically validated with ANOVA for repeated measures (ANOVA, p < 0.0001) (Fig. 4). Survival In Exp. 1, mice died from tumor progression 13 to 22 days after tumor inoculation in controls, at 13 to 18 days in MTD mice and at 17 to 26 days in MTL. The latter animals survived longer than ad libitum or MTD mice. The survival curves among the 3 groups differed with statistical significance (Log rank, p < 0.0001). In Exp. 2, tumor progression resulted in mouse death 12 to 15 days after inoculation in controls, 13 to 16 days in MTD and 14 to 17 days in MTL. The differences in survival curves were close to statistical significance, with MTL mice surviving longer than ad libitum or MTD animals (Log rank, p = 0.10).
Fig. 4. Tumor growth as a function of feeding schedule. Each daily point is mean tumor weight ( F SEM) of 16 or 19 mice (MTL and MTD, respectively) or ad libitum (n = 26). Pooled data from two experiments.
1190
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
Fig. 5. Survival curves of tumor-bearing mice as a function of feeding schedule in both experiments.
Survival data from pooled experiments further showed that survival was increased in MTL animals as compared to ad libitum or MTD mice. The survival curves differed with a high level of statistical significance (Log rank, p = 0.0035) (Fig. 5). Relation between body weight and tumor growth No significant correlation was found between body weight on the day preceding tumor inoculation and tumor weight on day 13 in each experiment for ad libitum mice (Exp. 1, r = 0.34, p = 0.41; Exp. 2, r = 0.25, p = 0.31), for MTD animals (Exp. 1, r = 0.18, p = 0.60; Exp. 2, r = 0.13, p = 0.76) or for MTD mice (Exp. 1, r = 0.15, p = 0.71; Exp. 2, r = 0.70, p = 0.08). Neither any significant correlation between both variables was found in pooled data from both experiments (ad libitum, r = 0.31, p = 0.12; MTL, r = 0.30, p = 0.26; MTD, r = 0.25, p = 0.30).
Discussion Both body temperature and rest-activity circadian rhythm were modified by restricting the feeding span, yet these alterations differed as a function of meal timing. The average circadian amplitude of body temperature was increased 1.9 fold MTD and 2.6 fold by MTL. MTD also enhanced the 24-h amplitude in activity with minimal alteration of its phase. These findings support a reinforcement of circadian coordination by MTD. In contrast, MTL also shifted the acrophase of body temperature by about 10 h and split the rest-activity rhythm. Such MTL-induced desynchronization was reported by several investigators. For instance, the 24-h rhythm in DNA synthesis in the tongue of ad libitum fed mice
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
1191
became bimodal following MTL. A first peak appeared to be cued by food presentation and the other one by the onset of D (Scheving et al., 1983). A 4 h span has been showed to be the lowest duration of food availability for mice to maintain their basic physiological function and metabolism (Hodgson et al., 1997). In a preliminary experiment (data not shown), we confirmed that meal feeding for 4 h during L or D resulted in a steady decrease in body weight over the initial 5 days, yet body weight remained stable thereafter as the mice learned how to increase the amount of ingested food during this limited feeding time. This observation was in line with earlier reports (Nelson and Halberg, 1986). Similar results were obtained in the current study, although the magnitude of the body weight loss produced by MTL slightly exceeded that resulting from MTD. Nevertheless, the mice on either feeding schedule ingested a similar estimated amount of food over the feeding time span. Meal feeding during L resulted in mean tumor weight being smaller by 5.1 fold in Exp. 1 and by 1.6 fold in Exp. 2 as compared to corresponding controls. These differences could relate to differences in mouse age (Exp.1, 6 weeks; Exp. 2, 4–5 weeks at reception), season of study (Exp. 1, winter; Exp. 2, summer) or meal feeding duration (Exp. 1, 4 h; Exp. 2, 6 h). Although tumor growth significantly faster in Exp. 1 as compared to Exp. 2, the effect of feeding schedule was consistent in both experiments (no schedule x Exp interaction at 2-way ANOVA). Meal feeding during D had little inhibitory effect on tumor growth and no effect on survival. Conversely, meal feeding during L largely and significantly slowed down malignant growth and prolonged survival in tumor-bearing mice, as compared with animals fed ad libitum or during D. The meal timing-dependent effect on tumor progression might result from a more pronounced underfeeding produced by MTL as compared to MTD, as calorie restriction was shown to slow down the growth of transplanted tumors (Hodgson et al., 1997; Kritchevsky, 2001). In the current study, the average body weight loss of MTL mice only exceeded that of MTD animals by 1.3 g in Exp. 1 and by 2.2 g in Exp. 2. For comparison, the average difference between ad libitum mice and MTD animals was 5 g in Exp. 1 and 3 g in Exp. 2. Thus the body weight loss produced by food restriction itself was much larger than that resulting from meal timing during L as compared to D. Furthermore, no significant correlation was found between individual body weight on the day preceding tumor inoculation and subsequent tumor growth. Although we cannot eliminate the possibility that losing body weight below a precise threshold may be needed for tumor growth to be slowed down by calorie restriction, the most likely explanation for the current results lies in the effects of meal timing on circadian rhythms. In the present study, MTL produced discrepant effects on mouse rest-activity and body temperature circadian rhythms. Indeed, this feeding schedule was shown to be able to shift by 10 h the circadian phase of peripheral oscillators, such as liver, colon, etc, without affecting that of the central pacemaker in rodents (Damiola et al., 2000; Hara et al., 2001; Stokkan et al., 2001; Rutter et al., 2002). Similarly, MTL advanced the circadian acrophases of body temperature, plasma insulin, serum glucose, liver glycogen or serum thyroxine (T4), while the rhythm in circulating white blood cell counts remained synchronized to the photoperiodic regimen (Nelson et al., 1975; Challet et al., 1997). A relation between meal timing and tumor growth regulation is further supported by the demonstration that MTL did modify relevant several host rhythms, such as hormonal secretions or DNA synthesis in healthy mouse tissues (Pauly et al., 1976, Lakatua et al., 1983). For instance, DNA synthesis in mouse colon mucosa was entrained by the feeding schedule rather than by the lighting regimen (Lakatua et al., 1983), a finding similar to that found for DNA synthesis in mouse tongue (Scheving et al., 1983). Conversely, the rhythm in tumor DNA synthesis remained similar in ad libitum or MTL mice (Lakatua et
1192
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
al., 1983). These observations suggest that meal timing can exert a differential resetting effect on healthy and malignant peripheral oscillators. Many cellular rhythmic processes including cell proliferation are controlled by clock genes (Reppert and Weaver, 2002; Schibler and Sassone-Corsi, 2002; Matsuo et al., 2003). Clock gene mPer2 was shown to exert tumor suppressor like properties (Fu et al., 2002; Fu and Lee, 2003). Malignant growth was accelerated in mice with ablated SCN or jet-lag induced alteration of clock gene transcription rhythms both in liver and in tumor (Filipski et al., 2002, 2003). The current findings suggest that meal timing during L may directly or indirectly modify circadian clock function or signaling pathways within host peripheral tissues and/or tumor cells, so that tumor growth is slowed down. We hypothesize that meal timing during L amplifies host rhythms and assigns their peak in a time window when the tumor is most susceptible to host-mediated control. Feeding schedules could then be used as an adjunct to further improve the benefit derived from circadian timing of cancer chemotherapy (Le´vi et al., 1997; Le´vi, 2002; Mormont and Le´vi, 2003).
Acknowledgements Program supported by: Programme de Recherches Avance´es de Coope´ration Franco-Chinoises (grant to M.W. W.), INSERM Action The´matique Concerte´e Nutrition and Association Internationale pour la Recherche sur le Temps Biologique et la Chronothe´rapie (ARTBC), Hoˆpital Paul Brousse, Villejuif (France) Clinical Key Program of the Ministry of Public Health of P.R. China (No. 20014058).
References Challet, E., Pevet, P., Vivien-Roels, B., Malan, A., 1997. Phase-advanced daily rhythms of melatonin, body temperature, and locomotor activity in food-restricted rats fed during daytime. Journal of Biological Rhythms 12 (1), 65 – 79. Damiola, F., Le Minh, N., Preitner, N., Kornmann, B., Fleury-Olela, F., Schibler, U., 2000. Restricted feeding uncouples circadian oscillators in peripheral tissues from the central pacemaker in the suprachiasmatic nucleus. Genes Development 14 (23), 2950 – 2961. Duffy, P.H., Feuers, R., Nakamura, K.D., Leakey, J., Hart, R.W., 1990. Effect of chronic caloric restriction on the synchronization of various physiological measures in old female Fischer 344 rats. Chronobiology International 7 (2), 113 – 124. Filipski, E., King, V.M., Li, X.M., Granda, T.G., Mormont, M.C., Liu, X.H., Claustrat, B., Hastings, M.H., Levi, F., 2002. Host circadian clock as a control point in tumor progression. Journal of the National Cancer Institute 94 (9), 690 – 697. Filipski, E., Delaunay, F., Greche-Cassiau, A., Li, X.M., King, V.M., Claustrat, B., Jasmin, C., Hastings, M.H., Levi, F., 2003. Chronic jet lag alters circadian genes expression and accelerates malignant growth in tumor-bearing mice. The 94th annual meeting of American Association for Cancer Research, Washington, D.C., USA, July 11 – 14, 2003. 2463 abs. Fu, L., Pelicano, H., Liu, J., Huang, P., Lee, C., 2002. The circadian gene Period2 plays an important role in tumor suppression and DNA damage response in vivo. Cell 111 (1), 41 – 50. Fu, L., Lee, C.C., 2003. The circadian clock: pacemaker and tumour suppressor. Nature Reviews. Cancer 3 (5), 350 – 361. Granda, T.G., D’Attino, R.M., Filipski, E., Vrignaud, P., Garufi, C., Terzoli, E., Bissery, M.C., Levi, F., 2002. Circadian optimisation of irinotecan and oxaliplatin efficacy in mice with Glasgow osteosarcoma. British Journal of Cancer 86 (6), 999 – 1005. Hara, R., Wan, K., Wakamatsu, H., Aida, R., Moriya, T., Akiyama, M., Shibata, S., 2001. Restricted feeding entrains liver clock without participation of the suprachiasmatic nucleus. Genes to Cells 6 (3), 269 – 278. Hodgson, D.M., Chiappelli, F., Morrow, N.S., Taylor, A.N., 1997. Chronic dietary restriction influences tumor metastasis in the rat: parametric considerations. Nutrition and Cancer 28 (2), 189 – 198.
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
1193
Hotz, M.M., Connolly, M.S., Lynch, C.B., 1987. Adaptation to daily meal-timing and its effect on circadian temperature rhythms in two inbred strains of mice. Behavior Genetics 17 (1), 37 – 51. Hursting, S.D., Perkins, S.N., Phang, J.M., 1994. Calorie restriction delays spontaneous tumorigenesis in p53-knockout transgenic mice. Proceeding of the National Academy of Sciences U S A 91 (15), 7036 – 7040. Koubova, J., Guarente, L., 2003. How does calorie restriction work?. Genes and development 17, 313 – 321. Kritchevsky, D., 2001. Caloric restriction and cancer. Journal of Nutritional Science and Vitaminology 47 (1), 13 – 19. Lakatua, D.J., White, M., Sackett-Lundeen, L.L., Haus, E., 1983. Change in phase relations of circadian rhythms in cell proliferation induced by time-limited feeding in BALB/c X DBA/2F1 mice bearing a transplantable Harding-Passey tumor. Cancer Research 43 (9), 4068 – 4072. Le´vi, F., Zidani, R., Misset, J.L., 1997. Randomized multicenter trial of chronotherapy with oxaliplatin, fluorouracil, and folinic acid in metastatic colorectal cancer. International Organization for Cancer Chronotherapy. Lancet 350 (9079), 681 – 686. Le´vi, F., 2002. From circadian rhythms to cancer chronotherapeutics. In: Smolensky, M., Rensing, L. (Eds.), Chronobiology International. Special issue on cancer Chronotherapeutics, vol. 19. Marcel dekker Inc., New York, pp. 1 – 19. McCay, C.M., Crowell, M.F., Maynard, L.A., 1935. The effect of retarded growth upon the length of life span and upon the ultimate body size. Journal Nutrition 10, 63 – 79. McCay, C.M., Maynard, L.A., Sperling, G., Barnes, L.L., 1939. Retarded growth, life span, ultimate body size and age changes in the albino rat after feeding diets restricted in calories. Journal of Nutrition 18, 1 – 13. Matsuo, T., Yamaguchi, S., Mitsui, S., Emi, A., Shimoda, F., Okamura, H., 2003. Control mechanism of the circadian clock for timing of cell division in vivo. Science 302 (5643), 255 – 259. Mormont, M.C., Le´vi, F., 2003. Cancer chronotherapy: principles, applications, and perspectives. Cancer 97 (1), 155 – 169. Nelson, W., Scheving, L., Halberg, F., 1975. Circadian rhythms in mice fed a single daily meal at different stages of lighting regimen. Journal of Nutrition 105 (2), 171 – 184. Nelson, W., Tong, Y., Lee, J.K., Halberg, F., 1979. Methods for cosinor rhythmometry. Chronobiologia 6, 305 – 323. Nelson, W., Halberg, F., 1986. Meal-timing, circadian rhythms and life span of mice. Journal of Nutrition 116 (11), 2244 – 2253. Pariza, M.W., 1986. Calorie restriction, ad libitum feeding, and cancer. Proceedings of the Society for Experimental Biology and Medicine 183 (3), 293 – 298. Pauly, J.E., Scheving, L.E., Burns, E.R., Tsai, T.H., 1976. Circadian rhythm in DNA synthesis in mouse thymus: effect of altered lighting regimens, restricted feeding and presence of Ehrlich ascites tumor. The Anatomical Record 184 (3), 275 – 284. Pollard, M., Luckert, P.H., Pan, G.Y., 1984. Inhibition of intestinal tumorigenesis in methylazoxymethanol-treated rats by dietary restriction. Cancer Treatment Reports 68 (2), 405 – 408. Reppert, S.M., Weaver, D.R., 2002. Coordination of circadian timing in mammals. Nature 418 (6901), 935 – 941. Roebuck, B.D., Baumgartner, K.J., MacMillan, D.L., 1993. Caloric restriction and intervention in pancreatic carcinogenesis in the rat. Cancer Research 53 (1), 46 – 52. Rutter, J., Reick, M., McKnight, S.L., 2002. Metabolism and the control of circadian rhythms. Annual Review of Biochemistry 71, 307 – 331. Scheving, L.E., Tsai, T.H., Scheving, L.A., 1983. Chronobiology of the intestinal tract of the mouse. American Journal of Anatomy 168 (4), 433 – 465. Schibler, U., Sassone-Corsi, P., 2002. A web of circadian pacemakers. Cell 111 (7), 919 – 922. Siegel, I., Liu, T.L., Nepomuceno, N., Gleicher, N., 1988. Effects of short-term dietary restriction on survival of mammary ascites tumor-bearing rats. Cancer Investigation 6 (6), 677 – 680. Stokkan, K.A., Yamazaki, S., Tei, H., Sakaki, Y., Menaker, M., 2001. Entrainment of the circadian clock in the liver by feeding. Science 291 (5503), 490 – 493.
Effects of meal timing on tumor progression in mice M.W. Wu a,b, X.M. Li a, L.J. Xian b, F. Le´vi a,* a
INSERM E 0354 ‘‘Chronothe´rapeutique des cancers’’, Hoˆpital Paul Brousse and Universite´ Paris XI, 94807 Villejuif Cedex, France b Cancer Center, Sun Yat-sen University, Guangzhou, China Received 2 December 2003; accepted 9 February 2004
Abstract Meal timing can reset circadian clocks in peripheral tissues. We investigated the effects of such non– photic entrainment on tumor growth rate. Two experiments involved a total of 61 male B6D2F1 mice synchronized with an alternation of 12 h of light (L) and 12 h of darkness (D) (LD12:12). Mice were randomly allocated to have access to food ad libitum, or restricted to 4 or 6 h during L or D. Rest-activity and body temperature, two circadian outputs, were monitored with an intra-peritoneal sensor. Glasgow osteosarcoma was inoculated into both flanks of each mouse ten days after meal timing onset. Before tumor inoculation, meal timing during D amplified the 24-h rhythms in rest-activity and body temperature with minimal phase alteration as compared to ad libitum feeding. Conversely, meal timing during L induced dominant 12-h or 8-h rhythmic components in activity, nearly doubled the 24-h amplitude of body temperature and shifted its acrophase (time of maximum) from f mid-D to f mid-L. Thirteen days after tumor inoculation, mean tumor weight ( F SEM, mg) was 1503 F 150 in ad libitum mice, 1077 F 157 in mice fed during D and 577 F 139 in mice fed during L (ANOVA, p < 0.0001). Overall survival was prolonged in the mice fed during L (median, 17.5 days, d) as compared with those fed during D (14.5 d) or ad libitum (14 d) (Log Rank, p = 0.0035). The internal desynchronization produced by meal timing during L slowed down tumor progression, an effect possibly resulting from improved host-mediated tumor control and/or altered tumor circadian clocks. D 2004 Elsevier Inc. All rights reserved. Keywords: Circadian rhythms; Tumor progression; Feeding schedule; Entrainment; Mouse
Introduction Tumor growth was reported to be hampered by underfeeding in mice with transplanted tumors nearly 100 years ago (reviewed in Kritchevsky, 2001). Thirty years later, food restriction was * Corresponding author. INSERM E 0354, Hoˆpital. Paul Brousse, 14 avenue P.V. Couturier, 94807 Villejuif Cedex, France. Tel.: +33-1-45-59-38-55; fax: +33-1-45-59-36-02. E-mail address: [email protected] (F. Le´vi). 0024-3205/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.lfs.2004.02.014
1182
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
proposed to lower the incidence of spontaneous tumors in rats (McCay et al., 1935, 1939), a finding subsequently confirmed by other investigators (Nelson and Halberg, 1986) and extended to chemically-induced tumors as well (Pollard et al., 1984; Roebuck et al., 1993). Food restriction was also shown to slow down the growth of transplanted tumors (Siegel et al., 1988; Hursting et al., 1994), with low calorie intake rather than vitamin or mineral deficiencies appearing as a critical factor (Pariza, 1986). Several mechanisms have been proposed as accounting for the tumor growth inhibition produced by calorie restriction, a dietary regimen low in calories yet not causing undernutrition (Koubova and Guarente, 2003). These include alterations of the secretory patterns of glucocorticoids, mammotrophic hormones or insulin, as well as reduced growth factor responsiveness of malignant cells or decreased malignant cell proliferation (Lakatua et al., 1983; Scheving et al., 1983; Hodgson et al., 1997). Nevertheless, most of these studies have not carefully attempted to differentiate the effect of calorie restriction from that of circadian entrainment by meal timing, as restricted feeding usually consisted of a limitation of the access to food to a few hours during the light span (Scheving et al., 1983; Hotz et al., 1987; Duffy et al., 1990). Mammalian circadian rhythms result from interlocked molecular loops involving specific genes, which have been found in almost all cells. These cellular clocks are coordinated at the organism’s level by the suprachiasmatic nuclei (SCN), a central pacemaker located in the hypothalamus. The SCN further helps the adjustment of the circadian time structure to the regular alternation of light and darkness over 24 hours, the main synchronizer of the endogenous circadian rhythms (Reppert and Weaver, 2002; Schibler and Sassone-Corsi, 2002). Nevertheless, the restriction of food availability to 4 hours during the light span was shown to entrain the rhythm in clock gene mPer1 expression in liver kidney, heart, pancreas, skeletal muscle and lung, while no such effect was found in the SCN (Damiola et al., 2000; Hara et al., 2001; Stokkan et al., 2001; Rutter et al., 2002). Thus, whereas peripheral clocks were synchronized with restricted feeding during light, a time when the animals normally rest, the central pacemaker remained entrained to the light-dark schedule. This resulted in internal desynchronization, as the physiological relation between several endogenous rhythms were lost (Hara et al., 2001). Recently, our team reported that malignant growth was accelerated by the disruption of the circadian coordination, which resulted from SCN destruction or repeat alterations of the light-dark schedule (Filipski et al., 2002, 2003). These findings have indicated that the circadian clock was a control point in tumor growth. They have called for investigating whether other modifications of circadian entrainment could rather slow down cancer progression. Here, we sought whether restricted feeding could alter the growth rate of a transplanted tumor in mice synchronized with an alternation of 12 h of light and 12 h of darkness, as a function of meal timing.
Materials and methods Animals and synchronization Two experiments (Exp.) involved male B6D2F1 mice (male DBA/2 female C57BL/6, Charles river, L’Arbresle, France). All the mice were singly housed and kept in an autonomous chronobi-
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
1183
ologic facility which is equipped with temperature control (23 F 1 jC) and comprises 6 compartments, each one being provided with filtrated air (100 F 10 l/min) and separate lighting regimens (Jouan, Saint-Herblain, France). The mice were synchronized with an alternation of 12 h of light (L) and 12 h of darkness (D) (LD 12:12) with water ad libitum. In Exp. 1, 28 male mice, aged 6 weeks, were randomly allocated to one of three groups, group A: food ad libitum (n = 8) with L, 9:00–21:00; group B: meal timing during L (MTL, n = 9) with L, 9:00–21:00; group C: meal timing during D (MTD, n = 11) with L, 21:00–9:00. In Exp. 2, 33 mice aged 4 to 5 weeks were randomly allocated to one of four groups, A: food ad libitum (n = 10) with L, 9:00–21:00; B: MTL (n = 7) with L, 9:00–21:00; C: MTD (n = 8) with L, 21:00–9:00; D: food ad libitum (n = 8) with L, 21:00–9:00 (Table 1). Two weeks following LD synchronization onset, food (A04–10, UAR, Villemoisson-Sur-Orge, France) was made available at specific daily times for both meal-fed groups. Feeding time was from 2 to 6 Hours After Light Onset - HALO - for group B (MTL) or from 14 to 18 HALO for group C (MTD) in exp. 1. For Exp. 2, feeding times were from 2 to 8 HALO for group B (MTL) or from 14 to 20 HALO for group C (MTD). Body temperature and locomotor activity monitoring In Exp. 2, an intraperitoneal sensor was implanted into 17 mice for continuous monitoring of body temperature and rest-activity. The procedure was performed 7 days before starting food restriction. The animals were anesthetized for 30 min with intraperitoneal injection of tribromoethanol (Fluka, Saint Quentin Fallavier Fallavier, France: 10 mg per mouse). A 2 cm-incision was made in the skin and the peritoneum, the telemetry sensor (PhysiolTel, TA10TA-F20, Data Sciences Inc., MN, USA) was implanted into the abdominal cavity and the abdominal wall sutured. After implantation, calibration values of each transmitter were entered into the DataquestR A.R.T. data acquisition system. Signals emanation from each transmitter were received by an antenna mounted on a receiver board (model RLA 1000, Data Sciences International) placed below the animal’s cage. Intraperitoneal temperature and locomotor activity of each animal were automatically monitored every 10 min during the entire study. Data were processed by a PC equipped with a dedicated recording and analysis system (DataquestR A.R.T. version 2.3, Data Sciences International). Table 1 Synchronization with feeding and photoperiodic schedules Experiment 1
2
Group A B C A B C D
Feeding schedule Ad libitum MTD MTL Ad libitum MTD MTL Ad libitum
Meal Timing during Darkness (MTD) or during Light (MTL). N.A., not applicable.
Feeding times (HALO)
Lighting times (clock hours)
Start
Onset
Offset
09:00 09:00 21:00 09:00 09:00 21:00 21:00
21:00 21:00 09:00 21:00 21:00 09:00 09:00
N.A. 02:00 14:00 N.A. 02:00 14:00 N.A.
Finish 06:00 18:00 08:00 20:00
1184
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
Body weight In Exp 1, body weight was measured both at 2 and 6 HALO for ad libitum mice and for MTL mice and both at 14 and 18 HALO for ad libitum mice and for MTD mice. In Exp. 2, body weight was measured at 2 and 8 HALO in ad libitum mice (group A) and in MTL mice and at 14 and 20 HALO in MTD mice (group C) and in mice fed ad libitum (group D). Tumor Glasgow osteosarcoma was obtained from Aventis Pharma S.A and maintained in C57BL/6 mice then transplanted into male B6D2F1 mice for the purpose of the experiment. The tumor was dissected from a single donor male B6D2F1 mouse for each experiment, and 3 3 3 mm3 tumor fragments were prepared and suspended in Hank’s buffer at 4 jC. A tumor fragment was implanted subcutaneously into both flanks of each mouse using a 12-gauge trocar within the 2 h following sampling. The tumor was inoculated to all the mice 10 days after food restriction (day 0). Two perpendicular diameters (mm) of each tumor were measured daily with a caliper. Tumor weight (mg) was computed as: tumor weight = (length width2)/2. Each mouse with tumor weight reaching 10% of the mouse body weight ( f 2500 mg) along the course of the study was sacrificed for ethical reasons and was considered as dead from tumor progression on this date (Granda et al., 2002). Tumor weight and mortality were recorded daily until the end of each experiment. Statistical analyses Temperature (jC) and activity data (arbitrary units) of each mouse were converted into ASCII format. Data were analyzed with cosinor analysis for periods of 24, 12 or 8 h (Nelson et al., 1979; Filipski et al., 2002). Mean and SEM were calculated for body weight and tumor weight and differences between groups or sampling days were analyzed by ANOVA. Survival data were analyzed with Log rank using the SPSS 11.5 software.
Results Body temperature and locomotor activity rhythms Both variables displayed similar circadian rhythms prior to the onset of food restriction in all three groups. A striking decrease in nadir body temperature was found over the initial 5 days of meal timing in the mice fed during L, which was not the case for mice fed during D. Activity and temperature data of each mouse were double plotted throughout the entire experimental span (Fig. 1) and averaged over the 3-day span preceding tumor inoculation (Fig. 2). These graphic displays clearly indicated that both activity and temperature patterns were altered with meal timing. Both MTD and MTL produced a marked activity bout near feeding onset, short after the beginning of darkness or light, respectively. Yet, a lower activity bout occurred after the end of the feeding span in the MTD mice, whereas some level of activity persisted after feeding and during D in the
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
1185
Fig. 1. Double-plotted representation of locomotor activity (left panel) and body temperature (right panel) in a mouse fed ad libitum, a mouse meal-fed during darkness (MTD) and a mouse meal-fed during light (MTL) in experiment 2. Two consecutive days are shown on the x-axis to visualize the regular alternation of high and low values of each variable from one day to the next. Open and black boxes in horizontal axis respectively represent 12 h of light and 12 h of darkness for photoperiodic synchronization. Time is expressed in hours after light onset (HALO). Shaded boxes represent times of food availability. The feeding condition is shown on the x-axis as follow: ad libitum food (top panel); MTD = meal timing during darkness (from 14 to 20 HALO, middle panel); MTL = meal timing during light (from 2 to 8 HALO, bottom panel).
1186
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
Fig. 2. Time course of locomotor activity (left panel) and body temperature (right panel) in ad libitum fed mice (top panel), in MTL mice (middle panel) or in MTD animals (bottom panel). Figure displays mean circadian pattern of each variable over 3 days before tumor inoculation. Each point represents mean F (SEM) of 3-5 animals. Shaded boxes represent time of food availability.
MTL animals. The body temperature pattern was reinforced by MTD, yet its phase was similar to that observed in ad libitum condition. Conversely, high temperature values were observed during L rather than during D, in MTL mice.
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
1187
Cosinor analysis of these data revealed that the acrophase of both variables were similar in ad libitum and in MTD mice, yet the circadian amplitude of both rhythms were significantly increased with meal timing during D as compared to ad libitum. Conversely MTL altered the 24-h period of the rest-activity cycle resulting in dominant 12-h or 8-h periodic components in activity (splitting). In these animals, the amplitude of the body temperature rhythm was nearly doubled and its acrophase was shifted from f 17 HALO to f 8 HALO. Quite interestingly, the amplitude increase was greater with MTL as compared to MTD (Table 2). Body weight change In both experiments, the body weight of the ad libitum mice remained stable over the time span preceding tumor inoculation, reaching an average value ( F SEM) of 27.2 F 0.2 g in Exp. 1 and to 26.6 F 0.4 g in Exp. 2. Conversely, all the meal-fed mice lost weight during the initial 5 days of meal timing. The mice subsequently increased the daily amount of ingested food up to 6 g, so that they maintained a stable or gradually increasing body weight. Thus, over the 4-day span preceding tumor inoculation, the average body weight gain during meal feeding was similar in MTD and in MTL mice (4.1 F 0.1% vs 4.4 F 0.2%, p from ANOVA = 0.20). On the day preceding tumor inoculation, mean body weight before feeding was 23.2 F 0.3 g for MTD mice and 21.3 F 0.4 g for MTL animals (p from ANOVA < 0.001). Thus, as compared to ad libitum fed mice, the MTD animals lost on average 10.7 to 16.6% of their body weight over 10 days, pending on experiment. The mean relative body weight loss of the MTL mice ranged from 19.4 to 21.3% (Fig. 3). Tumor weight In Exp. 1, tumor became palpable in 7 of 8 mice fed ad libitum eight days after tumor inoculation (87.5%). This was the case for 8 of 11 mice fed during D (72.7%) and for 4 of 9 mice fed during L (44.4%). Six days later (day 14), two ad libitum mice and one MTD mouse died from tumor progression Table 2 Mean circadian parameters F SEM or ultradian parameters of body temperature and locomotor activity in Exp. 2 Variable
Group
Period (H in h)
Amplitude F SEM (jC)
Acrophase (HALO) F SEM (min)
p
Body temperature
Ad libitum MTD MTL Ad libitum MTD MTL # 1 #2 #3
24 24 24 24 24 12 12 8
0.8 1.6 2.1 2.6 3.9 3.4 4.3 3.3
17:20 F 10 18:26 F 8 7:54 F 24 17:16 F 8 15:39 F 31 2:24 and 14:24 2:08 and 14:08 2:05, 10:05 and 18:05
< 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 0.003 < 0.0001 < 0.0001
Activity
F F F F F
0.1 0.2 0.12 0.2 1.4
Individual parameters were obtained with single cosinor analysis of data collected over the 3-day span preceding tumor inoculation. #1, 2 or 3 mean different individuals, with corresponding dominant periods.
1188
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
Fig. 3. Body weight ( F SEM) on the day preceding tumor inoculation, as a function of feeding schedule. Results from Exp. 1 (upper panel) and Exp. 2 (lower panel). Body weight measured at 2 HALO in ad libitum fed mice or before feeding in the MTL group (left panels) and at 14 HALO in ad libitum fed mice or before feeding in the MTD group (right panels).
as tumor weight reached 2500 mg. On this day, mean tumor weight ( F SEM) was 1432 F 421 mg in controls, 1140 F 224 mg in MTD mice and 280 F 96 mg in MTL animals (ANOVA, p < 0.0001). Thus tumor grew more slowly in the meal-fed mice as compared to the ad libitum ones (ANOVA, p < 0.0001), and even more so in the MTL animals as compared to the MTD ones (ANOVA, p < 0.0001). In Exp. 2, tumor growth was similar in both ad libitum groups (A and D) despite light onset occurred 12 h apart (ANOVA, p = 0.7). Thus, the data of both groups were pooled and considered as a single
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
1189
control group. Thirteen days after inoculation, the tumor weight of three ad libitum fed mice reached 2500 mg. On this day, mean tumor weight ( F SEM) was 1795 F 660 mg in controls, 1616 F 528 mg in MTD mice and 1114 F 368 mg in MTL animals (ANOVA, p < 0.0001). The inspection of the tumor growth curves of pooled experiments indicated that meal timing slowed down tumor growth, and even more so if it occurred during L. Thirteen days after tumor inoculation, a time when the first mouse died from tumor progression, mean tumor weight ( F SEM) was 1503 F 150 mg in controls, 1077 F 157 mg in MTD and 577 F 139 mg in MTL (2-way ANOVA: Feeding schedule, p < 0.001; experiment, p < 0.001; interaction, p = 0.99). The overall differences between the tumor growth curves were statistically validated with ANOVA for repeated measures (ANOVA, p < 0.0001) (Fig. 4). Survival In Exp. 1, mice died from tumor progression 13 to 22 days after tumor inoculation in controls, at 13 to 18 days in MTD mice and at 17 to 26 days in MTL. The latter animals survived longer than ad libitum or MTD mice. The survival curves among the 3 groups differed with statistical significance (Log rank, p < 0.0001). In Exp. 2, tumor progression resulted in mouse death 12 to 15 days after inoculation in controls, 13 to 16 days in MTD and 14 to 17 days in MTL. The differences in survival curves were close to statistical significance, with MTL mice surviving longer than ad libitum or MTD animals (Log rank, p = 0.10).
Fig. 4. Tumor growth as a function of feeding schedule. Each daily point is mean tumor weight ( F SEM) of 16 or 19 mice (MTL and MTD, respectively) or ad libitum (n = 26). Pooled data from two experiments.
1190
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
Fig. 5. Survival curves of tumor-bearing mice as a function of feeding schedule in both experiments.
Survival data from pooled experiments further showed that survival was increased in MTL animals as compared to ad libitum or MTD mice. The survival curves differed with a high level of statistical significance (Log rank, p = 0.0035) (Fig. 5). Relation between body weight and tumor growth No significant correlation was found between body weight on the day preceding tumor inoculation and tumor weight on day 13 in each experiment for ad libitum mice (Exp. 1, r = 0.34, p = 0.41; Exp. 2, r = 0.25, p = 0.31), for MTD animals (Exp. 1, r = 0.18, p = 0.60; Exp. 2, r = 0.13, p = 0.76) or for MTD mice (Exp. 1, r = 0.15, p = 0.71; Exp. 2, r = 0.70, p = 0.08). Neither any significant correlation between both variables was found in pooled data from both experiments (ad libitum, r = 0.31, p = 0.12; MTL, r = 0.30, p = 0.26; MTD, r = 0.25, p = 0.30).
Discussion Both body temperature and rest-activity circadian rhythm were modified by restricting the feeding span, yet these alterations differed as a function of meal timing. The average circadian amplitude of body temperature was increased 1.9 fold MTD and 2.6 fold by MTL. MTD also enhanced the 24-h amplitude in activity with minimal alteration of its phase. These findings support a reinforcement of circadian coordination by MTD. In contrast, MTL also shifted the acrophase of body temperature by about 10 h and split the rest-activity rhythm. Such MTL-induced desynchronization was reported by several investigators. For instance, the 24-h rhythm in DNA synthesis in the tongue of ad libitum fed mice
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
1191
became bimodal following MTL. A first peak appeared to be cued by food presentation and the other one by the onset of D (Scheving et al., 1983). A 4 h span has been showed to be the lowest duration of food availability for mice to maintain their basic physiological function and metabolism (Hodgson et al., 1997). In a preliminary experiment (data not shown), we confirmed that meal feeding for 4 h during L or D resulted in a steady decrease in body weight over the initial 5 days, yet body weight remained stable thereafter as the mice learned how to increase the amount of ingested food during this limited feeding time. This observation was in line with earlier reports (Nelson and Halberg, 1986). Similar results were obtained in the current study, although the magnitude of the body weight loss produced by MTL slightly exceeded that resulting from MTD. Nevertheless, the mice on either feeding schedule ingested a similar estimated amount of food over the feeding time span. Meal feeding during L resulted in mean tumor weight being smaller by 5.1 fold in Exp. 1 and by 1.6 fold in Exp. 2 as compared to corresponding controls. These differences could relate to differences in mouse age (Exp.1, 6 weeks; Exp. 2, 4–5 weeks at reception), season of study (Exp. 1, winter; Exp. 2, summer) or meal feeding duration (Exp. 1, 4 h; Exp. 2, 6 h). Although tumor growth significantly faster in Exp. 1 as compared to Exp. 2, the effect of feeding schedule was consistent in both experiments (no schedule x Exp interaction at 2-way ANOVA). Meal feeding during D had little inhibitory effect on tumor growth and no effect on survival. Conversely, meal feeding during L largely and significantly slowed down malignant growth and prolonged survival in tumor-bearing mice, as compared with animals fed ad libitum or during D. The meal timing-dependent effect on tumor progression might result from a more pronounced underfeeding produced by MTL as compared to MTD, as calorie restriction was shown to slow down the growth of transplanted tumors (Hodgson et al., 1997; Kritchevsky, 2001). In the current study, the average body weight loss of MTL mice only exceeded that of MTD animals by 1.3 g in Exp. 1 and by 2.2 g in Exp. 2. For comparison, the average difference between ad libitum mice and MTD animals was 5 g in Exp. 1 and 3 g in Exp. 2. Thus the body weight loss produced by food restriction itself was much larger than that resulting from meal timing during L as compared to D. Furthermore, no significant correlation was found between individual body weight on the day preceding tumor inoculation and subsequent tumor growth. Although we cannot eliminate the possibility that losing body weight below a precise threshold may be needed for tumor growth to be slowed down by calorie restriction, the most likely explanation for the current results lies in the effects of meal timing on circadian rhythms. In the present study, MTL produced discrepant effects on mouse rest-activity and body temperature circadian rhythms. Indeed, this feeding schedule was shown to be able to shift by 10 h the circadian phase of peripheral oscillators, such as liver, colon, etc, without affecting that of the central pacemaker in rodents (Damiola et al., 2000; Hara et al., 2001; Stokkan et al., 2001; Rutter et al., 2002). Similarly, MTL advanced the circadian acrophases of body temperature, plasma insulin, serum glucose, liver glycogen or serum thyroxine (T4), while the rhythm in circulating white blood cell counts remained synchronized to the photoperiodic regimen (Nelson et al., 1975; Challet et al., 1997). A relation between meal timing and tumor growth regulation is further supported by the demonstration that MTL did modify relevant several host rhythms, such as hormonal secretions or DNA synthesis in healthy mouse tissues (Pauly et al., 1976, Lakatua et al., 1983). For instance, DNA synthesis in mouse colon mucosa was entrained by the feeding schedule rather than by the lighting regimen (Lakatua et al., 1983), a finding similar to that found for DNA synthesis in mouse tongue (Scheving et al., 1983). Conversely, the rhythm in tumor DNA synthesis remained similar in ad libitum or MTL mice (Lakatua et
1192
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
al., 1983). These observations suggest that meal timing can exert a differential resetting effect on healthy and malignant peripheral oscillators. Many cellular rhythmic processes including cell proliferation are controlled by clock genes (Reppert and Weaver, 2002; Schibler and Sassone-Corsi, 2002; Matsuo et al., 2003). Clock gene mPer2 was shown to exert tumor suppressor like properties (Fu et al., 2002; Fu and Lee, 2003). Malignant growth was accelerated in mice with ablated SCN or jet-lag induced alteration of clock gene transcription rhythms both in liver and in tumor (Filipski et al., 2002, 2003). The current findings suggest that meal timing during L may directly or indirectly modify circadian clock function or signaling pathways within host peripheral tissues and/or tumor cells, so that tumor growth is slowed down. We hypothesize that meal timing during L amplifies host rhythms and assigns their peak in a time window when the tumor is most susceptible to host-mediated control. Feeding schedules could then be used as an adjunct to further improve the benefit derived from circadian timing of cancer chemotherapy (Le´vi et al., 1997; Le´vi, 2002; Mormont and Le´vi, 2003).
Acknowledgements Program supported by: Programme de Recherches Avance´es de Coope´ration Franco-Chinoises (grant to M.W. W.), INSERM Action The´matique Concerte´e Nutrition and Association Internationale pour la Recherche sur le Temps Biologique et la Chronothe´rapie (ARTBC), Hoˆpital Paul Brousse, Villejuif (France) Clinical Key Program of the Ministry of Public Health of P.R. China (No. 20014058).
References Challet, E., Pevet, P., Vivien-Roels, B., Malan, A., 1997. Phase-advanced daily rhythms of melatonin, body temperature, and locomotor activity in food-restricted rats fed during daytime. Journal of Biological Rhythms 12 (1), 65 – 79. Damiola, F., Le Minh, N., Preitner, N., Kornmann, B., Fleury-Olela, F., Schibler, U., 2000. Restricted feeding uncouples circadian oscillators in peripheral tissues from the central pacemaker in the suprachiasmatic nucleus. Genes Development 14 (23), 2950 – 2961. Duffy, P.H., Feuers, R., Nakamura, K.D., Leakey, J., Hart, R.W., 1990. Effect of chronic caloric restriction on the synchronization of various physiological measures in old female Fischer 344 rats. Chronobiology International 7 (2), 113 – 124. Filipski, E., King, V.M., Li, X.M., Granda, T.G., Mormont, M.C., Liu, X.H., Claustrat, B., Hastings, M.H., Levi, F., 2002. Host circadian clock as a control point in tumor progression. Journal of the National Cancer Institute 94 (9), 690 – 697. Filipski, E., Delaunay, F., Greche-Cassiau, A., Li, X.M., King, V.M., Claustrat, B., Jasmin, C., Hastings, M.H., Levi, F., 2003. Chronic jet lag alters circadian genes expression and accelerates malignant growth in tumor-bearing mice. The 94th annual meeting of American Association for Cancer Research, Washington, D.C., USA, July 11 – 14, 2003. 2463 abs. Fu, L., Pelicano, H., Liu, J., Huang, P., Lee, C., 2002. The circadian gene Period2 plays an important role in tumor suppression and DNA damage response in vivo. Cell 111 (1), 41 – 50. Fu, L., Lee, C.C., 2003. The circadian clock: pacemaker and tumour suppressor. Nature Reviews. Cancer 3 (5), 350 – 361. Granda, T.G., D’Attino, R.M., Filipski, E., Vrignaud, P., Garufi, C., Terzoli, E., Bissery, M.C., Levi, F., 2002. Circadian optimisation of irinotecan and oxaliplatin efficacy in mice with Glasgow osteosarcoma. British Journal of Cancer 86 (6), 999 – 1005. Hara, R., Wan, K., Wakamatsu, H., Aida, R., Moriya, T., Akiyama, M., Shibata, S., 2001. Restricted feeding entrains liver clock without participation of the suprachiasmatic nucleus. Genes to Cells 6 (3), 269 – 278. Hodgson, D.M., Chiappelli, F., Morrow, N.S., Taylor, A.N., 1997. Chronic dietary restriction influences tumor metastasis in the rat: parametric considerations. Nutrition and Cancer 28 (2), 189 – 198.
M.W. Wu et al. / Life Sciences 75 (2004) 1181–1193
1193
Hotz, M.M., Connolly, M.S., Lynch, C.B., 1987. Adaptation to daily meal-timing and its effect on circadian temperature rhythms in two inbred strains of mice. Behavior Genetics 17 (1), 37 – 51. Hursting, S.D., Perkins, S.N., Phang, J.M., 1994. Calorie restriction delays spontaneous tumorigenesis in p53-knockout transgenic mice. Proceeding of the National Academy of Sciences U S A 91 (15), 7036 – 7040. Koubova, J., Guarente, L., 2003. How does calorie restriction work?. Genes and development 17, 313 – 321. Kritchevsky, D., 2001. Caloric restriction and cancer. Journal of Nutritional Science and Vitaminology 47 (1), 13 – 19. Lakatua, D.J., White, M., Sackett-Lundeen, L.L., Haus, E., 1983. Change in phase relations of circadian rhythms in cell proliferation induced by time-limited feeding in BALB/c X DBA/2F1 mice bearing a transplantable Harding-Passey tumor. Cancer Research 43 (9), 4068 – 4072. Le´vi, F., Zidani, R., Misset, J.L., 1997. Randomized multicenter trial of chronotherapy with oxaliplatin, fluorouracil, and folinic acid in metastatic colorectal cancer. International Organization for Cancer Chronotherapy. Lancet 350 (9079), 681 – 686. Le´vi, F., 2002. From circadian rhythms to cancer chronotherapeutics. In: Smolensky, M., Rensing, L. (Eds.), Chronobiology International. Special issue on cancer Chronotherapeutics, vol. 19. Marcel dekker Inc., New York, pp. 1 – 19. McCay, C.M., Crowell, M.F., Maynard, L.A., 1935. The effect of retarded growth upon the length of life span and upon the ultimate body size. Journal Nutrition 10, 63 – 79. McCay, C.M., Maynard, L.A., Sperling, G., Barnes, L.L., 1939. Retarded growth, life span, ultimate body size and age changes in the albino rat after feeding diets restricted in calories. Journal of Nutrition 18, 1 – 13. Matsuo, T., Yamaguchi, S., Mitsui, S., Emi, A., Shimoda, F., Okamura, H., 2003. Control mechanism of the circadian clock for timing of cell division in vivo. Science 302 (5643), 255 – 259. Mormont, M.C., Le´vi, F., 2003. Cancer chronotherapy: principles, applications, and perspectives. Cancer 97 (1), 155 – 169. Nelson, W., Scheving, L., Halberg, F., 1975. Circadian rhythms in mice fed a single daily meal at different stages of lighting regimen. Journal of Nutrition 105 (2), 171 – 184. Nelson, W., Tong, Y., Lee, J.K., Halberg, F., 1979. Methods for cosinor rhythmometry. Chronobiologia 6, 305 – 323. Nelson, W., Halberg, F., 1986. Meal-timing, circadian rhythms and life span of mice. Journal of Nutrition 116 (11), 2244 – 2253. Pariza, M.W., 1986. Calorie restriction, ad libitum feeding, and cancer. Proceedings of the Society for Experimental Biology and Medicine 183 (3), 293 – 298. Pauly, J.E., Scheving, L.E., Burns, E.R., Tsai, T.H., 1976. Circadian rhythm in DNA synthesis in mouse thymus: effect of altered lighting regimens, restricted feeding and presence of Ehrlich ascites tumor. The Anatomical Record 184 (3), 275 – 284. Pollard, M., Luckert, P.H., Pan, G.Y., 1984. Inhibition of intestinal tumorigenesis in methylazoxymethanol-treated rats by dietary restriction. Cancer Treatment Reports 68 (2), 405 – 408. Reppert, S.M., Weaver, D.R., 2002. Coordination of circadian timing in mammals. Nature 418 (6901), 935 – 941. Roebuck, B.D., Baumgartner, K.J., MacMillan, D.L., 1993. Caloric restriction and intervention in pancreatic carcinogenesis in the rat. Cancer Research 53 (1), 46 – 52. Rutter, J., Reick, M., McKnight, S.L., 2002. Metabolism and the control of circadian rhythms. Annual Review of Biochemistry 71, 307 – 331. Scheving, L.E., Tsai, T.H., Scheving, L.A., 1983. Chronobiology of the intestinal tract of the mouse. American Journal of Anatomy 168 (4), 433 – 465. Schibler, U., Sassone-Corsi, P., 2002. A web of circadian pacemakers. Cell 111 (7), 919 – 922. Siegel, I., Liu, T.L., Nepomuceno, N., Gleicher, N., 1988. Effects of short-term dietary restriction on survival of mammary ascites tumor-bearing rats. Cancer Investigation 6 (6), 677 – 680. Stokkan, K.A., Yamazaki, S., Tei, H., Sakaki, Y., Menaker, M., 2001. Entrainment of the circadian clock in the liver by feeding. Science 291 (5503), 490 – 493.