Physiology & Behavior 173 (2017) 298–304

Contents lists available at ScienceDirect

Physiology & Behavior

journal homepage: www.elsevier.com/locate/phb

Time-restricted feeding on weekdays restricts weight gain: A study using

rat models of high-fat diet-induced obesity
Magnus Kringstad Olsen a,b, Man Hung Choi a, Bård Kulseng a,b,c, Chun-Mei Zhao a,b,c, Duan Chen a,b,c,⁎
a
Department of Cancer Research and Molecular Medicine, Norwegian University of Science and Technology, Trondheim, Norway
b
The Central Norway Regional Health Authority, Norway
c
Center for Obesity Research, Department of Surgery, St. Olav's Hospital, Trondheim University Hospital, Norway

H I G H L I G H T S

• A time-restricted feeding regimen (TRF) was tested, i.e., no food consumption during inactive phase daily for 5 weekdays, but not for weekend.
• This special TRF restricted weight gain in juvenile rats.
• Total calorie intake per week was not reduced by this special TRF regimen.

a r t i c l e i n f o a b s t r a c t

Article history: A recent study reported that a special weekly scheduled time-restricted feeding regimen (TRF), i.e., no food con-
Received 12 October 2016 sumption for 15 h during the light (inactive) phase per day for 5 weekdays, attenuated the outcome of diverse
Received in revised form 17 February 2017 nutritional challenges in response to high-fat diet in mice. In the present study, we wanted to further test wheth-
Accepted 23 February 2017
er this TRF could restrict body weight gain in both juvenile and adult animals when fed a high-fat diet. Fifty male
Available online 24 February 2017
Sprague-Dawley rats at ages from 5 to 27 weeks were used. First, we found that freely fed rats with 60% fat diet
Keywords:
gained weight significantly, which was associated with more calorie intake (particularly during light phase) than
Body weight those fed standard food (7% fat). Secondly, we found that TRF restricted high-fat diet-induced weight gain in both
Circadian rhythm groups of juvenile rats (5 and 13 weeks of age) compared to freely fed rats with high-fat diet, despite the same
Food intake levels of 24 h-calorie intake during either weekdays or the weekend. Thirdly, we found that TRF did not restrict
Feeding behavior high-fat diet-induce weight gain in adult rats (27 weeks of age). Thus, we suggest that this special TRF regimen
Obesity could be further tested in humans (particularly young adults) for the purpose of obesity prevention.
© 2017 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction
According to World Health Organization's report, N13% of the world
population are obese, which is a doubling over the last 35 years. Of the
world's 7.2 billion people, 25% are overweight or obese, N 50 million
children at the age of 5 are obese, and 266 and 375 million men and
woman are obese, respectively [1]. In Europe, the EU Action Plan on
Childhood Obesity 2014–2020 outlines actions targeted to make it eas-
ier for children to reduce their calorie intake [2]. In order for children to
have proper linear growth and development, one should consider
how to design new diet regimens for the purpose of obesity prevention
without causing malnutrition.
Circadian rhythms play an important role in the regulation of metab-
olism to optimize energy use and storage. Both rodent and human
⁎ Corresponding author at: Erling Skjalgssons Gate 1, Eastside 3rd Floor, Laboratory
Centre of St. Olav's Hospital, 7006 Trondheim, Norway.
E-mail address: duan.chen@ntnu.no (D. Chen).
studies have demonstrated that restricting food intake to the active
phase (nighttime in rodents and daytime in humans) limits metabolic
disturbances induced by high-energy diets, while eating during the in-
active/sleep phase (daytime in rodents and nighttime in humans)
leads to a worse metabolic outcome [3–8]. Hence, abolishment of eating
during the inactive/sleep phase should be an ideal target for obesity
prevention and treatment.
Animal research using different paradigms has often been per-
formed to characterize the effects of timing of food intake on metabolic
profiles in order to obtain the evidence to support a “best time” not to
eat. Previous studies have shown that a high-fat diet altered daily feed-
ing rhythm by increasing feeding during the inactive phase [9,10]. In-
deed, mice fed a high-fat diet during inactive phase gained more
weight than mice fed the same diet during the active phase [11]. Both
animal study using male Zucker rats and pilot clinical study in 8 obese
patients showed beneficial effects of time-restricted feeding (TRF) as
an anti-obesity strategy [5,7]. It should be noticed that most animal
studies of TRF were conducted in adult, but not juvenile rodents.

http://dx.doi.org/10.1016/j.physbeh.2017.02.032
0031-9384/© 2017 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 M.K. Olsen et al. / Physiology & Behavior 173 (2017) 298–304 299

Recently, Chaix et al. introduced a special weekly schedule of TRF regi- 2.3. Experimental design
men that allows only 9–12 h/day access to food for 5 days (i.e., during
the weekdays), but free access to food during the weekend. Based on Fifty rats were used in three experiments (Fig. 1). Experiment no. 1
their results using mouse models, they suggested that this TRF regimen was performed to obtain basic information. Two groups of age-
can be a preventative as well as therapeutic intervention against obesity matched rats (5–35 weeks of age) were given either high-fat diet
[4]. This TRF regimen sounds more adaptable to people's lifestyles than (HFD) to induce obesity (DIO) (n = 12) or were fed normal diet (ND)
current regimens including the so-called 5:2 diet and daily calorie re- (n = 8). In experiment no. 2, we wanted to test whether TRF could pre-
striction [12]. Thus, it has been particularly highlighted in Science as vent DIO. TRF was applied in juvenile rats fed HFD either starting at
“You are not just what, but when you eat” [13]. 5 weeks of age for 12 weeks (n = 6) or starting at 13 weeks of age for
In the study by Chaix et al., there was, however, no data showing 4 weeks (n = 6) vs. time-matched controls fed HFD ad libitum (n =
why calorie intake during inactive phase was the target, and whether 6). Male Sprague-Dawley rats at 4–5 weeks of age correspond to pre-
this specific TRF could prevent young animals from developing obesity adolescent in humans, as they would reach sexual maturation at
under high-fat diet. Further, the method of determining food intake by 6 weeks [16]. In experiment no. 3, we wanted to test whether TRF
Chaix et al. was performed by manually monitoring the food intake on could induce body weight loss in adult DIO rats. TRF was started at
a weekly basis. 18 weeks of age and lasted for 9 weeks (n = 12). The effect of TRF
In the present study, we wanted to expand these previous studies by was assessed in animals acting as their own control. The studies report-
analyzing the feeding rhythm and metabolic measurements in more de- ed here were conducted over a 3-year period in the same state-of-the-
tail (e.g., number of meals, meal size, intermeal interval, calorie intake, art facility, under controlled conditions and using animals sourced
and energy expenditure) using Comprehensive Laboratory Animal from a single supplier.
Monitoring System (CLAMS) animals [14]. We used rats and not mice,
because CLAMS is performed much more precisely in rats than in 2.4. Measurements of energy balance
mice. The aims of the present study were; 1) to demonstrate that calorie
intake during light (inactive) phase was an ideal target, particularly dur- The animals were acclimatized to the Comprehensive Laboratory
ing the course of high-fat diet; 2) to examine whether this special TRF Animal Monitoring System (CLAMS; Columbus Instruments Interna-
regimen might restrict the weight gain in young animals when fed a tional, Columbus, OH USA) for 24 h before data collection. CLAMS anal-
high-fat diet; and 3) to examine whether this TRF regimen might ysis was performed in the following experimental groups: ND and DIO
be used as a therapeutic intervention in older obese rats when fed rats at 35 weeks of age (experiment no. 1), HFD juvenile rats with and
high-fat diet. without TRF at 7 weeks of age (experiment no. 2) and, adult DIO rats be-
fore and during TRF (17 weeks and 24 weeks of age, respectively) (ex-
2. Methods periment no. 3). We did not perform CLAMS analysis of the second
TRF group (13 weeks, experiment no. 2), because we assumed it
2.1. Animals and high-fat diet would be the same as the first TRF group (5 weeks, experiment no. 2).
At data collection, the animals were kept in CLAMS for 72 h and data
Rats (Sprague-Dawley, four weeks old, male) were purchased from from the last 48 h were used for analysis. Animals were placed in
Taconic (Ejby, Denmark). The diet-induced obese (DIO) rats were put CLAMS with access to expanded high-fat diet or normal diet, and tap
on a 50:50 high-fat diet and normal diet for two weeks (from 5 weeks water. This system is composed of a four-chamber open circuit indirect
of age) before placed on 100% high-fat diet until the end of study [15]. calorimeter designed for continuous monitoring of individual animals.
The normal diet (ND) rats were kept on normal diet during the entire
study. All animals were housed three-four together in individually ven-
tilated cages on wood chip bedding with a 12 h light/dark cycle, room
temperature of 22 °C and 40–60% relative humidity. The standard hous-
ing conditions were specific pathogen free and in agreement with
FELASA (Federation of European Laboratory Animal Science Associa-
tion) recommendations. Throughout the experiment, all animals had
free access to tap water and food regardless of being in metabolic or
Makrolon cages. The high-fat diet (5.24 kcal/g metabolizable energy)
(D12492) was purchased from Research Diets Inc. (New Brunswick,
NJ, USA), and normal diet (7% fat, 2.57 kcal/g metabolizable energy)
(RM1 811004) was purchased from Scanbur BK AS (Sweden). Body
weight was measured every week from arrival until euthanization.
Body composition analysis to ensure obesity was performed using
dual energy X-ray absorptiometry (DXA), while CLAMS was used to an-
alyze eating behavior in obese and non-obese age-matched controls.
Animal experiments were performed according to the guidelines for
the design and statistical analysis of experiments using laboratory ani-
mals after being approved by the Norwegian National Animal Research
Authority (Forsøksdyrutvalget, FDU).

2.2. Time-restricted feeding

The rats placed on time-restricted feeding (TRF) regimen had freely

access to food for 9 h per day (21:00 to 06:00) during the dark phase
(19:00–07:00), for 5 weekdays. During these weekdays, food was man-
ually removed at 06:00 to ensure no food consumption for subsequent
15 h. However, during the weekend (2 days), rats were allowed to
have freely access to food for 24 h per day.
Fig. 1. Study design. ND: normal diet; HFD: high-fat diet; TRF: time-restricted feeding, i,e.,
no food consumption for 15 h during the light (inactive) phase per day for 5 weekdays;
Univ-ANOVA: univariate analysis of variance; RM-ANOV: repeated measures analysis of
variance.
300 M.K. Olsen et al. / Physiology & Behavior 173 (2017) 298–304
Eating behavior and metabolic parameters were recorded automatical-
ly. High-resolution feeding data was generated by monitoring all feeder
balances every 0.5 s. The end of an eating event (meal) was determined
when the balances were stable for N 10 s and a minimum of 0.05 g of
food were eaten. An air sample was withdrawn every 5 min. The energy
expenditure (EE) (kcal/h) was calculated according to this equation:
(3.815 + 1.232 RER) × VO2, where RER (respiratory exchange ratio)
was the volume of CO2 produced per volume of O2 consumed. VO2
was the volume of O2 consumed per h per kilogram of mass of the ani-
mal. Parameters that were obtained during daytime (07:00–19:00) and
nighttime (19:00–07:00) for each individual animal included number of
meals, meal size, meal duration, accumulated food intake, intermeal in-
terval, rate of eating, satiety ratio, drinking activity, energy expenditure
and ambulatory activity. The intermeal interval was defined as the in-
terval in minutes between two meals. Rate of eating was calculated by
dividing the average meal size by the average duration of a meal, and sa-
tiety ratio, an index of the non-eating time produced by each gram of
food consumed, was calculated by dividing the average intermeal inter-
val by the average meal size.
2.5. Body composition analysis
Body composition was determined using dual energy X-ray absorp-
tiometry (DXA) with small animal software (Hologic QDR 4500A,
Hologic Inc., Bedford, MA, USA). Fat percentage (%) was calculated
based on total fat mass (g) and total fat-free mass (g). DXA was per-
formed under anesthesia (isoflurane), 4% was used for induction and
3% for maintenance. Due to practical reason, we performed DXA analysis
only in experiment no. 1, i.e., ND and DIO rats at 30 weeks of age.
2.6. Statistical analysis
The data are expressed as means ± SEM. Statistical comparisons
were performed using independent t-test between the groups, and
paired t-test for comparisons within groups. One-way ANOVA with
Tukey post-hoc test was used for comparison of more than two indepen-
dent groups. Univariate analysis of variance was used to detect differ-
ences in body weight development and in eating patterns between
groups. Repeated measures ANOVA was used for comparison of body
weight after TRF in adult rats. A p-value of b0.05 (two tailed) was con-
sidered statistically significant. G*Power 3.1 was used for power analy-
sis, and 0.80 was considered as adequate. Randomization table was used
for allocation of rats to experimental groups. Test for normality was per-
formed by Shapiro-Wiik test. If normality assumption was not met,
Wilcoxon Signed Rank Test was used for paired comparisons (denoted
W), and Mann Whitney-U (denoted U) was used for independent com-
parisons. The data analysis was performed in SPSS version 22.0 and 23.0,
and GraphPad Prism was used to plot data.
3. Results
3.1. High-fat diet (HFD) vs. normal diet (ND)
At the age of 5 weeks, animals that were allocated to high-fat diet
(HFD) group started to have ad libitum a 50:50 mixture of HFD and nor-
mal diet (ND), and from the age of 7 weeks they had ad libitum access to
HFD only. All these individual animals gained body weight continuously
and although not all were heavier than ND controls, the mean body
weight was significantly higher compared to controls, and they all had
higher fat percentage, evidence of developing diet-induced obesity
(DIO) (Fig. 2A–C, Supplementary Table S1 & 2). DIO rats had increased
calorie intake per 100 g body weight, which was statistically significant
in the light phase (p b 0.05), but not during dark phase (p N 0.05) in
comparison with ND rats (Fig. 2E & K, Table S3 & 4). When expressed
as calorie intake per rat, DIO rats had an increased caloric intake
(p b 0.001) during both light and dark phase (Fig. 2D, Table S3 & 4).
Regression and correlation analyses revealed a significant relationship
between body weight and calorie intake (kcal) (p b 0.001), RER
(p b 0.001), and energy expenditure (kcal/h) (p b 0.001) (Fig. 2F–H,
Table S3). However, there was no relationship between body weight
and calorie intake per 100 g body weight or energy expenditure per
100 g body weight (Fig. S1C & D, Table S3). RER, but not energy expen-
diture, were lower in DIO rats compared to control rats (p b 0.001 and
p N 0.05, respectively) (Table S4). In addition, the DIO rats had increased
meal size (kcal/meal) and rate of eating (g/min) compared to controls
(p b 0.001). Meal duration (min/meal) was unchanged, while number
of meals was significantly reduced in DIO rats compared to controls
(p b 0.001) (Table S4). Further analysis showed that rats either fed
with ND or HFD displayed a normal 24-h eating pattern (Fig. 2I & J,
and Table S5). There was significant difference in eating pattern be-
tween ND and DIO rats, where the DIO rats had increased calorie in-
take/100 g body weight during the light phase compared to ND rats
(p b 0.05) (Fig. 2I–K, Table S4–6). CLAMS data collected during the
weekend did not differ from that collected during weekday (p N 0.05)
(Fig. S1A & B, Table S7).
3.2. HFD with vs. without TRF in juvenile rats
The rationale of TRF was to prevent food intake during the inactive
phase (light phase in rats). HFD-fed juvenile rats placed on TRF at
5 weeks of age gained significantly less weight over the course of
12 weeks compared to HFD-fed rats without TRF (p b 0.001) (Fig. 3A,
Table S8). Another group of HFD-fed rats were placed on TRF at
13 weeks of age, and also gained significantly less weight than age-
matched HFD-fed controls (p b 0.001) (Fig. 3B, Table S9). At 17 weeks
of age, HFD-fed control rats were significantly heavier than normal
diet (ND) rats (p b 0.05) and there were no differences between ND
rats and HFD-rats on TRF either starting 5 weeks or 13 weeks of age
(Fig. 3C, Table S10) (p N 0.05). HFD rats in juvenile control group
showed a similar 24-h eating pattern, i.e. eating most during the dark
phase, but also eating during the light phase (Fig. 3D). TRF was applied
to block the access to food for 15 h during the light phase on weekdays
(5 days), but not during the weekend (Fig. 3E and F, respectively). Uni-
variate analysis of variance revealed that the eating pattern during the
weekend in the HFD-rats (Fig. 3D) was similar to rats on TRF
(p N 0.05) (Fig. 3F) (Table S11–12).
Eating behavior analysis by CLAMS was performed only at 7 weeks
of age. In juvenile rats, 24-h calorie intake was unchanged during week-
days, which was because the elimination of food intake during light
phase was compensated by an increased food intake during dark
phase (Fig. 3G & H, Table S13). No change was seen in either RER or en-
ergy expenditure (Fig. 3I & J, Table S13). There was no difference in eat-
ing behavior parameters such as number of meals, meal size (kcal/
meal), meal duration (min/meal) or rate of eating (g/min) during the
weekdays compared to HFD-fed control rats (Table S14). During the
weekend, 24-h calorie intake was increased in TRF-treated rats com-
pared to the control rats (Fig. 3K & L, Table S15). TRF rats had higher
RER compared to HFD-fed control rats (p b 0.05) (Fig. 3M, Table S15),
and unchanged energy expenditure (Fig. 3N, Table S15). Calculations
of calorie intake per week revealed no statistically significant difference
between TRF and control group (p N 0.05) (Table S14). There was no dif-
ference in eating behavior parameters such as number of meals, meal
size (kcal/meal), meal duration (min/meal) or rate of eating (g/min)
during the weekend compared to HFD control rats (Table S16).
3.3. TRF in adult DIO rats
Adult DIO rats continued to gain weight after they were placed on
TRF at 18 weeks of age (DIO + TRF) (Fig. 4A, Table S17). Comparison
of absolute body weight at 18 weeks of age, and 1 week, 1 month and
2 months after TRF revealed that DIO + TRF rats had a significantly in-
crease in body weight during the period from 18 to 27 weeks of age
 M.K. Olsen et al. / Physiology & Behavior 173 (2017) 298–304 301

Fig. 2. Basic information of high-fat diet-induced obese (DIO) rats. A. Body weight development of individual normal diet (ND) rats (n = 8) and DIO rats (n = 12) compared to baseline. B.
Body weight of ND and DIO rats at 35 weeks of age. C. Fat percentage (%) of individual ND rats (n = 8) and DIO rats (n = 11). D. Calorie intake (kcal) per 24 h (ND; n = 7, DIO; n = 11). E.
Calorie intake (kcal/100 g BW) (ND; n = 7, DIO; n = 11). F. Regression and correlation between body weight (g) and calorie intake (kcal) per 24 h (ND; n = 7, DIO; n = 11). †: a rat
excluded from analysis due to technical error with the feeder. G. Regression and correlation between body weight (g) and RER per 24 h (ND; n = 7, DIO; n = 11). H. Regression and
correlation between body weight (g) and energy expenditure (kcal/h) per 24 h (ND; n = 7, DIO; n = 11). I. Eating pattern of ND rats (n = 7). J. Eating pattern of DIO rats (n = 11). K.
Calorie intake (kcal/100 g BW) during daytime and nighttime (ND; n = 7, DIO; n = 11). Body composition analysis was performed at 30 weeks of age, and CLAMS at 35 weeks of age.
Data are expressed as means ± SEM in B, D, E, I, J & K. *, **, ***: p b 0.05, 0.01, 0.001. NS: not significant. Independent t-test was used for comparison analysis in B–E and K, and univariate
analysis of variance in I & J. For values see Supplementary Table S1–6.

(Fig. 4A, Table S17). The rats showed a typical 24-h eating pattern, i.e.
eating most during the dark phase, but also eating during the light
phase (Fig. 4B, Table S18). TRF was applied to block the access to food
for 15 h during the light phase during weekdays (5 days), but not during
the weekend (Fig. 4C & D, Table S18). The eating pattern during the
weekend in DIO + TRF rats (Fig. 4D) was similar as before they were
placed on TRF (Fig. 4B, Table S18 and 19) (p N 0.05).
In adult DIO rats, 24-h, and dark phase calorie intake were un-
changed during weekdays, and 24-h, light and dark phase calorie intake
were unchanged during the weekend (Fig. 4E & F, I & J, Table S20–23).
TRF rats had lower RER and energy expenditure compared to control
during both the weekend and weekdays (Fig. 4G & H, K & L, Table S20
& 22). TRF did not reduce calorie intake per week (p N 0.05,
Table S21). TRF did not alter eating behavior parameters such as rate
of eating (g/min) or meal size (kcal/meal), however meal duration
(min/meal) was reduced compared to before TRF (p b 0.05) during
both weekday and the weekend (Table S21 and 23).
4. Discussion
The present study showed the circadian rhythm of eating behavior
in a high-fat diet-induced obese rat model and importantly that TRF
for 9 h/day during weekdays reduced obesity during adolescence and
early adulthood in rats. This corresponds well with our knowledge of
the eating behavior of obese patients who often eat late in the evening
and/or during the night [17]. The metabolic parameters showed that
diet-induced obese (DIO) rats had a reduced RER, probably suggesting
a lower oxidative capacity in muscle, i.e. more fat is being used for ener-
gy expenditure compared to normal diet (ND) rats [18]. This is true for
all DIO rats as displayed by correlation analysis between body weight
302 M.K. Olsen et al. / Physiology & Behavior 173 (2017) 298–304

Fig. 3. Effects of TRF in high-fat diet (HFD)-fed juvenile rats. A. Mean body weight development of juvenile rats (HFD without TRF and HFD + TRF from 5 weeks) compared to baseline
(control; n = 6, TRF; n = 6). Arrow indicates starting time point of TRF. B. Mean body weight development of juvenile rats (HFD without TRF and HFD + TRF rats from 13 weeks)
compared to baseline (control; n = 5, TRF; n = 6). Arrow indicates starting time point of TRF. C. Mean body weight (g) at end of TRF regimen (17 weeks of age) in normal diet (ND)
(n = 8), HFD-fed controls (n = 5), HFD + TRF at 5 weeks (n = 6), and 13 weeks (n = 6). D. Eating pattern of HFD control rats (n = 12). E. Eating pattern of HFD + TRF rats during
weekdays (n = 6). F. Eating pattern of HFD + TRF rats during the weekend (n = 6). G. Calorie intake in 24 h (kcal/100 g BW) during the weekdays (HFD without TRF; n = 12,
HFD + TRF; n = 6). H. Calorie intake (kcal/100 g BW) during daytime and nighttime during the weekdays (HFD without TRF; n = 12, HFD + TRF; n = 6). I. RER during the weekdays
(HFD without TRF; n = 12, HFD + TRF; n = 6). J. Energy expenditure (kcal/h/100 g BW) during the weekdays (HFD without TRF; n = 12, HFD + TRF; n = 6). K. Calorie intake in
24 h (kcal/100 g BW) during the weekend (HFD without TRF; n = 12, HFD + TRF; n = 6). L. Calorie intake (kcal/100 g BW) during daytime and nighttime during the weekend (HFD-
fed control; n = 12, HFD + TRF; n = 6). M. RER during the weekend (HFD without TRF; n = 12, HFD + TRF; n = 6). N. Energy expenditure (kcal/h/100 g BW) during the weekend
(HFD without TRF; n = 12, HFD + TRF; n = 6). CLAMS analysis was performed at 7 weeks of age. Data are expressed as means ± SEM. *,***: p b 0.05, 0.001. NS: not significant.
Univariate analysis of variance was used in A and B. ANOVA with Tukey post-hoc was used in C. Independent t-test was used for comparison analysis in G–N. Mann-Whitney U test
was used in J and N. For values see Supplementary Table S8–16.

and RER. The detailed analysis of the eating behavior further showed
that DIO rats ate significantly higher amount of calories per 100 g
body weight than the ND group, however, only during light phase, indi-
cating an excessive energy intake during light phase, which is likely to
be driving the development of obesity. This is in line with evidence
that whether food is ingested during the day or the night will lead to dif-
ferent metabolic consequences [3]. Correlation analysis between calorie
intake (kcal) and body weight (g) showed that the more the animal ate,
the more they weighed. Thus it is more feasible to use calories per 100 g
body weight when comparing amount eaten during day and nighttime
between ND and DIO rats. There was no difference in energy
expenditure between ND and DIO rats, and this may be explained by
the correlation analysis between energy expenditure and body weight.
Energy expenditure was related to body weight rather than fat percent-
age. Thus, obesity in DIO rats does not occur due to reduction in energy
expenditure. In addition, DIO rats have increased meal size and rate of
eating, and reduced number of meals per day compared to ND rats,
which is consistent with previous findings [15]. This also confirms the
reports showing that rapid eating and large meals were associated
with increased body weight and BMI [19,20]. Furthermore, total energy
intake in our DIO model was increased by 74% in terms of kcal/24 h or
41% in terms of kcal/100 g BW/24 h compared to ND control. This is
 M.K. Olsen et al. / Physiology & Behavior 173 (2017) 298–304 303

Fig. 4. Effects of TRF in DIO adult rats. A. Comparison of body weight between values obtained before, and 1 week, 1 month and 2 months after starting TRF (DIO + TRF; n = 12). B. Eating
pattern of DIO rats before TRF (n = 5). C. Eating pattern of DIO + TRF rats during the weekdays (n = 5). D. Eating pattern of DIO + TRF rats during the weekend (n = 5). E. Calorie intake in
24 h (kcal/100 g BW) during the weekdays (before TRF; n = 5, during TRF; n = 5). F. Calorie intake (kcal/100 g BW) during daytime and nighttime during the weekdays (before TRF; n = 5,
during TRF; n = 5). G. RER during the weekdays (before TRF; n = 5, during TRF; n = 5). H. Energy expenditure (kcal/h/100 g BW) during the weekdays (before TRF; n = 5, during TRF;
n = 5). I. Calorie intake in 24 h (kcal/100 g BW) during the weekend (before TRF; n = 5, during TRF; n = 5). J. Calorie intake (kcal/100 g BW) during daytime and nighttime during the
weekend (before TRF; n = 5, during TRF; n = 5). K. RER during the weekend (before TRF; n = 5, during TRF; n = 5). L. Energy expenditure (kcal/h/100 g BW) during the weekend (before
TRF; n = 5, during TRF; n = 5). CLAMS analysis was performed at 17 and 24 weeks of age. Data are expressed as means ± SEM. *, **, ***: p b 0.05, 0.01, 0.001. NS: not significant. Repeated
measures ANOVA was used for comparison of body weight before, during and after TRF in adult rats (A). Paired t-test was used for in-group comparison analysis (E–L). Wilcoxon Ranked
test was used for E and F. For values see Supplementary Table S17–23.

comparable with other studies. For instance, a DIO model in a report by
Bake et al. had about 50% increase in total energy intake compared to
controls [21]. It should be noticed that normal control diets used in
these two studies were different, i.e. 2.57 kcal/g in our study vs.
2.67 kcal/g in the other study [21]. It should also be noticed that by ap-
plying the modified feeding method as reported previously by our
group [15], DIO rats in the present study displayed much less variability
in individual responses to high-fat diet than those earlier reports
[22–25].
Based on these findings and the report by Panda's group [4], we fur-
ther tested whether this TRF that blocks food accessibility during the
daytime in rats (corresponding to nighttime in humans) would prevent
the body weight gain in rats that were still on HFD. In order to make this
regimen relevant to human lifestyles, we have also chosen the weekly
schedule of TRF, i.e. TRF for 5 days (i.e. weekdays) with 2 days of free-
feeding (i.e. the weekend) in juvenile rats from 5 and 13 weeks of age
and adult rats from 18 weeks of age for 12, 4 and 9 weeks, respectively.
Despite the same levels of 24 h–calorie intake during either weekdays
or the weekend, juvenile TRF rats gained less weight than those fed ad
libitum. This is consistent with previous studies showing that day-fed
rats gained more body weight than rats fed during the night or ad
libitum [26]. However, this only occurred in juvenile rats, but not in
adult rats. This is in contrast with the previous study that showed a re-
striction in body weight gain in DIO mice when ad libitum feeding was
304 M.K. Olsen et al. / Physiology & Behavior 173 (2017) 298–304
switched to the 5 days per week TRF regimen [4]. It was shown in the
previous study that HFD mice on TRF gained significantly less body
weight, which was in parallel with reduced fat mass compared to
mice fed HFD ad libitum [4]. In the present study, we were unable to
compare body composition between HFD rats with and without TRF
due to practical reasons. TRF was without effect on RER and energy ex-
penditure (EE) in juvenile rats, but reduced both RER and EE in adult
DIO rats, particularly during the weekdays, probably suggesting in-
creased fat oxidation and/or a lower oxidative capacity in the muscle
during TRF [27]. Switching from ad libitum feeding regimen to TRF did
not alter feeding behavior such as meal size, satiety ratio or number of
meals on a day-to-day basis in juvenile rats. Thus, the causal factor for
weight loss in juvenile rats was most likely due to food restriction per
se. The reduction in meal duration per meal by TRF in adult rats might
suggest that the rats learned that the food access would be restricted,
and therefore they ate quicker while food was available. DIO adult rats
had positive relationships between body weight and calorie intake or
EE but negative relationship with RER. TRF in DIO adult rats was without
effect on calorie intake and reduced EE as well as RER, together leading
to unchanged body weight. Recently, we have shown that when a body
weight plateau was reached following either HFD-induced weight gain
or gastric bypass-induced weight loss in adult rats, a new steady-state
energy homeostasis occurred by alterations in EE and RER rather than
total calorie intake [14]. Taken together, it may explain why TRF restrict-
ed HFD-induced body weight gain in younger rats but not adult DIO rats.
In general, body weight loss induced by dietary intervention differs
depending on age and the degree of overweight and obesity. An impor-
tant consideration is that children need sufficient calories and nutrition
for proper linear growth [28,29]. Since this special TRF regimen works
based on when to eat, and not what to eat, we may suggest that this spe-
cial TRF is safe for the juvenile and adolescent population without
risking disruption in growth. However, more data are needed in the fu-
ture, particularly when considering clinical trials of childhood obesity
development and prevention.
5. Conclusions
In the present study, we demonstrated that diet-induced obese rats
ate significantly higher amount of calories per 100 g body weight only
during light phase in comparison with the normal diet rats. Further-
more, we demonstrated that the TRF regimen, by restricting feeding to
a 9 h interval during the dark phase on 5 weekdays per week, restricted
high-fat diet-induced weight gain in juvenile rats without reducing total
calorie intake per week. We may suggest a new strategy, i.e., the earlier
to apply this new way of dieting the better to prevent obesity
development.
Conflict of interest statement
All authors declare no conflict of interest.
Acknowledgements
The research leading to these results has received funding from
European Union Seventh Framework Programme (FP7/2007-2013, no.
266408), from the Joint Programme of the Medical Faculty of Norwe-
gian University of Science and Technology (NTNU) and St. Olav's Uni-
versity Hospital, the Liaison Committee between the Central Norway
Regional Health Authority and NTNU.
MKO, BK, CM-Z, and DC generated the study design; MKO and MHC
carried out the experiments and gathered data; MKO and MHC analyzed
the data; MKO, CM-Z and DC interpreted the data. MKO generated
figures. All authors were involved in writing the paper and approved
the final submitted version.
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.physbeh.2017.02.032.
References
[1] NCD Risk Factor Collaboration, Trends in adult body-mass index in 200 countries
from 1975 to 2014: a pooled analysis of 1698 population-based measurement stud-
ies with 19·2 million participants, Lancet 387 (10026) (2016) 1377–1396.
[2] S.E. Barlow, C. Expert, Expert committee recommendations regarding the preven-
tion, assessment, and treatment of child and adolescent overweight and obesity:
summary report, Pediatrics 120 (Suppl. 4) (2007) S164–S192.
[3] S. Moran-Ramos, et al., When to eat? The influence of circadian rhythms on meta-
bolic health: are animal studies providing the evidence? Nutr. Res. Rev. (2016)
1–14.
[4] A. Chaix, et al., Time-restricted feeding is a preventative and therapeutic interven-
tion against diverse nutritional challenges, Cell Metab. 20 (6) (2014) 991–1005.
[5] R.E. Mistlberger, H. Lukman, B.G. Nadeau, Circadian rhythms in the Zucker obese rat:
assessment and intervention, Appetite 30 (3) (1998) 255–267.
[6] M. Hatori, et al., Time-restricted feeding without reducing caloric intake prevents
metabolic diseases in mice fed a high-fat diet, Cell Metab. 15 (6) (2012) 848–860.
[7] S. Gill, S. Panda, A smartphone app reveals erratic diurnal eating patterns in humans
that can be modulated for health benefits, Cell Metab. 22 (5) (2015) 789–798.
[8] M.J. Duncan, et al., Restricting feeding to the active phase in middle-aged mice at-
tenuates adverse metabolic effects of a high-fat diet, Physiol. Behav. 167 (2016) 1–9.
[9] A. Kohsaka, et al., High-fat diet disrupts behavioral and molecular circadian rhythms
in mice, Cell Metab. 6 (5) (2007) 414–421.
[10] J.S. Pendergast, et al., High-fat diet acutely affects circadian organisation and eating
behavior, Eur. J. Neurosci. 37 (8) (2013) 1350–1356.
[11] D.M. Arble, et al., Circadian timing of food intake contributes to weight gain, Obesity
(Silver Spring) 17 (11) (2009) 2100–2102.
[12] M. Garaulet, P. Gomez-Abellan, Timing of food intake and obesity: a novel associa-
tion, Physiol. Behav. 134 (2014) 44–50.
[13] L.B. Ray, You are not just what, but when you eat, Science 347 (6217) (2015) 39.
[14] M.K. Olsen, et al., Steady-state energy balance in animal models of obesity and
weight loss, Scand. J. Gastroenterol. (2016) 1–8.
[15] M.W. Furnes, C.M. Zhao, D. Chen, Development of obesity is associated with in-
creased calories per meal rather than per day. A study of high-fat diet-induced obe-
sity in young rats, Obes. Surg. 19 (10) (2009) 1430–1438.
[16] E.M. Lewis, et al., Sexual maturation data for Crl Sprague-Dawley rats: criteria and
confounding factors, Drug Chem. Toxicol. 25 (4) (2002) 437–458.
[17] M.E. Gluck, et al., Nighttime eating: commonly observed and related to weight gain
in an inpatient food intake study, Am. J. Clin. Nutr. 88 (4) (2008) 900–905.
[18] A. Ramos-Jimenez, et al., The respiratory exchange ratio is associated with fitness in-
dicators both in trained and untrained men: a possible application for people with
reduced exercise tolerance, Clin. Med. Circ. Respirat. Pulm. Med. 2 (2008) 1–9.
[19] J.H. Ledikwe, J.A. Ello-Martin, B.J. Rolls, Portion sizes and the obesity epidemic, J.
Nutr. 135 (4) (2005) 905–909.
[20] T. Ohkuma, et al., Association between eating rate and obesity: a systematic review
and meta-analysis, Int. J. Obes. 39 (11) (2015) 1589–1596.
[21] T. Bake, D.G. Morgan, J.G. Mercer, Feeding and metabolic consequences of scheduled
consumption of large, binge-type meals of high fat diet in the Sprague-Dawley rat,
Physiol. Behav. 128 (2014) 70–79.
[22] B.E. Levin, et al., Resistance to diet-induced obesity: food intake, pancreatic sympa-
thetic tone, and insulin, Am. J. Phys. 252 (3 Pt 2) (1987) R471–R478.
[23] S. Chang, et al., Metabolic differences between obesity-prone and obesity-resistant
rats, Am. J. Phys. 259 (6 Pt 2) (1990) R1103–R1110.
[24] M.J. Pagliassotti, et al., Time course of adaptation to a high-fat diet in obesity-
resistant and obesity-prone rats, Am. J. Phys. 267 (3 Pt 2) (1994) R659–R664.
[25] B.E. Levin, et al., Differential stress responsivity in diet-induced obese and resistant
rats, Am. J. Phys. Regul. Integr. Comp. Phys. 279 (4) (2000) R1357–R1364.
[26] R. Salgado-Delgado, et al., Food intake during the normal activity phase prevents
obesity and circadian desynchrony in a rat model of night work, Endocrinology
151 (3) (2010) 1019–1029.
[27] M. Rosenkilde, et al., Fat oxidation at rest predicts peak fat oxidation during exercise
and metabolic phenotype in overweight men, Int. J. Obes. 34 (5) (2010) 871–877.
[28] K. Sonneville, in: C. Duggan, J.B. Watkins, W.A. Walker (Eds.), Popular Diets for Obe-
sity, in Nutrition in Pediatrics: Basic Science, Clinical Applications, BC. Decker Inc.
2008, pp. 467–469.
[29] S. Adu-Afarwuah, et al., Randomized comparison of 3 types of micronutrient supple-
ments for home fortification of complementary foods in Ghana: effects on growth
and motor development, Am. J. Clin. Nutr. 86 (2) (2007) 412–420.