Physiology & Behavior 105 (2012) 376–387

Contents lists available at SciVerse ScienceDirect

Physiology & Behavior

j ourna l homepage: www.e lsev ie r .com/ locate /phb

Physiological and behavioral responses to intermittent starvation in C57BL/6J mice

Li-Na Zhang a, Sharon E. Mitchell a, Catherine Hambly a, David G. Morgan b,John C. Clapham b, John R. Speakman a,⁎a Institute of Biological and Environmental Sciences, University of Aberdeen, Aberdeen, Scotland, UKb Research and Development (CVGI), AstraZeneca, Alderley Park, Cheshire, UK

⁎ Corresponding author. Tel.: +44 1224 272879; fax:E-mail address: J.Speakman@abdn.ac.uk (J.R. Speakm

0031-9384/$ – see front matter © 2011 Published by Eldoi:10.1016/j.physbeh.2011.08.035

a b s t r a c t

a r t i c l e i n f o

Article history:Received 24 May 2011Received in revised form 8 August 2011Accepted 26 August 2011

Keywords:Intermittent starvationPhysical activityBody temperatureC57BL/6J mice

The dual intervention point model states that body mass is controlled by upper and lower intervention points,above and below which animals (and humans) intervene physiologically to bring their body mass back into theacceptable range. It has been further suggested that the lower intervention point may be defined by the risk ofstarvation, while the upper intervention point may be defined by the risk of predation. The objective of the pre-sent study was to test whether the risk of starvation determines the lower intervention point and to examine thephysiological and behavioral mechanisms that underpin the regulation of body mass, when the risk of starvationis increased. Sixty-four mice were exposed to random days of complete fasting or 50% food restriction and theirbody mass and fat mass responses were measured. Food intake, physical activity and body temperature weremeasured throughout the experiment. In addition, plasma leptin and insulin, triglyceride and non-esterifiedfatty acids, along with hypothalamic neuropeptides gene expression in the arcuate nucleus were assessed after13 and 42 days of treatment. We found that C57BL/6J mice increased body mass and fatness in response to ashort-term (13 days) intermittent fasting,whichwas restored to baseline as the treatmentwas prolonged. In con-trast, intermittently 50% food restrictedmice showed no significant changes in bodymass or fatness. Over thefirst13 days of treatment the data were consistent with the dual intervention point model as the mice showed bothincreased body mass and adiposity over this period. Over the more protracted period of 42 days the effectwaned and was therefore inconsistent with themodel. The bodymass and fat mass gains in intermittently fastedmiceweremainly accounted for by increased food intake. ElevatedNPY gene expression after 13 days (three 24 hfasting events) may have driven the increase in food intake. However, no changes were observed in such neuro-peptides as POMC, CART, AgRP, Ob-Rb and SOCS 3 or circulating levels of leptin, insulin, NEFA and TG. Hypother-mia during fasting daysmay have also contributed to the increase in body mass. Over 42 days of treatment (nine24 h fasting events) cumulative food intake was not affected by intermittent starvation. However physical activ-ity, mainly activity during the light phase was lowered suggesting an adaptation to unpredictable starvation.Overall, mice exhibited different behavioral and physiological responses to intermittent starvation dependingon the duration of treatment.

+44 1224 272396.an).

sevier Inc.

© 2011 Published by Elsevier Inc.

1. Introduction

1.1. Body mass regulation

The global epidemic of obesity has raised the need for a better un-derstanding of the mechanisms that regulate body and fat mass. Abalance between energy intake and energy expenditure is necessaryto maintain a stable body mass. It is well established that the centralnervous system (CNS) regulates food intake and energy expenditurein response to neuronal, hormonal and nutrient signals [1–3]. The hy-pothalamus is the most studied area in CNS with respect to the regu-lation of energy homeostasis. The arcuate nucleus in particular is

known to play an important role in energy homeostasis because itcontains two distinct populations of neurons: pro-opiomelanocortin(POMC) neurons that express the endogenous anorectic melanocor-tin receptor agonist (α-melanocyte stimulating hormone, α-MSH)along with cocaine-and-amphetamine-regulated transcript (CART),and neurons that express neuropeptide Y (NPY) and also the endog-enous orexigenic melanocortin receptor antagonist, agouti-relatedprotein (AgRP) [4–6]. These neuronal populations respond to periph-eral signals, such as leptin, insulin and gastrointestinal hormones bymodifying the production of the above neuropeptides that modulateenergy balance [7]. Moreover, leptin signaling within neurons is de-pendent on the presence of the long form of the leptin receptor(Ob-Rb) [8], which signals principally via the Janus kinase 2 (JAK2)/signal transducer and activator of transcription 3 (STAT3) pathway[9]. The suppressor of cytokine signaling 3 (SOCS3) is a negative reg-ulator of the leptin receptor which is stimulated by the JAK/STAT2

377L.-N. Zhang et al. / Physiology & Behavior 105 (2012) 376–387

pathway [10–13]. Gene expression levels of Ob-Rb and SOCS3 aremarkers of leptin sensitivity [14].

There have been several theoretical models that have attemptedto conceptualize the mechanisms involved in the regulation of bodymass and fatness. Kennedy (1953) proposed the lipostatic set pointmodel which suggested that the size of body fat depots is sensed bya ‘lipostat’, which adjusts food intake and energy metabolism tomaintain the body and fat masses at a set-point [15]. Although leptinhas been often interpreted as the molecular manifestation of the lipo-static fat signal [16], this model is in conflict with the evidence in pat-terns of changes in animal and human body mass [17,18]. Analternative interpretation emerged suggesting that body mass is notregulated by a set-point, but rather is controlled by upper and lowerintervention points, above and below which animals (and humans)intervene physiologically to bring their body mass back into the ac-ceptable range [19,20]. In humans the upper intervention point maybe located at different positions in different individuals explainingwhy some individuals become obese when exposed to environmentswith readily available food supplies, but others are able to regulatetheir body weights at normal levels. Based on data from small mam-mals and birds, Speakman (2007) has further suggested that thelower intervention point may be defined by the risk of starvationwhile the upper intervention point may be defined by the risk of pre-dation [20]. In this scenario, one would expect that an increased riskof starvation would increase the lower intervention level and animalswould gain fat and body mass.

1.2. Intermittent starvation

Periods of negative energy balance arising from restricted feedingor total starvation are common events [21]. Throughout their lives,small mammals must face periodic food shortages interspersed by pe-riods of food abundance.

Stochastically imposed intermittent periods of fasting (completeabsence of food) or periodic food restriction (involving reduced levelsrelative to habitual intake, but not complete absence) may mimic theunpredictable food availability in the wild. However, few studies haveinvestigated the physiological and behavioral responses to stochasticfood exposure. Swiss mice showed increased food intake anddecreased energy expenditure on days that intervened between24 h fasting events over 4 weeks treatment (including 3 or 4 fastingdays and 4 or 3 feeding days each week), and they decreased overallbody mass [22]. However in another study the same group found that4 weeks treatment including 3 fasting days and 4 feeding days eachweek had no effect on overall body mass [23]. Alternate day fasting(ADF) is a similar and widely studied regimen, however, most of thestudies have been focused on the beneficial effects on aging and theeffect of ADF on body weight is less often reported. Body mass hasbeen shown to be highly variable in response to ADF in both humansand in animal models. In Fisher rats, when ADF regimens wereapplied in the short term, no effect on body weight was observedafter 2 weeks [24], whereas gains in body mass were noted inC17BL/10 mice after 8 weeks [25]. However, in other studies, whenADF was administered for 12 weeks, body mass decreased in bothC57BL/6 mice and rats. This variability in response is confusing andits relationship to the dual-intervention model is uncertain becausein ADF the fasting days come at predictable intervals, and hence theuncertain risk of starvation is not altered by this treatment. Even inthe studies performed by Zhao and colleagues [22,23] the occurrenceof fasting days is so frequent that it may be predictable — leading todifferent responses from those predicted by the dual-interventionpoint model.

In the present study we aimed firstly to test whether starvationrisk defines the lower intervention point for body mass and adiposityas suggested in the dual intervention point model, by experimentallyincreasing starvation risk over a period of 42 days, by exposing mice

to nine random days of complete fasting or food restriction, and re-cording their body mass and fat mass responses. Second we investi-gated the physiological and behavioral mechanisms that underpinthe regulation of body mass during intermittent fasting or foodrestriction, by measuring food intake, physical activity and body tem-perature throughout the experiment. In addition, NPY, AgRP, POMC,CART, Ob-Rb and SOCS3 gene expression in the arcuate nucleus ofthe hypothalamus were assessed after 13 days and 42 days of treat-ment. We also measured plasma leptin and insulin, as well as triglyc-eride (TG) and non-esterified fatty acids (NEFA). We hypothesizedthat if the dual intervention point model is correct randomly imposedintermittent fasting and food restriction would cause body mass andfat mass to increase as a result of an increase in the lower interven-tion point. Secondly, we hypothesized that physiological and behav-ioral compensation mechanisms would be employed to adjust bodymass and fatness within the range and thirdly, that leptin, insulinand other potential signals would change in response to intermittentfasting and food restriction and neuropeptides sensitive to leptinwould be involved in the body mass response.

2. Material and methods

2.1. Animal housing and intermittent starvation regimen

Sixty four female C57BL/6 mice aged 6–8 weeks were purchasedfrom Charles River (Charles River UK Ltd, Kent, UK) and housed insingle cages (M3 cage 48×15×13 cm, NKP Cages, Kent, UK) in a tem-perature controlled room (21±1 °C) under a 12:12-hour light:darkphotoperiod with the lights coming on at 06:00 and a “dawn/dusk”period of 20 min at either end of the light period. Wood shavingsand shredded paper bedding were provided for enrichment. Ataround age 9–10 weeks mice were implanted with transmittersthat measured their body temperatures and physical activity levels(details below) and were then fed a standard control diet(D12450B, 10% kcal/fat, Research Diet, New Brunswick, NJ, USA) adlibitum until 22 weeks of age when they were randomly assignedinto three groups: a control group AL, fed ad libitum (n=20); IF, in-termittently fasted (n=22); IR, intermittently 50% food restricted(n=22). All mice had free access to water throughout the study. Allprocedures were reviewed by a local ethical committee and per-formed in accordance with UK home office regulations under licensePPL 60/3707 held by JRS.

Mice were divided into two cohorts containing 32 mice in each AL(n=10); IF (n=11); IR (n=11). Note the sample sizes in each groupand cohort were unbalanced because the number of recording padsfor the implanted transmitters (see below) was fixed at 64. Bodymass and food intake were recorded for 3 days on a daily basis priorto the intermittent starvation regime. Animals were stochastically ex-posed to 24 h fasting or restriction on 9 occasions over a period of42 days. Each starvation day was always followed by a non-starvationday. Probability of next day would be starvation was set at 0.21 andwas decided using random integers. During the treatment, days 4, 7,10, 14, 16, 20, 26, 29 and 39 were assigned as fasting/restrictiondays. On each treatment day, IF mice were completely deprived offood, while IR mice were given 50% of their average daily food intakemeasured over the baseline period at the start of the experiment.Mice were provided with ad libitum food between fasting/restrictiondays. Body mass and food intake were measured at 1500 h every daywhen food was given or removed. The first cohort of mice was killedon day 13 of the treatment, while the second cohort was killed on day42.

2.2. Physical activity and body temperature

Prior to experimentation the mice were implanted with a teleme-try transmitter to monitor body temperature (Tb) and physical

Table 1Details of primers used, the fragment size and Accession numbers.

Gene Primers 5′ to 3′ bp Accession number

AgRP TCCTTCCCCAATCCCAATCC 205 NM_007427CGCCCTACCTTCTACTTTATCC

CART CCGAGCAGCGAGGAGGTCCAGAA 372 NM_013732GGACAGTCACACAGCTTCCCGATCC

NPY ATGGACAAGATCCGGGACAG 227 NM_010934TAAAAGATGGGGTTGACGCAGGTGG

POMC GGGCAAGCGCTCCTACTCCAT 348 NM_008895CACTGGCCCTTCTTGTGCG

ObRb GTCCTCTTCTTCTGGAGC 634 NM_146146GTGTGAGCATCTCTCCTGGAG

SOCS-3 CGCCTCAAGACCTTCAGCTCCAA 565 NM_007707GGGTCACTTTCTCATAGGAGT

Agouti-related peptide, (AgRP); cocaine- and amphetamine-regulated transcript(CART); neuropeptide Y (NPY); pro-opiomelanocortin (POMC); signaling form ofleptin receptor (obRb); suppressor of cytokine signaling 3 (SOCS 3).

378 L.-N. Zhang et al. / Physiology & Behavior 105 (2012) 376–387

activity (PA) (Model PDT-4000 E-Mitter, Mini-Mitter, Bend, OR, USA).The surgery was carried out under anesthesia induced by a mixedflow of isofluorane and oxygen. An incision of approximately 1.0 cmwas made in the ventral skin, then a second smaller incision madethrough the peritoneal wall. The transmitter was inserted intraperito-neally, and the two layers were sutured separately (Ethicon Vicryl:W9982; W&J Dunlops Ltd., Dumfries, UK). The surgical proceduretook around 15–20 min/mouse and mice were given 2 weeks withminimal handling to allow them to recover from the surgery. A re-ceiver pad (ER-4000 Receiver, Mini-Mitter, Bend, OR, USA) underthe cage where the mouse was housed received PA and Tb informationfrom the transmitter which was collected every minute by a WindowsPC-based data acquisition system (VitalView™: Mini-Mitter, Bend,OR, USA). PA and Tb recordings were averaged for each hour to showdaily activity and Tb patterns. Daily means, and means during darkand light phaseswere calculated throughout the baseline and treatmentperiods.

2.3. Body fatness and plasma metabolites levels

Body fat mass (FM) was determined using dual energy X-ray ab-sorptiometry (DXA; PIXImus2 Series Densimeters, GE Medical SystemsUltrasound and BMD, Bedford, UK). Mice were anesthetized during thescan. Data generated from the DXA software were corrected using anequation developed specifically for our machine [26]. DXA was per-formed to assess body fatness at baseline in both cohorts, and addition-ally on day 12 for cohort 1 and ondays 24, 30, 37 for cohort 2.Miceweresacrificed after fasting for 4 h. Blood samples were collected by cardiacpuncture for the determination of circulating hormone and metabolitelevels (leptin, insulin, TG and NEFA). Fat depots (epidydimal fat, perire-nal fat, mesenteric fat and subcutaneous fat) were dissected andweighed. All blood samples were placed in EDTA-treated vials and cen-trifuged at 4 °C at 10000 rpm for 10 min. Blood plasma was stored at−80 °C until analysis.

Plasma leptin and insulin were measured with ELISA-basedmethods using commercially available kits (mouse leptin ELISA kit,Millipore Billerica, MA, USA; mouse insulin ELISA kit, Crystal ChemInc, Downers Grove, IL, USA, respectively), while plasma NEFA andTG were measured using colorimetric assays (NEFA-HR (2), WakoChemicals, Neuss, Germany; Triglyceride and Free Glycerol Kit,Sigma-Aldrich, Kent, UK, respectively).

2.4. Gene expression of neuropeptides

2.4.1. Tissue collection and preparationMicewere killed by CO2 inhalation. Their brainswere immediately re-

moved, frozen in isopentane over dry ice and stored at −80 °C. Coronal20 μm sections of the hypothalamic region were cut on a cryostat(Leica, CM3050S, Milton Keynes, UK) and thaw mounted on poly-L-lysine-coated slides. 20 slides (with approximately 6 sections on eachslide) through the hypothalamus were collected per animal, and theywere numbered as 1–10, 11–20. Two slides numbered as 1 and 11 (or 2and 12, etc.) were used for one specific neuropeptide.

2.4.2. Riboprobe synthesisTemplate DNA, complimentary to the mRNA sequence of interest,

was gifted from Dr. Sharon Mitchell. Briefly fragments were amplifiedfrom mouse brain cDNA by polymerase chain reaction (PCR). Primers(Eurofins MWG Operon, London, UK), fragment size and Accessionnumbers are detailed in Table 1. PCR products were cloned into eitherthe pCR-Script Amp SK (+) vector (Stratagene, CA, USA) or thepGMET Easy Vector (Promega, Southampton, UK) and transformedinto competent cells. Plasmid DNA was isolated using the QIAprepSpin Miniprep kit (Qiagen, Germany) and linearized with appropriaterestriction enzymes (Promega, Southampton, UK). All sequenceswere verified by Eurofins Sequencing Services (London, UK). The

linearized DNA was transcribed using T7, T3 or SP6 RNA polymerasepromoters matched to a restriction site contained within the vector.Riboprobes were labeled with 35S-UTP (Perkin Elmer, UK) and unin-corporated label removed by spinning through Chromospin columns(BD Biosciences, UK). Finally riboprobes were made up to a final con-centration of 1×106 c.p.m ml−1 in hybridization buffer. Antisenseand sense riboprobes were tested for specificity and backgroundbinding.

2.4.3. In situ hybridizationLabware was baked at 200 °C and all solutions made up in 0.1%

diethylpyrocarbonate (DEPC) to prevent RNase contamination. Sectionswere fixed in 4% paraformaldehyde (Sigma-Aldrich, UK)/0.1 M phos-phate buffer (144 mM NaH2PO4 and 5.6 mM Na2HPO4, pH 7.4; PB) for20 min on ice andwashed in 2×0.1 M PB. To eliminate background sig-nal and inactivate RNases, sections were first immersed in 0.1 mMtriethanolamine (TEA) (Sigma-Aldrich, UK) for 2 min before acetylationin 0.1 mM TEA/0.25% acetic anhydride (Sigma-Aldrich, UK) for 10 minat room temperature. Sections were again washed in 0.1 M PB for4 min, dehydrated through increased concentrations of ethanol (50%,70%, 90% and 100%) and dried under vacuum. Labeled probe waspipetted onto cover slips, annealed to sections and sealed with DPXbefore an overnight hybridization at 58 °C. Post-hybridization, coverslipswere removed by soaking slides in 4 X saline-sodium citrate buffer(150 mM sodium chloride and 15 mM sodium citrate, pH7; SSC). Non-specific hybridization was dissociated by increasing the temperature,up to 60 °C for 30 min. Unhybridized RNA was digested with RNasefor 30 min. Sections were desalted over a series of SSC solutions: 2 XSSC, 1 X SSC, 0.5 X SSC and 0.1 X SSC and finally dehydrated throughgraded increased concentrations of 50%, 70%, 90% and 100% ethanol.The slides were air-dried and exposed to Kodak BioMax Film (Sigma,UK) for time appropriate to specific riboprobe. Autoradiographs of sec-tions including microscale standards were scanned on Umax PowerLook II (UMAX Data System, Fremont, CA, USA), and gene expressionwas taken as the integrated optical density (IOD) in the area of interestusing ImageJ software system (WinZip Computing Inc., USA). The slideswere observed by a blinded observer.

2.5. Statistical analysis

All data were expressed as means±SD. General linear modeling(GLM) with individual ID included as a random factor nested withinthe treatment to account for the repeated measures was used to com-pare changes of body mass, body fat mass, food intake, PA and Tbthroughout the experiment. One-way analysis of variance (ANOVA)with post hoc Tukey tests was used to test the between-group differ-ences of body mass, body fat mass, food intake, cumulative food

379L.-N. Zhang et al. / Physiology & Behavior 105 (2012) 376–387

intake, PA and Tb at different time points where group differenceswere significant over the whole time course. The two cohorts wereanalyzed separately. Analysis of covariance (ANCOVA) was per-formed to examine differences body fat mass using body mass as a co-variate and ANCOVAwas also used to test differences in plasma leptinlevels using fat mass as a covariate. Group differences in plasma insu-lin, TG, NEFA levels and fat depots were examined with one-wayANOVA. Group differences in neuropeptides gene expression wereexamined by comparison of the total integrated optical density(IOD) for each gene using ANOVA. Gene expression data were pre-sented as percentage of the mean for AL control. P valuesb0.05were considered statistically significant. All data were analyzedusing SPSS 18.0 statistical package for Windows (SPSS Inc., Chicago,IL, USA).

3. Results

3.1. Body mass and fatness

3.1.1. Cohort 1Prior to the start of intermittent starvation regime, animals had

similar body mass on day 0 (IF, 24.46±1.15 g; IR, 24.28±1.24 g;AL, 24.29±1.27 g; ANOVA: F2,29=0.08, P=0.926; Fig. 1A). The inter-mittently fasted (IF) mice significantly decreased body mass on thefirst (day 4) and second (day 7) days of fasting (ANOVA: day 4,F2,29=7.00, P=0.003; day 7, F2,29=4.71, P=0.017; Fig. 1A). The IFmice did not exhibit a significant decrease in body mass on thethird day of fasting (day 10) (ANOVA: F2,29=1.96, P=0.159). After3 days of fasting and 7 days of free feeding, on days 11–13, IF micemaintained a significantly higher body mass in comparison to bothAL control and the intermittently restricted (IR) mice (ANOVA: day11, F2,29=4.18, P=0.025; day 13, F2,29=3.89, P=0.032; day 13,F2,29=3.78, P=0.035; Fig. 1A).

3.1.2. Cohort 2There were no significant differences in body mass between groups

before the treatment started (ANOVA: F2,29=0.01, P=0.989; Fig. 1B).On day 0, the average body mass of IF, IR and AL groups were 23.11±1.19 g, 23.12±1.21 g and 23.19±1.21 g, respectively. The responsesof the second cohort over the first 13 days of treatment were almostidentical to those observed in cohort 1. After the first starvation day,IF mice showed a significantly lowered bodymass compared to AL con-trol (Post hoc Tukey test: Pb0.001) and IR mice (Post hoc Tukey test:P=0.045), whereas no differences were detected between IR miceand AL control (Post hoc Tukey test: PN0.05) (ANOVA: F2,29=11.22,Pb0.001; Fig. 1B). The body mass of IF mice returned to the AL controllevel after one-day of free feeding (P=0.605). The second starvationday resulted in a similar response in IFmice exemplified by a significantreduction in body mass (ANOVA: F2,29=11.45, Pb0.001), which wasfollowed by an immediate increase after refeeding (ANOVA:F2,29=1.31, P=0.285). Moreover, after the second fasting event the IFmice showed a trend for bodymass to increase above that of the AL con-trol animals (0.69 g higher) although this difference was not statistical-ly significant (ANOVA: F2,29=2.39, P=0.109). Different from the firsttwo starvation days, no significant bodymass loss relative to themassesof the other groups was not observed in IF mice after the third fastingday and no between groups difference in body mass was detected(ANOVA: F2,29=2.70, P=0.08). Free feeding following the third starva-tion day caused a pronounced weight increase in the IF mice above ALcontrol levels which lasted for 3 continuous free feeding days(ANOVA: F2,29=4.57, P=0.019; F2,29=4.10, P=0.027; F2,29=4.20,P=0.025; respectively. Post hoc Tukey test: IF vs AF, Pb0.05; IR vs AF,PN0.05). As the treatment progressed, this response of elevated bodymass weakened. Significant mass losses in IF mice occurred on fastingdays 4 to 9 (ANOVA, Pb0.05) while body mass on the free feedingdays did not increase significantly above the levels of AL control or IR

mice (ANOVA, PN0.05). Over the 42 days of treatment, there was asignificant effect of intermittent starvation on body mass withingroups but not between groups (GLM repeated measures: day of treat-ment, F2,29=63.29, Pb0.001; day of treatment∗group, F2,29=12.80,Pb0.001; group, F2,29=0.48, P=0.622; Fig. 1B).

All mice in cohort 1 and cohort 2 had similar fatness before thetreatment started (IF: 3.52±0.50 g; IR: 3.44±0.57 g; AL: 3.64±0.70 g: ANCOVA: F2,62=0.18, P=0.836; Fig. 1C). After 3 starvationdays and 9 free feeding days on day 12, there was a significant differ-ence in body fat mass between three groups (ANCOVA: F2,29=4.24,P=0.025; Fig. 1C). IF mice had higher body fat than AL control(Post hoc Tukey test: P=0.020) while no difference was found be-tween IR mice and AL control (P=0.068) or between IF and IR(P=0.816)(Fig. 1C). On day 24 (after 6 starvation days and 18 freefeeding days), the fat mass of IF mice were significantly higher thanAL control (ANCOVA: F2,29=5.02, P=0.014; Post hoc Tukey test: IFvs AL, P=0.01; Fig. 1C). No differences were found between the IRgroup and the AL control or between the IR and IF groups (Post hocTukey test: P=0.162; P=0.384, respectively). However, the differ-ences in body fatness observed at days 11–13 had disappeared byday 30 (after 8 starvation days and 22 free feeding days) and day 37(after 8 starvation days and 29 free feeding days) (ANCOVA:F2,29=1.66, P=0.209; F2,29=1.56, P=0.227, respectively; Fig. 1C).

3.2. Food intake

3.2.1. Cohort 1Before intermittent starvation treatment started, all mice had a

similar amount of food intake (IF: 2.64±0.31 g; IR: 2.67±0.51 g;2.59±0.33 g; ANOVA: F2,29=0.12, P=0.899; Fig. 2A). The IF micein cohort 1 displayed hyperphagia on the first two free feeding daysafter the fasting days (ANOVA: days 5–6, days 8–9, day 11:pb0.001; day 12, p=0.012; Fig. 2A). By the end of this treatment(day 13), cumulative food intake differed significantly between thethree groups (ANOVA: F2,29=5.73, P=0.008; Fig. 2B). Overall, IRmice consumed 2.79 g (9%) less food (P=0.012) than AL controlswhile the amount of cumulative food intake consumed by IF micewas similar to that of AL control (P=0.902) over the course of thetreatment (Fig. 2B).

3.2.2. Cohort 2Under the free feeding baseline condition, therewas no difference in

food intake between groups (ANOVA: PN0.05). On the first free feedingday following the first starvation (day 5), IF and IR mice consumed sig-nificantly more food than the AF control (ANOVA: F2,29=12.41,Pb0.001; Post hoc Tukey test: Pb0.001, P=0.029, respectively;Fig. 2C). On the second free feeding day (day 6), the food intake of IFmice was still higher than AL control (ANOVA: F2,29=11.917,Pb0.001; Post hoc Tukey test: P=0.002; Fig. 2C) while the food intakeof the IRmice returned to the similar level of AL control (Post hoc Tukeytest: P=0.858). Compared to AL controls, no increase in food intake fol-lowing the treatment days was found in IR mice from the second re-striction day onwards (Post hoc Tukey test: PN0.05 in all cases). Incontrast, IF mice showed a hyperphagic response to starvation on thefirst two free feeding days that followed both the second and third re-strictions (days 8–9 and days 11–12) (ANOVA: day 8, F2,29=12.58,Pb0.001; day 9, F2,29=6.42, P=0.005; day 11, F2,29=33.83, Pb0.001;day 12, F2,29=5.11, P=0.013; Post hoc Tukey test: IF vs AL, Pb0.05for all cases; Fig. 2C). Themagnitude of hyperphagiawas less on the sec-ond refeeding day than that on the first refeeding day. The patterns incohort 2 matched those of cohort 1 over the first 13 days of treatmentalmost exactly. From days 14–38, hyperphagia following fasting in theIF group was present only on the first free feeding days after starvation(ANOVA: day 15, F2,29=11.21, Pb0.001; day 17, F2,29=16.86, Pb0.001;day 21, F2,29=18.44, Pb0.001; day 27, F2,29=4.56, P=0.019; day 30,F2,29=10.51, Pb0.001; Post hoc Tukey test: IF vs AF, Pb0.05 for all

A

B

C

18192021222324252627

-2 0 2 4 6 8 10 12

Bod

y m

ass

(g)

Days of treatment

IF IR AL

* *

* * *

18

19

20

21

22

23

24

25

26

Bod

y m

ass

(g)

Days of treatment

IF IR AL

* *

**

* * * * *

*

Bod

y fa

t m

ass

(g)

AL IR IF

ab

aba

ab

-2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42

0

1

2

3

4

5

6

Day0 (Cohort1+2) Day12 (Cohort1) Day24 (Cohort2) Day30 (Cohort2) Day37 (Cohort2)

Fig. 1. Effects of intermittent fasting/food restriction on body mass and fatness in female C57BL/6J mice. A. Body mass over 13 days of intermittent fasting/food restriction treatment(cohort 1). B. Body mass over 42 days of intermittent fasting/food restriction treatment (cohort 2). C. Body fat mass at different time points during intermittent fasting/food restric-tion treatment (a combination of cohorts 1 and 2). Filled triangles represent intermittently fasted mice (IF), Filled squares represent intermittently food restricted mice (IR), opensquares represent control fed ad libitum (AL). Filled bars represent intermittently fasted mice (IF), Slashed bars represent intermittently food restricted mice (IR), open bars rep-resent control fed ad libitum (AL). *, Pb0.05. Different letters over bars on the same treatment day indicate significant difference.

380 L.-N. Zhang et al. / Physiology & Behavior 105 (2012) 376–387

cases; Fig. 2C). Unexpectedly, on thefirst free feedingday after theninthstarvation (day 40), IF and IR mice consumed significantly less foodthan AL control (ANOVA: F2,29=29.19, Pb0.001; Post hoc Tukey test:IF vs AL, Pb0.001; IR vs AL, P=0.01; Fig. 2C). There was no significantdifference between three groups in cumulative food intake over42 days of the treatment (ANOVA: F2,29=2.56, P=0.095; Fig. 2D).

3.3. Physical activity

Fig. 3A illustrates physical activity over a free feeding day prior tothe experimental treatment starting. Mice from all groups showed asimilar circadian pattern in physical activity with a low level of activityduring the light phase (punctuated by a brief rise when disturbed to befed at 1500 h) and a high level of activity during the dark phase. On thefirst fasting/restriction day (day 4), there was no significant differencein 24 h physical activity between the three groups (GLM repeated

measures: day 4, F2,29=0.28, P=0.758; Fig. 3B). When the treatmentproceeded to the second fasting/restriction day, 24 h physical activitylevels tended to differ between the three groups although the differencedid not reach significance (GLM repeated measures: day 7, F2,29=3.08,P=0.062; Fig. 3C). From the third fasting/restriction day (day 10) andonwards, there was a significant effect of fasting/restriction on 24 hphysical activity (GLM repeated measures: day 7, F2,29=9.42,P=0.001; Post hoc Tukey test: IF vs AL, P=0.001, IR vs AL, P =0.104;IF vs IR, P=0.086, Fig. 3D; data on days 14, 16, 20, 26, 29 and 39not shown). In addition to changes in 24 h activity, there was no differ-ence in daily average activity between the three groups on the firsttwo fasting/restriction days (ANOVA: PN0.05 for both cases; Fig. 3E).However, IF mice showed a significant decrease in daily averagephysical activity on the third to ninth fasting/starvation days (ANOVA,Pb0.05, Post hoc Tukey test, IF vs AL, Pb0.05) and IR mice significantdecreased daily average activity on the fifth to ninth fasting/starvation

A B

C

D

-2 0 2 4 6 8 10 12

Dai

ly f

ood

inta

ke (

g)

Days of treatment

IF IR AL

*

* *

*

**

*

*

* *

AL IR IFCum

ulat

ive

food

inta

ke (

g) a b b

Dai

ly f

ood

inta

ke (

g)

Days of treatment

IF IR AL

*

*

**

* *

*

* **

*

*

*

* *

**

*

* *

* *

*

AL IR IFCum

ulat

ive

food

inta

ke (

g) a ab

00.5

11.5

22.5

33.5

44.5

5

0

20

40

60

80

100

120

0

1

2

3

4

5

6

-2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40

05

10152025303540

Fig. 2. Effects of intermittent fasting/food restriction on food intake in female C57BK/6Jmice. A. Daily food intake over 13 days of intermittent fasting/food restriction treatment (cohort 1).B. Cumulative food intake over 13 days of intermittent fasting/food restriction treatment. C. Daily food intake over 42 days of intermittent fasting/food restriction treatment (cohort 2).D. Cumulative food intake over 42 days of intermittent fasting/food restriction treatment. Filled triangles represent intermittently fastedmice (IF), Filled squares represent intermittentlyfood restricted mice (IR), open squares represent control fed ad libitum (AL). Filled bars represent intermittently fasted mice (IF), Slashed bars represent intermittently food restrictedmice (IR), open bars represent control fed ad libitum (AL). *, Pb0.05. Different letters over bars on the same treatment day indicate significant difference.

381L.-N. Zhang et al. / Physiology & Behavior 105 (2012) 376–387

days (ANOVA, Pb0.05, Post hoc Tukey test, IR vs AL, Pb0.05) (Fig. 3E).Furthermore, day of treatment representing the number of events hada significant effect on physical activity over the 13 fasting/restrictiondays (GLM repeated measures: group, F2,29=7.27, P=0.003; group∗-day of treatment, F2,16=3.31, Pb0.001; Fig. 3E).

The first day of refeeding (day 5) following the first fasting/restric-tion (day 4) 24 h physical activity levels differed significantly be-tween groups (GLM repeated measures: F2,29=24.62, Pb0.001). IFand IR mice displayed lower activity levels than AL control (Posthoc Tukey: IF vs AL, Pb0.001; IR vs AL, P=0.001; IF vs IR, P=0.025;Fig. 3F). Similar responses persisted on the second day of refeedingon day 6 (GLM repeated measures: F2,29=7.20, P=0.003; Post hocTukey: IF vs AL, P=0.002; IR vs AL, P=0.312; IF vs IR, P=0.076)even though the magnitude of responses became weaker. On thethird day of refeeding (day 13) after the third fasting/restrictionevent (day 10), the difference in physical activity between groups to-tally disappeared (GLM repeated measures: F2,29=0.80, P=0.462;Fig. 3G). Additionally, IF mice and IR mice displayed a decrease indaily physical activity on the first refeeding days following all

fasting/restriction events (ANOVA: Pb0.05) except the fourth oneon day 14 (ANOVA: P=0.081) and on the second refeeding daysafter the first-third, fifth, seventh and ninth fasting/restriction events(ANOVA: Pb0.05) (Fig. 3H). Moreover, day of treatment (represent-ing number of events) had a significant effect on physical activity dur-ing all free feeding and refeeding days (GLM repeated measures:group, F2,29=3.74, P=0.037; group∗day of treatment F2,64=3.39,Pb0.001). Overall, total activity levels of IF mice were significantlylowered by 17% in comparison with AL control over 42 days of treat-ment compared with AL controls (ANOVA, F2,29=4.79, P=0.017;Post hoc Tukey test, IF vs AL, P=0.017) whereas no differenceswere found between IR mice and AL control (P=0.82) (Fig. 3I). Con-sistent with changes in total daily activity, there were significant dif-ferences between three groups in activity during the light phase(ANOVA: F2,29=11.13, Pb0.001). However, activity levels wereunchanged during the dark phase across the three groups (ANOVA:F2,29=2.22, P=0.128). In cohort 2, intermittent fasting and restric-tion had exactly the same impact on physical activity as during thefirst 13 days of treatment in cohort 1.

A B C

D E F

G H I

18:0

0

20:0

0

22:0

0

00:0

0

02:0

0

04:0

0

06:0

0

08:0

0

10:0

0

12:0

0

14:0

0

16:0

0

IF

IR

AL

18:0

0

20:0

0

22:0

0

00:0

0

02:0

0

04:0

0

06:0

0

08:0

0

10:0

0

12:0

0

14:0

0

16:0

0

18:0

0

20:0

0

22:0

0

00:0

0

02:0

0

04:0

0

06:0

0

08:0

0

10:0

0

12:0

0

14:0

0

16:0

0

18:0

0

20:0

0

22:0

0

00:0

0

02:0

0

04:0

0

06:0

0

08:0

0

10:0

0

12:0

0

14:0

0

16:0

0

0

2

4

6

8

10

12

1 2 3 4 5 6 7 8 9

Dai

ly a

vera

ge p

hysi

cal a

ctiv

ity

(AU

)

Days of fasting/restriction

**

** *

* *

18:0

0

20:0

0

22:0

0

00:0

0

02:0

0

04:0

0

06:0

0

08:0

0

10:0

0

12:0

0

14:0

0

16:0

0

Phy

sica

l act

ivit

y (A

U)

18:0

0

20:0

0

22:0

0

00:0

0

02:0

0

04:0

0

06:0

0

08:0

0

10:0

0

12:0

0

14:0

0

16:0

0

Dai

ly a

ctiv

ity

(AU

)

Days of treatment

** ** ** ** * ** * **

IF IR AL

Acc

umul

ated

phy

sica

l act

ivit

y (A

U)

a ab

1 3 5 7 9 111315171921232527293133353739410

50100150200250300350400450

0

5

10

15

20

25

0

2

4

6

8

10

12

0

5

10

15

20

25

0

5

10

15

20

25

30

0

5

10

15

20

25

30

35

0

5

10

15

20

25

0

5

10

15

20

25

30

Phy

sica

l act

ivit

y (A

U)

Phy

sica

l act

ivit

y (A

U)

Phy

sica

l act

ivit

y (A

U)

24h PA at baseline 24h PA on 1st starving day 24h PA on 2nd starving day

24h PA on 3rd starving day Daily PA on 9 starving days 24h PA on 1st refeeding day

24h PA on 3rd refeeding day Daily PA on refeeding days Accumulated PA over 42 days

Fig. 3. Effects of intermittent fasting/food restriction on physical activity in female C57BL/6Jmice. A. Representative graphof 24 h physical activity at baseline. B. 24 h physical activity on thefirst fasting/food restriction day (GLM repeatedmeasures: F2,29=0.28, P=0.758). C. 24 h physical activity on the second fasting/food restriction day (GLM repeatedmeasures: F2,29=3.08,P=0.062). D. 24 h physical activity on the third fasting/food restriction day (GLM repeatedmeasures: F2,29=9.42, P=0.001). E. Daily average physical activity on 9 fasting/food restrictiondays (GLM repeatedmeasures: group, F2,29=7.27, P=0.003; group∗day of treatment, F2,16=3.31, Pb0.001). F. 24 h physical activity on thefirst day of refeeding following thefirst fasting/restriction (GLM repeated measures: F2,29=24.62, Pb0.001). G. 24 h physical activity on the third day of refeeding after the third fasting/restriction event (GLM repeated measures:F2,29=0.80, P=0.462). H. Daily average physical activity on refeeding days. I. Accumulated physical activity over 42 days of intermittent fasting/food restriction treatment. Filled trianglesrepresent intermittently fasted mice (IF), Filled squares represent intermittently food restricted mice (IR), open squares represent control fed ad libitum (AL). Filled bars represent inter-mittently fasted mice (IF), Slashed bars represent intermittently food restricted mice (IR), open bars represent control fed ad libitum (AL). *, Pb0.05. PA, physical activity. Different lettersover bars on the same treatment day indicate significant difference.

382 L.-N. Zhang et al. / Physiology & Behavior 105 (2012) 376–387

3.4. Body temperature

All mice showed a similar circadian pattern of body temperaturewhen fed ad libitum (Fig. 4A). On day 4, 24 h fasting and food restric-tion both led to pronounced hypothermia (GLM repeated measures:F2,29=32.52, Pb0.001; Post hoc Tukey test: IF vs AL, Pb0.001; IR vsAL, P=0.002; Fig. 4B). IF mice had a significantly lower body temper-ature compared with IR mice (Post hoc Tukey test: IF vs IR, Pb0.001).The effect of fasting/restriction on daily average temperature was ob-served on each fasting/restriction day (ANOVA: Pb0.001 in all 9cases; Fig. 4C). Compared with control fed ad libitum, average bodytemperature in IF and IR mice on fasting/restriction days was de-creased by 2.48 °C and 0.90 °C, respectively.

Body temperature was returned to the AL control levels after 1 dayof refeeding (GLM repeated measures: F2,29=1.21, P=0.313;

Fig. 4D). There were no significant differences in daily average tem-perature across the three groups on all refeeding days (ANOVA:PN0.05 in all cases; Fig. 4E). In cohort 2, the effects of intermittentfasting and restriction on body temperature were exactly the sameas in cohort 1 for the first 13 days of treatment.

3.5. Circulating hormone and metabolite levels

Circulating levels of leptin were positively correlated with body fatmass on both day 13 and day 42 of the treatment, (Linear regression:day 13, F1,30=7.34, P=0.011, adjusted R2=0.17; Fig. 5A; day 42,F1,30=12.13, P=0.001, adjusted R2=0.288; Fig. 5B). On day 13, therewas no significant difference in circulating leptin levels across thethree groups (ANCOVA using fat mass as a covariate: F2,28=0.139,P=0.87; Fig. 5C). However on day 42 IF mice had a significantly

A B

C D

E

Bod

y te

mpe

ratu

re (

°C)

IF IR AL

25

27

29

31

33

35

37

39

Bod

y te

mpe

ratu

re (

°C)

Days of fasting/restriction

***

Bod

y te

mpe

ratu

re (

°C)

Dai

ly a

vera

ge T

b (°

C)

Dai

ly a

vera

ge T

b (°

C)

Days of treatment

25

27

29

31

33

35

37

39

25

27

29

31

33

35

37

39

25

27

29

31

33

35

37

39

2527293133353739

18:0

0

20:0

0

22:0

0

00:0

0

02:0

0

04:0

0

06:0

0

08:0

0

10:0

0

12:0

0

14:0

0

16:0

0

18:0

0

20:0

0

22:0

0

00:0

0

02:0

0

04:0

0

06:0

0

08:0

0

10:0

0

12:0

0

14:0

0

16:0

0

1 2 3 4 5 6 7 8 9

18:0

0

20:0

0

22:0

0

00:0

0

02:0

0

04:0

0

06:0

0

08:0

0

10:0

0

12:0

0

14:0

0

16:0

0

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41

Fig. 4. Effects of intermittent fasting/food restriction on body temperature in female C57BK/6J mice. A. Representative graph of 24 h body temperature at baseline. B. 24 h bodytemperature on a typical fasting/food restriction day (GLM repeated measures: F2,29=32.52, Pb0.001). C. Daily average body temperature on 13 fasting/food restriction days.D. 24 h body temperature on a typical refeeding day. E. Daily average body temperature on refeeding days. Filled triangles represent intermittently fasted mice (IF), Filled squaresrepresent intermittently food restricted mice (IR), open squares represent control fed ad libitum (AL). Filled bars represent intermittently fasted mice (IF), Slashed bars representintermittently food restricted mice (IR), open bars represent control fed ad libitum (AL). ***, Pb0.001.

383L.-N. Zhang et al. / Physiology & Behavior 105 (2012) 376–387

elevated plasma leptin levels comparedwith the AL control and IRmice(ANCOVA using fat mass as a covariate: F2,28=4.92, P=0.015; ANOVApost hoc Tukey: IF vs AL, P=0.043, IF vs IR, P=0.002; Fig. 5C). Circulat-ing insulin were unchanged on days 13 and 42 (ANOVA, F2,29=1.60,P=0.220; F2,29=1.65, P=0.209; Fig. 5D). Intermittent fasting andfood restriction had no effect on TG or NEFA levels (ANOVA: PN0.05for both TG and NEFA on days 13 and 42; Fig. 5E,F).

3.6. Hypothalamic neuropeptides gene expression

On day 13 (cohort 1) after 3 starvation days and 10 free feedingdays, levels of NPY gene expression in arcuate nucleus was signifi-cantly elevated by 49% in IF mice compared with AL control(ANOVA: F2,29=3.39, P=0.047; Post hoc Tukey test: IF vs AL,P=0.039; Fig. 6A). No differences were detected between IR miceand AL control at this time point (Post hoc Tukey test: IR vs AL,P=0.257; Fig. 6A). However, the difference in NPY had disappearedafter 42 days of treatment (cohort 2) (ANOVA: F2,29=0.22,P=0.805; Fig. 6A). There were no effects of intermittent fasting/res-triction on the levels of expression of POMC, CART, AgRP, Ob-Rb andSOCS 3 (ANOVA: PN0.05 for all on days 13 and 42; Fig. 6B–F).

4. Discussion

4.1. Effects of intermittent starvation on body mass and fatness in mice

We examined the changes in body mass and fatness in response toelevated starvation risk. Our finding that after 13 days of treatmentthere was a significant increase in both body mass and fat mass inthe group that received intermittent complete food removal supportsthe suggestion that starvation risk defines the lower interventionlevel, according to the dual intervention points model [20]. However,mice that were stochastically exposed to only 50% food restrictionshowed no difference in body mass compared to controls that werefed ad libitum. This lack of response does not support the suggestionthat starvation risk increases the lower intervention point. Moreover,when intermittent starvation was prolonged for a greater duration,the effects on body mass and adiposity were reduced and not signifi-cantly greater than the AL controls. The absence of significant eleva-tion in body mass in response to the longer-term intermittentfasting is also at odds with the prediction from the dual interventionpoint model. It seemed that the immediate response of the mice to in-creased starvation risk was indeed to elevate fatness, but that as timewent on different strategies were employed. One possibility is that a

A B

C D

E F

Pla

sma

lept

in (

ng/m

l)

Fat mass (g)

IF IR AL

Pla

sma

lept

in (

ng/m

l)

Fat mass (g)

IF IR AL

Day 13 Day 42

Pla

sam

lept

in (

ng/m

l) IF IR AL

a

b b

Day 13 Day 42P

lasa

m in

sulin

(ng

/ml)

IF IR AL

Day 13 Day 42

Pla

sam

TG

(ng

/ml)

IF IR AL

Day 13 Day 42

Pla

sam

NE

FA

(ng

/ml)

IF IR AL

0

5

10

15

20

0

2

4

6

8

10

12

14

16

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0

2

4

6

8

10

12

14

16

0

0.5

1

1.5

2

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1 1.5 2 2.5 3 3.5 1 1.5 2 2.5 3

Fig. 5. Effects of intermittent fasting/food restriction on circulating hormones and metabolites in female C57BK/6J mice. A. Linear regression plots of plasma leptin levels againstbody fat mass on day 13 of intermittent fasting/food restriction treatment (Linear regression: F1,30=7.34, P=0.011, adjusted R2=0.17). B. Linear regression plots of plasma leptinlevels against body fat mass on day 42 of intermittent fasting/food restriction treatment. (Linear regression: F1,30=12.13, P=0.001, adjusted R2=0.288). C. Plasma leptin levels onday 13 and day 42 of intermittent fasting/food restriction treatment (ANCOVA using fat mass as a covariate: F2,28=4.92, P=0.015). D. Plasma insulin levels on days 13 and 42 ofintermittent fasting/food restriction treatment. E. Plasma triglyceride (TG) levels on days 13 and 42 of intermittent fasting/food restriction treatment. F. Plasma non-esterified fattyacid (NEFA) levels on days 13 and 42 of intermittent fasting/food restriction treatment. Filled triangles represent intermittently fasted mice (IF), Filled squares represent intermit-tently food restricted mice (IR), open squares represent control fed ad libitum (AL). Filled bars represent intermittently fasted mice (IF), Slashed bars represent intermittently foodrestricted mice (IR), open bars represent control fed ad libitum (AL). Different letters over bars on the same treatment day indicate significant difference.

384 L.-N. Zhang et al. / Physiology & Behavior 105 (2012) 376–387

short term intermittent starvation might provoke a more efficientcompensation to defend the negative energy balance, thus generatingover compensation in body mass.

Our result from long-term intermittent starvation is consistent withthe finding that Swiss mice and striped hamsters (C. barabensis) alsoshowed no change in body mass after exposure to stochastic food sup-plies for 4 weeks [23]. In their study, however, Swiss mice and stripedhamsters showed no immediate response of bodymass gain that atten-uated over time. However, in this latter study the occurrence of fastingdays was far more frequent than in the present study. Nevertheless,similar findings have been seen in humans. Healthy women subjectsundergoing a restricted diet on 4 days each week exhibited no substan-tial weight change after 4 weeks [27]. However, in other studies whereADF regimens were administered for 12 weeks, bodyweight was foundto decrease [28,29], but, whenADF regimenswere applied for 16 weeks,no effect on bodyweightwas observed [30,31]. Such findings imply thatanimals were unable to consume twice their daily food intake on therefeeding day for longer periods, which resulted in body weight loss.

There clearly is variability in the capacity of animals to compensate fora fast day on a fed day [32] particularly if fasting days are as frequentlyencountered as the fed days, and this may compromise their capacity toelevate body fatness. Furthermore, there could be longitudinal changesin this capacity to compensate dependent on the duration of treatment.

4.2. Physiological and behavioral responses to intermittent starvation

4.2.1. Food intakeIn the present study, mice experiencing intermittent fasting dis-

played pronounced hyperphagia on the first one or two days ofrefeeding days after the fasting day, which largely accounted for theweight restoration/elevation. It seemed necessary to have at least2 days between the fasting days to build up a fat reserve and hencethe difference between the current study and the previous studiesby Zhao and colleagues may rest primarily in the frequency of fasting.If this frequency is too high there may be limited capacity to compen-sate on the fed days. However, the magnitude of hyperphagia was

A B

C D

E F

Day 13 Day 42

NP

Y e

xpre

ssio

n (%

of

AL

con

trol

)

IF IR ALa

ab

b

Day 13 Day 42

PO

MC

exp

ress

ion

(% o

f A

L c

ontr

ol)

Day 13 Day 42CA

RT

exp

ress

ion

(% o

f A

L c

ontr

ol)

Day 13 Day 42AgR

P e

xpre

ssio

n (%

of

AL

con

trol

)

Day 13 Day 42ObR

b ex

pres

sion

(%

of

AL

con

trol

)

Day 13 Day 42SOC

S3 e

xpre

ssio

n (%

of

AL

con

trol

)

0

50

100

150

200

250

0

20

40

60

80

100

120

140

160

180

200

0

20

40

60

80

100

120

140

160

180

200

0

20

40

60

80

100

120

140

160

0

20

40

60

80

100

120

140

160

180

200

0

50

100

150

200

250

Fig. 6. Effects of intermittent fasting/food restriction on gene expression of neuropeptides in the hypothalamus in female C57BK/6J mice. A. Neuropeptide Y. B. Pro-opiomelanocor-tin (POMC). C. Cocaine- and amphetamine-regulated transcript (CART). D. Agouti-related peptide, (AgRP). E. Signaling form of leptin receptor (Ob-Rb). F. Suppressor of cytokinesignaling 3 (SOCS 3). Filled bars represent intermittently fasted mice (IF), Slashed bars represent intermittently food restricted mice (IR), open bars represent control fed ad libitum(AL). Different letters over bars on the same treatment day indicate significant difference.

385L.-N. Zhang et al. / Physiology & Behavior 105 (2012) 376–387

associated with the severity of food deprivation since IF mice con-sumed more food than IR mice on the refeeding days. So the failureof IR mice to deposit a fat reserve was clearly not because theydid not have the capacity to eat sufficient food on the days whenfood was available. Post fasting or restriction food compensationhas been observed in a number of previous studies e.g. [33–35]. Forinstance, rats showed a dramatic post fast hyperphagia and conse-quently a rapid recovery of body weight lost during starvation[36,37]. Similarly, an initial hyperphagic response following refeedingwas observed in C57BL/6 mice previously on 40% food restriction [33].Hambly et al. (2007) further found that the magnitude of the hyper-phagic response at refeeding was independent of the prior length ofcaloric restriction [34]. In contrast, golden hamsters (Mesocricetusauratus) of both sexes showed little or no post fast food compensation

after 24 h of food deprivation and their daily food intake was nogreater than daily intake during baseline testing [38]. Despite ahyperphagic response on refeeding days, IF mice consumed a similartotal amount of food over the 13 days of intermittent fasting treatment,while IR mice consumed significantly less food compared to AL con-trols. However, with an even longer duration of treatment, the differ-ence in accumulated food intake between IR mice and the AL controldisappeared and mice from all the three groups consumed similartotal amount of food over 42 days of intermittent fasting/restrictiontreatment. It has been shown that the amount of food consumed byrodents maintained on an ADF regimen varies depending on the strainand, in some cases (e.g., C57BL/6 mice), the animals gorge during thenon-fasting time period to an extent that overall food intake is essen-tially equivalent to that of animals fed ad libitum [32]. Wan et al.

386 L.-N. Zhang et al. / Physiology & Behavior 105 (2012) 376–387

(2003) reported that rats on ADF regimen consumed 30% less food overtime compared with rats fed ad libitum [39]. The intermittent starva-tion regimen provides stochastic food availability together with foodabundance allowing compensation. Our results suggested that the re-sponses to refeeding varied with the extent of starvation. IF micewere able to compensate for 24 h complete food deprivation on refeed-ing days and maintained cumulative food intake at the same level asAL control. However the responses to refeeding in IR mice changedover time. During the first 13 days of treatment, the decrease in cumu-lative food intake with unchanged body mass in IR mice implies adecrease in energy expenditure. On day 42, cumulative food consump-tion in IR mice was normalized to AL control level indicating other ad-aptations e.g. decreased energy expenditure might also be restored.When intermittently 50% food restricted, mice might employed differ-ent responses dependent on the duration of treatment.

4.2.2. Physical activityIn contradiction to previous studies demonstrating an elevation in

physical activity level which occurred in several species such as le-murs (Lemur catta) [40], Siberian hamsters (Phodopus sungorus) [41]and mice [42] in response to food restriction, we observed a decreasein physical activity level in mice when fasted or 50% food restrictedfor 24 h, suggesting a response to decrease energy expenditure andconserve energy. This is consistent with the finding that mice exposedto 20% food restriction showed decreased activity which contributedto 75.5% of altered energy expenditure to compensate for reduced en-ergy intake [43]. Similarly in female monkeys (Macaca mulatta), arapid decrease in physical activity was observed to counteract diet-induced weight loss [44]. Since the energy cost of physical activity isan important component of energy expenditure, this physiological re-sponse to reduce energy expenditure is potentially an integral factorcontributing to protection against excessive weight loss during calo-ric restriction and potentially predisposes to weight re-gain in post-obese individuals [45]. Further analyses of activity data showed thatthe decrease was mainly due to activity during the light phase coinci-dent with the occurrence of hypothermia. However, IF and IR micehad similar levels of activity during the dark phase compared to thecontrols fed ad libitum. Increased activity during food restriction hasbeen previously interpreted to reflect an increase in foraging behav-ior, which may enhance the chances of survival in a free living animal[46]. This inconsistency could be explained by the short duration ofstarvation (24 h), which failed to induce food foraging behavior inmice. Additionally, Gutman et al. (2007) observed that desert goldspiny mice (Acomys russatus) employed two strategies for copingwith food shortage, some significantly reduced activity level concen-trating their activity around feeding time lowering energy expendi-ture while others significantly increased activity level searching forfood and thus losing body mass rapidly [47]. It is currently unclearwhether and how the response in activity is correlated with the se-verity or duration of starvation. On refeeding days after starvation,our mice still maintained a lowered level of physical activity, whichalso contributed to the recovery and the increased body mass of theIF mice. Although it has been suggested that hypoleptinemia caninfluence starvation induced hyperactivity [48,49], in our study thedecreased physical activity during intermittent starvation was ac-companied by either unaltered (13 days) or elevated (42 days) leptinlevels, indicating that lowered leptin was unlikely to be the stimulusof the activity changes.

4.2.3. Body temperatureMice are documented as an attractive model for studying the re-

sponse of body temperature to dietary restriction because body tem-perature of mice is sensitive to reduced food intake [50]. Indeed micefrom our study demonstrated a decrease in body temperature in re-sponse to 24 h of complete fasting or 50% food restriction. Notably,hypothermia was only present when food was deprived or restricted

and refeeding immediately restored body temperature to controllevels. Caloric restriction is associated with lower mean body temper-ature in most homeotherms that have been examined including mice,rats, monkeys and humans [49]. The use of torpor has been describedas one of the key strategies for coping with food shortage in rodents[51,52]. In the present study, to what extent the reduced body tem-perature contributed to the saving of energy and thus the preventionof weight loss remained unclear, particularly when the reduction inbody temperature was restored rapidly following refeeding.

4.3. Effects of intermittent fasting/restriction on neuroendocrine status

Leptin is primarily known for its role in regulating food intake andenergy homeostasis [53]. At the end of 12 days of intermittent fastin-g/restriction, no changes in circulating leptin and insulin levels weredetected. This was unexpected given both the increased body fatnessand food intake in the IF mice. Moreover, after a long-term intermit-tent starvation, IF mice exhibited elevated leptin levels while theirbody fatness and food intake were restored to the same level as ALcontrols. It is possible that leptin sensitivity might have been im-paired; however, the absence of changes in gene expression of Ob-Rb and SOCS 3 suggested this was not the case.

With respect to other neuropeptides involved in eating behavior,NPY gene expression was increased after a short-term intermittentstarvation. NPY and AgRP neurons of the hypothalamic arcuate nucle-us are thought to promote feeding and consistent with an orexigenicrole, fasting induced a 4-fold increase in the basal action potential fre-quency of NPY/AgRP neurons [54]. It seems likely that the elevation inNPY expression stimulated the increased food intake in the IF mice.Similarly, an increase in NPY expression and a parallel decrease inPOMC expression were observed in the hypothalamic arcuate nucleusof rats killed just before rebound hyperphagia following a restrictedfeeding/repletion protocol [55].

5. Conclusions

C57BL/6J mice displayed an increased body mass and fatness in re-sponse to 13 days of intermittent fasting, which were both restoredto baseline levels when the treatment was prolonged to 42 days.Over the first 13 days of treatment the data were consistent withthe dual intervention point model as the mice showed both increasedbody mass and adiposity, however the effect disappeared over themore protracted period of 42 days. Whether the data are consideredconsistent with the dual intervention point model therefore dependscritically on the time course of the experiment. Over 13 days the re-sponse was consistent but over 42 days it was not. In the shortterm, an increase in food intake, decreased physical activity as wellas a reduction in body temperature contributed to the elevatedbody mass and fatness. In contrast, the body mass of mice treatedwith intermittent fasting for 42 days were not significantly elevated.This finding was inconsistent with the dual intervention pointmodel. Intermittent starvation resulted in unaltered cumulativefood intake and meanwhile a decreased physical activity level, sug-gesting an absence of overcompensation in energy intake but a de-crease in energy expenditure in coping with stochastic food.

Overall, mice exhibited different behavioral and physiological re-sponses to intermittent starvation which were critically dependenton the duration of treatment.

Acknowledgment

This workwas funded by a studentship fromUniversity of Aberdeenand AstraZeneca. The authors wish to thank Dr Cathy Wyse for herpatient help with cryostat brain sectioning and for her valuable com-ments to the manuscript. The authors also would like to acknowledgeRachel Sinclair and Dr Lobke Vaanholt for their help with animal work

387L.-N. Zhang et al. / Physiology & Behavior 105 (2012) 376–387

and statistical analyses. Thanks to the staff in animal house for takingcare of the mice.

References

[1] Mercer JG, Speakman JR. Hypothalamic neuropeptide mechanisms for regulatingenergy balance: from rodent models to human obesity. Neurosci Biobehav Rev2001;25(2):101–16.

[2] Morton GJ, Cummings DE, Baskin DG, Barsh GS, Schwartz MW. Central nervoussystem control of food intake and body weight. Nature 2006;443(7109):289–95.

[3] Schwartz MW, Woods SC, Porte Jr D, Seeley RJ, Baskin DG. Central nervous systemcontrol of food intake. Nature 2000;404(6778):661–71.

[4] Coll AP. Effects of pro-opiomelanocortin (POMC) on food intake and body weight:mechanisms and therapeutic potential? Clin Sci 2007;113(3–4):171–82.

[5] Ilnytska O, Argyropoulos G. The role of the agouti-related protein in energy balanceregulation. Cell Mol Life Sci 2008;65(17):2721–31.

[6] Cone RD. Anatomy and regulation of the central melanocortin system. Nat Neurosci2005;8(5):571–8.

[7] Archer ZA, Mercer JG. Brain responses to obesogenic diets and diet-induced obesity.Proc Nutr Soc 2007;66(1):124–30.

[8] Mercer JG, Hoggard N, Williams LM, Lawrence CB, Hannah LT, Trayhurn P. Locali-zation of leptin receptor mRNA and the long form splice variant (ob-rb) in mousehypothalamus and adjacent brain regions by in situ hybridization. FEBS Lett1996;387(2–3):113–6.

[9] Hübschle T, Thom E, Watson A, Roth J, Klaus S, Meyerhof W. Leptin-induced nucleartranslocation of STAT3 immunoreactivity in hypothalamic nuclei involved in bodyweight regulation. J Neurosci 2001;21(7):2413–24.

[10] Tups A, Ellis C, Moar KM, Logie TJ, Adam CL, Mercer JG, et al. Photoperiodic regu-lation of leptin sensitivity in the siberian hamster, phodopus sungorus, is reflectedin arcuate nucleus SOCS-3 (suppressor of cytokine signaling) gene expression.Endocrinology 2004;145(3):1185–93.

[11] Bjørbæk C, Lavery HJ, Bates SH, Olson RK, Davis SM, Flier JS, et al. SOCS3 mediatesfeedback inhibition of the leptin receptor via Tyr

985. J Biol Chem 2000;275(51):

40649–57.[12] Howard JK, Cave BJ, Oksanen LJ, Tzameli I, Bjoørbæk C, Flier JS. Enhanced leptin sen-

sitivity and attenuation of diet-induced obesity in mice with haploinsufficiency ofSocs3. Nat Med 2004;10(7):734–8.

[13] Kievit P, Howard JK, Badman MK, Balthasar N, Coppari R, Mori H, et al. Enhancedleptin sensitivity and improved glucose homeostasis in mice lacking suppressor ofcytokine signaling-3 in POMC-expressing cells. Cell Metab 2006;4(2):123–32.

[14] Król E, Duncan JS, Redman P, Morgan PJ, Mercer JG, Speakman JR. Photoperiodregulates leptin sensitivity in field voles, microtus agrestis. J Comp Physiol B2006;176(2):153–63.

[15] Kennedy GC. The role of depot fat in the hypothalamic control of food intake inthe rat. Proc R Soc Lond B Biol Sci Jan 15 1953;140(901):578–96.

[16] Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, Friedman JM. Positional cloningof the mouse obese gene and its human homologue. Nature 1994;372(6505):425–32.

[17] Wirtshafter D, Davis JD. Set points, settling points, and the control of body weight.Physiol Behav 1977;19(1):75–8.

[18] BerthoudHR. Homeostatic and non-homeostatic pathways involved in the control offood intake and energy balance. Obesity (Silver Spring) 2006;14(Suppl 5).

[19] Levitsky DA. Putting behavior back into feeding behavior: a tribute to george collier.Appetite 2002;38(2):143–8.

[20] Speakman JR. A nonadaptive scenario explaining the genetic predisposition toobesity: the “predation release” hypothesis. Cell Metab Jul 2007;6(1):5–12.

[21] Wang T, Hung CCY, Randall DJ. The comparative physiology of food deprivation:from feast to famine; 2006 [cited 31 October 2008].

[22] Cao J, Zhang L-, Zhao Z-. Trade-off between energy budget, thermogenesis and be-havior in swiss mice under stochastic food deprivation. J Therm Biol 2009;34(6):290–8.

[23] Zhao Z-, Cao J. Plasticity in energy budget and behavior in Swiss mice and stripedhamsters under stochastic food deprivation and refeeding. Comp Biochem PhysiolA Mol Integr Physiol 2009;154(1):84–91.

[24] Siegel I, Liu TL, Nepomuceno N, Gleicher N. Effects of short-term dietary restrictionon survival of mammary ascites tumor-bearing rats. Cancer Invest 1988;6(6):677–80.

[25] Krizova E, Simek V. Effect of intermittent feeding with high-fat diet on changes ofglycogen, protein and fat content in liver and skeletal muscle in the laboratorymouse. Physiol Res 1996;45(5):379–83.

[26] Johnston SL, Peacock WL, Bell LM, Lonchampt M, Speakman JR. PIXImus DXA withdifferent software needs individual calibration to accurately predict fat mass.Obes Res 2005;13(9):1558–65.

[27] Laessle RG, Platte P, Schweiger U, Pirke KM. Biological and psychological correlates ofintermittent dieting behavior in youngwomen. Amodel for bulimia nervosa. PhysiolBehav 1996;60(1):1–5.

[28] Hsieh EA, Chai CM,HellersteinMK. Effects of caloric restriction on cell proliferation inseveral tissues in mice: role of intermittent feeding. Am J Physiol Endocrinol Metab2005;288(5):51–5.

[29] Ahmet I, Wan R, Mattson MP, Lakatta EG, Talan M. Cardioprotection by intermittentfasting in rats. Circulation 2005;112(20):3115–21.

[30] Descamps O, Riondel J, Ducros V, Roussel A-. Mitochondrial production of reactiveoxygen species and incidence of age-associated lymphoma in OF1 mice: effect ofalternate-day fasting. Mech Ageing Dev 2005;126(11):1185–91.

[31] Mager DE, Wan R, Brown M, Cheng A, Wareski P, Abernethy DR, et al. Caloric re-striction and intermittent fasting alter spectral measures of heart rate and bloodpressure variability in rats. FASEB J 2006;20(6):631–7.

[32] Anson RM, Guo Z, de Cabo R, Iyun T, Rios M, Hagepanos A, et al. Intermittent fast-ing dissociates beneficial effects of dietary restriction on glucose metabolism andneuronal resistance to injury from calorie intake. Proc Natl Acad Sci U S A May 132003;100(10):6216–20.

[33] Clinthorne JF, Adams DJ, Fenton JI, Ritz BW, Gardner EM. Short-term re-feeding ofpreviously energy-restricted C57BL/6 male mice restores body weight and body fatand attenuates the decline in natural killer cell function after primary influenza in-fection. J Nutr 2010;140(8):1495–501.

[34] Hambly C, Mercer JG, Speakman JR. Hunger does not diminish over time in miceunder protracted caloric restriction. Rejuvenation Res Dec 2007;10(4):533–42.

[35] Evans SA, Messina MM, Knight WD, Parsons AD, Overton JM. Long-Evans andSprague–Dawley rats exhibit divergent responses to refeeding after caloric re-striction. Am J Physiol Regul Integr Comp Physiol Jun 2005;288(6):R1468–76.

[36] Müller-Riemenschneider F, Reinhold T, Berghöfer A, Willich SN. Health-economicburden of obesity in europe. Eur J Epidemiol 2008;23(8):499–509.

[37] Robin J-, Decrock F, Herzberg G, Mioskowski E, Le Maho Y, Bach A, et al. Restorationof body energy reserves during refeeding in rats is dependent on both the intensityof energy restriction and the metabolic status at the onset of starvation. J Nutr2008;138(5):861–6.

[38] Silverman HJ, Zucker I. Absence of post fast food compensation in the goldenhamster (Mesocricetus auratus). Physiol Behav 1976;17(2):271–85.

[39] Wan R, Camandola S, MattsonMP. Intermittent fasting and dietary supplementationwith 2-deoxy-D-glucose improve functional andmetabolic cardiovascular risk factorsin rats. FASEB J 2003;17(9):1133–4.

[40] Geiser F. Metabolic rate and body temperature reduction during hibernation anddaily torpor; 2004 [cited 13 January 2011].

[41] Ruf T, Klingenspor M, Preis H, Heldmaier G. Daily torpor in the Djungarian hamster(Phodopus sungorus): interactions with food intake, activity and social behaviour. JComp Physiol B 1991;160(6):609–15.

[42] Williams TD, Chambers JB, Henderson RP, Rashotte ME, Overton JM. Cardiovascularresponses to caloric restriction and thermoneutrality in C57BL/6J mice. Am J PhysiolRegul Integr Comp Physiol 2002;282(5):51–5.

[43] Hambly C, Speakman JR. Contribution of different mechanisms to compensationfor energy restriction in the mouse. Obes Res 2005;13(9):1548–57.

[44] Sullivan EL, Cameron JL. A rapidly occurring compensatory decrease in physicalactivity counteracts diet-induced weight loss in female monkeys. Am J PhysiolRegul Integr Comp Physiol 2010;298(4).

[45] Leibel RL, Rosenbaum M, Hirsch J. Changes in energy expenditure resulting fromaltered body weight. N Engl J Med 1995;332(10):621–8.

[46] Cornish ER, Mrosovsky N. Activity during food deprivation and satiation of sixspecies of rodent. Anim Behav 1965;13(2–3):242–8.

[47] Gutman R, Yosha D, Choshniak I, Kronfeld-Schor N. Two strategies for coping withfood shortage in desert golden spiny mice. Physiol Behav 2007;90(1):95–102.

[48] Hillebrand JJG, Kas MJH, van Elburg AA, Hoek HW, Adan RAH. Leptin's effect onhyperactivity: potential downstream effector mechanisms. Physiol Behav2008;94(5):689–95.

[49] Exner C, Hebebrand J, Remschmidt H, Wewetzer C, Ziegler A, Herpertz S, et al.Leptin suppresses semi-starvation induced hyperactivity in rats: implications foranorexia nervosa. Mol Psychiatry 2000;5(5):476–81.

[50] Rikke BA, Johnson TE. Physiological genetics of dietary restriction: uncoupling thebody temperature and body weight responses. Am J Physiol Regul Integr CompPhysiol Oct 2007;293(4):R1522–7.

[51] Gutman R, Choshniak I, Kronfeld-Schor N. Defending body mass during food re-striction in Acomys russatus: a desert rodent that does not store food. Am J PhysiolRegul Integr Comp Physiol Apr 2006;290(4):R881–91.

[52] Génin F, Perret M. Daily hypothermia in captive grey mouse lemurs (Microcebusmurinus): effects of photoperiod and food restriction. Comp Biochem Physiol BBiochem Mol Biol 2003;136(1):71–81.

[53] Leibel RL. The role of leptin in the control of body weight. Nutrition Reviews2002;60(10 I).

[54] Takahashi KA, Cone RD. Fasting induces a large, leptin-dependent increase in theintrinsic action potential frequency of orexigenic arcuate nucleus neuropeptide Y/Agouti-related protein neurons. Endocrinology 2005;146(3):1043–7.

[55] Lauzurica N, García-García L, Pinto S, Fuentes JA, Delgado M. Changes in NPY andPOMC, but not serotonin transporter, following a restricted feeding/repletion protocolin rats. Brain Res 2010;1313(C):103–12.