N U T R I T I O N R E S E A R C H 3 3 ( 2 0 1 3 ) 1 0 9 – 1 1 9

Ava i l ab l e on l i ne a t www.sc i enced i r ec t . com

www.n r j ou rna l . com

Time-restricted feeding of rapidly digested starches causesstronger entrainment of the liver clock in PER2::LUCIFERASEknock-in mice

Misa Itokawaa, Akiko Hiraoa, Hiroki Nagahamaa, Makiko Otsukaa, Teiji Ohtsua,Naoki Furutani a, Kazuko Hiraob, Tamao Hatta c, Shigenobu Shibataa,⁎a Laboratory of Physiology and Pharmacology, School of Advanced Science and Engineering, Waseda University, Tokyo, Japanb Aikoku Gakuen Junior College, Edogawa, Tokyo, Japanc National Institute of Animal Health, National Agriculture and Food Research, Tsukuba City, Japan

A R T I C L E I N F O

Abbreviations: AUC, area under the bloorestricted feeding; ZT, Zeitgeber time.⁎ Corresponding author. Department of Elect

University, Wakamatsu-cho 2-2, Shinjuku-kuE-mail address: shibatas@waseda.jp (S. S

0271-5317/$ – see front matter © 2013 Elsevihttp://dx.doi.org/10.1016/j.nutres.2012.12.004

A B S T R A C T

Article history:Received 27 September 2012Revised 20 December 2012Accepted 27 December 2012

Keywords:Mouse

Restricting feeding to daytime can entrain circadian clocks in peripheral organs of rodents,and nutrients that rapidly increase the blood glucose level are suitable for inducingentrainment. However, dietetic issues, for example, whether or not the diet comprisesheated food, have not been fully explored. We therefore hypothesized that rapidly digestedstarch causes stronger entrainment than slowly digested starch. The entrainment ability ofthe liver clock in PER2::LUCIFERASE knock-in mice, blood glucose levels, insulin levels, andacute changes in liver clock gene expression were compared between a β-starch (native)–substituted AIN-93M standard diet and an α-starch (gelatinized)–substituted diet. β-Cornand β-rice starch induced larger phase delays of the liver clock, larger blood glucoseincreases, and higher Per2 gene expression in the liver compared with β-potato starch.Starch granule size, as examined by electron microscopy, was larger for β-potato starchthan for β-corn or β-rice starch. After heating, we obtained gelatinized α-potato, α-corn, andα-rice starch, which showed destruction of the crystal structure and a high level ofgelatinization. No difference in the increase of blood glucose or insulin levels was observedbetween β-corn and α-corn starch, or between β-rice and α-rice starch. In contrast, α-potatostarch caused higher levels of glucose and insulin compared with β-potato starch. An α-potato starch–substituted diet induced larger phase delays of the liver clock than did β-potato starch. Therefore, rapidly digested starch is appropriate for peripheral clockentrainment. Dietetic issues (heated vs unheated) are important when applying basicmouse data to humans.

© 2013 Elsevier Inc. All rights reserved.

StarchCircadian rhythmDigestionInsulinJet lag syndromeLuminescent measurements

d glucose concentration-time curve; FF, free-feeding; PER2::LUC, PER2::LUCIFERASE; RF,

rical Engineering and Biosciences, School of Advanced Science and Engineering, Waseda, Tokyo 162-8480, Japan. Tel.: +81 3 53697318; fax: +81 3 33419815.

hibata).

er Inc. All rights reserved.

110 N U T R I T I O N R E S E A R C H 3 3 ( 2 0 1 3 ) 1 0 9 – 1 1 9

1. Introduction

Endogenous circadian rhythms such as body temperature,food intake, and the sleep-wake cycle are widely observed inanimals includinghumans [1,2]. Themammalian clock systeminvolves clock genes of which Per1/Per2 and Cry1/Cry2 arenegative regulators and Clock/Bmal1 is a positive regulator ofthe transcriptional and translational feedback loop, whichmediates 24-hour RNA and protein level rhythms [1,2]. Normalcircadian rhythm is important for normal physiologicalfunctions, as shown, for example, by the metabolic syndromephenotype of Clock mutant mice [3]. In addition, humanepidemiologic studies involving rotating shift workers suchasnurseshave shown that such individuals are at an increasedrisk for metabolic syndrome [4], diabetes [5], and cancer [6]. Inmodern society, individuals tend to eat a small breakfast orskip it entirely, and then eat a late dinner after work. Thecombination of a late dinner and short sleep duration is clearlyassociated with the risk of obesity in humans [7]. In addition,this risk of obesity has been associatedwith eating dinner after20:00 hours [8]. Thus, understanding the relationship betweencircadian rhythm and feeding habits and/or nutrition isimportant for human health.

The circadian rhythm of clock gene expression can beentrained not only by light-dark signals but also by signalsfrom daily feeding restricted to daytime [9–11]. The expres-sion rhythm of clock genes such as Per1 and Per2 in most ofthe peripheral/nonsuprachiasmatic nucleus tissue can beentrained to restricted-feeding (RF) stimuli [12,13]. Under RFschedules, numerous physiological and metabolic functionsincluding locomotor activity, body temperature, insulinrelease, and corticosterone release become entrained tofood availability.

Nutrient factors associated with RF-induced entrainmentof peripheral clocks have been examined. The sugar compo-nent of the diet has been shown to play an important role inperipheral clock phase shifts [14], and daily injection of insulininto mice was shown to entrain liver circadian rhythm inPER2::LUCIFERASE (PER2::LUC) knock-in mice [15]. In anattempt to elucidate the mechanism of RF-induced entrain-ment, several studies have reported acute increases in Per2and/or Dec1 gene expression in the liver after refeeding a dietunder fasting conditions [13,15,16], suggesting that acuteincreases of Per2 and/or Dec1 gene expression may be animportant first step in inducing RF-induced entrainment. InRF schedules of 2 meals per day, a meal after a 16-hour fastcaused a stronger entrainable effect than another meal afteran 8-hour fast [16], suggesting that food intake after a longerfasting period (ie, breakfast) increases insulin release morethan food intake after a shorter fasting period, and a higherquantity of insulin may cause phase shifts. Taken together,the evidence suggests that glucose-induced insulin release isan important step in the phase shifts of peripheral clocksunder RF conditions.

In rodent experiments, the starch component of thestandard AIN-93M diet is provided in the form of raw starchsuch as β-corn starch. Humans generally consume cookedstarch balanced with protein and fat as opposed to simplenutrients such as glucose. Thus, the present experimental

protocol was prepared in an attempt to model humannutrition, and the current rodent results can be easily appliedto human physiology and society. In this study, we prepared β-potato starch as a representative slowly digested starch and α-potato starch as a representative rapidly digested starch. Wethen compared the entrainment ability of the liver clock,glucose, and insulin increases, as well as acute changes ingene expression of Per2, Bmal1, and Dec1 in the liver using anAIN-93M diet in which β- and α-corn starch components weresubstituted by β- or α-potato starch components.

We hypothesized that rapidly digested starch causesstronger entrainment of the peripheral clock than slowlydigested starch. In previous RF experiments in mice and rats,food was given during the daytime, which is an inactive phase[9,10,14]. In the present study, the RF protocol was used toadvance rhythms, such as abnormal feeding habits, and thenrefeed at the normal feeding time to test reentrainment inresponse to rapidly or slowly digested starches.

2. Methods and materials

2.1. Animals

PER2::LUCIFERASE knock-in mice [17] were bred in-house.We prepared PER2::LUC homozygous male mice with a C57/BL6J background to mate with Institute of Cancer Research-strain female mice because we wanted a larger body size.From this crossing, we obtained F2 hybrid PER2::LUChomozygous males that weighed 30 to 40 g each at thestart of the experiment. No differences were observed inthe peak time of bioluminescence in the livers of theoriginal male or F1 hybrid mice, as mentioned previously[14]. The animal room was maintained at a controlledtemperature of 22°C ± 2°C, humidity of 60% ± 5%, and a 12-hour light/12-hour dark cycle (lights on from 08:00 to 20:00hours). Zeitgeber time (ZT) of ZT 0 and ZT 12 was used as thelights-on and lights-off time. The light intensity at thesurface of the cages was approximately 100 lux. Before theRF experiment, mice were fed an AIN-93M diet and providedwith water ad libitum. Experimental animal care wasconducted with the permission of the Animal WelfareCommittee of Waseda University (permission no. 2011-A061).

2.2. Preparation of starches

β-Potato, β-corn, and β-rice starch were obtained from WakoPure Chemical Industries, Ltd (Osaka, Japan). α-Starch wasprepared according to the following procedure. First, β-starchin water (10% wt/vol) was mixed for 3 minutes in a glass tube,and the tube was covered with aluminum foil and autoclavedat 120°C for 20 minutes. The tube was then stored at −80°C for2 hours and freeze-dried for 3 days. The freeze-dried starchwas crushed with a mixer and used as α-starch.

2.3. Preparation of food tablets for RF and the RF protocol

For the RF experiments, we prepared diet tablets of 1 to 4 gwith a tableting machine (HANDTAB-100; Ichihashi-Seiki,

111N U T R I T I O N R E S E A R C H 3 3 ( 2 0 1 3 ) 1 0 9 – 1 1 9

Kyoto, Japan). For the control, an AIN-93M formula diet(Oriental Yeast Co Ltd, Tokyo, Japan) with the followingcomposition was prepared: 14% casein, 0.18% L-cystine, 47%corn starch, 15% gelatinized corn starch, 10% sucrose, 4%soybean oil, 5% cellulose powder, 3.5% AIN-93M mineralmixture, 1% AIN-93 vitamin mixture, 0.25% choline bitar-trate, and 0.0008% tert-butyl hydroquinone. For the substi-tution experiments, starch components (β-corn starch and α-corn starch) were substituted with various starches. Theingredient composition of the standard and substituted dietsis shown in Table.

Mice were subjected to RF (AIN-93M) from ZT 0 to ZT 4 for1 week and then refed a food tablet (1-4 g) at ZT 12 after a 32-hour fasting period. The next day, mice were killed forassessment of bioluminescence rhythm. Food consumptionwas checked at ZT 16 in all groups. In some experiments, 1 gof α-potato starch or 1 g of β-potato starch was given. Micethat did not eat all of the food within 4 hours were excludedfrom analysis.

The maximum concentration of the water suspension ofAIN-93M for oral administration via syringe was 0.1 g/mL. Ourpreliminary experiments revealed that neither the 0.1 g tabletnor the 0.1 g suspension/mouse caused a phase shift. Wetherefore used the tablet diet (1-4 g/mouse) and not thesuspension diet (0.1 g/mouse) for phase-shift experiments.

2.4. Estimation of digestion

We did not calculate the overall digested values (kilocaloriesper gram) for the β-potato, β-corn, or β-rice starch. Instead, weexamined the volume of food and dried feces in the 3 groupsfor 4 days after free-feeding (FF) for 7 days. Digestion values (inpercent) were estimated by the following formula: [(foodvolume − feces volume)/food volume] × 100.

2.5. Recording of bioluminescence rhythm

The day after refeeding, PER2::LUC mice were killed at ZT 3by ether anesthesia to record bioluminescence rhythmicityin the liver. A block of liver was rapidly removed from eachmouse and placed in ice-cold Hank balanced salt solution(pH 7.2; Sigma-Aldrich, St Louis, MO, USA). Protocol and

Table – Ingredient composition of the diets for each experimen

Composition (g/100 g) AIN-93M corn starchw

Starch (β-starch) 46.5692 6Gelatinized starch (α-starch) 15.5 0Casein 14.0 1Sucrose 10.0 1Cellulose powder 5.0 5Soybean oil 4.0 4AIN-93M mineral mixture 3.5 3AIN-93 vitamin mixture 1.0 1Choline bitartrate 0.25 0L-Cystine 0.18 0tert-Butyl hydroquinone 0.0008 0

medium composition were the same as those in our previousstudies [14,18]. The original data (1-minute bins) weresmoothed by an adjusting-averaging method with 2-hourrunning means, as described previously [14,18]. The data setwas then detrended by subtracting the 24-hour runningaverage from the raw data with R software (R developmentCore Team; http://www.r-project.org/). Peaks were defined aspoints at which bioluminescence was greater than that onboth sides of each point and were confirmed from thewaveform (Fig. 1).

2.6. Time-course measurements of blood glucose level,insulin level, and liver gene expression

In time-course experiments, feeding time and duration areimportant potential confounding factors that may alter thetime course. To avoid problems related to these factors, we fedthe mice an oral suspension diet administered via syringe, asreported previously [14,15].

The AIN-93M diet substituted with various starches wasdissolved in water at 0.1 g/mL. The suspension diet wasadministered orally (1 mL/mouse) at ZT 12, and blood glucoselevel was measured after 15, 30, 60, 90, 120, 180, and 360minutes (Fig. 2) or after 15, 30, 60, and 90 minutes (Fig. 6). Thearea under the blood glucose concentration-time curve (AUC)was calculated for each group. In some experiments, theblood insulin level was measured 60 minutes after adminis-tration of the suspension diet. Because released insulin mayaffect clock gene expression, mice were killed 120 minutesafter being administered the suspension diet for measure-ment of clock gene expression.

2.7. Measurement of insulin and glucose concentration

We collected blood by vein puncture of the eye underanesthesia with ether. Blood was collected into tubescontaining heparin as an anticoagulant, and the plasmafraction was separated. Samples were stored at 2°C to 8°C forup to 24 hours. Plasma insulin and glucose concentrationwere measured using an Ultra Sensitive Mouse Insulinenzyme-linked immunosorbent assay kit (Mercodia AB,Uppsala, Sweden) and Glucose PILOT kit (Aventir Biotech,

t

AIN-93M diet substitutedith β-potato/β-corn/β-rice

AIN-93M diet substitutedwith α-potato

2.0692 062.0692

4.0 14.00.0 10.0.0 5.0.0 4.0.5 3.5.0 1.0.25 0.25.18 0.18.0008 0.0008

Day 1

RF ZT0-4 for 1 week

re-feeding

ZT3 Sampling

Day 8

Day 9

0 12A

300

200

100

Cou

nts/

Sec

ond

80

0

-80

12 24 36 48 60

Dev

iatio

n fr

om

Mov

ing

Ave

rage

Time lapse [ hr ]0

0 2 4 6 8 10 12

(13)

(14)

(15)

(7)

[ pZT ]

*

**

E( 0g )

B

C

( 0g ) ( 2g )

( 4g )

0 2 4 6 8 10 12 14

D

[ pZT ]

(3)

(10)

(6)

(4)

(4)

(7)

( RF only )

( 0g )

( 1g )

( 2g )

( 3g )

( 4g )

( FF ) (4)

*

*

**

**

( 0g ) ( 2g )

( 4g )

Fig. 1 – Effects of volume of AIN-93M diet and effects of AIN-93Mdiet containing various starches on liver circadian clock phase.A, Experimental protocol. Micewere fed anAIN-93Mdiet tablet at ZT 0 to ZT 4 for 1week. On day 8,micewere refedwith 0 to 4 gAIN-93M diet, and bioluminescence rhythm was recorded on day 9. B and C, Representative raw data and detrended data ofPER2::LUC bioluminescence rhythms in the liver after refeeding the AIN-93M diet. D, Magnitude of phase delay induced byvarious doses of AIN-93M diet. *P < .05, **P < .01 (vs FF; Tukey-Kramer test). E, Magnitude of phase delay induced by AIN-93Mdiet containing β-potato, β-corn, or β-rice starch. Data of the 0-g group are the same as those in panel D. These data wereexcluded from statistical analysis. *P < .05, **P < .01 (vs β-potato; Tukey-Kramer test). Data are presented as means ± SEMvalues. Daytime RFmicewere fed fromZT 0 to ZT 4, and FFmicewere fed ad libitum. The horizontal axis indicates projected ZT(pZT) at the peak of the bioluminescence rhythm. ZT 0 was lights-on time, and ZT 12 was lights-off time in the housing roombefore killing the mice. Numbers in parentheses indicate the number of mice tested.

112 N U T R I T I O N R E S E A R C H 3 3 ( 2 0 1 3 ) 1 0 9 – 1 1 9

LLC, Carlsbad, CA, USA). The range of blood glucose concen-tration that can be measured by this kit is 20 to 600 mg/dL.

2.8. Isolation of total RNA and real-time,reverse-transcription polymerase chain reaction

Tissue messenger RNA was measured by real-time reverse-transcription polymerase chain reaction (RT-PCR), as de-scribed previously [15,16]. Fifteen nanograms of total RNAwas reverse transcribed and amplified with a One-Step SYBRRT-PCR kit (TaKaRa, Otsu, Japan) in an Step One Plus (LifeTechnologies Japan, Tokyo, Japan).

Specific primer pairswere designed based onpublished datafor the 18srRNA, Per2, Bmal1, Ror-γ, Dec1, and Gck genes. The18srRNA, Per2, Bmal1,Ror-γ,Dec1, andGckprimersweredesignedto cross exon-intron boundaries. The primer sequenceswere as follows: mouse 18srRNA: 5′-ggggagtatggttgcaaagc-3′,5′-tgtcaatcctgtccgtgtcc-3′; mouse Per2: 5′-tgtgtgctta-cacgggtgtccta-3′, 5′-acgtttggtttgcgcatgaa-3′; mouse Bmal1: 5′-ccacctcagagccattgataca-3′, 5′-gagcaggtttagttccactttgtct-3′;mouse Ror-γ: 5′-tgcaagactcatcgacaagg-3′, 5′-aggggattcaacat-cagtgc-3′; mouse Dec1: 5′-atcagcctcctttttgccttc-3′, 5′-agcatttctc-

cagcataggcag-3′; mouse Gck: 5′-tccctgtaaggcacgaagac-3′, 5′-acgatgttgttcccttctgc-3′. Real-time RT-PCR was performed asdescribed in our previous studies [15,16]. The relative levels ofthe target genePCRproductwerenormalized to thoseof18srRNA.Data were analyzed by the delta-delta Ct method. A melt curveanalysis was performed to check for nonspecific products.Results indicated no amplification of nonspecific products.

2.9. Scanning electron microscopy

Dried starch powder samples weremounted directly on roundaluminum stubs using double-sided tape, coated with 12-nmgold, and examined and photographed in a JSM-5600LVscanning electron microscope (JEOL Ltd, Tokyo, Japan) at anaccelerating voltage of 5 kV.

2.10. Measurement of particle size distribution

The particle size distribution of the starch granules wasmeasured by dry-type method in a Laser Micronsizer LMS-2000e (Seishin Enterprise Co, Ltd, Tokyo, Japan). Measurementwas performed with a dispersion unit equipped with a 1-shot

Day 0

RF ZT0-4 for 1 week

Oral administration

Day 8

0 12 A

0

50

100

150

200

250

0 60 120 180 240 300 360

Blo

od g

luco

se [

mg/

dL ]

Time lapse [ min ]

water

β-potato

β-corn

β-rice

B

0

0.2

0.4

0.6

0.8

Blo

od in

sulin

[ µg

/L ]

(5) (9)

(7)

(6)

* **

0

2000

4000

6000

8000

10000

12000

water β-potato β-corn β-rice

AU

C [

min

mg/

dL ]

(5) (5)

(5)

(5)

##

##

*

water β-potato β-corn β-rice

DC

Fig. 2 – Effects of AIN-93M diet containing various starches on blood glucose and insulin levels. A, Experimental protocol.White triangles show the blood sampling time for glucose level, and black triangles show the blood sampling time forinsulin level. B, Time course of glucose level after oral administration of suspension of AIN-93M substituted with β-potato,β-corn, or β-rice starch. C, Area under the blood glucose concentration-time curve values for blood glucose. D, Seruminsulin level 60 minutes after oral administration of suspension diet. *P < .05, **P < .01 (vs β-potato starch; Tukey-Kramertest). ##P < .01 (vs water; Tukey-Kramer test). Data are presented as means ± SEM values. Numbers in parentheses indicatethe number of mice tested.

113N U T R I T I O N R E S E A R C H 3 3 ( 2 0 1 3 ) 1 0 9 – 1 1 9

nozzle (dispersion pressure, 0.3 MPa). Average particle sizesare expressed according to median diameter (d50).

2.11. Assessment of x-ray diffraction pattern

Each sample was prepared by suspending starch granules inwater, placing them on a slide with water, and drying. Thesamples were then observed under transmitted light. X-raypowder diffraction patterns of starches were determinedusing a RAD-X system (Rigaku Corp, Tokyo, Japan) under thefollowing conditions: 40 kV; 5 mA; scan step, 0.02°; scan

speed, 2°/min; slits, 1° (divergence slit); 0.3 mm (receivingslit); and 1° (scattering slit).

2.12. Statistical analyses

Data are presented as means ± SEM values. Statisticalsignificance between the 2 groups was determined using theStudent t test. Statistical analysis formultiple comparisonswasperformed by 1-way or 2-way analysis of variance (ANOVA)with Tukey-Kramer post hoc comparisons. Differences be-tweenmeanswere considered statistically significant at P < .05.

0

1

2

3

4

5

Rel

ativ

e m

RN

A L

evel

Per2

0

0.5

1

1.5

2

Rel

ativ

e m

RN

A L

evel

Bmal1

0

0.5

1

1.5

2

Rel

ativ

e m

RN

A L

evel

Ror-γ

0

1

2

3

4

5

6

Rel

ativ

e m

RN

A L

evel

Dec1

01234567

Rel

ativ

e m

RN

A L

evel

Gck

(4)

(4)

(3)

(3) ##

#

(4) (4)

(4)

(4)

(4) (4)

(4)

(4)

(3) (3) (3)

(3)

(3)

(3) (3) (3)

B

Day 0

RF ZT0-4 for 1 week

Oral administration

Day 8

0 12 A

Fig. 3 – Effect of AIN-93M diet containing various starches onclock gene expression in the liver. A, Experimental protocol.The black triangles show sampling time of the liver. B, Micewere killed by ether anesthesia 120 minutes after oraladministration of suspension of AIN-93M diet substitutedwith β-potato, β-corn, or β-rice starch. Clock gene expressionlevels were measured. The relative levels of target genepolymerase chain reaction products were normalized tothose of the 18srRNA gene. #P < .05, ##P < .01 (vswater; Tukey-Kramer test). Data are presented as means ± SEM values.Numbers in parentheses indicate the number of mice tested.

114 N U T R I T I O N R E S E A R C H 3 3 ( 2 0 1 3 ) 1 0 9 – 1 1 9

3. Results

3.1. AIN-93M diet substituted with potato starch induceda weak phase delay of the liver clock by refeeding

In previous experiments, we found that the liver clock wasphase advanced and showed a food volume dependenceunder daytime RF conditions [14]. In the present experimentalprotocol, the effect of food volume on phase shift of the liverclock was examined. The experimental protocol is shown inFig. 1A. Representative raw data for bioluminescence rhythmare shown in Fig. 1B and C; a food volume–dependent phasedelay was observed. Summarized data are shown in Fig. 1D,and a significant volume dependency was observed (F6,31 = 80,P < .01 by ANOVA, P < .05 vs FF). Body weight (average 35.0 g)decreased to 29.1 g in the 0-g fed group, 29.6 g in the 1-g fedgroup, 30.0 g in the 2-g fed group, 31.6 g in the 3-g fed group,and 33.4 g in the 4-g fed group. Thus, body weight after fastingand refeeding was 83% to –95% of baseline values.

Because the magnitude of phase shifting caused by 3-g or4-g food volume was close to that obtained with FF, wedecided to use 2-g food tablet (85% control bodyweight) for thenext experiment. We then examined the effect of AIN-93Msubstituted with 3 kinds of starch on phase shift (Fig. 1E). β-Corn and β-rice starch showed a larger phase delay than β-potato starch (F2,39 = 7.2, P < .01 by ANOVA, *P < .05 and **P < .01vs β-potato starch).

3.2. Blood glucose concentration in response to AIN-93Mdiet substituted with various starches

Each type of starch had a different effect on the magnitude ofphase shift of the liver clock. We hypothesized that thisdifferencewas caused by digestibility; therefore, wemeasuredthe blood glucose level after oral administration of a suspen-sion diet (Fig. 2A). The time course for blood glucose level isshown in Fig. 2B. The onset of glucose increase (15 minutesafter administration) was similar among all starch groupsbecause the diet included 5% glucose in all groups. However,from 30 to 90 minutes after administration, the blood glucoselevel in the β-potato starch group showed a slower increasecompared with that in the β-corn or β-rice starch group. TheAUC values for β-corn and β-rice starch were larger than thosefor β-potato starch (F3,16 = 18.1, P < .01 by ANOVA, *P < .05 and**P < .01 vs β-potato starch; Fig. 2C). Blood insulin levels afteroral administration of β-corn starch were significantly higherthan after oral administration of β-potato starch (F3,23 = 4.3, P <.05 by ANOVA, *P < .05 vs β-potato starch; Fig. 2D).

Digestion values (in percent) were high for the β-corn (92%± 0.3%, n = 6) and β-rice (92% ± 0.8%, n = 6) starch groups andlow for the β-potato starch group (34% ± 5.4%, n = 6). Thus, lowdigestive ability of β-potato starch may cause the low levels ofblood glucose and insulin.

3.3. Expression of Per2, Bmal1, Ror-γ, Dec1, and Gckmessenger RNA

Refeeding after fasting is reported to increase the expressionof clock genes such as Per2, Bmal1, Ror-γ, Dec1 and the

gluconeogenesis gene Gck in the liver and to decrease theexpression of Rev-erbα [15,16]. We therefore examined theeffect of 3 different starches on clock gene expressions in theliver. The experimental protocol is shown in Fig. 3A. The micewere killed 120 minutes after the oral administration of each

0

200

400

600

0 10 20 30 40 50

Inte

nsity

[ cp

s ]

0

200

400

600

0 10 20 30 40 50

0

200

400

600

0 10 20 30 40 50

A

05

101520253035404550

Mea

n pa

rtic

le s

ize

[ µm

]

B

**

**

**

(3)

(3)

(3)

0

100

200

300

400

0 10 20 30 40 50

0

200

400

6

800

00

0 10 20 30 40 50

C

0

200

400

600

0 10 20 30 40 50

Inte

nsity

[ cp

s ]

[ ° ] [ ° ]

Fig. 4 – Results of scanning electronmicroscopy. A Scanning electronmicroscopy images of β-potato,β-corn,β-rice, α-potato, α-corn, and α-rice starch. Scale: 10 μm. B, Average particle size of β-potato, β-corn, and β-rice starch granules. **P < .01 (Tukey-Kramer test). Data are presented as means ± SEM values. Numbers in parentheses indicate the repeat number of the test. C, X-ray diffraction pattern of β-potato, β-corn, β-rice, α-potato, α-corn, and α-rice starch.

115N U T R I T I O N R E S E A R C H 3 3 ( 2 0 1 3 ) 1 0 9 – 1 1 9

(9)

(10)

(7)

(6)

(6) (5)

**

0

0.2

0.4

0.6

0.8

1

potato corn rice

Blo

od in

sulin

[µg/

L]

RF ZT0-4 for 1 week

Oral administration

0 12

α-starch

β-starch

Day 0

Day 8

B

A

Fig. 5 – Effect of AIN-93M diet containing β-potato, β-corn, β-rice,α-potato,α-corn, orα-rice starchonblood insulin level.A,Experimental protocol. Black triangles show blood samplingtime. B, Blood insulin levels were measured 60 minutes afteroral administration of suspension of AIN-93M diet substitut-ed with β- or α-starch. **P < .01, (vs β-potato starch; Tukey-Kramer test). Data are presented as means ± SEM values.Numbers in parentheses indicate the number of mice tested.

116 N U T R I T I O N R E S E A R C H 3 3 ( 2 0 1 3 ) 1 0 9 – 1 1 9

suspension diet. The expression of Per2 and Gck was signifi-cantly increased in β-corn starch group compared with thecontrol (water) group (F3,10 = 6.3, P < .05 by ANOVA for Per2; F3,10= 7.0, P < .01 for Gck; #P < .05 and ##P < .01 vs water; Fig. 3B).However, in the current experimental protocol, no increase inBmal1 or Ror-γ or a decrease in Rev-erbα expression in the liverwas detected after administration of the AIN-93M diet.

3.4. Starch granule size and shape and x-ray powderdiffraction patterns

β-Corn and β-rice starch had a larger effect on phase shift andclock gene expression in the liver than β-potato starch,indicating that these events are dependent on whether or notthe starch is rapidly digestible.We assumed that any differencein digestion and absorption would be dependent on starchgranule size and shape. Granule sizewas significantly larger forβ-potato starch than for β-corn or β-rice starch by scanningelectron microscopy and assessment of particle size distribu-tion (F2,6 = 6700, P < .01 byANOVA; Fig. 4A, B). The granule shapeof β-potato starch formed a prolate spheroid. The granule shapeof β-corn and β-rice starchwas polygonal. After gelatinization ofβ-potato, β-corn, and β-rice starch, we found that granules werecompletelybrokendown(Fig. 4A). X-raydiffractionanalysiswasapplied to determine if the crystal structure of the starchgranule had been broken down. Gelatinization was regarded ascomplete if the starch granule of α-potato, α-corn, and α-ricestarch had been destroyed (Fig. 4C).

3.5. Comparison of β- and α- starch on insulin level

Given the differences in granule size and shape between β-and α-starch, we compared blood insulin levels in response tofeeding with β- and α-starch. The experimental protocol isshown in Fig. 5A. Two-way ANOVA revealed a significantdifference between β- and α-starch (F1,37 = 8.9, P < .01), but nointeractions between β- and α-starch and other starches (F2,37= 2.0, P = .15). Insulin levels were high in all α-starches, andalmost no difference in insulin levels was observed between β-and α-corn starch, or between β- and α-rice starch. In contrast,α-potato starch significantly increased insulin levels com-pared with β-potato starch (**P < .01 vs β-starch, Student t test;Fig. 5B). A large difference in insulin levels between β- and α-starch was seen in potato starch, but not in corn or rice starch.Therefore, a further experimentwas conducted using β- and α-potato starch.

3.6. Comparison of β- and α-potato starch

Because there was a significant difference in insulin responsebetween mice fed with β- and α-potato starch, we comparedglucose increase and clock gene expression in the liver inresponse to feeding with β- or α-potato starch. To determinewhether α-potato starch is rapidly digested compared with β-potato starch, blood glucose level was assessed 15, 30, 60, and90 minutes after oral administration of suspension of AIN-93M substituted with each starch. The experimental protocolis shown in Fig. 6A. Two-way ANOVA revealed a significantdifference in interactions (time course × β/α starch; F4,25 = 7.3, P< .01; Fig. 6B). The AUC for blood glucose level for α-potato

starchwas significantly higher than that for β-potato starch (*P< .05 vs β-starch, Student t test; Fig. 6C). Similar to the resultspresented in Fig. 3B, the expression levels of clock genes wereexamined after oral administration of each starch. Per2 andRor-γ expression was increased in response to administrationof α-potato starch (*P < .05 vs β-starch, Student t test; Fig. 6D).

3.7. Refeeding induced phase shift of the liver clock by β-or α-potato starch

We compared the magnitude of phase delay in the liver clockin response to intake of anAIN-93Mdiet tablet containing β- orα-potato starch. Before examining the effect of β- or α-potatostarch on phase shift, we determined the eating speedbecause experimental results might be affected by rate offood intake. We measured the remaining food volume 30, 60,90, 120, 180, 240, and 300 minutes after intake of 2 g of β- or α-potato starch tablets at ZT 6 (Fig. 7A). Two-way ANOVArevealed significant differences in β- and α-potato starch (F1,63= 49, P < .01) and also in time course (F6,63 = 8.1, P < .01), but notin interactions (F6,63 = 1.3, P = .25). Mice ate α-potato starchslower than β-potato starch. As shown in Fig. 7A, 2 g of AIN-93M containing β-potato starch was almost fully eaten within120 minutes. However, 0.5 g of AIN-93M containing α-potatostarch remained 300 minutes after intake. Given that 1-gtablets of both were consumed within 120 minutes, wedecided to use 1-g tablets for this experiment. The experi-mental protocol is shown in Fig. 7B. Food intake of 1 g of β- orα-potato starch caused phase delay of liver clock compared

00.5

11.5

22.5

3

Rel

ativ

e m

RN

A L

evel

Per2 D

0

50

100

150

200

250

0 30 60 90

Blo

od g

luco

se [

mg/

dL ]

Time lapse [ min ]

B

( 3 )

( 4 )

0

2000

4000

6000

8000

10000

AU

C [

min

mg/

dL ]

C *

(3)

(4)

0

0.5

1

1.5

Rel

ativ

e m

RN

A L

evel

Bmal1

0

0.5

1

1.5

2

Rel

ativ

e m

RN

A L

evel

Ror-

0

0.5

1

1.5

2

2.5

Rel

ativ

e m

RN

A L

evel

Dec1

0

1

2

3

4

Rel

ativ

e m

RN

A L

evel

Gck* *

(3)

(4)

(3) (3) (3)

(3)

(4)

(4)

(4) (4)

Day 0

RF ZT0-4 for 1 week

Oral administration

Day 8

0 12 A

** ** **

Fig. 6 – Effect of AIN-93M diet containing β- or α-potato starch on blood glucose level and clock gene expression in the liver. A,Experimental protocol. White triangles show the blood sampling time for glucose level, and black triangles show the samplingtime for liver. B, Time course of blood glucose level after oral administration of suspension of AIN-93M diet substituted with β-or α-potato starch. **P<.01, (vs β-potato starch; Student t test). C, Area under the blood glucose concentration-time curve valuesfor blood glucose. D, Mice were killed by ether anesthesia 120 minutes after oral administration of the diet. Clock geneexpression levels were measured. The relative levels of the target gene polymerase chain reaction product were normalized tothose of the 18srRNA gene. *P < .05, (vs β-potato starch; Student t test). Data are presented as means ± SEM values. Numbers inparentheses indicate the number of mice tested.

117N U T R I T I O N R E S E A R C H 3 3 ( 2 0 1 3 ) 1 0 9 – 1 1 9

with the no-food (0 g) group (Fig. 7C). The magnitude of phasedelay induced by α-potato starch was larger than that inducedby β-potato starch (**P < .01, Student t test; Fig. 7C).

4. Discussion

We first demonstrated that β-potato starch substituted forcorn starch in the AIN-93M diet has a weak effect on RF-induced entrainment of the liver clock, diet-induced Per2 geneexpression in the liver, and diet-induced increases in bloodglucose and insulin levels compared with β-corn or β-ricestarch. Crystal structure and granule size were examined bymicroscopy and particle size counting, respectively. Thegranule size for β-potato starch was larger than that for β-corn or β-rice starch. Thus, digestion of β-potato starchrequired more time than that of β-corn or β-rice starch. Inour previous experiment, we reported that high amylose cornstarch causesweak entrainment of the liver clock, with aweakincrease in blood glucose level [14]. Therefore, speed of

digestion of starch may be an important factor for entrain-ment of the liver clock. The granule size for β-corn starch wassmaller than that for β-potato starch, supporting our previousobservation that both β-corn starch and α-corn starch inducelarge phase shifts and blood glucose increases, with nosignificant differences between groups [14]. In the presentstudy, we examined the effect of the AIN-93M diet in whichthe starch component was substitutedwith other starches; wedid not examine the effect of a 100% starch diet. When peopleeat meals, they generally eat many nutrient componentstogether, not each nutrient alone. Another reason why we didnot use 100% starch is that mice did not consume a sufficientvolume of pellets of 100% starch within 4 hours (data notshown). Therefore, we evaluated the effect of each starch inthe AIN-93M diet with substitution of starch components.

In the subsequent experiments, we prepared β- and α-potato starch and compared them accordingly. The largestarch granules and crystal structure of β-potato starch weredestroyed by heating, yielding highly gelatinized α-potatostarch. α-Potato starch showed a stronger effect on phase

118 N U T R I T I O N R E S E A R C H 3 3 ( 2 0 1 3 ) 1 0 9 – 1 1 9

entrainment of the liver clock, increases in blood glucose andinsulin, and increases in Per2 and Ror-γ gene expression in theliver. Thus, α-potato starch had a similar or greater ability toinduce entrainment of the liver clock compared with β-corn orβ-rice starch. Interestingly, mice took more time to eat AIN-

0

0.5

1

1.5

2

0 60 120 180 240 300

Am

ount

of f

ood

[ g ]

Time lapse [ min ]

β-potato ( 6 )

α-potato ( 5 )

**

** ** *

A

Day 1

RF ZT0-4 for 1 week

re-feeding ( 1g )

ZT3 Sampling

Day 8

Day 9

0 12 B

0 2 4 6 8 10 12 [ pZT ]

(5)

(5)

(7)

**

( β-potato )

( α-potato )

( 0g )

C

Fig. 7 – Refeeding induced phase shift of the liver clock by α-or β-potato starch. A, Time course of remaining food volumein the trough after 2-g AIN-93M diet substituted with α- or β-potato starch. *P < .05, **P < .01 (vs β-potato starch; Student ttest). B, Experimental protocol. Mice were fed an AIN-93Mdiet tablet at ZT 0 to ZT 4 for 1 week. On day 8, mice wererefed a 1-g AIN-93M diet substituted with α- or β-potatostarch, and bioluminescence rhythm was recorded on day 9.C, Magnitude of phase delay of PER2 rhythm by the 1-g AIN-93M diet substituted with α- or β-potato starch. Data of 0-ggroup are the same as those in Fig. 1D. These data wereexcluded from statistical analysis. **P < .01 (vs β-potatostarch; Student t test). Data are presented as means ± SEMvalues. The horizontal axis indicates projected ZT (pZT) atthe peak of the bioluminescence rhythm. ZT 0 was lights-ontime, and ZT 12 was lights-off time in the housing roombefore killing the mice. Numbers in parentheses indicate thenumber of mice tested.

93M containing α-potato starch compared with AIN-93Mcontaining β-potato starch. For the α-potato starch diet, theytook 1 hour for consumption of 0.5 g, 2 hours for 1 g, and 4 to 5hours for 1.5 g, whereas for the β-potato starch diet, they took1 hour for consumption of 1.5 g and 2 hours for 2 g. Although α-potato starch may be more digestible than β-potato starch, ittakes longer to eat. We suspect that α-potato starch powderbecomes sticky and may be more difficult for mice to eat.Actually, mice ate α-potato starch containing pellets contin-uously during the observation period. Therefore, we com-pared β- and α-potato starch with a 1-g feeding schedule inthis entrainment experiment. The results of our study supportour hypothesis because α-potato (rapidly digested starch)produced a large phase delay compared with β-potato (slowlydigested starch).

The present experimental protocol for entrainment of theliver clock was different from that of our previous study [14],in which we gave approximately 2 g of food at ZT 6 for 2 daysafter a 24-hour fasting period and killed the mice the nextday at ZT 3. Results showed a 2- to 3-hour phase advance ofthe liver clock compared with FF mice. In the present study,we first entrained mice to an RF schedule of 4 hours from ZT0 to ZT 4 for 7 days, and then we gave 1 to 4 g of food on thenext day at ZT 12 after a 32-hour fasting period. Resultsshowed a 6- to 8-hour phase delay of the liver clock, similarto the phase observed under FF conditions. The presentexperimental protocol has 2 advantages for circadian rhythmexperiments. One is that mice subjected to RF for 4 hoursfrom ZT 0 to ZT 4 are forced to phase advance in liverrhythm under the light-dark cycle, exhibiting the “jet lagsyndrome.” Given that food intake at ZT 12 may normalizeliver rhythm, we evaluated an experimental protocol thatcould normalize the circadian rhythm. The other advantageis that food intake at ZT 12 for 1 day caused a large phasedelay (6-8 hours). Therefore, we were able to evaluate smalldifferences among the experimental groups.

The mechanism underlying the larger phase shift by α-potato starch compared with β-potato starch may refer to themagnitude of insulin release induced by these diets. The α-potato starch increased the blood glucose and insulin levelsin the present study. We have reported that insulin releasein response to refeeding after fasting for 24 hours is animportant step for entrainment of the liver clock, and insulinadministration in vivo and in vitro produce phase shift ofcircadian rhythms [18]. Refeeding and insulin injectionincrease Per2 gene expression in the liver [13,15,16,19].Therefore, an acute increase in Per2 gene expression maybe important for RF-induced entrainment of the liver clock.Per2 gene expression in the liver was strongly increased notonly by β-corn and β-rice starch but also by α-potato starch,whereas β-potato starch induced a weak increase in Per2gene expression in the liver. The present results support theimportance of insulin-induced change in expression of clockgenes for entrainment of the liver clock.

In the present study, oral administration of an α-potatostarch–containing suspension diet produced not only ahigh level of insulin but also a high level of glucosecompared with β-potato starch. Secreted insulin maypromote the uptake of a large level of glucose into livercells, and increased intracellular glucose may cause phase

119N U T R I T I O N R E S E A R C H 3 3 ( 2 0 1 3 ) 1 0 9 – 1 1 9

delay of the liver clock. In behavioral experiments, glucoseapplication canmimic the feeding-induced anticipatory activityrhythm [20]. However, we considered that such a mechanismmight not be involved in insulin-induced phase delay of theliver clock. Glucose application in vitro does not produce anincrease but, rather, a decrease in Per1 and Per2 gene expressionin rat-1 cells [21]. We demonstrated that glucose itself does notcause phase shift of the liver PER2::LUC rhythm [14]. Further-more, glucose administration with insulin injection in strepto-zotocin-treated mice does not further potentiate an insulin-induced increase in Per2 gene expression [15]. Taken together,activation of the insulin signaling pathway but not the insulin-induced glucose increase within cells might be important forRF-induced liver clock entrainment.

One limitation of the present study is thatwe evaluated theeffect of different starches on the phase shift produced by asingle-meal schedule, which contrasts to the 3-meal-per-dayschedule in humans. Another limitation is that the experi-mental data were obtained without precisely controllingfeeding/digestion rate.

In summary,we found that gelatinized starch can be used asan appropriate diet for peripheral clock entrainment, comparedwith native starch. In addition, we suggest that mouse diets infuture studies shouldmimic humandiet as closely as possible ifwe want to apply data from mouse experiments to humandietetic studies. In fact, wedemonstrated that dietetic issues (ie,heated vs unheated food) may be important for entrainment ofperipheral clocks. Digestible starch is a good food material forentrainment of the liver peripheral clock.

Acknowledgment

This work was supported to by the Grants-in-Aid for ScientificResearch (23300278,23659126), the Fuji Foundation for ProteinResearch (2010, 2012), and the Program for Promotion of Basicand Applied Researches for Innovations in Bio-orientedIndustry (given to S.S.).

We declare there is no conflict of interest that could beperceived as prejudicing the impartiality of the researchreported.

R E F E R E N C E S

[1] Dibner C, Schibler U, Albrecht U. The mammalian circadiantiming system: organization and coordination of central andperipheral clocks. Annu Rev Physiol 2010;72:517–49.

[2] King DP, Takahashi JS. Molecular genetics of circadianrhythms in mammals. Annu Rev Neurosci 2000;23:713–42.

[3] Turek FW, Joshu C, Kohsaka A, Lin E, Ivanova G, McDearmonE, et al. Obesity and metabolic syndrome in circadian clockmutant mice. Science 2005;308:1043–5.

[4] Lin YC, Hsiao TJ, Chen PC. Persistent rotating shift-workexposure accelerates development of metabolic syndrome

amongmiddle-aged female employees: a five-year follow-up.Chronobiol Int 2009;26:740–55.

[5] Pan A, Schernhammer ES, Sun Q, Hu FB. Rotating night shiftwork and risk of type 2 diabetes: two prospective cohortstudies in women. PLoS Med 2011;8:e1001141.

[6] Bonde JP, Hansen J, Kolstad HA, Mikkelsen S, Olsen JH, BlaskDE, et al. Work at night and breast cancer–report on evidence-based options for preventive actions. Scand J Work EnvironHealth 2012;38:380–90.

[7] Hsieh SD, Muto T, Murase T, Tsuji H, Arase Y. Association ofshort sleep duration with obesity, diabetes, fatty liver andbehavioral factors in Japanese men. Intern Med 2011;50:2499–502.

[8] Baron KG, Reid KJ, Kern AS, Zee PC. Role of sleep timing incaloric intake and BMI. Obesity (Silver Spring) 2011;19:1374–81.

[9] Mistlberger RE. Food-anticipatory circadian rhythms: con-cepts and methods. Eur J Neurosci 2009;30:1718–29.

[10] Hughes ME, Di Tacchio L, Hayes KR, Vollmers C, Pulivarthy S,Baggs JE, et al. Harmonics of circadian gene transcription inmammals. PLoS Genet 2009;5:e1000442.

[11] Shibata S, Tahara Y, Hirao A. The adjustment and manipu-lation of biological rhythms by light, nutrition, and abuseddrugs. Adv Drug Deliv Rev 2010;62:918–27.

[12] Hara R, Wan K, Wakamatsu H, Aida R, Moriya T, Akiyama M,et al. Restricted feeding entrains liver clock without partici-pation of the suprachiasmatic nucleus. Genes Cells 2001;6:269–78.

[13] Wu T, Ni Y, Kato H, Fu Z. Feeding-induced rapid resetting ofthe hepatic circadian clock is associated with acute inductionof Per2 and Dec1 transcription in rats. Chronobiol Int 2010;27:1–18.

[14] Hirao A, Tahara Y, Kimura I, Shibata S. A balanced diet isnecessary for proper entrainment signals of the mouse liverclock. PLoS One 2009;4:e6909.

[15] Tahara Y, Otsuka M, Fuse Y, Hirao A, Shibata S. Refeedingafter fasting elicits insulin-dependent regulation of Per2 andRev-erbαwith shifts in the liver clock. J Biol Rhythms 2011;26:230–40.

[16] Hirao A, Nagahama H, Tsuboi T, Hirao M, Tahara Y, Shibata S.Combination of starvation interval and food volume de-termines the phase of liver circadian rhythm in Per2::Lucknock-in mice under two meals per day feeding. Am J PhysiolGastrointest Liver Physiol 2010;299:G1045–53.

[17] Yoo SH, Yamazaki S, Lowrey PL, Shimomura K, KoCH, Buhr ED,et al. PERIOD2::LUCIFERASE real-time reporting of circadiandynamics reveals persistent circadian oscillations in mouseperipheral tissues. Proc Natl Acad Sci USA 2004;101:5339–46.

[18] Tahara Y, Hirao A, Moriya T, Kudo T, Shibata S. Effects ofmedial hypothalamic lesions on feeding-induced entrain-ment of locomotor activity and liver Per2 expression in Per2::luc mice. J Biol Rhythms 2010;25:9–18.

[19] Kawamoto T, Noshiro M, Furukawa M, Honda KK, NakashimaA, Ueshima T, et al. Effects of fasting and re-feeding on theexpression of Dec1, Per1, and other clock-related genes. JBiochem 2006;140:401–8.

[20] Stephan FK, Davidson AJ. Glucose, but not fat, phase shiftsthe feeding-entrained circadian clock. Physiol Behav 1998;65:277–88.

[21] Hirota T, Okano T, Kokame K, Shirotani-Ikejima H, Miyata T,Fukada Y. Glucose down-regulates Per1 and Per2mRNA levelsand induces circadian gene expression in cultured Rat-1fibroblasts. J Biol Chem 2002;277:44244–51.