Physiology & Behavior 107 (2012) 338–345

Contents lists available at SciVerse ScienceDirect

Physiology & Behavior

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

No effects of monosodium glutamate consumption on the body weight orcomposition of adult rats and mice

Michael G. Tordoff ⁎, Tiffany R. Aleman, Michelle C. MurphyMonell Chemical Senses Center, Philadelphia, PA 19104, United States

H I G H L I G H T S

► Drinking monosodium glutamate solution had no effect on the development of diet obesity.► This was true for Sprague Dawley rats, obesity-prone rats and C57BL/6J mice.► This was true for several diet formulations.► MSG is not a useful anti-obesity supplement nor is it responsible for exacerbating obesity.

⁎ Corresponding author at: Monell Chemical Senses CentPA 19104‐3308, United States. Tel./fax: +1 267 519 4805.

E-mail address: tordoff@monell.org (M.G. Tordoff).

0031-9384/$ – see front matter © 2012 Elsevier Inc. Allhttp://dx.doi.org/10.1016/j.physbeh.2012.07.006

a b s t r a c t

a r t i c l e i n f o

Article history:Received 8 June 2012Accepted 25 July 2012

Keywords:Dietary obesityObesity-prone ratC57BL/6J mouseHigh-fat dietUmami

Monosodium glutamate (MSG) is pervasively consumed as a flavor enhancer so there are important implica-tions to understanding its physiological actions, particularly its effects on body weight. Previous studies sug-gest that MSG increases, decreases, or has no effect on the body weight of rodents. However, most of thesestudies involved administration of MSG to immature rodents and consequently may not be relevant for un-derstanding human obesity. We report here five experiments in which we measured the body weights of atotal of 32 groups of 10–12 adult rats or mice given various diets to eat and MSG to eat or drink. We found noevidence that MSG influenced body weight, energy intake, or body composition. To the extent that experimentsin rodents illuminate mechanisms involved in human obesity and body weight control, our results suggest thatMSG is unlikely to be a useful anti-obesity supplement but neither is it responsible for exacerbating obesity.

© 2012 Elsevier Inc. All rights reserved.

1. Introduction

Monosodium glutamate (MSG) is widely used as a flavor enhancerso reports that it can influence body weight raise intense public con-cern. It was discovered almost 50 years ago that the repeated admin-istration of high doses of MSG to fetal or infant rodents produceslife-long increases in body weight and fat mass [1–4]. These effectsare secondary to glutamate-induced brain damage occurring beforethe rodent blood–brain barrier has fully developed [e.g., [4,5]]. Conse-quently, their significance for human obesity is tenuous. There areseveral reports showing that when rodents consume MSG (ratherthan when it is administered by intubation or injection) it has no ef-fect on body weight. For example, adding MSG to food (up to 20% ofthe diet) did not alter the body weight of rats [6] even when thiswas done over multiple generations [7]. Similarly, rats or mice given2% MSG to drink from weaning until 14 weeks old had normal bodyweights and carcass lipid contents [2]. Feeding large quantities ofMSG to adult humans or gerbils also had no effects on body weight [8].

er, 3500 Market St, Philadelphia,

rights reserved.

Thus, the consensus opinion is that early fears that MSG consumptionmight cause obesity were misguided [review [9]].

However, recent work has rekindled interest in the possibility thatMSG might influence body weight. One study found a higher preva-lence of overweight in Chinese MSG consumers than nonusers[10,11]. However, this did not replicate in another sample [12] andthe interpretation of both these correlational studies is disputed[13,14]. More troubling, Collison and colleagues found that C57BL/6Jmice fed a high trans-fatty acid diet and given 0.064% MSG solutionto drink had greater increases in abdominal girth than did controls[15,16]. However, virtually the opposite results were obtained byKondoh and Torii [[17] review [18]]. In one of their experiments,rats were given voluntary access to 1% MSG (and water) from age 3to 18 weeks while they ate one of four diets that differed primarilyin fat content. Irrespective of diet, the rats that drank MSG gainedless weight and had modest reductions in intraabdominal and subcu-taneous fat relative to rats without MSG. In another experiment, adultrats fed a high-fat diet and given 1% MSG to drink gained ~10% lessbody weight than did controls fed the same diet but not given MSG.This effect of MSG to reduce body weight is congruent with evidencethat MSG stimulates thermogenesis [17–19], which in the absence ofcompensatory increases in food intake would be expected to lead toweight loss. It is also broadly consistent with a recent study showing

Table 1Parameters of experiments designed to investigate the influence of MSG consumption on body weight.

Exp. Strainand species

Age atstart, wk

No. ofgroups

No.per group

Diet MSG accessconditions

MSG accessduration, wk

Body weight and intakemeasures

1 CD rats 7 8 10 AIN-76A orhigh-energy

None1% MSG in water1% MSG in food3% MSG in food

8 Body weights and intakesevery 2 days

2 C57BL/6Jmice

8 8 10 AIN-76A orhigh-energy

None1% MSG in water1% MSG in food3% MSG in food

12 Body weights and intakesweeklyBody composition at end

3 CD rats 8 8 10 AIN-76ALow-fatHigh-fatUltrahigh-fat

None1% in water

18 Body weights and intakesevery 2 days

4 C57BL/6Jmice

12 4 10 Low-fatUltrahigh-fat

None1% in water

8 Body weights and intakesweeklya

Body composition at end5 OP rats 8b 4 12 AIN-76A or high-energy None

1% in water26 MSG solution intakes daily,

food intakes and body weightsevery 3–4 daysBody composition at end

Notes: All rats and mice were males. CD rats=Sprague Dawley CD rats, OP rats=Obesity-prone Sprague Dawley-derived rats. Diet information is provided in Table 2. Body com-position was assessed at the end of the experiment. Exp.=Experiment number, wk=week. low-fat and ultrahigh fat are identical to “mouse control” and “mouse high-energy”diets prepared by Research Diets.

a We did not measure food intake in this experiment.b MSG access began at 8 weeks, experimental diets began at 9 weeks.

339M.G. Tordoff et al. / Physiology & Behavior 107 (2012) 338–345

that human infants are satiated more rapidly by a formula containingMSG than by isocaloric control formulas [20]. It therefore seemedworthwhile to investigate the parameters supporting this effect morethoroughly. We describe below a series of five experiments that weredesigned to observe—but failed to find—an effect of voluntaryMSG con-sumption on body weight.

2. Methods

2.1. General methods

Table 1 provides critical methodological parameters of each of the5 experiments we conducted. The protocols of all experiments wereapproved by the IACUC of the Monell Chemical Senses Center.

2.1.1. Subjects and maintenanceRats in Experiments 1 and 3 were outbred male Sprague Dawley

CDs obtained from Charles River [Crl:CD(SD) IGS; Strain Code 001;from Portage, MI]. Experiment 5 used obesity-prone (OP) SpragueDawley rats from Charles River [OP-CD, Crl:OP(CD)]. This strain wasselectively bred by Levin and colleagues for its proclivity to gainweight when fed a high-fat diet [21]. Founders were from Crl:CD(SD) stock, and the OP line is maintained as an outbred strain byCharles River.

Mice were used in Experiments 2 and 4. These were male inbredC57BL/6J mice obtained from The Jackson Laboratory (Bar Harbor,ME). Experiment 4 used C57BL/6J mice that were “pre-fattened” atThe Jackson Laboratory by feeding them a high-fat diet (ResearchDiets, D12492; Table 2) according to procedures published on-line(http://jaxmice.jax.org/diomice/index.html).

In all experiments, rats or mice were assigned to equal-sized treat-ment (i.e., MSG access) and control groups by matching subjects withsimilar body weights so the groups had almost identical means andlow variance at the time MSG access began.

2.1.2. Housing conditionsAll animals in an experiment were tested in a single replication

and were housed in the same vivarium on the same or adjacentracks. Rats and mice were housed in different vivariums. Each vivari-um was maintained at 23 °C with fluorescent illumination between

0600–1800 h or 0700–1900 h (depending on the experiment anddaylight savings time).

Each rat was individually housed in a 25×18×19 cm hangingcage, with stainless steel back and side walls and a mesh grid frontwall and floor. Unless otherwise noted, pelleted AIN-76A diet wasavailable from a hopper or glass jar attached to the front wall, orscattered on the cage floor (depending on the experiment). Card-board sheets under the rats' cages collected feces and spillage, andwere changed every 1–3 days. Fluids were available from one(water only) or two (water and MSG) 300-mL glass bottles equippedwith neoprene stoppers and stainless steel sippers. The inverted bot-tles were attached to the front wall with stainless steel springs.

Each mouse was housed individually in a plastic “tub” cage(26.5 cm×17 cm×12 cm) with a stainless steel wire lid, and woodshavings scattered on the floor. The lid included space for pelletedfood and a water bottle [see [22] for details]. The mouse was trans-ferred to a clean cage with fresh bedding once every week, generallyat the time it was weighed (see below).

2.1.3. Diets and fluidsDiet compositions are provided in Table 2. All experiments in-

volved a comparison of animals fed a high-carbohydrate, low-fatdiet with others fed various high-energy diets. In most experimentsthe high-carbohydrate, low-fat diet was AIN-76A diet. We preferthis semisynthetic diet to the more recent AIN‐93 formulation be-cause it is less powdery and thus spillage is easier to collect whenmeasuring food intakes. In Experiments 1, 2 and 5, the high-energydiet was one we have used previously to dissect the contribution offat, carbohydrate and energy content to dietary obesity [23]. It has aratio of 16 kcal% protein, 40 kcal% carbohydrate, and 44 kcal% fat,which reflects human macronutrient intake. In Experiment 3, weattempted to replicate the work of Kondoh and Torii [17] and soused the diet formulations they provided. In Experiment 4, somemice were “pre-fattened” by The Jackson Laboratory by feedingthem a high-fat diet made by Research Diets (New Brunswick, NJ).We continued to feed these mice this diet when they arrived in ourfacility. It was serendipity that the “low-fat” and “high-fat” dietsused by Kondoh and Torii were the same composition as those usedby The Jackson Laboratory.

Table 2Composition and energy content of diets used in each experiment.

Diet name AIN-76A High-energy Low-fat High-fat Ultrahigh fat

Dyets catalogue no.: 100000 112162 112251 102629 112252

Used in experiment: 1,2,3,5 1,2,5 3,4 3 3,4

Ingredient (g/kg)Casein 200.0 200.0 189.6 233.1 258.5L-Cysteine 0 0 2.8 3.5 3.9Methionine 3.0 3.0 0 0 0Corn starch 150.0a 376.1a 298.6 84.8 0Maltodextrin 0 103.9 33.2 116.6 161.5Sucrose 500.0 0 331.8 201.4 88.9Corn oil 50.0 220.0 0 0 0Soybean oil 0 0 23.7 29.1 32.3Lard 0 0 19.0 206.9 316.6Cellulose 50.0 50.0 47.4 58.3 64.6Minerals 35.0b 35.0c 9.5d 11.7d 12.9d

Vitaminse 10.0 10.0 9.5 11.7 12.9Choline bitartrate 2.0 2.0 1.9 2.3 2.6Dicalcium phosphate 0 0 12.3 15.2 16.8Calcium carbonate 0 0 5.2 6.4 7.1Potassium citrate.H2O 0 0 15.6 19.2 21.3Total energy (kcal/kg) 3774 4492 3654 4591 5122

Percent energy asProtein % 19 16 19 19 19Carbohydrate % 69 40 71 35 20Fat % 12 44 11 46 61

Notes: Some diets also contained b100 mg/kg FD&C dye to aid recognition by color. The low-fat, high-fat, and ultrahigh-fat diets contained 5 mg/kg t-butylhydroquinone as a pre-servative. aWhen 1% or 3% MSG was added to these diets, corn starch was reduced by 10 or 30 g/kg, respectively.In Experiment 4, we used “mouse control” (D12350B) and “mouse high-fat” (D12492) diets from Research Diets. These have identical ingredients to diets 112251 and 112252 fromDyets Inc, with the exception that the Dyets formulation uses “Dyetrose”, a proprietary form of maltodextrin that reduces pellet crumbling. The two vendors use slightly differentvalues to convert ingredients into energy values. We have used the values published by Dyets Inc. The low-fat, high-fat, and ultrahigh fat diets are identical to those published byKondoh and Torii ([17] Table 1) but amounts have been adjusted here to provide ingredient weights per kilogram diet and to use the energy values used by Dyets Inc.bMineral mix #200000, cMineral mix #213758, dMineral mix #210088, eVitamin mix #300050.

340 M.G. Tordoff et al. / Physiology & Behavior 107 (2012) 338–345

All diets were prepared and pelleted (~2 cm long pellets) by DyetsInc (Bethlehem, PA) or Research Diets. When MSG was mixed in thediet, we provided the MSG (Affymetrix-USB Corp, catalogue no.16245) to Dyets Inc to pellet, and used MSG from the same batch tomake MSG solution. To compensate for the 10 or 30 g/kg diet ofMSG added to these diets, corn starch was reduced by 10 or 30 g/kg.

All animals had ad libitum food to eat and deionized water todrink. In most experiments, half the animals also had 1%MSG solutionavailable to drink. We used 1% w/v MSG (10 g/L; 59 mM) to followthe work of Kondoh and Torii [17]. This concentration is highly pre-ferred by rodents; they typically drink 85–95% of their daily fluid asMSG when given this choice with water [17,24,25]. Water and MSGbottles were refilled as needed. For this purpose, 1% MSG solutionwas prepared in 3 L batches in a plastic carboy and stored until need-ed in the vivarium. Deionized water was delivered either from aplastic carboy or from a dedicated deionized water source in thevivarium.

2.1.4. Body weight and intake measurementsRats and mice were weighed using top-loading balances with ei-

ther 0.1 g or 0.01 g precision. Food and fluid(s) were measured andreplenished at the same time that body weight was measured. Thefood intake of rats was determined by collecting remaining foodpellets from the cage floor or weighing each rat's food hopper(depending on the experiment) at the start and end of each measure-ment period. Food intake of mice was determined by weighing eachcage lid, which had a built-in food hopper containing food, at thestart and end of each measurement period. In experiments withrats, corrugated cardboard absorbent sheets were placed in traysunder the hanging cages to collect spillage, and this was collectedand accounted for. In experiments with mice, the food droppedonto pine shavings making collection of all spillage impossible butany large pieces of spilled food were weighed. Fluid intakes were

measured by recording the weight of the water bottle(s) and theircontents at the beginning and end of each measurement period.

2.1.5. Body composition analysisThe body composition of mice was measured at the end of Experi-

ment 2 and 4 using a PIXImus II mouse densitometer (#51601; GEMed-ical Systems Lunar, Madison, WI). The PIXImus II uses low energyX-rays to produce high-resolution (0.18×0.18 mm pixel) images. Soft-ware included with the machine identifies and estimates the densityand weight of lean and fat tissue. Each dead mouse was laid out in theprone position on a plastic tray. The limbs were held splayed apartwith tape. The tail was displaced to themouse's left so that it waswithinthe X-ray field but did not overlap a limb. Because the cone-beamX-rayfield is only 80×65 mm and thus too small to fit a large adult mouse,the head of each mouse was excluded from analysis; it was eitherplaced outside the X-ray field or masked using software. The PIXImusII was calibrated daily using a plastic “mouse phantom” provided bythe manufacturer. Assessments of body fat made by the machine arestrongly replicable (r=0.98) and strongly congruent (r=0.95) withbody fat measured by chemical extraction [24].

The X-ray field of the PIXImus is too small to analyze rats. On the lastday of thefinal experiment,we obtained a BrukerOptics LF110 apparatus,which uses time-domain nuclear magnetic resonance to analyze bodycomposition. Each unanesthetized rat was placed into the coil of the ana-lyzer using the provided probe (which is designed to hold rats up to 1 kg)and proprietary software estimated whole-animal lean and fat mass.

2.2. Statistical analyses

Food intakes were converted from gram intakes to energy intakesusing the energy density of the diets listed in Table 2. Fluid intakeswere converted from weights to volumes assuming that 1 g ofwater or MSG solution had a volume of 1 mL. Intakes collected over

Fig. 1. Mean±SE body weights of rats fed AIN-76A (left) or high-energy diet (right) given MSG as well as drinking water or two concentrations of MSG in food (Experiment 1).Standard error bars (vertical lines) obscuring symbols are not shown.

341M.G. Tordoff et al. / Physiology & Behavior 107 (2012) 338–345

several days were divided by the number of days to obtain averagedaily intakes, and these were used in analyses. We initially analyzedintakes using three-way analyses of variance (ANOVAs) with factorsof Diet×MSG availability×Time, but there were no concerted effectsinvolving the Time factor. Consequently, and for brevity, we presenthere only the average daily intakes, which we analyzed with two-way ANOVAs (Diet×MSG availability).

Analysis of body weights is not straightforward because as ani-mals grow, between-subject variability tends to increase (leading toheterogeneity of variance) and growth rates are not independent (vi-olating the requirement for independent observations). This pre-cludes using mixed-design ANOVAs with time as a within-subjectfactor. The alternative, which involves conducting independentANOVAs at each measurement period, leads to concerns about multi-ple testing. Here we avoided these issues by using a two-step process.First, we analyzed body weight change over the course of the entireexperiment (i.e., final weight–weight when first given MSG) using atwo-way ANOVA with factors of Diet×MSG availability. This had theadvantage of matching the information and analyses conducted onintakes and body composition (which also involved Diet×MSG

Fig. 2. Mean±SE body weights of mice fed AIN-76A (left) or high-energy diet (right) givenStandard error bars (vertical lines) obscuring symbols are not shown.

availability). Next, to be sure that there were no transient differencesin body weight we plotted mean body weights and standard errorsover time (Figs. 1–5) and used these to indicate whether analyses atintermediate times were warranted. In no case did we observe a sig-nificant main effect or interaction involving MSG access.

3. Procedures and results

3.1. Experiment 1 and 2. Influence of drinking or eating MSG on the bodyweight of rats and mice

3.1.1. Rationale, design and proceduresWe began by attempting to build on the findings of Kondoh and

Torii that drinking 1% MSG reduced the body weight gain of rats fedhigh-fat diet [17]. We wanted to determine whether for MSG to in-fluence bodyweight it had to be presented as a voluntarily-ingested so-lution, which we thought might involve cognitive factors or favor ataste-related or osmotic mechanism. Based on previous research[25,26], we knew the approximate amounts of 1% MSG that rats andmice were likely to consume and calculated that this would be

MSG as well as drinking water or two concentrations of MSG in food (Experiment 2).

Fig. 3. Mean±SE body weights of rats fed one of four diets with or without access to 1% MSG solution as well as drinking water. Standard error bars (vertical lines) obscuring sym-bols are not shown.

342 M.G. Tordoff et al. / Physiology & Behavior 107 (2012) 338–345

equivalent to somewhere between 1% and 3% MSG in the diet. Conse-quently, we conducted parallel experiments with rats and mice, witheach experiment involving a 2×4 design: Animals fed AIN-76A dietwere compared with those fed a high-energy diet; this was crossed

Fig. 4. Mean±SE body weights of mice fed control diet or high-energy, high-fat diet.Half the mice fed each diet had access to 1% MSG solution in addition to water todrink. Standard error bars (vertical lines) obscuring symbols are not shown.

with animals given (a) no MSG, (b) 1% MSG solution, (c) 1% MSGadded to food, or (d) 3% MSG added to food. We used ahigh-energy diet rather than the high-fat diet used by Kondoh andTorii because we had no reason to suspect that the diet ingredientswere critical for observing an effect of MSG on body weight, and weconsidered the high-energy diet with a 16/40/44 protein/carbohydrate/fat ratio a closer reflection of the diet of obese humans. The rats were

Fig. 5. Mean±SE body weights of obesity-prone (OP) rats fed AIN-76A control diet orhigh-energy diet. Half the rats fed each diet had access to 1% MSG solution in additionto water to drink. Standard error bars (vertical lines) obscuring symbols are not shown.

Table 3MSG intake, body weight gain, and energy intake of rats or mice with or without MSG to consume.

Exp. Strain andspecies

MSG access,weeks

Diet MSG intake,mg/d

Body weight gain, g/d Energy intake, kcal/d

no MSG 1% MSGsolution

1% MSG infood

3% MSG infood

no MSG 1% MSGsolution

1% MSG infood

3% MSG infood

1 CD rats 8 AIN-76A 398±481 2.30±0.14 2.32±0.09 2.29±2.34 2.34±0.09 89±3 89±3 101±4 104±4CD rats 8 High-energy 321±161 2.62±0.09 2.86±0.19 2.74±0.12 2.88±0.15 97±2 100±4 105±4 108±4

2 B6 mice 12 AIN-76A 56±31 0.13±0.01 0.12±0.01 0.13±0.01 0.14±0.01 12.9±0.2 12.8±0.3 12.8±0.2 14.3±0.5B6 mice 12 High-energy 45±31 0.21±0.02 0.23±0.01 0.22±0.01 0.18±0.01 17.5±1.2 17.4±1.0 16.8±0.9 16.7±0.6

3 CD rats 18 AIN-76A 432±17 1.21±0.05 1.29±0.08 87±3 93±3CD rats 18 Low-fat 433±46 1.39±0.07 1.33±0.09 85±2 91±2CD rats 18 High-fat 351±29 1.67±0.09 1.73±0.06 94±3 93±3CD rats 18 Ultrahigh-fat 322±29 1.36±0.09 1.42±0.08 93±4 100±5

4 B6 mice 8 Low-fat 52±4 0.07±0.01 0.08±0.01 n/m n/mB6 mice 8 High-fat 56±4 0.17±0.02 0.17±0.02 n/m n/m

5 OP rats 26 AIN-76A 444±25 1.84±0.05 1.85±0.05 101±2 99±2OP rats 26 High-energy 442±42 2.34±0.12 2.40±0.09 113±4 112±3

Notes: Values in body of table are means±SEs (n=10 or 12/group). 1Intake of group with solution. MSG intakes of groups with MSG in food are given in text. n/m=not measured.B6=C57BL/6J.

343M.G. Tordoff et al. / Physiology & Behavior 107 (2012) 338–345

monitored every two days for 8 weeks and the mice were monitoredweekly for 12 weeks (Table 1 provides parameters of each experiment).

3.1.2. ResultsTherewere no effects in either species of consumingMSG either as a

drink or mixed with food on body weight (Figs. 1 and 2), body weightgain (Table 3) or energy intake (Table 3). There were also no effects ofconsuming MSG on body composition of the mice (Table 4).

The animals given MSG solution to drink consumed it. In both ex-periments, animals had MSG intakes that were bracketed by theequivalent diet groups receiving MSG in food. The rats fed AIN-76Aand drinking MSG ingested significantly more MSG (398±48 mg/d)than did rats fed 1%MSG in the diet (238±11 mg/d) and significantlyless than did those fed 3% MSG in the diet (794±32 mg/d); the cor-responding values for rats fed the high-energy diet were 321±16,229±8, and 719±28 mg/d [main effect of MSG route, F(2,54)=525.4, pb0.0001]. In the experiment with mice, the correspondingvalues were: AIN-76A diet; 56±3, 34±1 and 122±3 mg/d;high-energy diet; 45±3, 37±2 and 109±4 mg/d, F(2,54)=412.7,pb0.0001. In both experiments, animals fed high-energy diet con-sumed slightly but significantly less MSG than did animals fedAIN-76A diet [rats, F(1,54)=5.21, p=0.0263; mice, F(1,54)=7.42,p=0.0087].

The high-energy diet had its intended effect of increasing bodyweight. Animals fed high-energy diet gained significantly moreweight than did those fed AIN-76A diet [rats, F(1,72)=24.1,pb0.0001, mice, F(1,72)=99.4, pb0.0001] and they ate significantlymore energy [rats, F(1,72)=6.85, p=0.0108, mice, F(1,72)=60.4,pb0.0001]. The mice fed high-energy diet also had significantlymore of both lean, F(1,72)=36.7, pb0.0001, and fat tissue,F(1,72)=98.0, pb0.0001.

Table 4No effects of consuming MSG on lean or fat tissue weights.

Exp. Strain andspecies

MSG access,weeks

Diet Lean tissue weight, g

noMSG

1% MSGsolution

1%fo

2 B6 mice 12 AIN-76A 25±1 24±1 23B6 mice 12 High-energy 26±1 28±0 26

4 B6 mice 8 Low-fat 24±0 23±0B6 mice 8 High-fat 30±1 29±1

5 OP rats 26 AIN-76A 428±6 428±7OP rats 26 High-energy 447±9 442±7

Notes: Values in body of table are means±SEs (n=10 or 12/group). There were no signifi

3.2. Experiment 3. Influence of drinking MSG on the body weight of ratsfed one of four diets

3.2.1. Rationale, design and proceduresWe were puzzled that we had not observed an effect of MSG on

body weight given the work of Kondoh and Torii [17] so we nextattempted a direct replication. Kondoh and Torii found that drinking1% MSG reduced the body weight and fat stores of young rats fedfour diets in one experiment, and adult rats fed a high-fat diet in an-other. The effects of drinking MSG on adults fed the high-fat diet werethe largest: At the end of testing, the group given MSG weighed ~70 g(~10%) less than did the control group so we chose to follow the pro-cedures of this experiment. However, to be thorough, we expandedthe design to include tests of additional adult rats that were fedAIN-76A diet (the staple in our laboratory) and the ultrahigh-fatdiet Kondoh and Torii used to fatten some of the young rats(Table 1). Thus, we employed a 4×2 design, with half the rats fedeach of 4 diets receiving MSG to drink (and the other half not havingMSG). Whereas Kondoh and Torii stopped testing after 15 weeks wetested for 18 weeks.

3.2.2. ResultsBody weights collected during the secondweek of MSG access were

lost due to a technical error. Nevertheless, it was clear that there was noeffect of drinking MSG on body weight at any time (Fig. 3). There alsowere no effects of drinkingMSG on body weight gain (Table 3) or ener-gy intake (Table 3).

Rats fed the high-fat diet gained more weight over the 18-week ex-periment than did rats fed the other diets, F(3,71)=12.3, pb0.0001.Body weights of rats fed the AIN-76A, low-fat, and ultrahigh-fat dietsdid not differ.

Fat tissue weight, g

MSG inod

3% MSG infood

no MSG 1% MSGsolution

1% MSG infood

3% MSG infood

±0 24±1 11±1 11±1 11±1 14±1±1 26±1 18±1 19±1 17±1 16±1

9±1 10±122±1 22±1

137±7 137±6201±13 201±10

cant effects of MSG access on body composition. B6=C57BL/6J.

344 M.G. Tordoff et al. / Physiology & Behavior 107 (2012) 338–345

MSG intakes depended on diet, F(3,36)=3.08, p=0.0395 (Table 3).The group fed the ultrahigh-fat diet drank significantly less MSG thandid the groups fed the AIN-76A or low-fat diets, but not the group fedthe high-fat diet. The other three diet groups did not differ from eachother in MSG intake.

3.3. Experiment 4. Influence of drinking MSG on the body weight of obesemice

3.3.1. Rationale, design and proceduresThe previous experiments found no evidence that the consump-

tion of MSG influenced body weight. However, we were concernedthat our subjects were rapidly growing male rats or mice so it waspossible that this rapid growth might obscure subtle effects of MSGon body fat stores. To obviate this, in Experiment 4, we tested micethat were already-obese. We used a 2×2 design: normal-weightmice fed a low-fat control diet were compared with already-obesemice fed a high-fat diet; half the mice fed each diet had access to 1%MSG to drink. The mice fed high-fat diet were given this diet fromweaning, while at the Jackson Laboratory (see 2.1.3). When they ar-rived in our facility, all mice were given 6 days to adapt to vivariumconditions. They continued to be fed the low-fat or high-fat dietsthey received at The Jackson Laboratory (Table 1). We measured in-takes and body weights during the following week, and then gavehalf the mice fed each diet access to 1% MSG for 8 weeks, continuing tomeasure fluid intakes and body weights weekly. After the first week,we abandoned measuring food intake in this experiment because thehigh-fat diet food pellets crumbled through the cage lid bars, making itimpractical to accurately assess consumption.

3.3.2. ResultsDrinking MSG had no effects on body weight (Fig. 4), body weight

gain (Table 3) or body composition (Table 4). Relative to mice fed thecontrol diet, those fed high-fat diet since weaning were heavier at thestart of the test (29±0 vs. 42±1 g), gained more weight during thetest [F(1,36)=37.3, pb0.0001; Fig. 4, Table 3] had more lean mass[F(1,36)=127.4, pb0.0001; Table 4] and had more than double thefat mass [F(1,36)=328.7, pb0.0001; Table 4]. Diet had no effect onMSG intake.

3.4. Experiment 5. Influence of drinking MSG on the body weight of ratswith a proclivity for obesity

3.4.1. Rationale, design and proceduresIn this experiment, we used rats from the obesity-prone (OP)

strain. We thought this strain would be more likely than others to re-spond to MSG because its body weight is influenced by what it eats: Itmaintains a normal body weight when fed low-fat chow but becomesobese when fed high-fat diets [21].

When the OP rats arrived from Charles River, they were fed Teklad8604 rat pellets for two weeks (until 8 weeks old). Then, 24 of the 48rats were randomly assigned to receive 1% MSG solution in additionto water to drink. After a week to collect baseline intakes and bodyweights, the rats were assigned to four groups matched for bodyweight, with the two groups that had MSG also matched for MSG in-take. The rats were then fed either AIN-76A diet or high-energy dietfor the next 25 weeks, with fluid intakes measured daily, and food in-takes and body weights collected twice a week.

3.4.2. ResultsOP rats with access toMSG did not differ from thosewithoutMSG in

body weight (Fig. 5), body weight gain (Table 3), energy intake(Table 3) or carcass composition at the end of the experiment(Table 4). Relative to rats fed AIN-76A diet, those fed the high-energydiet had significantly higher values of all measures except MSG intake[Tables 3 and 4; body weight gain, F(1,44)=43.7, pb0.0001; energy

intake, F(1,44)=22.9, pb0.0001; lean tissue weight, F(1,44)=5.71,p=0.0211; fat tissue weight, F(1,44)=45.1, pb0.0001].

4. Discussion

In a series of five experiments involving two species, three strains,five diets, and a total of 32 groups of 10–12 animals, therewere no effectsof consuming MSG on body weight, body weight gain, energy intake, orbody composition. Providing MSG in food (Experiments 1 and 2) was asineffective as voluntarily ingested 1% MSG solution at influencing bodyweight. This was true even when animals consumed 3% MSG in food,which more than doubled their daily MSG consumption relative toanimals that drank 1% MSG solution voluntarily. Thus, we conclude thatvoluntary consumption of 1% MSG solution, and involuntary ingestionof greater amounts of MSG in food, neither increases nor decreasesbody weight or body fat stores.

It is unclear why we failed to replicate the findings of Kondoh andTorii that rats given 1% MSG to drink gained less weight and less fatthan did controls without MSG [17]. In Experiment 3, we followedtheir methodology as closely as was practical, including using identi-cal diets, but inevitably there are idiosyncratic differences among lab-oratories that can influence behavior [27]. One striking difference wasthat Kondoh and Torii's control rats gained ~430 g over 15 weeks(i.e., ~4 g/day) whereas the equivalent group in our experimentgained only 227 g over 18 weeks (i.e., ~1.8 g/day). Kondoh andTorii did not include groups fed low-fat diet in their experimentwith adult rats, so it is unclear whether this difference between theexperiments in weight gain is due to differences in growth or to dif-ferences in the rate of deposition of adipose tissue. We cannot ruleout a genetic contribution to the difference in weight gain: Both stud-ies used Sprague Dawley rats obtained from Charles River but the an-imals came from colonies on different continents. We also wonderabout the possible contribution of litter effects to the results observedby Kondoh and Torii given the relatively small number of rats theyused (9 per group×2 groups). Although the authors did not reportthe parentage of rats used in their experiment involving adult rats,the other experiments in the same paper involved groups formedfrom only a few litters. Litter characteristics, particularly litter size,can have life-long influences on body weight [e.g., [28]] so an unfor-tunate allocation of animals to groups could produce a spuriousfinding.

Our results also differ from those obtained by Collison et al., whoreported that C57BL/6J mice fed a diet high in vegetable shorteningand forced to drink 0.064% MSG solution had greater abdominalgirth than did controls [or more accurately, a greater increase in ab-dominal girth between the 6th and 32nd week of MSG exposure[15,16]]. Our methods are sufficiently different from theirs that sever-al factors could plausibly explain the difference in general outcomes.However, we note some serious problems with their work. First, theauthors do not disclose that the same data form the basis of both pa-pers—and also Refs. [29] and [30]]. Second, the experiment(s) used aDiet×MSG factorial design but the results were analyzed withone-way ANOVAs, which do not allow inferences to be made aboutinteractions. Third, Collison et al used change in abdominal girth tomeasure “central obesity” but we are skeptical that this is valid be-cause (a) the procedure requires using calipers to measure thewidth of mice, which has a subjective component, and (b) some ofthe results reported seem counterintuitive. For example, abdominalgirth change was lower in females fed the high-fat diet than thosefed chow, and serum leptin levels did not track abdominal girth;they were decreased by MSG in females but, if anything, increasedby MSG in males [15]. There was also no effect of MSG on visceral ad-ipocyte volume [16]. In the light of our multiple failures to observe aneffect of drinking MSG on body fat stores, and that Collison et al.reported effects with mice drinking 0.064% MSG solution, which ledto a daily exposure that was ~20 times lower than ours (~2.75 vs.

345M.G. Tordoff et al. / Physiology & Behavior 107 (2012) 338–345

~50 mg MSG/day in mice), we do not find the evidence that MSGcauses obesity in adult animals to be compelling.

While our work was near completion, Ren et al. [31] reported astudy showing no effect of drinking 1% MSG on the body weight ofC57BL/6J mice. Starting at 8-wk old, the mice were fed either a low-fator high-fat diet similar in composition to the low-fat and high-fatdiets used here. Over the next 15 weeks, half the mice had access to1% MSG solution in addition to water; the others had access to twodrinking tubes of water (n=9 per group). There was no difference inbody weight gain between mice with or without MSG, although micefed the high-fat diet gained more weight than did those fed thelow-fat diet. At the end of the experiment, blood leptin, insulin, choles-terol, other blood factors, liver glycogen, and respiratory gas exchangeshowed the expected effects of diet but no effects of MSG. Thus, thisstudy is an independent confirmation of our findings that voluntary in-gestion of MSG has no effect on the body weight of C57BL/6J mice.

The present results using animal models to investigate the influ-ence of MSG on body weight provide sober perspective for propo-nents of both sides of the argument that MSG “causes” or “cures”human obesity. Our results do not, and cannot, prove the null hypoth-esis that MSG voluntarily ingested by adult animals does not affectbody weight or obesity, but we believe that they encompass themodels with most obvious face validity, including those where an ef-fect would most likely be observed, if there was one. It will be incum-bent on others to explain how MSG can influence body weight undersome conditions but not the response to the common and straightfor-ward manipulations used here.

Acknowledgments

Funding for this work was received from the Ajinomoto AminoAcid Research Program. MCM received support from the MonellNIH-NIDCD P30 Training Grant. Body composition was measuredusing equipment provided by the Monell Behavioral and Physiologi-cal Phenotyping Core, which is supported, in part, by funding fromthe NIH-NIDCD P30 Core Grant DC-011735.

References

[1] Olney JW. Brain lesions, obesity, and other disturbances in mice treated withmonosodium glutamate. Science 1969;164:719-21.

[2] Bunyan J, Murrell EA, Shah PP. The induction of obesity in rodents by means ofmonosodium glutamate. Br J Nutr 1976;35:25-39.

[3] Dawson Jr R, Lorden JF. Behavioral andneurochemical effects of neonatal administra-tion of monosodium L-glutamate in mice. J Comp Physiol Psychol 1981;95:71-84.

[4] Tanaka K, Shimada M, Nakao K, Kusunoki T. Hypothalamic lesion induced by in-jection of monosodium glutamate in suckling period and subsequent develop-ment of obesity. Exp Neurol 1978;62:191-9.

[5] Dawson Jr R. Acute and long lasting neurochemical effects of monosodium gluta-mate administration to mice. Neuropharmacology 1983;22:1417-9.

[6] Prosky L, O'Dell RG. Effect of dietary monosodium L-glutamate on some brain andliver metabolites in rats. Proc Soc Exp Biol Med 1971;138:517-22.

[7] Anantharaman K. In utero and dietary administration of monosodium l-glutamateto mice: reproductive performance and development in a multigeneration study.

In: Filer LJ, Garattini S, Kare MR, Reynolds WA, Wurtman RJ, editors. Glutamic Acid:Advances in Biochemistry and Physiology. New York: Raven Press; 1972. p. 231-53.

[8] Bazzano G, D'Elia JA, Olson RE. Monosodium glutamate: feeding of large amountsin man and gerbils. Science 1970;169:1208-9.

[9] Walker R, Lupien JR. The safety evaluation of monosodium glutamate. J Nutr2000;130:1049S-52S.

[10] He K, Zhao L, Daviglus ML, Dyer AR, Van Horn L, Garside D, et al. Association ofmonosodium glutamate intake with overweight in Chinese adults: the INTERMAPStudy. Obesity 2008;16:1875-80.

[11] He K, Du S, Xun P, Sharma S, Wang H, Zhai F, et al. Consumption of monosodiumglutamate in relation to incidence of overweight in Chinese adults: China Healthand Nutrition Survey (CHNS). Am J Clin Nutr 2011;93:1328-36.

[12] Shi Z, Luscombe-Marsh ND, Wittert GA, Yuan B, Dai Y, Pan X, et al. Monosodiumglutamate is not associated with obesity or a greater prevalence of weight gainover 5 years: findings from the Jiangsu Nutrition Study of Chinese adults. Br J Nutr2010;104:457-63.

[13] Bursey RG,Watson L, SmrigaM. A lack of epidemiologic evidence to link consumptionof monosodium L-glutamate and obesity in China. Am J Clin Nutr 2011;94:958-60.

[14] Samuels A. Monosodium glutamate is not associated with obesity or a greaterprevalence of weight gain over 4 years: findings from the Jiangsu NutritionStudy of Chinese adults. Br J Nutr 2010;104:1729.

[15] Collison KS, Makhoul NJ, Inglis A, Al-Johi M, Zaidi MZ, Maqbool Z, et al. Dietarytrans-fat combined with monosodium glutamate induces dyslipidemia and im-pairs spatial memory. Physiol Behav 2010;99:334-42.

[16] Collison KS, Maqbool Z, Saleh SM, Inglis A, Makhoul NJ, Bakheet R, et al. Effect ofdietary monosodium glutamate on trans fat-induced nonalcoholic fatty liver dis-ease. J Lipid Res 2009;50:1521-37.

[17] Kondoh T, Torii K. MSG intake suppresses weight gain, fat deposition, and plasmaleptin levels in male Sprague–Dawley rats. Physiol Behav 2008;95:135-44.

[18] Kondoh T, Torii K. Brain activation by umami substances via gustatory and viscer-al signaling pathways, and physiological significance. Biol Pharm Bull 2008;31:1827-32.

[19] Smriga M, Murakami H, Mori M, Torii K. Use of thermal photography to explorethe age-dependent effect of monosodium glutamate, NaCl and glucose onbrown adipose tissue thermogenesis. Physiol Behav 2000;71:403-7.

[20] Ventura AK, Beauchamp GK, Mennella JA. Infant regulation of intake: the effect offree glutamate content in infant formulas. Am J Clin Nutr 2012;95:875-81.

[21] Levin BE, Dunn-Meynell AA, Balkan B, Keesey RE. Selective breeding fordiet-induced obesity and resistance in Sprague–Dawley rats. Am J Physiol RegulIntegr Comp Physiol 1997;273:R725-30.

[22] Tordoff MG, Bachmanov AA. Monell Mouse Taste Phenotyping Project. MonellChemical Senses Center; 2001. www.monell.org/MMTPP.

[23] Ramirez I, Friedman MI. Dietary hyperphagia in rats: role of fat, carbohydrate, andenergy content. Physiol Behav 1990;47:1157-63.

[24] Reed DR, Bachmanov AA, Tordoff MG. Forty mouse strain survey of body compo-sition. Physiol Behav 2007;91:593-600.

[25] Bachmanov AA, Tordoff MG, Beauchamp GK. Intake of umami-tasting solutions bymice: a genetic analysis. J Nutr 2000;130:935S-41S [supplement].

[26] Tordoff MG, Alarcon LK, Lawler MP. Preferences of 14 rat strains for 17 taste com-pounds. Physiol Behav 2008;95:308-32.

[27] Crabbe JC, Wahlsten D, Dudek BC. Genetics of mouse behavior: interactions withlaboratory environment. Science 1999;284:1670-2.

[28] Remmers F, Fodor M, Delemarre-van de Waal HA. Neonatal food restriction per-manently alters rat body dimensions and energy intake. Physiol Behav 2008;95:208-15.

[29] Collison KS, ZaidiMZ,Maqbool Z, Saleh SM, Inglis A, Makhoul NJ, et al. Sex-dimorphismin cardiac nutrigenomics: effect of trans fat and/or monosodium glutamate consump-tion. BMC Genomics 2011;12:555.

[30] Collison KS, Maqbool ZM, Inglis AL, Makhoul NJ, Saleh SM, Bakheet RH, et al. Effectof dietary monosodium glutamate on HFCS-induced hepatic steatosis: expressionprofiles in the liver and visceral fat. Obesity 2010;18:1122-34.

[31] Ren X, Ferreira JG, Yeckel CW, Kondoh T, de Araujo IE. Effects of ad libitum inges-tion of monosodium glutamate on weight gain in C57BL6/J mice. Digestion2011;83(Suppl. 1):32-6.