DDC

SPa

ab

Pc

Pd

B

AgbiatpeccbeptFmqcatsFdtiftisnI

Ki

*2EAiG

Neuroscience 152 (2008) 859–866

0d

IFFERENT TYPES OF NUTRITIONAL DEFICIENCIES AFFECTIFFERENT DOMAINS OF SPATIAL MEMORY FUNCTION

HECKED IN A RADIAL ARM MAZE

ImnroiIdsf(d4mStswccda2mme

ntun(Miami(aatsceeto

. C. RANADE,a* A. ROSE,a M. RAO,a J. GALLEGO,b,c

. GRESSENSb,c,d AND S. MANIa

National Brain Research Centre, National Highway 8, Manesar, Hary-na 122050, India

INSERM U 676, Hôpital Robert Debré, 48 BD Sérurier, F-75019aris, France

Université Paris 7, Faculté de Médecine 16, Rue Huchard, 75018aris, France

AP HP, Hôpital Robert Debré, Service de Neurologie Pédiatrique, 48D Serurier, 75019 Paris, France

bstract—Several studies using animal models have sug-ested that the effects of nutritional insult on the developingrain are long-lasting and lead to permanent deficits in learn-

ng and behavior. Malnutrition can refer to the availability ofll the nutrients but in insufficient quantities or it may implyhat one or more of essential nutrients is either missing or isresent, but in the wrong proportions in the diet. The hypoth-sis addressed in this study is that different domains ofognitive functioning can be affected by malnutrition and thisan be related to the type of nutritional deficiency that therain has been exposed to during development. To study theffect of nutritional deprivation during brain development, aaradigm of maternal malnutrition during the period of gesta-

ion and lactation was used and its effects were studied on the1 offspring using Swiss albino mice. Three different types ofalnutrition were used, that involve, caloric restriction, inade-uate amount of protein in the diet and condition of low ironontent. Our results show that the domain of spatial memoryffected in the F1 generation depended on the kind of malnu-rition that the mother was subjected to. Further our studyhows that although hippocampal volume was reduced in all1 pups, hippocampal subregions of the F1 animals wereifferentially vulnerable depending on type of malnutritionhat the mother was subjected to. These results highlight themportance of qualifying the kind of malnutrition that is suf-ered by the mother during the period of gestation and lacta-ion as it has consequences for the cognitive domain affectedn the offspring. Awareness of this should inform preventiontrategies in trying to reverse the effects of adverse maternalutrition during critical periods in brain development. © 2008

BRO. Published by Elsevier Ltd. All rights reserved.

ey words: maternal malnutrition, reference memory, work-ng memory, hippocampal volume.

Corresponding author. Tel: �91-124-2338921-26; fax: �91-124-338921-10.-mail address: sayali@nbrc.res.in (S. C. Ranade).bbreviations: CED, chronic energy deficiency; DG, dentate gyrus; ID,

tron deficiency; IDECG, International Dietary Energy Consultancy

roup; PD, protein deficiency; RAM, radial arm maze.

306-4522/08$32.00�0.00 © 2008 IBRO. Published by Elsevier Ltd. All rights reseroi:10.1016/j.neuroscience.2008.01.002

859

n developing countries in utero under-nourishment andalnutrition are a major cause of low birth weight (Inter-ational Dietary Energy Consultancy Group (IDECG) Annl.eport, 1997; Mavalankar et al., 1994). Dietary deficiencyf the mother has been associated with increased morbid-

ty and mortality in the offspring (Mavalankar et al., 1994).n economically developed countries malnutrition and un-er-nutrition continue to be a problem in certain sections ofociety (IDECG Annual Report, 1997). Under-nutrition re-ers to insufficient caloric intake (chronic energy deficiencyCED), with BMI �18.5 kg/m2), while malnutrition refers toeficiency in a macronutrient or a micronutrient. Almost6% of Indian children under the age of 3 suffer fromalnutrition according to a recent National Family Healthurvey conducted by the Indian Health Ministry in conjunc-

ion with UNICEF, the United Nations children’s agency. Inub-Saharan Africa this figure is 35%. The diet of peopleho live below the poverty line is not adequate in terms ofalorific value and in addition it is highly imbalanced andontains very little protein content. In addition the inci-ence of anemia among pregnant women can be as highs 80% in certain developing countries (Sharma et al.,003). In this study we focus on CED, protein energyalnutrition (protein deficiency, PD) as an example ofacronutrient deficiency, and iron deficiency (ID) as anxample of micronutrient deficiency.

We will use the term malnutrition to refer to both under-utrition and malnutrition of the kinds that have been men-ioned above. Animal models of malnutrition have beensed extensively in order to understand the effect of mal-utrition on brain development and cognitive functioningPortman, 1987; Tonkiss et al., 1993; Morgane et al., 1993;anjarrez et al., 1998). These studies have validated an-

mal models of prenatal malnutrition and suggest that inddition to understanding the effect of malnutrition, theseodels can also lead to insights important in understand-

ng intra-uterine growth retardation (IUGR) in humansManjarrez et al.1998). Further, it is evident from the liter-ture that these types of deficiencies do not result in grossbnormalities in brain anatomy or in gross mental retarda-ion or psychopathology, but rather result in permanentuboptimal development leading to long term learning andognitive deficiencies. Since malnutrition may be of sev-ral different types that include CED, PD or deprivation ofssential micronutrients such as iron, a critical questionhat remains unanswered is whether these different typesf malnutrition affect similar developmental pathways that

hen result in similar deficits in cognitive performance.ved.

mittgmC(TtofpfflswrmCnvtmcndvfigtd

G

Scg(ttgldcvodtfwNmoftmwT

twrmNiCo

R

Tsewfs3avdtttdpettiwtptttdaeso

D

Fcneobfim

S

Tlefit(tfUadcs

S. C. Ranade et al. / Neuroscience 152 (2008) 859–866860

In this study, we address whether cognitive perfor-ance in the F1 offspring is differentially affected depend-

ng on the type of malnutrition that the mother is subjectedo. We used three different protocols of maternal malnutri-ion that lead to CED, PD or ID in the mother duringestation and lactation. As a measure of cognitive perfor-ance the F1 offspring from these groups (F1-CTL, F1-ED, F1-PEM, F1-ID) were tested on the radial arm maze

RAM) to assess spatial memory and working memory.he RAM was preferred to the standard Morris water maze

est for spatial memory function since more complex datan working and reference memory error can be extracted

rom this test. In addition, the RAM does not have a com-onent of fear and aversion that could be a potential con-

ounder for working memory deficits. Our results show theollowing. 1) Maternal malnutrition during the gestational–actational period has an adverse effect on measures ofpatial learning in the F1 offspring. 2) Reference memoryas most affected in F1-ID and there was sparing of

eference memory in F1-CED. 3) Working memory wasost impaired in F1-PD and least impaired in F1-ID. 4)ompared with F1-CTL, hippocampal volumes were sig-ificantly different, reduced in all F1 groups. These obser-ations suggest that the general rubric of maternal malnu-rition warrants further classification in terms of the type ofaternal deprivation that occurred. This is important since

ognitive impairments may depend on the type of maternalutritional deprivation. This suggests that regions of theeveloping brain in utero and after birth are differentiallyulnerable to nutritional inadequacies of the mother. Thesendings have important implications in developing strate-ies for reversal of suboptimal performance due to nutri-ional inadequacies suffered during critical periods in brainevelopment.

EXPERIMENTAL PROCEDURES

eneration of deficient animals

wiss albino female mice 6–8 weeks old were taken from inbreedolony at National Brain Research Centre. The females wereiven either control feed [starch (50%), sucrose (8%), casein24.2%), cellulose (6%), refined groundnut oil (7%), mineral mix-ure (3.5%), vitamin (1%), L-cystine (0.3%)] (group 1/control) orhe iron deficient feed [skim milk powder (55%), sucrose (33%),roundnut oil (5%), salt mixture (4%), vitamin mixture (1%), cel-

ulose (2%), methionine (2%)] (group 2/iron deficient) or proteineficient feed [starch (65.1%), sucrose (10%), casein (7.1%),ellulose (6%), refined groundnut oil (7%), mineral mixture (3.5%),itamin mixture (1%), L-cystine (0.3%)] (group 3/protein deficient)r on calorie restricted diet (group 4/CED). For CED animals theiraily feed intake was monitored and then was reduced to 70% ofheir daily intake. The animals were maintained on reduced feedor the entire experimental duration. All the different types of feedere manufactured at and purchased from the National Institute ofutrition (NIN) (Hyderabad, India). The group 2/iron deficient fe-ales were given 1 �g/g body weight of iron. This concentrationf iron was maintained by giving weekly dose of iron syrup. Theemales were maintained for 6 weeks on this deficient diet. Duringhis period their body weights as well as their food intake waseasured. Females were then kept for mating. The deficient dietas continued throughout gestation as well as during lactation.

he pups (F1) were weaned on P21. After weaning, pups from all v

he groups were maintained on the control diet. The F1 animalsere tested for their spatial learning and memory skills when they

eached 8 weeks of age on an eight-arm RAM. All animals wereaintained in specific pathogen-free conditions according toBRC guidelines that follow the NIH guidelines. All animal exper-

ment protocols were approved by the Internal Animal Ethicsommittee. Experiments were designed to minimize the numberf animals used and their suffering.

AM

he apparatus used was a RAM (similar to that previously de-cribed by Olton and Samuelson, 1976) with eight identical andqually spaced arms radiating from the central platform. The armsere 42 cm long and 11.5 cm wide with a recessed food cup 1 cm

rom the end and 1.5 cm deep. Each arm and central platform hadeparate transparent lids. The central platform diameter was2 cm. The whole apparatus was mounted 28 cm from the groundnd placed in a small, well-lit room that contained a number ofisual cues that remained invariant for the testing period. Foodeprivation was introduced 2 weeks before habituation. This en-ailed monitoring the food intake of the animals for 5 days prior tohis time point and reducing the amount of food given to animalso 85% of their initial intake. Food pellets were weighed out everyay and put in their home cages during the food deprivationeriod. The weights of the animals were monitored daily till thend of the habituation period. Habituation lasted for 8 days. Forhe first 3 days (acclimatization) animals were allowed to explorehe maze for 5 min. No food pellets were given to animalsn the maze. On the subsequent 5 days (pre-training) animalsere given food pellets. On the first day they were scattered on

he platform, on the second day food pellets were placed in theroximal one third of all the eight arms and on subsequent dayshey were placed farther and farther away from the platform intohe arms. All experimental groups performed equally well duringhe pre-training sessions. Acquisition lasted for 20 consecutiveays with one trial per day. The trial was terminated when eitherll the food pellets were consumed or at the end of 5 min, which-ver came first. For each animal four arms were randomly as-igned to be the baited arms, with restriction that no more thanne pair of adjacent arms was baited or un-baited.

ata analysis

ive dependant measures were obtained on every trial: time toomplete a trial (maximum of 300 s), latency to first arm entry,umber of reference memory errors (defined as the number of firstntries into an un-baited arm, maximum of four per trial), numberf working memory correct errors (defined as re-entries into aaited arm) and number of working memory incorrect errors (de-ned as re-entries into un-baited arms). The distinction of workingemory errors follows that of Jarrad (1993).

tatistics for RAM data

he scores used for analysis were the time taken for trial, theatency period, the reference memory errors, the working memoryrrors (correct), and the working memory errors (incorrect). Theseve scores were subjected to separate analyses of variances withreatment (four levels: protein-deprived animals (PD), controlsCTL), iron-deprived (ID), and caloric energy–deprived (CED), ashe between-subject factor and days (1–20) as the within-subjectactor (Statview 5 Software, Abacus Concepts, Berkeley, CA,SA). To take into account the heterogeneity of correlationsmong repeated measurements, we adjusted the degrees of free-om using the Greenhouse-Geisser factor, which is a conservativeorrection procedure (Keselman and Keselman, 1984). Within-ubject main effects and interactions are reported together with P

alues based on these adjusted degrees of freedom. When ap-

pbtrVm

S

UtoftmTpaet(

V

Tmitbtsas

itktrseuvrmt(tVf

wap

gbtur

S

Two

Eo

TmsdlitwaTst2mT0(c

T

TwDpmFtgta�r2

FfitbRtoa

S. C. Ranade et al. / Neuroscience 152 (2008) 859–866 861

ropriate, learning rates were assessed by the score improvementetween the first and the last session, expressed as percentage ofhe first day value (i.e. 100�(Score20�Score1)/Score1). Bonfer-oni/Dunn tests were used for group comparisons of these rates.alues are means�standard deviations (S.D.) in the text andean�standard eErrors (S.E.) in figures.

tereological analysis

sing the Neurolucida program (MicroBrightField version 7.0),racings were made from the anterior through the posterior extentf the hippocampal formation. The tracing were done on everyourth section. In the tracings, the CA1 and CA2/3 fields andhe dentate gyrus were included. A total of 39–41 tracings wereade per hippocampal formation, per hemisphere in each animal.he traced sections were evenly spaced 80 �m apart. The firstlane was a randomly chosen section within 180 �m of the mostnterior plane of the hippocampus. The last traced section forvery animal was the last plane in which the hippocampal forma-ion was present [see Plate 44, left side] in Paxinos and Watson1986).

olumetric estimation

o obtain the total volume of the hippocampus, we first esti-ated the volume shrinkage factor (Uylings et al., 1986), since

f this correction for shrinkage is not performed the volume ofhe region of interest may be over- or underestimated. Onerain of an age-matched animal that had undergone RAMraining was sectioned fresh-frozen on a cryostat. The 60 �mections were made through the entire hippocampal formation,nd the tissue was immediately mounted onto slides and lightlytained.

This unprocessed tissue was compared with the fixed andmmunocytochemically processed tissue. In stereology, calcula-ion of regional volume using Cavalieri estimation is based uponnowing the actual thickness of each section on the slide (h), as inhe formula V(ref)��nAi�h, where V(ref) is the volume of theegion of interest, and �n Ai is the cross-sectional area of the ithection of the region of interest (for n sections). However, becauserrors inherent to determining h for each section may occur, wesed the equation V (ref)�(SF) v�t�nAi, where (SF) v is theolume shrinkage factor, t is the constant section thickness (asecorded from the cryostat), and n is the number of sectionseasured (Uylings et al., 1986). As evident from these two equa-

ions, (SF) v[tims]t�h. We calculated (SF) v using the equationSF) v�Vf/Vs, where Vf is the volume of the region of interest inhe experimental brain (perfused, stained, and coverslipped), ands is the volume of the region of interest in the control (fresh-

rozen) brain (Nunez and MacCarthy, 2003).This value was used in the equation V (ref)�(SF) v�t�nAi,

here t�60 �m, n�7–8 sections, and Ai is the cross-sectionalrea of each region in a given section. Ai was measured using therogram Neuroexplorer (MicroBrightField, version 2.01).

For hippocampal subdivision analysis, CA1, CA3 and dentateyrus (DG) regions were distinguished in the Nissl-stained slidesy referring to the mouse atlas (Paxinos and Franklin, 2004) and

raced individually. The procedure for tracing was the same as thatsed for the total hippocampal volume. Shrinkage factor for eachegion is calculated as described above.

tatistical analysis of volumetric data

he hippocampal volumes of all the three malnourished groupsere compared with that of control hippocampal volumes by using

ne-way ANOVA. w

RESULTS

ffect of maternal malnutrition on the weightsf the F1 pups

he decrement in the weights of all the F1 pups which wasonitored before and during the RAM training did not show

tatistically significant difference between the groups. Tenays before trial started all the malnourished groups had

ower body weights compared with control, however, dur-ng the “weight reduction” period before RAM training allhe groups showed an even decrease in their bodyeights. Therefore during the pre-training and trial periodll the groups had weights comparable to that of controls.he differences in weight between groups remained verymall throughout the trial time (Fig. 1 mean weights overhe entire trial: Controls 20.4�0.97; PD: 19.80�0.80; ID:0.81�0.62, and CED: 20.36�1.04) although repeatedeasures ANOVA revealed a significant main effect forrial Time (from day 1 to day 10, F(9,315)�2.93, P�.0467) and a significant Group by Time interactionF(27,315)�2.53, P�0.0001) due a faster decrease inontrols compared with ID and CED.

ime taken for trials

he animals were first acclimatized for 3 days at the end ofhich all the animals were freely exploring the maze.uring the pretrial period all groups readily consumed theellets available. Time taken for trial decreased in all treat-ent groups as a function of time (main effect for Time:(19,608)�43.18, P�0.0001). However treatments by

ime interaction were significantly different betweenroups: (F(57,608)�5.12. This was due to the difference inotal time taken for trials between day 1 and day 20 of thecquisition period (�75%�8 in F1-CTL versus �34%�22,32%�7, and �45%�16 in F1-CED, F1-PD, and F1-ID

espectively; P�0.0001 for all pair-wise comparisons) [Fig.A]. When averaged across time, the time taken for trials

ig. 1. Weight changes of F1 pups before and during RAM trials. Thegure shows changes in the weights before and during the RAMraining. The x axis shows days. The negative numbers indicate daysefore RAM were started. The day 0 indicates commencement ofAM. Day 0–day 3 is acclimatization period, day 4–day 7 is pre-

raining and the trials begin on day 8. The y axis shows mean weightf all the animals in particular group in g. The standard error is plotteds error bar.

as significantly different between the treatment groups

(F

t

FTsrigosgoirmiTp that part

S. C. Ranade et al. / Neuroscience 152 (2008) 859–866862

main effect for Treatment: F(3, 608)�36.34, P�0.0001).

ig. 2. Effect of different types of nutritional deficiencies on F1 pups ootal trial time in RAM. The x axis shows time in days for the entire traolid diamonds represent control group, while CED, PD and ID are reprepresents mean�S.E. value of all the animals in that group on that pan RAM. The x axis shows time in days for the entire training session. Throup, while CED, PD and ID are represented by solid squares, solidf all the animals in that group on that particular day. (C) The effects ohows time in days for the entire training session. The y axis shows mroup, while CED, PD and ID are represented by solid squares, solidf all the animals in that group on that particular day. (D) The effects

n RAM. The x axis shows time in days for the entire training session. Tepresent control group, while CED, PD and ID are represented by sean�S.E. value of all the animals in that group on that particular da

ncorrect errors in RAM. The x axis shows time in days for the entire trhe solid diamonds represent control group, while CED, PD and ID areoint represents mean�S.E. value of all the animals in that group on

1-CTL had on the average faster total trial time values P

han F1-CED, F1-PD, and F1-ID groups (P�0.0001,

AM performance. (A) The effects of different types of malnutrition onsion. The y axis shows time required to complete trial in minutes. Theby solid squares, solid triangles and asterisks respectively. Each pointay. (B) The effects of different types of malnutrition on latency periodhows latency period in seconds. The solid diamonds represent controland asterisks respectively. Each point represents mean�S.E. valuetypes of malnutrition on reference memory errors in RAM. The x axiser of reference memory errors. The solid diamonds represent controland asterisks respectively. Each point represents mean�S.E. value

nt types of malnutrition on number of working memory correct errorsshows mean no. working memory errors (correct). The solid diamondsres, solid triangles and asterisks respectively. Each point representse effects of different types of malnutrition on no. of working memoryssion. The y axis shows mean no. working memory errors (incorrect).nted by solid squares, solid triangles and asterisks respectively. Eachicular day.

n their Rining sesesentedrticular de y axis striangles

f differentean numbtrianglesof differehe y axisolid squay. (E) Th

aining sereprese

�0.0001, and P�0.0027, respectively). Within the treat-

maf

L

LcentabsPlF

R

RfPfWsgPb(fn

W

Wa2aWhwPn[

W

Wa3fFtsFAfFWh(

Pw(

Hd

HwFp

Dd

TvrasrrscDvFa(c[ot

Tdb

Fpgmpp

S. C. Ranade et al. / Neuroscience 152 (2008) 859–866 863

ent groups the F1-ID had the fastest total trial time valuess compared with F1-CED and F1-PD groups (P�0.0001or either comparison).

atency period

atency period for entry into first arm significantly de-reased as a function of time in all treatment groups (mainffect for Time; F(19,589)�18.18, P�0.0001) and this wasot significantly affected by the different treatment condi-ions (Treatment by Time interaction: ns). When averagedcross time, latency scores showed significant differencesetween groups: F1-CED latency values were significantlyhorter than F1-CTL, F1-PD, and F1-ID values (P�0.003,�0.0008 and P�0.0020, respectively) and F1-ID had the

ongest latency score when compared with F1-CTL and1-PD (P�0.0007 and P�0.0020, respectively) [Fig. 2B].

eference memory errors

eference memory errors significantly decreased as aunction of time, [main effect for Time: F(19,608)�22.58;�0.0001] but this time effect was not significantly af-

ected by treatments (Treatment by Time interaction: ns).hen averaged across time, reference memory errors

howed significant differences between the experimentalroups (main effect for Treatment: F(3,608)�15.65,�0.0001). F1-PD and F1-ID had significantly higher num-er of reference memory errors as compared with F1-CTLP�0.0012 and P�0.0001, respectively), however the dif-erence between F1-CTL and F1-CED did not reach sig-ificance; (P�0.131) [Fig. 2C].

orking memory errors (correct)

orking memory errors (correct) significantly decreaseds a function of time, [main effect for Time: F(19,608)�9.55 P�0.0001] but this time effect was not significantlyffected by treatments (Treatment by Time interaction: ns).hen averaged across time, F1-CED, F1-PD, and F1-ID

ad higher number of working memory errors (correct)hen compared with F1-CTL (P�0.0008, P�0.0001, and�0.0095, respectively). F1-ID had significantly fewerumber of working memory errors than F1-PD (P�0.0031)Fig. 2D].

orking memory errors (incorrect)

orking memory errors (incorrect) significantly decreaseds a function of time [main effect for Time: F(19,608)�2.71 P�0.0001]. This time effect was significantly af-

ected by treatments (Treatment by Time interaction:(57,608)�2.15, P�0.0001). This is reflected in the fact

hat the decrease in error rates over time was significantlylower in F1-CED, F1-PD and F1-ID when compared with1-CTL (P�0.0001, P�0.0001, P�0.0229, respectively).mong the different F1 malnourished groups F1-ID had a

aster decrease in error rates across time compared with1-CED and F1-PD (0.0447, and P�0.0258, respectively).hen averaged across time, F1-CED, F1-PD, and F1-ID

ad significantly greater number of working memory errors

incorrect) than the F1-CTL (P�0.0028, P�0.0008, and t

�0.0085, respectively). F1-ID made significantly fewerorking memory errors (incorrect) than the F1-PDs

P�0.0031) [Fig. 2E].

ippocampal volume differs significantly betweenifferent treatment groups

ippocampal volume estimation showed that comparedith F1-CTL, F1-CED (P�0.001), F1-PD (P�0.05) and1-ID (P�0.001) had a significant reduction in hippocam-al volume [Fig. 3].

ifferent subdivisions of hippocampus areifferentially affected in the different F1 groups

he total volume of the hippocampus was further subdi-ided into the CA1, CA3 and DG regions. This analysisevealed that the hippocampal regions were differentiallyffected in the different F1 conditions. F1-CED showed themallest reduction in hippocampal volume in all the sub-egions compared with F1-CTL. F1-PD showed a 60%eduction in DG volume (P�0.0001) whereas F1-IDhowed a 44% (P�0.0001) reduction DG volume whenompared with F1-CTL. Similar to the observations in theG region, F1-PD also had the greatest reduction in theolume of the CA1 region [66% reduction compared with1-CTL; (P�0.0001)]. Interestingly, in contrast to thebove, the CA3 region was reduced by only 38% in F1-PDP�0.0001) whereas the percentage decrease in volumeompared with F1-CTL was only 10% in F1-ID (P�0.0562)Fig. 4A and 4B]. This shows that the different sub-regionsf the hippocampus are differentially vulnerable to condi-ions of maternal malnourishment.

DISCUSSION

his study looks at the effect of maternal malnutritionuring a critical period in brain development. Any pertur-ation during this period of development has the capacity

ig. 3. The effect of different types of nutritional deficiencies on hip-ocampal volume of F1 pups. The x axis shows different malnutritionroups. The y axis shows corrected hippocampal volume in cubicicrons. Each column represents mean value of total corrected hip-ocampal volume for that particular group. The standard error islotted as error bars.

o modulate the development of nervous system and is

ltarwd1mWntaiTwwm

woo

cmanwifbwdsfalwdp

dtstndp5taedsegandewhagafcadapTidFgmw

tc

Fuhpapvpvssh

S. C. Ranade et al. / Neuroscience 152 (2008) 859–866864

ikely to have long lasting effects that continue throughouthe life of an individual. Nutritional insults are not usuallyssociated with gross anatomical defects but can lead toeduction in regional brain volumes that can be associatedith long-term cognitive impairments (Peterson, 2003),eficits in learning (Isaacs et al., 2001; Lefebvre et al.,988; Lloyd et al., 1988; Larroque and Samain, 2001) andemory impairments (Briscoe et al., 2001; Isaacs, 2000).ith the broad spectrum of deficits observed in variety of

utritional insults, it is important as evident from our resultso check whether a specific nutritional deficiency can bessociated with a specific cognitive parameter that may

ndicate defects in a particular developmental pathway.herefore, the primary motivation behind the current studyas to see if specific cognitive deficits can be associatedith specific conditions of maternal malnutrition. For this aouse animal model was used.

In our experimental paradigm, dams were deprived 6eeks prior to conception in order that the embryos devel-ped during their entire gestational period under conditions

ig. 4. The effect of different types of nutritional deficiencies on vol-me of different subdivisions of hippocampus. (A) The changes in theippocampal subdivision volume due to maternal malnutrition in the F1ups is shown. The x axis shows different malnutrition groups. The yxis shows corrected volumes of different subdivisions of hippocam-us in cubic microns. Each column represents mean value of theolume of particular subdivision for that particular group. The S.E. islotted as error bars. (B) The percent changes in hippocampal subdi-ision volume of malnourished brains with respect to control brains ishown. The x axis shows different malnourished groups. The y axishows percent changes in the volume of different subdivisions of theippocampus.

f malnutrition. In this paradigm it is possible that pre- d

onception deprivation of the dams may lead to adaptiveechanisms in the mother. Therefore the present studylso includes the effect of this maternal adaptive mecha-ism, if any, on the developing fetus. This study designas chosen since it closely mimicked the realistic situation

n populations that live below the poverty line and thereforeemales who suffer from various conditions of malnutritionefore they conceive. This was also the rationale behindhy the condition of maternal malnutrition was continueduring the period of lactation. Our results most importantlyhow that there was a differential effect on cognitive per-ormance depending on the type of malnutrition that thenimal was subjected to in utero and during the period of

actation. Future studies using this model will determinehether specific intervention strategies can be tailoredepending on the type of malnutrition that is prevalent in aarticular geographical area.

The confounding factors that could account for theeficits in the performance of the experimental groups inhe RAM are motivation and motor deficits. In the currenttudy, since food deprivation is used as a motivating factor,he effect of this on the different experimental groupseeds to be ruled out. Our evidence suggests that motoreficits are not the prime reason for the difference inerformance since all groups finished their trial in less thanmin, although there were significant differences between

he groups on total trial time. Therefore, the memory errorsre unlikely to be accounted for by motor deficits. How-ver, one cannot rule out the possibility that some of theifference in total trial time between groups may be due toubtle motor deficits. Fig. 1A shows that although thexperimental groups were lower in weight than the controlroup, during the food deprivation period the weights ofnimals in all the groups dropped such that by the begin-ing of the acclimatization period in the RAM there was noifference in the weight of the animals between the differ-nt experimental groups. In addition, it was seen that thereas no difference in the amount of food consumed in theirome cages (data not shown). In addition, during the RAMcquisition period by day 15 all the animals in all theroups finished the pellets kept in the baited arms. Thebove results show that no consistent trend of differentialood consumption or weight loss could be detected thatould account for the differences in the memory scorescross the different experimental groups. Therefore theifferences in cognitive performance between the groupsre not likely to arise from either the differences in feedingattern of the various groups or due to motivation issue.hus, it can be reasonably concluded that the differences

n the performances observed are indeed due to memoryeficits. Interestingly, data from Fig. 1 also shows that the1 malnourished groups were consuming more food perram of body weight suggesting a long lasting effect on theetabolism of F1 animals of malnourished mothers. Thisas not investigated further in the current study.

Spatial representation is one aspect of cognition wherehe abilities of humans and animals have been extensivelyompared (Tolman, 1948; Poucet, 1993). The RAM task

esigned in our experiments had four baited/four unbaited

artafROnrlmcoa

mimtafaIttirmrCTmdamds

faaahgp(2tiivtghgttid

tswae

tasrsddKwtpgaep(vmi

Ibitjivttcrso

AtKapRrId(

BB

S. C. Ranade et al. / Neuroscience 152 (2008) 859–866 865

rms. This design enables a direct comparison betweeneference memory and working memory performancehrough a within task within subject comparison (Poucetnd Buhot, 1994). In this task, memory is usually inferredrom the day-to-day improvement in performances in theAM that is indicative of reference memory of the animal.ur results show that there is clear-cut reduction in theumber of reference memory errors over the training pe-iod in the control group suggesting that these mice hadearned to locate and not to enter a never-baited arm. This

easure is severely affected in the F1-ID animals as indi-ated by smallest reduction in number of reference mem-ry errors during the acquisition period. F1-PD and F1-IDnimals were moderately affected in this measure.

Working memory is defined as the ability to retain andanipulate mnemonic information to guide ongoing behav-

or (Baddeley, 1986). An important component of workingemory is the short term storage of trial-unique informa-

ion (Goldman-Rakic, 1987), whereby unique informationbout specific stimuli (e.g. spatial location and object in-ormation) is retained briefly in a short-term memory buffernd discarded after an appropriate response is executed.n our RAM task, working memory is measured by countinghe number of working memory errors within a particularrial. The working memory errors were further classifiednto working memory correct and working memory incor-ect errors which reflect the animal’s ability to learn andemorize previous entries into baited or unbaited arms

espectively. In this measure of memory, F1-PD and F1-ED animals were affected more than F1-ID animals.herefore our behavioral data show that the three experi-ental F1 groups are compromised in different memoryomains and moreover the degree to which the domainsre affected depends on the type of maternal malnourish-ent. This suggests that the morphological changes un-erlying the differential deficits in memory may also bepecific for each nutritional condition.

There is widespread agreement that the hippocampalormation plays a critical role in normal memory processingnd hippocampal damage in rats causes deficits in thecquisition of various learning tasks (Hirsh, 1974; O’Keefend Nadel, 1978; Olton, 1978). Recent imaging studies inumans has also revealed that the hippocampus under-oes selective volume reduction in stress-related neuro-sychiatric disorders such as recurrent depressive illnessSheline et al., 1996; Sapolsky, 2000; Bremner et al.,000). It is also been shown in studies with children thatheir vulnerability to hypoxic, metabolic, and nutritionalnsults leads to reduced hippocampal volumes and to def-cits in memory (Isaacs et al., 2000). In this context, ourolumetric data analysis showed a significant reduction inhe hippocampal volume in all the three malnutritionroups. The overall decrease in hippocampal volume isighest in the F1-PD group followed by F1-ID and F1-CEDroups respectively. Therefore the deficits that are seen inhe RAM performance could be attributable to changes inhe hippocampal volume, a hypothesis that needs furthernvestigation. However the differences that are seen in the

ifferent error measurements of RAM performance among

he F1 malnutrition groups are unlikely to be attributableolely to the reduction in hippocampal volume comparedith the control group. We therefore did a subdivisionalnalysis of the hippocampus to see if there were differ-nces between the F1 groups.

Although all groups show a decrease in the volume ofhe DG, the decrease was most pronounced in the F1-PDnimals (60% reduction). In addition, the F1-PD animalshowed a reduction of 66% in the volume of the CA1egion. In this context selective lesions of DG and CA1ubregions of the hippocampus have been shown to in-uce a deficit in the acquisition of the task with short-termelays and impair performance in the RAM (Lee andesner, 2003). Whether the worse performance in theorking memory task of the F1-PD group can be attributed

o the dramatic reduction in these regions of the hippocam-us needs further investigation. Interestingly the CA3 re-ion is selectively and specifically spared in the F1-IDnimals. In contrast, the F1-CED animals showed a mod-rate and evenly distributed reduction over all the hip-ocampal subdivisions DG (30%), CA1 (22%) and CA327%). The underlying developmental cause for this obser-ation and whether it is related to the pattern of spatialemory deficits seen in these groups remain to be

nvestigated.

CONCLUSION

n conclusion, our study highlights the long lasting effect ofrain development under conditions of maternal malnour-

shment. Further, it emphasizes the importance of defininghe type of nutritional deficiency that the mother was sub-ected to during the gestation and lactation period. Finally,t demonstrates that functional pathways are differentiallyulnerable to nutritional status during the development ofhe brain. Unraveling the specific developmental pathwayshat are affected by nutritional deprivation and finding theritical period during which these cognitive deficits may beeversed are important goals for the future. It is hoped thatuch studies would help in reversing the deleterious effectsf malnourishment suffered by millions across the world.

cknowledgments—We thank Dr. Shikha Yadav for expert assis-ance with care of the experimental animals. We also thank Dr.alyansunderam from the National Institute of Nutrition, Hyder-bad, India for preparation of the different types of animal feeds aser our specifications. We thank Dr. Neeraj Jain, National Brainesearch Centre, India with help with the Neuroleucida and Neu-

oexplorer programs. This work was supported by an Indo-FrenchCMR-INSERM grant award (P.G. and S.M.) and by the post-octoral scholarship program of the French Embassy in IndiaS.C.R.).

REFERENCES

addeley AD (1986) Working memory. Oxford: Clarendon.remner JD, Narayan M, Anderson ER, Staib LH, Miller HL, Charney

DS (2000) Decreased benzodiazepine receptor binding in prefron-tal cortex in combat-related posttraumatic stress disorder. Am J

Psychiatry 157:115–117.

B

G

H

I

I

I

I

J

K

L

L

L

L

M

M

M

N

O

O

O

P

P

P

P

P

P

S

S

S

T

T

U

S. C. Ranade et al. / Neuroscience 152 (2008) 859–866866

riscoe J, Gathercole SE, Marlow N (2001) Everyday memory andcognitive ability in children born very prematurely. J Child PsycholPsychiatry 42:749–754.

oldman-Rakic PS (1987) Development of cortical circuitry and cog-nitive function. Child Dev 58:601–622.

irsh R (1974) The hippocampus and contextual retrieval of informa-tion from memory: A theory. Behav Biol 12:421–444.

DECG (International Dietary Energy Consultancy Group) Annual Re-port, 1997.

saacs EB, Edmonds CJ, Lucas A, Gadian DG (2001) Calculationdifficulties in children of very low birthweight: a neural correlate.Brain 124:1701–1707.

saacs EB, Lucas A, Chong WK, Wood SJ, Johnson CL, Marshall C,Vargha-Khadem F, Gadian DG (2000) Hippocampal volume andeveryday memory in children of very low birth weight. Pediatr Res47:713–720.

saacs ND (2000) Recovered memory therapy: a dubious practicetechnique. Soc Work 45(2):189.

arrad LE (1993) On the role of Hippocampus in learning and memoryin the rat. Behav Neural Biol 60:9–26.

eselman HJ, Keselman JC (1984) The analysis of repeated mea-sures designs in medical research. Stat Med 3:185–195.

arroque B, Samain H (2001) Groupe Epipage. Epipage study: mor-tality of very premature infants and state of progress at follow up.J Gynecol Obstet Biol Reprod (Paris) 30:S33–S41.

ee I, Kesner RP (2003) Differential roles of dorsal hippocampalsubregions in spatial working memory with short versus interme-diate delay. Behav Neurosci 117:1044–1053.

efebvre F, Bard H, Veilleux A, Martel C (1988) Outcome at schoolage of children with birthweights of 1000 grams or less. Dev MedChild Neurol 30:170–180.

loyd BW, Wheldall K, Perks D (1988) Controlled study of intelligenceand school performance of very low-birth weight children from adefined geographical area. Dev Med Child Neurol 30:36–42.

anjarrez G, Contreras JL, Chagoya G, Hernandez RJ (1998) Freetryptophan as an indicator of brain serotonin synthesis in infants.Pediatr Neurol 18(1):57–62.

avalankar DV, Trivedi CC, Gray RH (1994) Maternal weight, heightand risk of poor pregnancy outcome in Ahmedabad, India. IndianPediatr 31(10):1205–1212.

organe PJ, Austin-LaFrance RJ, Bronzino J, Tonkiss J, Diaz-CintraS, Cintra L, Kemper T, Galler JR (1993) Prenatal malnutrition and

development of the brain. Neurosci Biobehav Rev 17:91–128.

unez JL, MacCarthy MM (2003) Estradiol exacerbates hippocampaldamage in a model of preterm infant brain injury. Endocrinology144(6):2350–2359.

’Keefe J, Nadel L (1978) The hippocampus as a cognitive map.London: Oxford UP.

lton DS (1978) The function of septo-hippocampal connections inspatially organized behavior. Functions of the septo-hippocampalsystem. Amsterdam: Cuba Foundation Symposium, Elsevier.

lton DS, Samuelson RJ (1976) Remembrance of places passed:Spatial memory in rats. J Exp Psychol Anim Behav Proc 2:97–116.

axinos G, Watson C (1986) The rat brain in stereotaxic coordinates.Sydney, Australia: Academic Press.

axinos G, Franklin KBJ (2004) The mouse brain in stereotaxic coor-dinates. Amsterdam: Academic Press.

eterson BS (2003) Brain imaging studies of the anatomical andfunctional consequences of preterm birth for human brain devel-opment. Ann N Y Acad Sci 1008:219–237.

ortman OW (1987) Effects of maternal and long-term postnatal pro-tein malnutrition on brain size and composition in rhesus monkeys.J Nutr 117(11):1844–1851.

oucet B (1993) Spatial cognitive maps in animals: New hypotheseson their structure and neural mechanisms. Psychol Rev 100:163–182.

oucet B, Buhot MC (1994) Effects of medial septa1 or unilateralhippocampal inactivations on reference and working spatial mem-ory in rats. Hippocampus 4(3):315–321.

apolsky RM (2000) Glucocorticoids and hippocampal atrophy in neu-ropsychiatric disorders. Arch Gen Psychiatry 57:925–935.

harma JB, Soni D, Murthy NS, Malhotra M (2003) Effect of dietaryhabits on prevalence of anemia in pregnant women of Delhi. JObstet Gynaecol Res 29(2):73–78.

heline Y, Wang P, Gado M, Csernansky J, Vannier M (1996) Hip-pocampal atrophy in recurrent major depression. Proc Natl AcadSci U S A 93:3908–3913.

olman EC (1948) Cognitive maps in rats and men. Psychol Rev55:189–208.

onkiss J, Galler J, Morgane PJ, Bronzino JD, Austin-LaFrance RJ(1993) Prenatal protein malnutrition and postnatal brain function.Ann N Y Acad Sci 678:215–227.

ylings HB, Van Eden CG, Hofman MA (1986) Morphometry of size/volume variables and comparison of their bivariate relations in thenervous system under different conditions. J Neurosci Methods

18:19–37.

(Accepted 2 January 2008)(Available online 9 January 2008)