Nutrient Requirements

and Interactions

Maternal Dietary Carbohydrate Restriction and Mildto-Moderate Exercise during Pregnancy Modify Aspects of Fetal Development in Rats1'2'3
MONA COBRIN4 AND KRISTINE G. KOSKI5
School of Dietetics and Human Nutrition, McGill University, Quebec H9X 3V9, Canada Montreal,

ABSTRACT To determine whether acute bouts of ex ercise during pregnancy would predispose the fetus to increased risk if maternal dietary carbohydrate were re stricted, untrained pregnant rats were randomly as signed to a 0% (low), 12% (moderate) or 60% (high) glucose diet, and either rested or exercised daily for 20 min from d 16 to term on a rodent treadmill at a mild (15.5 m/min) or moderate (24.3 m/min) intensity. A 3 x 3 nested factorial model with and without food intake as a covariate was employed. Both greater exercise in tensity and the lower levels of dietary carbohydrate in dependently decreased term maternal liver and plantaris glycogen concentrations and increased plasma lactate concentrations. However, significant differences due to exercise disappeared (except for plasma lactate) with food intake controlled for in the model, indicating that energy deficits modulated these exercise effects. In con trast, for the offspring, when food intake was controlled for, a restricted level of maternal dietary carbohydrate significantly lowered fetal weight, plasma glucose and insulin concentrations and liver glycogen concentrations measured at term. Exercise alone did not reduce mean fetal weight if nested weights within a litter were used in the statistical analysis. Mild to moderate maternal ex ercise lowered only fetal plasma glucose concentrations and only if maternal food intake was not controlled for. These results indicate that acute exercise during preg nancy can have detrimental effects on fetal development only if dietary glucose is severely restricted. Otherwise, adequate glucose and energy in the maternal diet in untrained pregnant rats during repeated bouts of acute exercise seem to protect the fetus. J. Nutr. 125: 1617-1627, 1995.
INDEXING KEY WORDS:

glycogen, augments maternal skeletal muscle glycogen content (Mottola and Christopher 1991) and compromises glucose intake by the fetus (Treadway and Young 1989). Adequate glucose delivery to the fetus is critical because glucose is the principal meta bolic fuel metabolized by the growing embryo and fetus (Battaglia and Meschia 1978). Glucose is re quired for the in utero glycogen reserves that must be deposited in the fetal liver and heart during the latter part of gestation (Shelley 1961). Deficits in these glycogen deposits have been associated with increased perinatal mortality (Shelley and Nelligan 1966). More recently, it was reported that the absence of a specific source of dietary glucose limited in utero glycogen deposition, leading to high rates of neonatal mortality that were reversed only if the level of glucose was increased in the maternal diet (Koski and Hill 1986 and 1990, Lanoue et al. 1992). The purpose of this study was to determine whether different levels of dietary carbohydrate com bined with various exercise intensities would perturb carbohydrate homeostasis in pregnancy and alter fetal growth and biochemical development. The hypothesis was that acute bouts of exercise combined

Downloaded from jn.nutrition.org by guest on November 11, 2013

tissue glycogen plasma glucose rats energy intake

'Presented at the 36th Annual Meeting of the Canadian Feder ation of Biological Sciences, June 1993, Windsor, Ontario (Cobrin, M. &. Koski, K. G. (1993) The interaction of dietary carbohydrate and exercise intensity during pregnancy on fetal growth and de velopment. Proc. Can. Fed. Biol. Soc. 36: 396 (abs.)j. 2The financial support of the Natural Sciences and Engineering Research Council of Canada (NSERC #OGP0003623) is gratefully acknowledged. 'The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 USC section 1734

Exercise during pregnancy may induce a compe tition for glucose between exercising muscles and the growing fetus; previous studies have shown that maternal exercise diminishes maternal (Carlson et al. 1986, Gorski 1983) and fetal (Gorski 1983) liver

solely to indicate this fact. 4M. Cobrin was the recipient of postgraduate student fellowship from the Fonds pour la Formation de Chercheurs et l'Aide du Qubec(FCAR|, a Qubecprovincial funding agency. 5To whom correspondence and reprint requests should be ad dressed.

0022-3166/95 $3.00 1995 American Institute of Nutrition. Manuscript received 12 July 1994. Initial review completed 24 August 1994. Revision accepted 22 December

1994.

1617

1618

COBRIN AND KOSKI

with restricted maternal dietary carbohydrate intake during pregnancy could compromise glucose delivery to the developing fetus during late gestation in un trained pregnant rats. The specific objectives were 2) to investigate whether graded levels of maternal di etary glucose combined with various degrees of ex ercise intensity in pregnant rat dams would provoke biochemical alterations in maternal plasma glucose, lactate and insulin and/or liver and muscle glycogen concentrations; 2} to determine whether changes in the maternal system, which also included measures of amniotic fluid glucose and lactate, would be reflected in fetal indices such as litter size, pup weight, liver and heart glycogen, and/or fetal plasma glucose, lactate and insulin concentrations; and 3) to establish if any measured metabolic changes were mediated through differences in energy intake or were a result of a specific dietary glucose deficit. The present study is unique in that the level of maternal dietary carbo hydrate was altered, food intake (which is a proxy for energy intake in isoenergetic diets) was measured, and significantly different effects of exercise are reported when food intake was and was not controlled for as a covariate in the statistical model.

MATERIALS AND METHODS

All procedures conformed to guidelines for ex perimental procedures set forth by the local animal care committee of McGill University and by the Canadian Council on Animal Care (1984). Experimental design. A 3 x 3 factorial design was used to study the interactive effects of varying ex ercise intensity and level of dietary carbohydrate during pregnancy on maternal and fetal measures. Time-bred Sprague-Dawley pregnant rats (180-200 g, Charles River, St. Constant, Qubec, Canada) were received from d 0 (day following impregnation) to d 2 of pregnancy (gestation length 22.5 d) and were im mediately housed individually in suspended wire cages in a temperature-controlled room (20C) with fluorescent lighting (0700-1700 h). Upon arrival, pregnant rats were randomized into three dietary treatments (0, 12 or 60% dietary glucose). These diets were fed throughout pregnancy, and daily food intake and body weight gain were measured. Within each dietary group, dams were again randomized to three daily exercise protocols (20 min/d from d 16 to d 21 of pregnancy): not exercised (rest), exercised at a mild intensity (15.5 m/min) on a rodent treadmill, or exer cised at a moderate intensity (24.3 m/min). All rats were rested from the day they were received until d 13 of pregnancy, when the acclimatization protocol began. The groups of exercised rats were habituated, following minor modifications to a previously reported acclimatization protocol (Carlson et al. 1986, Gorski 1983), by running on the rodent treadmill

from d 13 to 15 of pregnancy inclusive at 8.10 m/min for 10 min/d. Exercise protocol. Our exercise protocol was modified from those employed in previous pregnancyexercise studies (Carlson et al. 1986, Gorski 1983) in which the exercise was limited to one 60-143-min bout near term; our exercising regimen involved daily exercise (from d 16 to term) and spanned the entire period of fetal liver glycogen accumulation, which takes place from d 18 to term in pregnant rats (Shelley and Neligan 1966) and is known to be sensitive to dietary carbohydrate restriction (Koski et al. 1986). Pregnant rats were exercised on a rodent treadmill (from d 16 to term inclusive) at an intensity of either 15.5 or 24.3 m/min for 20 min at 0% grade. These exercise intensities were selected after a pilot study determined that the maximum running speed of un trained Sprague-Dawley pregnant rats at term was 24.3 m/min for 20 min at 0% grade; at this intensity our untrained term pregnant rats would not run beyond the 20 min due to lack of training and despite aversive stimulation. Using the calculation for work in kg-m [(% grade) x (m/min) x (durationmin) x body mass (g)]with 6 kg-m = 1 W, we determined that rats exercised at 15.5 m/min performed 16 W and those exercised at 24.5 m/min completed 28 W of work on average at term. Diet selection and composition. The carbohydratefree, triglyceride-based diet and the control carbohy drate-rich diet are described in Table 1. These same levels of carbohydrate (0, 12 and 60%) were used in previous investigations (Koski et al. 1986, Koski and

Downloaded from jn.nutrition.org by guest on November 11, 2013

TABLE

Composition of control and carbohydrate-restricted, triglyceride-based glucose diets1 Glucose level Ingredient Glucose2Soybean oil3Cellulose4Casein5Vitamin 0% 12% 60%

mix6Mineral mix6DL-MethionineSodium

bicarbonateTotal, gMetabolizable energy, kj/g039.6441.3611.01.25.50.341.0100.0417.31234.934.1011.0 'Dry weight basis (g). 2Dextrose (anhydrous) (ICN Biochemicals Canada, Montral,

Qubec,Canada!. 3Degummed soybean oil (Canada Packers, Montral,Qubec]. 4Alphacel, ICN Biochemicals Canada. High nitrogen casein containing 86.5% protein (N x 6.25) or 89.1% protein (N x 6.71) |ICN Biochemicals Canada). 6See Table 1 in Lanoue et al. (1992) for complete ingredient list.

DIETARY CARBOHYDRATE

AND EXERCISE DURING

PREGNANCY

1619

Hill 1986 and 1990) that established the essentiality of maternal dietary carbohydrate for pregnancy. Each isoenergetic diet was calculated to contain 17.3 kj of metabolizable energy per gram of dry matter, which met the NRC requirement for pregnant rats (NRC 1978). Protein was present at a minimal but adequate level of 11 % to supply the levels of protein, nitrogen and essential amino acids required by pregnant dams and to avoid any excess protein that would provide supplementary glucogenic amino acids that would reduce the severity of the carbohydrate deficiency,- in earlier experiments this level of protein produced pups with weight at term not significantly different from that for controls fed higher levels of protein (Koski et al. 1986). As with protein, the quantities of vitamins and minerals were held constant and were supplied at four times the NRC requirements for the water-soluble vitamins and 1.5 to two times the NRC requirements for the fat-soluble vitamins; these amounts permitted adequate consumption if food intake were voluntarily restricted. To maintain glucose as the first limiting nutrient, the low carbohydrate, high lipid diets were formu lated from the high carbohydrate reference control by substituting lipid (soybean oil, 38.6 kj/g) isoenergetically for glucose (13.3 kj/g) and adding cellulose to maintain a constant metabolizable energy for each kilogram of diet. Triglycridewas chosen as the lipid base because it serves as a natural fat source, and the use of a soybean oil preserves the glycerol component, which can act as a glucose precursor. Thus, the use of intact triglycrideas soybean oil provided 4% glucose equivalents as glycerol in the glucose-free (0%) diet, which is the minimal amount required to support implantation and maintain pregnancy to term (Koski et al. 1986). The intermediate 12% glucose diet was chosen because this restricted level of maternal di etary carbohydrate has produced fetal liver glycogen concentrations that are approximately half those of pregnant control animals (Koski et al. 1986). The 60% glucose diet is the standard control diet used in reproductive studies. The experimental diets and water were freely available, and daily food intake and food spillage were recorded. Total food intake was then calculated by summing the daily food intakes. Thus, cumulative food intake represented total energy intake in our isoenergetic diets. Tissue collection. Dams were delivered of fetuses by caesarian section on gestational d 22.5 (term). All dams were killed in a post-absorptive (fed) state after they were either exercised at a mild or moderate intensity or rested. Exercised rats were killed immedi ately (t < 5 min) after completing their designated exercise protocol. Dams were anesthetized with ketamine-HCl (Rogarsetic, 30 mg/kg, Rogar/STB, London, Ontario, Canada) injected into the jugular region so as to avoid anesthetizing the fetuses. Maternal blood (5-7 mL) was withdrawn by cardiac puncture. All blood was centrifuged and the plasma

stored

frozen

(-20C) until

analyzed.

Following

cardiac puncture, the abdomen was opened and the liver removed, freeze-clamped in liquid nitrogen and frozen at -80C. All uteri were examined for implan tation and rsorption sites, then removed from the abdomen but kept intact. A tuberculin syringe (1 cc, 27 gauge) was used to collect amniotic fluid, which was stored frozen at -20C until analyzed. Individual fetuses and placentas were then removed from the mother and weighed, and live fetuses were killed by exsanguination. Fetal blood was drained from the ax illary artery via heparinized capillary tubes. The blood was centrifuged and the plasma stored frozen at -20C until analyzed. All fetal livers and hearts were removed, freeze-clamped in liquid nitrogen and stored at -80C. The maternal soleus and plantaris muscles were removed from the hind leg, weighed, freezeclamped in liquid nitrogen and stored at -80C until analyzed. Analytical procedures. All tissue glycogen concen trations were determined by the method of Lo et al. (1970). Maternal and fetal plasma glucose and am niotic fluid glucose were measured by hexokinase determination (kit 16-UV, Sigma Chemical, St. Louis, MO) using the Abbott VP Super System (Irving, Texas). The procedure is similar to that described by Bondar and Mead (1974). Maternal and fetal plasma lactate and amniotic fluid lactate were measured by lactate dehydrogenase determination (kit no. 826-UV, Sigma Chemical) using the Abbott VP Super System. This procedure is based on that of Henry (1964). Maternal and fetal plasma insulin were quantitatively determined by RIA in which the antigen competes with a constant amount of radioactive tracer (I125labeled insulin) for the binding sites of the antibody using Serono kits (RIA, Biodata, NCS Diagnostics, Mississauga, Ontario, Canada). The insulin concen tration was obtained by interpolating from a standard curve. Statistical analysis. The experimental design was a 3x3 factorial randomized block design because there were three levels of dietary glucose (0, 12 and 60%) and three levels of exercise intensity (rest, mild and moderate). Maternal data were analyzed by ANO VA and analysis of covariance (SAS software, version 6.0; SAS Institute, Cary, NC) in which cumulative maternal food intake, a proxy for energy in isoener getic diets, was treated as the covariate. Analysis of covariance was necessary because food intake varied according to the level of dietary carbohydrate. Results of main effects (diet, exercise, diet x exercise) were presented with and without food intake as a covariate so that the impact of energy intake on modifying the significance of the main effects of diet and exercise on outcome measures could be evaluated. A limitation of earlier exercise studies (Carlson et al. 1986, Gorski 1983, Mottola and Christopher 1991, Treadway and Young 1989) was that they did not measure food intake or statistically control for energy and/or carbo hydrate intake; therefore, they could not differentiate between these dietary deficits.
Downloaded from jn.nutrition.org by guest on November 11, 2013

1620

COBRIN AND KOSKI

When there was no statistically significant inter action between dietary glucose level and exercise in tensity, the main effects of diet and of exercise were analyzed without an interaction term in the model statement. A significant interaction was found only for maternal liver weight and fetal plasma lactate. When fetal data were recorded by individual fetus (i.e., fetal and placenta weight, fetal liver and heart glycogen), the data were nested. The nesting of data meant that although the data were examined by in dividual fetus, the variability of these data was at tributed to the litter. Thus each pup was not treated as an independent observation because the dam was the experimental unit in this study. However, al though unnested data are not statistically relevant to our model, they are reported here so that comparisons can be made to previous investigations that have reported results using unnested fetal data. All values were considered statistically significant at the P < 0.05 level. Values in the text are means SEM.

RESULTS
Food intake Our exercise protocols did not significantly (P 0.9547) alter maternal food intake. However, as the dietary glucose level of the pregnant rats was in creased from 0 to 12%, there was a significant in

crease in food intake (P < 0.0001), but no further significant increase occurred (Table 2). Having established that only the level of dietary carbohydrate, and not exercise, significantly affected cumulative food intake (P < 0.0001), we analyzed all maternal and fetal measures with and without food intake as a covariate to differentiate between those differences attributable to a maternal energy and those attributable to a specific dietary carbohydrate deficit. Results with and without food (and hence energy) intake as a covariate in the statistical model were included because they allow comparison of our results with those of previous studies that did not control for food intake and permit us to establish the importance of energy intake to the interpretation of the results. Table 3 summarizes the significance for the main effects of diet and exercise for each maternal and fetal variable when food intake, which controls for energy intake in our isoenergetic diets, was and was not included in the statistical model as a covariate. These findings are summarized as follows: 1) Without food intake in the statistical model, maternal liver and muscle weights and their glycogen concentrations, maternal plasma lactate and insulin concentrations, and amniotic glucose and lactate con centrations were significantly modified by the level of glucose in the maternal diet; the only maternal variable not altered by the level of maternal dietary glucose was plasma glucose, which was also not af fected by exercise. Only liver and plantaris glycogen concentrations and maternal plasma lactate were sig nificantly changed by the exercise regimens.

Downloaded from jn.nutrition.org by guest on November 11, 2013

TABLE 2 The effect of graded levels of dietary glucose and exercise intensity on maternal food intake and weight gain in pregnant rats1'2
Statistical Exercise intensityCumulative (%)0(9) glucose level effects4DietExerciseD analysis

gRestMildModeratePooled3Cumulative food intake, 20(15) 327 1513) 318 16(37) 339 10a(9) 328

20(16) 382 1511) 375 18(36) 373 10b(9) 377

19(11) 399 18(11) 405 18(32) 384 10b(10) 396

11(42) 370 935) 366 10(28) 366

EDietExerciseD x

gRestMildModeratePooled3Dietary weight gain, 9(15) 73 9(16) 123 7(13) 61 7(10) 116 7(37) 59 8(35) 112 64 4a12(9) 117

8(11) 152 8(11) 136 8(32) 116 5b60(10) 135

5(42)A116 4(34)A104 B96 5Main 5CPooled3(28)

x EF12.070.050.4364.854.470.84P0.00010.9

'Values are means SEM,with number of dams in parentheses. 2Within a row, values are tested for main effect of diet; different lower-case letter superscripts

indicate significant differences. Within a

column, values are tested for main effect of exercise; different capital letter superscripts indicate significant differences. 3When there was no interaction between diet and exercise (D x E), the statistical analysis treated all dietary groups at each level of exercise as a single group and treated all exercise groups at each level of carbohydrate as a single group. Therefore, the significance is reported for these pooled means. 4Results from ANOVA: diet, main effect of levels of dietary carbohydrate; exercise, main effect of exercise intensity; D x E, interaction of the two main effects (diet and exercise).

DIETARY CARBOHYDRATE

AND EXERCISE DURING

PREGNANCY

1621

TABLE 3

Comparison of the significance of main effects (diet, exercise, diet x exercise) on maternal and fetal metabolic measures in pregnant rats and their offspring with and without food intake as a covariate in the statistical model1 intakeDependent variableMaternalLiver (g)Glycogen weight (mg/g|Heart (g|Soleus weight (g)Glycogen weight (mg/g)Plantaris (g)Glycogen weight (mg/g)PlasmaGlucose (mmol/L)Lactate |mmol/L)Insulin (pmol)AmnioticGlucose
Downloaded from jn.nutrition.org by guest on November 11, 2013

Without

food

intakeExerciseNSNSNS0.0776NSNSNSNS0.00 food

E0.0747NSNSNSNSNS0.0822NSNS0.0835NSNSNS0.009NSNANADiet0.0001NSNS0.0 x E0.0479NSNSNSNSNSNS x

(mmol/L|Lactate (mmol/L)FetalPlasmaGlucose

(mmol/L)Lactate (mmol/L)Insulin (pmol)Liver (mg/g)Heart glycogen glycogen |mg/g)Diet0.00010.00010.02030.00360.00010.01970.0001NS0.00270.00010.00010.03400.0001NS0.0201NA2NAExerciseNS20.0583NSNSNSNS0.0101N 1Results (P values) from ANOVA: diet, main effect of level of dietary carbohydrate (0, 12, 60% glucose); exercise, main effect of exercise intensity (rest, mild (15 m/min), moderate (23 m/min|]; D x E, interaction of two main effects of diet and exercise. "With food intake" represents the level of significance when energy was treated as a covariate in the model: exercise, diet, D x E, food intake. 2NS = nonsignificant (assumes P > 0.100). NA = not applicable.

2) When energy intake was included in the statistical model, liver and plantaris glycogen concen trations were not significantly lowered by our ex ercise protocols; only maternal plasma lactate con centration was modified by exercise. Similarly, maternal liver, soleus, and plantaris glycogen concen trations and maternal plasma lactate were not signifi cantly affected by the level of carbohydrate in the maternal diet if food intake was in the statistical model. In contrast, significant reductions in liver and soleus weight, maternal plasma insulin, and amniotic fluid glucose and lactate concentrations remained when food intake was treated as a covariate, in dicating that differences for these latter variables were produced by the specific maternal dietary carbo hydrate deficit whereas differences in the former, par ticularly the effect of exercise regimens on liver and plantaris glycogen concentrations, were produced by the energy deficit. 3) As for fetal measurements, increased maternal exercise and dietary glucose restriction modified fetal plasma glucose in spite of the fact that the dam was euglycemic. With food intake as a covariate, restric tions in maternal dietary glucose resulted in signifi cantly lower fetal plasma glucose and insulin whereas exercise had no effect, indicating that insufficient

maternal energy intake had produced the earlier sig nificant change in fetal plasma glucose following mild to moderate exercise in the pregnant rat dam.

Maternal body weights

Both the level of dietary carbohydrate (P < 0.0001) and exercise intensity (P < 0.0139) significantly in fluenced maternal weight gain,- cumulative weight gain decreased as the level of dietary carbohydrate decreased, and dams exercised at a moderate intensity weighed significantly less than those rested or exer cised at a mild intensity (Table 2). Initial body weights (mean = 200 g) had not been significantly different among the treatment groups.

Reproductive performance
Reproductive performance of the dams was measured by the following indicators: the number of implantations (12 1), the rsorption rate (1 I/ litter), the stillbirth rate (0/litter) and the number of live fetuses (11 I/litter). Neither dietary carbohy drate level nor exercise intensity influenced any of

1622

COBRIN AND KOSKI

these variables; the values were not different among the groups (data not shown), indicating that none of the significant differences were caused by variation in litter size. Results of the effects of dietary carbohydrate level and exercise intensity on fetal weights are summa rized in Table 4. The significance of both nested and unnested pup weight at term (d 22.5) is reported so that comparisons can be made to previous investiga tions, all of which have used unnested fetal data only, thus assuming that each fetus represented an in dependent observation. When our data were unnested, both diet (P < 0.0001) and exercise (P < 0.0002) were found to have significant effects on term fetal weights. However, after nesting, only carbohydrate level produced significant differences in fetal weights among all dietary treatment groups. The lack of sig nificant differences from our exercise protocols once the data were nested indicated that our exercise regimens produced higher variability among pups within a litter than among litters.

Maternal metabolic variables

Plasma. Maternal plasma glucose was not signifi cantly altered by diet or by exercise (Table 5). However, maternal plasma lactate was significantly higher in exercised vs. rested dams, whereas maternal plasma insulin was affected by diet and not exercise, but not in a dose-dependent manner as might be expected. Dams fed the 12% glucose diet had signifi cantly lower non-fasting plasma insulin concentra tions than dams fed the 0 and 60% diets,- concentra tions for the latter groups did not significantly differ.

These individual differences existed whether food intake was or was not included in the statistical model (Table 5). Plasma lactate concentrations, as expected, were significantly higher in exercised vs. rested dams. All plasma measurements were per formed in post-absorptive (fed) pregnant rat darns. Tissue. The glycogen concentrations of two exer cising muscles showed contrasting effects of diet and exercise and, depending on whether food intake was a covariate, exercise and/or the specific dietary carbo hydrate deficit modified the statistical significance of the main effects of diet and exercise (Table 5). The soleus muscle with its predominantly slow twitch oxidative fibers was not affected by the acute exercise protocols, and only when food intake was not in cluded as a covariate did a severe (0 or 12%) dietary carbohydrate restriction significantly lower the con centration of glycogen in this muscle. For the predominantly fast twitch plantaris muscle, which relies on the short-term glycolytic system for energy, the glycogen concentration was affected by both ex ercise and diet, with the highest intensity exercise and the lower levels of glucose in the maternal diet producing significantly lower plantaris muscle glycogen concentrations, but only if food intake was not controlled for in the statistical analysis. Thus, the elimination of significant exercise and diet effects, once food intake was included in the statistical model, demonstrated that for plantaris glycogen con centrations, the absence of sufficient food intake and not the level of carbohydrate in the maternal diet or the exercise intensity per se had produced the earlier significant differences. Maternal liver glycogen concentrations (Table 5) were not significantly lowered by the mild-moderate

Downloaded from jn.nutrition.org by guest on November 11, 2013

TABLE 4 The effect of graded levis of dietary glucose and exercise intensity on individual fetal weights in term rat pups1'2
Statistical Exercise intensityRest glucose0(103) dietary (%)g analysis

3.8 0.04 (127) 4.8 0.02 0.09065 0.00026 Mild (173) 3.3 0.06 1185) 5.0 0.04 (USI 5.2 0.04 (473) 4.5 0.02 Exercise ' (123) 4.8 0.07 0.3730sUnnested0.00016 0.75946 D x EPNested0.0001s Moderate (135) 3.3 0.07 (1261 5.2 0.05 (384) 4.5 0.02effects4Diet Pooled3Maternal 0.02a(130)1438)125.1 5.0 0.02blevel(368)605.4 0.02CPooled3(360) (411) 3.5 0.07 5.3 0.04 'Values are least square means SEM,with number of fetuses in parentheses. 2Within a row, values are tested for main effect of diet: different lower-case letter superscripts indicate significant differences. Within a column, values are tested for main effect of exercise; different capital letter superscripts indicate significant differences. 3When there was no significant interaction effect of diet and exercise (D x E), the statistical analysis treated all dietary carbohydrate groups at each level of exercise as a single group and treated all exercise groups at each level of carbohydrate as a single group; the significance is reported for these pooled means when food intake was included as a covariate. 4Results from ANOVA: diet, main effect of levels of dietary carbohydrate; exercise, main effect of exercise intensity,- D x E, interaction of the two main effects (diet and exercise). sSignificance (P value) when individual Significance (P value) when individual fetal weights are nested within a litter and the dam is treated as the experimental unit. fetal weights are not nested and each fetus is considered as an independent observation.

DIETARY CARBOHYDRATE

AND EXERCISE DURING

'S es d r+1+1
PO uo O\

PREGNANCY
food intake when analyzedajwer-case ample was vo

1623

rs j= Ovddov t -H -HcscStSOeSeS des dd d POd es d o d in+1 +1+1VOun +1 +1 +1+1 +1 +1 +1 +1 +1 +1 +1 ,*ett

*H vo TfSin OPO Ov vo O ^f eS ^ ro eS O PO

dvO

un -^-o unesJ3_PO-H eses eSPO

^ r^- *-; ^ -H PO o POCSPOPOPOPOPO

o\o eses es do

r^^

fet: individual s letter supers each

*i -HO
+H+I

intensity2'3Xpj?_jJj Ttddov m+I+I+H <s 2SB,5iCIO VOp unovo oo -H' ve

PO o\POPOPO PO
Tt<

-neSeSeSOeSeS dd opo'desdod

-H
+I+H +I+H+I+I+I+I+I +H+I

vocs OPO

poooo Tt in T o O 1^ O vo eS
-H es o -H es es

PO rfO
-H

Ov vo PO

POPO

*cS Tf Tj- PO Ti- PO

POeS

Ov

|| 7.

^^ "~ ^ d d vo dPO dd mTt-H vom un-Hi^ un00 OPO

1^es
'tPO

esooo es es es d POd es d o d -H poesoo T- es ON^doveNO^t

-H PO O

PO d i*-

VOPO <

'"*
Aa >

es

-H

PO POOV -o M POO es 2 '" PO SS E

-H es

0 ^ R.a u

Downloaded from jn.nutrition.org by guest on November 11, 2013

POO^ PO es CS-H

es es es es es es es

Sg l'a *o 2.s
Pe J3

o P -*rf u o 4^ e o.~ a -c S3
UcQ vo O
-H-H

U ja

PO ><;' do'vo -+1

+1 +1
fc,

CJ< es dd
+1+1

pa"ueS^o esooOajopodesdod

es^

o es

03"ja eSg,;d'*
-H -H

* ^S - a
Sf<i

+1 +1 -H -H -H -H -H

0~ 0g r^u

v> M Z l1

00<t

af-S u
-H es Ovlv. CS-H -.
i

S^

OvTfvoun
IN31si<a

oS-S5 S w
OveS

sT3
<S

csescs 'S-C.
"8Si1Gi

TtPO eses

rtOOiO-H-H PO-H POCSPOPOPOPOPO -o *

BQ _Q C^ eq Jl -C (S-^O.^ N^ (S rCOiSOd

*U O J OJ OHi C O ^^

_C fO 00 fO 'Bf ^<N Os -CSC3

COr**,

X CS 3 O

*T in <T>

i~
N>-y 'S

i||

i(SfCfO r--.

^oo

h--r-)c30

^H

\Q

i/

s '^i

r^ccf*3

CS(N

rCfOecrorO^tS

(ScS

diet exercise, exer x of effects die main ana;n for diet exercise r or ar of samples fetal
^ f*5 ~ O

1Moderate2S"PASCSoVariable
l""t<f OO un^^H ro ^H i i c^ 00 O ^ cS ^ cS OrCiNCJO ^f-^^OO OO OO O O O ^ VO ^^ O ^K i^ O CS

c^l^ Ol^-

of dams,
B,11areise
iTi ^^ O f1^ VO<Nf^ro CO COl/1 ^3 ?O Tf CS ^N IO \OO\ C^ ^ C*l

esco

fOtScOfOroroeS l^co

tsS

li3-t |E8 8
S u ^ U H -C

aa g .2 a

%llil |r1

s41 "uig S -11 II S nS Se8-al . "

StillDts
C>H0 -J3Ti C

c c S *ills-l3 a > i.
C n C C

2> .>.3 3 2 K C'* o Z ^Z " cS PC *&.2

!

1624

COBRIN AND KOSKI

exercise protocols. However, when food intake was not in the statistical analysis, the restriction of glucose in the maternal diet resulted in significantly lower liver glycogen concentrations, indicating, as previously shown for plantaris glycogen concentra tions, that adequate energy intake played an im portant role in maintaining these maternal tissue glycogen concentrations.

Fetal metabolic variables

Plasma. When individual values were compared (Table 5), fetal plasma glucose was significantly lower in dams fed 0% dietary glucose than in dams fed 12 or 60% dietary glucose, whereas fetal glucose was higher, but still euglycemic, in offspring from exer cised vs. rested dams. However, when food intake was treated as a covariate, only dietary carbohydrate modified fetal plasma glucose, with significantly lower concentrations occurring in fetuses from dams fed the 0% glucose diet. A significant interactive effect of diet and exercise was found for fetal plasma lactate (Table 6) such that pups from dams fed 12% glucose and mildly exercised had lower plasma lactate concentrations than pups from dams rested or moder ately exercised, whereas pups from dams fed the 60% glucose diet and moderately exercised had concentra tions significantly lower than those of all other groups. Fetal insulin responded to dietary carbohy drate level but not to exercise intensity (Table 5). Fetuses from dams fed 0% dietary carbohydrate had significantly lower insulin concentrations than did fetuses from dams fed 60% dietary carbohydrate, but

neither group differed from the values for fetuses of dams fed the 12% glucose diet. Tissue. Fetal liver and heart glycogen were measured as indicators of fetal carbohydrate stores (Table 5). Fetal liver glycogen concentration was in fluenced by the level of carbohydrate in the maternal diet but not by mild to moderate exercise and only when food intake was a covariate in the model. Fe tuses of dams fed 60% carbohydrate had liver glycogen levels significantly higher than those of dams fed 12% carbohydrate, which in turn had fetal liver glycogen values significantly higher than those of fetuses of dams fed 0% carbohydrate. Similarly, fetal heart glycogen was also significantly affected by the level of dietary glucose, but not by our exercise protocols when food intake was a covariate; fetuses from dams fed 0 or 12% dietary glucose had signifi cantly lower levels of heart glycogen than fetuses from dams fed the control (60%) glucose diet.

Downloaded from jn.nutrition.org by guest on November 11, 2013

DISCUSSION
In our study we attempted to differentiate between those effects that were mediated by energy restriction and those that could be attributed either to exercise and/or to the level of dietary carbohydrate. We also chose to use untrained pregnant rats to eliminate any possible metabolic adaptation resulting from training (e.g., enhanced glycogen repletion following high in tensity exercise). As a main treatment effect, our exercise protocol produced significantly lowered glycogen concentration only in the fast twitch plan taris muscle, which is as expected for the non-

TABLE 6 The effect of graded levels of dietary glucose and exercise intensity on fetal plasma lclate1'2
Statistical (%)01260Main dietary glucose level Exercise intensityMaternal effectsF*P4mmol/LRestMildModerate(9) 0.6|8) 4.9 0.7(11) 5.5 4.8 0.6a(5) A6.9(10) 0.8 C4.4(6| 0.6 0.7(7) Vi 0.7(6) A6.0 B3.2 0.8bDietExerciseD analysis3

A6.6 0.8ab(7)

x E1.32(0.61)1.67(0.30)3.69(2.60)0.2739(0.5490)0.1967(

'Values are means SEM,with number of dams in parentheses. ^Within a row, values are tested for main effect of diet; different lower-case letter superscripts indicate significant differences. Within a column, values are tested for main effect of exercise; different capital letter superscripts indicate significant differences. ^Results from ANOVA: diet, main effect of levels of dietary carbohydrate; exercise, main effect of exercise intensity; D x E, interaction of the two main effects (diet x exercise). 4Numbers in parentheses represent F and P values when energy was treated as a covariate in the model: exercise, diet, diet x exercise, food intake.

DIETARY CARBOHYDRATE

AND EXERCISE DURING

PREGNANCY

1625

pregnant animal. However, when food intake was controlled for in the statistical model, this difference was nonsignificant. The only other variable to be consistently altered by exercise was plasma lactate, which is expected in exercising animals and has been reported by others (Carlson et al. 1986). For most maternal variables, including maternal liver and muscle glycogen concentrations, our results showed that a significant main effect of diet was nonsig nificant when food intake was treated as a covariate, indicating that the reduction in food intake caused an energy deficit that resulted in low muscle and liver glycogen concentrations in rats fed the carbohydraterestricted diets. These results suggest that the primary concern in the maternal system is for ade quate energy and that with adequate energy intake, acute daily exercise late in pregnancy produces no deleterious maternal metabolic defects. In the fetal system, however, outcomes that were significantly altered by diet (plasma glucose and in sulin and liver and heart glycogen concentrations) were still significantly lower when food intake was treated as a covariate. This demonstrates that for the fetus, maternal dietary carbohydrate intake is as im portant as energy, and an adequate level of maternal dietary glucose is critical for fetal development. We suggest that an adequate energy intake of a carbohy drate-restricted diet during pregnancy will not supply sufficient glucose for fetal development, whereas daily bouts of mild to moderate exercise in untrained pregnant rats will not be deleterious to fetal de velopment if maternal energy intake is adequate. Previous experiments using single bouts of acute exercise near term in fed pregnant rats have presented conflicting results. Both Gorski (1993) and Carlson et al. (1986) observed that exercise lasting 60 min at 12 m/min, 0% grade (-35 W) caused no significant de cline in maternal blood glucose (rest vs. 60 min) but did result in significantly lower maternal liver glycogens on d 21 of gestation. However, lower fetal plasma glucose and fetal liver glycogen were reported by Gorski (1983) but not by Carlson et al. (1986). Our results, using sequential daily bouts of mild to moderate exercise (16-28 W), supported the obser vation of Carlson et al. (1986) that maternal plasma glucose concentrations and maternal and fetal liver glycogen concentrations were unchanged by exercise,however, we did observe a significant decrease in fetal plasma glucose that vanished once maternal food intake was considered as a covariate. To explain their earlier contrasting results (Carlson et al. 1986 vs. Gorski 1983) the authors had suggested interstrain differences in response to exercise and/or differences in experimental protocol... We could add the fol lowing comments in light of our in utero findings. It is possible that the comparison of near-term [i.e., gestational days 20.5 and 21.5 (Carlson et al. 1986)] with term data [gestational length 22.5 days, Gorski

(1989) and present study] may contribute to some of the confusion in data interpretation because it is ac cepted that fetal liver glycogen may be mobilized in utero, but only at term. Hence, the absence of differ ences in glycogen concentrations in the fetal livers in the Carlson study may be due to performing measure ments following only a single bout of exercise prior to, and not at, term. The absence of an effect of exercise on glycogen concentrations measured at term in our study (which did modify fetal, but not maternal plasma glucose) may result from the mildness of the exercise intensity even though it was performed at term. Therefore, it is still possible that high intensity exercise during the entire period of fetal glycogen accumulation could significantly modify tissue stores, as observed in the Gorski (1983) study, in which the fed dams exercised to exhaustion at term had pro found hypoglycemia, very low maternal liver glycogen levels, and low fetal liver glycogen concen trations. Another consideration in the interpretation of the existing studies concerns the sampling proce dures and the subsequent statistical analyses. Carlson et al. (1986) and Gorski (1983) used multiple t tests in which each fetus was treated as an individual obser vation even though the treatment has been applied to the dam. There is high individual variation within a litter, and consequently the lack of a random selection of a representative sample from each litter and the lack of application of correct statistics (oneor two-way ANOVA followed by a more conservative multiple comparison test, given the nature of the experimental designs) may have resulted in accepting significant differences that did not exist. In the present study, moderate exercise during pregnancy resulted in significantly lower glycogen concentrations in the maternal plantaris muscle and greater maternal lactate concentrations, indicating that the exercise protocol was sufficiently intensive to produce the commonly expected observations. However, the elimination of the significant effects on glycogen concentrations when food intake was con trolled for in the statistical model demonstrates an important role for energy in modulating muscle glycogen levels. In contrast to our observation, others (Baldwin et al. 1973, Blake and Hazelwood 1971, Mottola and Christopher 1991) have reported in creased glycogen concentrations in the gastrocnemius muscle following exercise; however, this occurred several hours after exercise, whereas our measure ments were made immediately after the exercise bout, before any repletion, thus explaining the lower than average reported values (Mottola and Christopher 1991) in our study. There was no effect of exercise on reproductive performance in our study or the studies by Carlson et al. (1986) or Gorski (1983). This contrasts with the results of studies by Garris et al. (1985) and Wilson

Downloaded from jn.nutrition.org by guest on November 11, 2013

1626

COBRIN AND KOSKI

and Gisolfi (1980), in which rat dams exercised regu larly during pregnancy had a significant decrease in the number of live births per litter, decreased placental weights, suppression of uteroplacental blood flow, and an increased fetal morality rate in exercised groups. Our failure to show a response to regular daily exercise may be due to the fact that the intensity and/ or duration of exercise was not as high or as long as that reported in the studies of Garris et al. (1995) or Wilson and Gisolfi (1980), in which the pregnant rats exercised for 1 h/d for a longer time span during pregnancy. Previously, a study from this laboratory (Koski and Fergusson 1992) found that amniotic fluid glucose and lactate increased significantly as the level of carbohy drate increased in the maternal diet. Although there was a significant effect of diet on amniotic fluid glucose and lactate in the present study, acute mild to moderate exercise did not produce any significant changes in these two variables. On the basis of this exercise study, we would suggest that amniotic fluid glucose or lactate would not be a sensitive indicator of mild to moderate exercise stress during pregnancy. Although maternal weight gain was significantly lower with increased exercise intensity in this study, fetal weight, when individual pup data were nested, was not significantly altered by exercise. This was surprising because it has been assumed that maternal weight loss, even if precipitated by increased exercise, would be reflected in the fetus. However, in two different studies (Mottola and Christopher 1991, Mottola et al. 1989), fetal body weight from exercised dams was either not different or increased in the exercised dams when compared with the nonexercised control group. In contrast, several other studies reported a reduction in fetal body weight as a result of maternal exercise (Gilbert et al. 1981; Longo et al. 1978, Nelson et al. 1983; Terada 1974). These obser vations are similar to our present findings when pup weights, which were unnested, were significantly lo wered by both dietary carbohydrate restriction and increased exercise intensity. These disparate results suggest that fetal weight may not be a reliably sen sitive indicator of exercise stress, and without uniform methodological and statistical controls, the issue cannot be fully resolved at this time. This study showed that fetal and maternal com partments do not respond similarly to the diet and exercise effects, as demonstrated by differing responses in maternal and fetal weight gain and plasma glucose, lactate and insulin. We suggest that maternal and fetal responses should be examined separately, and suggest that caution be used in the extrapolation of results from one compartment to another, because it cannot be assumed that fetal responses are reflective of those occurring in the maternal system. Because a significant modifying effect of food intake was found in this study, we also

suggest that exercise effects reported in previous studies could have resulted from inadequate energy intake. We conclude that mild to moderate exercise alone, when accompanied by adequate energy and carbohydrate intake, does not pose a threat to the mother or the developing fetus. Further investigations are needed to determine if this is also true when exercise is of a higher intensity.

LITERATURE CITED
Baldwin, K. M., Reitman, J. S., Terjung, R. L., Winder, W. W. & Holloszy, J. O. (1973| Substrate depletion in different types of muscle and liver during prolonged running. Am. J. Physiol. 225: 1045-1050. Battaglia, F. C. & Meschia, G. (1978) Principal substrates of fetal metabolism. Physiol. Rev. 58: 499-527. Blake, C. A. & Hazelwood, R. L. (1971) Effect of pregnancy and exercise on actomyosin, nucleic acid and glycogen content of the rat heart. Proc. Soc. Exp. Biol. Med. 136: 632-636. Bondar, P.J.L. & Mead, D. C. (1974) Evaluation of glucose-6-phosphate dehydrogenase from leuconostoc mesenteroides in the hexokinase method for determining glucose in serum. Clin. Chem. 20: 586-590. Canadian Council on Animal Care (1984) Guide to the Care and Use of Experimental Animals, Volumes I and II. National Li brary of Canada, Ottawa, Ontario, Canada. Carlson, K. I., Yang, H. T., Bradshaw, W. S., Conice, R. K. & Winder, W. W. (1986) Effect of maternal exercise on fetal liver glycogen late in gestation in the rat. J. Appi. Physiol. 60: 1254-1258. Garris, D. R., Kasperek, G. J., Overton, S. V. & Alligood, G. R. (1985) Effects of exercise on fetal-placental growth and uteroplacental blood flow in the rat. Biol. Neonate 47: 223-229. Gilbert, R. D., Nelson, P. & Longo, L. D. (1981) Long term maternal exercise in guinea pigs: effects of fetal growth and development and placental diffusing capacity. Am. J. Obstet. Gynecol. 140: 123-127. Gorski, J. (1983) Effect of exercise on metabolism of energy sub strates in pregnant mother and her fetus in the rat. In: Biochem istry of Exercise (Knuttgen, H. G., Vogel, J. A. &. Portmans, J., eds.), pp. 229-233. Human Kinetics Publishers, Champaign, IL. Henry, R. J. (1964) Clinical ChemistryPrinciples and Technics, pp. 664-666. Harper and Row, New York, NY. Koski, K. G. & Fergusson, M. A. (1992) Amniotic fluid composition responds to changes in maternal dietary carbohydrate and is related to metabolic status in term fetal rats. J. Nutr. 122: 385^92. Koski, K. G., Hill, F. W. &. Hurley, L. S. (1986) Effect of low carbohydrate diets during pregnancy on embryogenesis and fetal growth and development in rats. J. Nutr. 116: 1922-1937. Koski, K. G. & Hill, F. W. (1986) Effect of low carbohydrate diets during pregnancy on parturition and postnatal survival of the newborn rat pup. f. Nutr. 116: 1938-1948. Koski, K. G. & Hill, F. W. (1990) Evidence for a critical period during late gestation when maternal dietary carbohydrate is essential for survival of newborn rats. J. Nutr. 120: 1016-1027. Lanoue, L., Miniaci, S. & Koski, K. G. (1992] Placental composition does not respond to changes in maternal dietary carbohydrate intake in rats. }. Nutr. 122: 2374-2382. Lo, S., Russell, J. C. & Taylor, A. W. (1970) Determination of glycogen in small tissue samples. J. Appi. Physiol. 28: 234-236. Longo, L. D., Hewitt, C. W. & Loring, R.H.W. (1978) To what extent does maternal exercise affect fetal oxygnation and uterine blood flow? Fed. Proc. 37: 905 (abs.). Mottola, M. F., Bagnali, K. M. & Beicastro, A. N. (1989) Effects of

Downloaded from jn.nutrition.org by guest on November 11, 2013

DIETARY CARBOHYDRATE

AND EXERCISE DURING

PREGNANCY

1627

strenuous maternal exercise on fetal organ weights and skeletal development in rats. J. Dev. Physiol. 11: 111-115. Mottola, M. F. & Christopher, P. D. (1991) Effects of maternal exercise on liver and skeletal muscle glycogen storage in pregnant rats. J. Appi. Physiol. 71: 1015-1019. National Research Council |1978) Nutrient requirements of the laboratory rat. In: Nutrient Requirements of Laboratory Animals, pp. 7-37. National Academy of Sciences, Washington, DC. Nelson, P. S., Gilbert, R. D. & Longo, L. D. (1983) Fetal growth and placental diffusing capacity in guinea pigs following long-term maternal exercise. J. Dev. Physiol. 5: 1-10.

Shelley, H. J. (1961) Glycogen reserves and their changes at birth and in anoxia. Br. Med. Bull. 17: 137-143. Shelley, H. J. & Nelligan, G. A. (1966) Neonatal hypoglycemia. Br. Med. Bull. 22: 34-39. Terada, M. (1974) Effect of physical activity before pregnancy on fetuses of mice exercised forcibly during pregnancy. Teratology 10: 141-144. Treadway, J. L. & Young, J. C. (1989) Decreased glucose uptake in the fetus after maternal exercise. Med. Sci. Sports Exercise 21: 140-145. Wilson, N. C. & Gisolfi, C. V. (1980) Effects of exercising rats during pregnancy, f. Appi. Physiol. 48: 34-40.

Downloaded from jn.nutrition.org by guest on November 11, 2013