Main

Epidemiologic evidence suggests that maternal diet and body composition may be key factors in the links between low birth weight and cardiovascular disease, hypertension, and insulin-independent diabetes in adult life (1–3). These population studies are strongly supported by animal studies. In the rat, a maternal low-protein diet or global undernutrition in pregnancy is associated with reduced birth weight, and elevated blood pressure or impaired glucose tolerance in adult offspring (4–9). The mechanisms that underlie this effect are not known, but their elucidation will be important for future public health measures aimed at improving the nutrition of women of reproductive age.

The provision of nutrients to the fetoplacental unit depends partly on uterine blood flow. Uterine blood flow is increased during normal pregnancy because of vasodilation (10) and vascular remodeling and has been shown to be reduced when pregnant rats are fed a globally restricted diet (11).

VEGF is a specific mitogen (12), which promotes vasodilation and neovascularization, and it has been identified in fetal tissues, the placenta, and also in adult blood vessels (13, 14). Vasodilation in response to VEGF is mediated by the production of NO and PGI2 through stimulation of endothelial flt-1 and Flk-1/KDR receptors (15). Recently it has been reported in the rat that VEGF mRNA levels increase in the uterine tissue in pregnancy (16). In humans, it has been shown that the maternal plasma concentration of VEGF is increased during pregnancy (17, 18).

The present study was designed to investigate the hypothesis that a maternal low-protein diet affects the uterine circulation during pregnancy through a change in the vasodilator response to VEGF. Experiments were also conducted to determine whether any effect on the response to VEGF was dependent on reduction of NO or PGI2 production. Preliminary data showed that VEGF-induced NO release is reduced in uterine arteries of the low-protein group (Brawley, Itoh, and Clough, unpublished observation). In the present study we therefore concentrated on examining the effect of inhibition of NO synthase in isolated uterine arteries by removing the prostacyclin component of the VEGF-induced relaxation first using INDO. To determine whether any effects were confined to the uterine artery, the response of the mesenteric artery to VEGF was also studied in the same animals.

METHODS

Animals and dietary protocols.

All animal experimentation was conducted in accordance with the U.K. Home Office Animals (Scientific Procedures) Act 1986, and this study was approved by the ethical review process. Twenty-one virgin female Wistar rats, (Harlan, Bicestor, Oxon, U.K.) were studied. Rats were randomly divided into control and low-protein groups. Before pregnancy they were fed standard laboratory chow (CRMX, Special Diets Services, Cambridge, U.K.). Feeding of the synthetic diets began on the day of conception (d 0 of gestation), which was confirmed by observation of semen plugs on the floor of the mating cage. Ten rats in the control group were fed a diet containing 18% casein by weight, and 11 rats were fed a low-protein diet containing 9% casein by weight. All diets contained 5 g/kg methionine (5). In addition, 0.5 g/kg magnesium sulfate was included to prevent magnesium deficiency (Table 1). All rats had free access to food and water and were housed individually in cages in a room with a 12-h light/dark cycle and maintained at 22°C. Animals were killed by CO2 inhalation on d 18 or 19 of gestation. For each litter, five pups and their placentas were taken at random and weighed after the dissection of vessels.

Table 1 Composition of the 18% and 9% diets
Full size table

Determination of vascular function.

The uterus and intestine were removed and immediately placed in cold PSS (in mM: NaCl, 119; KCl, 4.7; CaCl2, 2.5 mM; MgSO4, 1.17; NaHCO3, 25; KH2PO4, 1.18; EDTA, 0.026; and glucose, 5.5). Segments of the main uterine artery (diameter approximately 400 μm) and mesenteric artery (diameter approximately 300 μm) were dissected free from adjoining connective tissue and mounted as a ring preparation on a small vessel wire myograph (Multi Myograph Model 610M, J.P. Trading, Aarhus, Denmark) (19). The arteries were bathed in PSS gassed with a mixture of 95% O2 and 5% CO2 (pH 7.4 and 37°C). The passive wall tension-internal circumference relationship was determined by incremental increases in tension, regulated by a microprocessor, to achieve an IC100 (using the Laplace relationship), and the arteries were set to a diameter equal to 0.9 × IC100. To test viability, vessels were exposed four times to KPSS (125 mM K, equimolar substitution of NaCl with KCl in PSS). Arteries that produced tension equivalent to <100 mm Hg in response to KPSS were excluded from subsequent study.

The following protocol was performed in the uterine artery. Cumulative dose-responses to PE (10−8 to 10−5 M) and KCl (10−3 to 75 × 10−3 M) were conducted. Cumulative relaxation dose-responses to ACh (10−8 to 10−5 M) and VEGF (10−11 to 3 × 10−8 M; human recombinant VEGF, Genentech Inc., San Francisco, CA, U.S.A.) were then assessed after preconstriction with PE. The PE concentration used for preconstriction was that required to produce 80% of the maximal response (pEC80) induced by KPSS. To assess the role of NO synthase, the ACh dose-response curve was repeated after preincubation with the NO synthase inhibitor, l-NAME (10−4 M) for 30 min. To assess the role of NO synthase and cyclooxygenase, the VEGF dose-response curve was performed after preincubation with the cyclooxygenase inhibitor INDO (10−5 M) and then repeated after preincubation with a combination of INDO and l-NAME for 30 min. In the mesenteric artery, the cumulative dose-response curve to PE (10−8 to 3 × 10−5 M) was established, and after preconstriction with PE (pEC80), VEGF dose-response curves (10−10 to 10−8 M) were studied.

All dose-response curves were conducted with additions of increasing concentration at 2-min intervals, and 20 min was allowed between measurements of each response curve. All drugs and chemicals except VEGF were obtained from Sigma Chemical Co. (Poole, U.K.).

Data analysis.

Values are given as mean ± SEM. Tension was expressed as a percentage of maximal constriction to KPSS, to correct for any small differences in vessel diameter. Relaxation was expressed as a percentage of the initial preconstriction tension. The maximal values of the PE and KCl responses were calculated by least squares nonlinear regression analysis, and the pEC50 was calculated (Prism 3.0, Graph Pad Software Inc., San Diego, CA, U.S.A.) if appropriate. These values were compared by Mann-Whitney U test (SPSS 9.0J, SPSS Inc., Chicago, IL, U.S.A.). If calculation of pEC50 was not appropriate, concentration-response curves were compared by two-way repeated-measures ANOVA (Prism 3.0, Graph Pad Software Inc.). Significance was accepted at the level of p < 0.05.

RESULTS

Animal Data

The maternal body weight on d 18 or 19 of gestation, the rate of increase of maternal body weight during pregnancy, and the number of fetuses were not different between the 18% and 9% casein groups (Table 2). Averages of fetal and placental weights from each individual litter were not different between the two groups (Table 2).

Table 2 Maternal and fetal weight
Full size table

Vascular Responses

The normalized internal diameters and the maximal tension in the uterine artery induced by 125 mM KPSS were not different between the two groups (Table 3). In the mesenteric artery, there was no significant difference in the KPSS response between groups (18%versus 9%, maximal tension, 17.5 ± 0.7 and 18.3 ± 1.00 mN/mm, respectively).

Table 3 Uterine vascular function of control and protein-restricted dams
Full size table

Constriction responses.

The uterine artery responses to KCl were not different between groups (Table 3 and Fig. 1A). There were also no significant differences in the PE dose-response curve between the groups (Fig. 1B). In the mesenteric artery, there was no significant difference in the PE response between groups (data not shown).

Figure 1
figure 1

A, dose-response curves to KCl of the uterine artery in the 18% (•, n = 9) and 9% casein (○, n = 9) groups. B, dose-response curves to PE of the uterine artery in the 18% (•, n = 10) and the 9% casein (○, n = 10) groups.

Full size image

Relaxation responses.

The maximal relaxation of the uterine artery to ACh (3 × 10−6 M) was not different between the two groups (Table 3 and Fig. 2). In the presence of l-NAME, the maximal relaxation to ACh was significantly reduced in both groups (Table 3 and Fig. 3). VEGF caused relaxation in the uterine artery in both dietary groups, however the overall relaxant response (p < 0.01, two-way ANOVA) and maximal relaxation were significantly reduced in the rats fed 9% casein (Table 3 and Fig. 4A). VEGF caused significant relaxation of the mesenteric artery in the both dietary groups. Although the effect was less in the 9% casein group, this was not significant (maximal relaxation, p > 0.05; two-way ANOVA, p = 0.05;Fig. 4B). The uterine artery was noted to be more sensitive to VEGF than the mesenteric artery (uterine artery, maximal relaxation attained with 5 × 10−10 M; mesenteric artery, maximal relaxation attained with 1 × 10−8 M;Fig. 4C). At high concentrations the VEGF-induced relaxation was attenuated in the uterine artery (Fig. 4C).

Figure 2
figure 2

Dose-response curves to ACh of the uterine artery in the 18% (•, n = 7) and the 9% casein (○, n = 6) groups.

Full size image
Figure 3
figure 3

The effect of l-NAME on the response to ACh of the uterine artery in the 18% (•: without inhibitors, n = 7; ▴: with l-NAME, n = 5) and the 9% casein groups (○: without inhibitors, n = 6; ▵: with l-NAME, n = 6).

Full size image
Figure 4
figure 4

A, dose-response curves to VEGF of the uterine artery in the 18% (•, n = 10) and the 9% casein (○, n = 11) groups. *p = 0.041, Mann-Whitney U test, for significant difference in the maximal relaxation between the 18% and the 9% casein groups. †p = 0.004, significant difference in the overall relaxant response assessed by two-way ANOVA between the 18% and the 9% casein groups. B, dose-response curves to VEGF of the mesenteric artery in the 18% (•, n = 9) and the 9% casein (○, n = 8) groups. ¶p = 0.05, difference in the overall relaxant response assessed by two-way ANOVA between the 18% and the 9% casein groups. C, dose-response curves to VEGF of the uterine (•, n = 10) and mesenteric (○, n = 9) artery in the 18% casein group. §p = 0.004, significant difference in the overall relaxant response assessed by two-way ANOVA between the VEGF response in the uterine and mesenteric arteries in the 18% casein group.

Full size image

In the 18% casein group, the maximal relaxation to VEGF (5 × 10−10 M) in the uterine artery was significantly reduced both by INDO alone (p < 0.05 versus VEGF) and by INDO and l-NAME together (p < 0.05 versus VEGF + INDO;Figs. 5A and 6A). In the 9% casein group, the maximal relaxation to VEGF (1 × 10−9 M) was significantly attenuated by INDO alone (p < 0.05 versus VEGF), but was not further reduced by INDO and l-NAME together (Figs. 5B and 6B).

Figure 5
figure 5

Dose-response curves showing the effects of INDO and combination of INDO and l-NAME on the response to VEGF of the uterine artery in the 18% (A, (•: without inhibitors, n = 10; ▪: with INDO, n = 9; ▴: with INDO/l-NAME, n = 7) and the 9% casein (B, ○: without inhibitors, n = 11; □: with INDO, n = 10; ▵: with INDO/l-NAME, n = 10) groups.

Full size image
Figure 6
figure 6

Histogram showing the effect of inhibitors on the maximal VEGF responses in the 18% (A; *p < 0.05 versus control VEGF, ¶p < 0.05 versus INDO) and the 9% casein groups (B, *p < 0.05 versus control VEGF).

Full size image

DISCUSSION

Maternal protein restriction in the rat has been associated with hypertension, impaired glucose tolerance, and 2reduced longevity in adult offspring (4–8). The effects on the pregnant dam have, however, been little studied. Our study has shown that dietary protein restriction is associated with significant blunting of the vasodilatory response of the uterine artery to VEGF. A recent study has also shown that VEGF induces relaxation in rat thoracic aortic rings (20). We have shown for the first time that VEGF dilates the mesenteric artery. However, the sensitivity to VEGF in the mesenteric artery was less than that in the uterine artery, and this is also a novel observation. At high concentrations there is an attenuation of VEGF-induced vasodilation in the uterine artery, not seen in the mesenteric artery. We cannot explain why this attenuation only occurs in the uterine vasculature.

Uterine blood flow is greatly increased in pregnancy, the change being associated with a marked increase in arterial diameter and reversible hypertrophy (21, 22). Enhancement of NO and PGI2 synthesis as well as remodeling are likely to contribute to the increase in diameter and fall in vascular resistance (23). However, the interaction between a low-protein diet and the physiologic effects in the uterine artery during pregnancy is unknown.

Although earlier reports have documented augmentation of the maximal constriction and sensitivity to PE during rat pregnancy, associated with increased diameter of the uterine artery (24), we did not detect any differences in maximal constriction or sensitivity to PE and KCl between the 18% and the 9% groups. Thus, any differences between the groups do not appear to be related to an increase in vascular smooth muscle cell growth.

Pregnancy is associated with greater serum concentrations of VEGF in humans (15) and VEGF mRNA in the uterine tissue of the rat (17, 18), with an increase in the response to VEGF in the uterine artery (16). It therefore seems reasonable to propose that VEGF plays an important role in dilation of the uterine artery during pregnancy. In addition, the concentration range of VEGF selected included 10−10 M VEGF, which could be readily achieved locally under normal physiologic conditions during pregnancy in uterine arteries, and concentrations of this order have been recorded for 106 vascular smooth muscle cells in culture (25).

A key finding was the attenuation of maximal relaxation to VEGF in the uterine artery in the 9% casein group compared with the 18% casein group. A study of cultured endothelial cells has shown that the vasodilation to VEGF is induced by PGI2 and NO via activation of flt-1 and Flk-1/KDR receptors, which are specific receptors for VEGF (15). Flt-1 and Flk-1/KDR receptors are present on endothelial cells and on smooth muscle cells in porcine uterine artery (26). The pivotal role of the endothelium was demonstrated by Ni et al.(16), who showed complete inhibition of relaxation to VEGF in rat uterine artery vessels denuded of endothelium. Because pregnancy augments the expression of flt-1 and Flk-1/KDR receptors on the endothelial cells in porcine uterine artery (26), the greater vascular response to VEGF in the uterine artery compared with the mesenteric artery observed in this study may be related to an increase in endothelial cell VEGF receptor expression. Among the factors reported to alter VEGF receptor expression, an increase in serum estrogen may play an especially important role in pregnancy (27, 28). In the bovine aorta, VEGF receptor occupation on the endothelium activates phospholipase C, thereby increasing intracellular 1,4,5-inositol trisphosphate and diacylglycerol. The rise in 1,4,5-inositol trisphosphate elevates intracellular calcium, which activates NO synthase, producing NO. The increase in diacylglycerol and intracellular calcium activates protein kinase C and, via subsequent activation of phospholipase A2, increases PGI2(15, 29). In the present study, the maximal relaxation to VEGF in the normal diet group (18% casein) was reduced by INDO, and this reduction was greater than that after superimposed l-NAME. It therefore appears that VEGF-induced PGI2 production predominates in the rat uterine artery. Like Ni et al.(16), we also found an NO-mediated component of VEGF-induced relaxation in the rat uterine artery. However, the magnitude was less in our study, and this contrasting result with that of Ni et al.(16) may be attributed to a difference in the experimental protocol used. We used 100 μM l-NAME, whereas Ni et al.(16) used 1 mM NG-nitro-l-arginine. NG-nitro-l-arginine is known to be a potent nonselective NO synthase inhibitor compared with other arginine analogs (30); therefore, greater NO inhibition would be expected in the study by Ni et al.(16). The differences in NO synthase inhibition of VEGF-induced relaxation between studies may thus be attributed to either a difference in concentration or type of inhibitor used.

The inhibition of VEGF-induced relaxation by l-NAME was significantly less in the 9% group, suggesting that reduced NO production may explain the blunted vasodilation to VEGF in the rats fed low-protein diets. ACh produced relaxation of the preconstricted uterine arterial segments, thereby confirming that the endothelium was intact. Interestingly, no change in the response to ACh was noted on the low-protein diet. In a similar study, Koumentaki et al.(31) reported that dietary protein restriction attenuated ACh-induced relaxation in small mesenteric arteries. This further confirms our findings that nutritional restriction in pregnancy induces vascular abnormalities that vary between uterine and mesenteric vessels. In the present study, ACh-induced relaxations were greatly reduced with l-NAME, confirming that NO is likely to account for ACh dilation in the uterine artery (32). The contrast with VEGF responses may indicate altered flt-1 and Flk-1/KDR receptor density in the low-protein group as opposed to perturbation in the NO signal transduction pathway or in vascular smooth muscle sensitivity to NO. Inasmuch as a low-protein diet may induce serum estrogen deficiency in rat (33, 34), this may arise from lowering of the plasma concentration of this steroid. Further studies are indicated in which the serum estrogen concentration and receptor density are simultaneously determined.

In summary, we have shown that the vasodilatory response to VEGF is reduced in the uterine artery of pregnant rats fed a low-protein diet, possibly owing to attenuation of the NO component of VEGF-induced vasorelaxation. Preliminary studies further support this hypothesis, indicating that VEGF-induced NO release is reduced in uterine arteries from protein-restricted dams compared with control rats (Brawley, Itoh, and Clough, unpublished observation). Because VEGF is likely to contribute to the increment in uterine blood flow during pregnancy, this attenuation of VEGF dilation might be a factor in the reduction in uterine blood flow that occurs on a globally restricted diet (11). In view of the many reports of adulthood cardiovascular dysfunction in the offspring of rats fed a low-protein diet, and because of the central importance of uteroplacental blood flow in fetal well-being, the data presented here suggest a role for altered uterine artery function in this model of fetal programming.