Endocrinology and Metabolism


Serine plays an important role in intermediary metabolism as a source of one carbon pool for nucleotide biosynthesis, as a precursor for glycine and glucose, and as a contributor to cysteine biosynthesis. A unique serine-glycine cycling between the liver and the placenta has been demonstrated in the sheep fetus. We hypothesized that, because of serine's role in growth and development, significant changes in serine metabolism will occur in pregnancy with advancing gestation. The rate of appearance (Ra) of serine and its metabolism were quantified in healthy women longitudinally through pregnancy with a [2-15N13C]serine tracer. The contribution of serine N to urea and the rate of oxidation of serine were measured using the precursor-product relation. Plasma serine concentrations and serine Ra were lower in pregnant (P) women, in both early and late gestation, compared with nonpregnant (NP) women [plasma serine: NP, 113 ± 24.5; P early, 71.9 ± 6.2; P late, 68.5 ± 9.6 μmol/l; serine Ra: NP (n = 7), 152.9 ± 42.8; P early (n = 12), 123.7 ± 21.5; P late (n = 8), 102.8 ± 18.2 μmol · kg−1 · h−1]. Serine contributed ∼6% to urea N and 15–20% to the plasma glycine pool, and oxidation of serine represented ∼8% of Ra. There was no significant difference between P and NP subjects. Glucose infusion, at 3 mg · kg−1 · min−1in P subjects, resulted in a decrease in serine Ra and an increase in oxidation. The decrease in serine turnover in pregnancy may represent a decrease in α-amino nitrogen turnover related to a decreased rate of branched-chain amino acid transamination and caused by pregnancy-related hormones aimed at nitrogen conservation and accretion.

  • stable isotopes
  • urea

pregnancy in humans and animals is associated with profound physiological and metabolic adaptation in the mother to support the needs of the growing conceptus. In this context, although changes in glucose and fat metabolism parallel the increase in energy requirements with advancing gestation, adaptation in protein and nitrogen metabolism is evident early in gestation, well before there is any significant increase in the mass of the conceptus (22, 23). Studies of nitrogen metabolism in humans and animals show that there is gestation-induced hypoaminoacidemia during fasting that is evident early in pregnancy, persists throughout pregnancy (14, 22, 24), and has been attributed to the increase in pregnancy-related hormones (24). During fasting, a greater reduction in glucogenic amino acids—alanine, glutamine, glutamic acid, serine, and threonine—has been observed during pregnancy in human studies (14, 17). In addition, studies in late gestation in humans show a diminished plasma amino acid response to enteral mixed nutrient administration, suggesting an increased splanchnic uptake of amino acids (17, 31, 36).

Data from studies by us in human pregnancy (26, 27) and data from animal studies (2, 32, 33) have shown a decreased rate of urea synthesis during fasting and in response to protein or amino acid load. The rate of urea synthesis has been shown to be positively related to the rate of transamination of branched-chain amino acids in human pregnancy (26). The rate of transamination of leucine is lower during pregnancy, both in early and in late gestation, than in nonpregnant women. These changes have been observed in the presence of unchanged weight-specific rate of whole body leucine carbon kinetics or the whole body rate of protein turnover (26, 41).

Most of the work in human pregnancy thus far has been focused on whole body nitrogen metabolism and on metabolism of essential amino acids, with little attention to metabolism of nonessential amino acids. These amino acids, which are synthesized in vivo, form the large bulk of nitrogen transferred to the fetus (29). In this context, the serine-glycine pair is of particular interest. The transport of serine across the placenta in human pregnancy remains uncertain on the basis of umbilical artery and vein concentration differences, as no net transfer is suggested in some studies (9) and significant transfer is suggested in others (20, 37). In contrast, the data in sheep suggest no significant uptake of serine by the fetus late in gestation (18, 34) in the presence of a very high utero-placental uptake of serine (10, 21). Additionally, a unique interorgan flux of glycine and serine has been documented in the sheep, wherein glycine is taken up by the fetus from the placenta and converted to serine in the fetal liver (7, 8). The importance of serine metabolism during development is underscored by its role as precursor for nucleotide biosynthesis during the rapid growth phase in neonatal rats (38, 39) and as a source for the one carbon pool for nucleotide synthesis and during cell proliferation (40). In addition, biosynthesis of serine has been shown to be metabolically coupled to its utilization for nucleotide precursor formation (16). Finally, clinical studies in humans have shown a strong positive correlation between maternal plasma serine levels and neonatal birth weight (28).

There are no studies reported that examine in human pregnancy the kinetics of serine and its metabolism with advancing gestation. In the present study, we have quantified the kinetics and metabolism of serine longitudinally during human pregnancy by use of stable isotope tracer [2-13C, 15N]serine. Our data show that the plasma serine levels are lower during pregnancy than in nonpregnant women and that a significant decrease in serine turnover is evident in late gestation. Approximately 6% of urea N and 15–20% of the plasma-glycine pool were derived from serine N, with no significant difference between pregnant and nonpregnant women. There was no correlation between serine turnover and infant weight at birth in this small group of healthy women.


Twelve healthy nonobese women who had no associated medical problems were recruited early in pregnancy. Repeat studies were performed on eight of them during the third trimester of pregnancy. Seven healthy nonpregnant women were studied as control subjects. Their clinical characteristics are displayed in Table1. None of the subjects had a family history of diabetes nor were receiving any medications other than routine prenatal vitamins. They did not have any medical or pregnancy-related illness. Written informed consent was obtained from each subject and her spouse (when available) after full explanation of the procedure. The protocol was approved by the Institutional Review Board. All studies were performed in the General Clinical Research Center.

View this table:
Table 1.

Clinical characteristics of subjects

l-[2-13C15N]serine (99 atom %13C, 99 atom % 15N) and sodium [13C]bicarbonate (99 atom % 13C) were obtained from Isotec (Miamisburg, OH). Each batch of the tracer was dissolved in isotonic saline, sterilized by Millipore filtration (0.22 mm), and tested for sterility and pyrogenicity as previously described (25).

For 7 days before the tracer study, all subjects were placed on an isocaloric diet containing 1.5 g of protein/kg body weight and ≥150 g carbohydrate per day. Each subject was interviewed and instructed by a nutritionist. The participants kept a diary record of their food intake for 7 days, and dietary compliance was evaluated from this record. Actual intake was estimated from the self-recorded diaries and calculated using Data Bank Nutritionist 5, V2.2 software (First Databank, San Bruno, CA).

Tracer isotope studies were performed after an overnight fast (∼10 h). The subjects reported to the General Clinical Research Center in the morning. Three 21-gauge indwelling cannulas were placed in the superficial veins on the dorsum of the hand or lower forearm. Arterialized blood samples were obtained from the dorsal vein of the hand. The hand was kept warm by placing it in a box heated to 55°C. The sampling site was kept patent by a constant infusion of isotonic saline at 20 ml/h. The other indwelling cannulas were used for the infusion of isotopic tracer and 10% dextrose.

A weighed amount ofl-[2-13C15N]serine dissolved in isotonic saline was infused at a constant rate of 5 μmol · kg−1 · h−1. A priming dose of the serine tracer, 5 μmol/kg, was given at the start of the constant rate infusion. In addition, a priming dose of 126 μmol of sodium [13C]bicarbonate was given to achieve an early isotopic steady state in the whole body bicarbonate pool. Blood and breath samples were obtained before the start of the tracer infusion and at 30-min intervals throughout the study. After 150 min of a basal study period, intravenous glucose (Dextrose 10%) was infused at a constant rate of 3 mg · kg−1 · min−1. Blood and breath samples were obtained at 30-min intervals for the next 3 h. The rate of tracer infusion was confirmed gravimetrically at the end of the study with the same tubing, cannula, and infusion pump. A sample of infusate was obtained for quantitative analysis and also sent for tests of sterility.

Respiratory calorimetry measurements were performed intermittently throughout the study with an open-circuit system, as described previously (1). The rates of oxygen consumption (V˙o 2) and CO2 production (V˙co 2) were measured at hourly intervals by placing a ventilated hood over the subject's head. Recordings were obtained for a period of ≥15 min. The analyzer was calibrated using a gravimetrically measured standard mixture of oxygen and carbon dioxide. The accuracy and precision of the respiratory calorimetry system were checked by measuring the respiratory quotient of absolute alcohol and were within 2% of expected value.

The kinetic data obtained during the basal study period (0–150 min) are designated “Fasting 12 h” (Table2). The effect of glucose infusion at 3 mg · kg−1 · min−1was examined in six women in early pregnancy, in seven women in late pregnancy, and in two nonpregnant women. These data are presented under “glucose” in the tables. To be certain that the observed changes in serine metabolism during glucose infusion were not the result of prolonged fast, six women in early pregnancy, one woman in late pregnancy, and five nonpregnant women received only tracer infusion and were not infused with glucose. These data are designated “Fasting 15 h.”

View this table:
Table 2.

Serine kinetics during pregnancy

Analytical procedures.

Plasma glucose was measured by the glucose oxidase method on a commercial analyzer (Beckman Instruments, Fullerton, CA). Plasma amino acids were quantified by high-performance liquid chromatography with a fluorescent detector. Precolumn derivatization of plasma amino acid was performed, and an OPDA derivative was used for analysis (45).

GC-MS analysis.

Plasma amino acids were isolated from a deproteinized sample by ion exchange chromatography. An N-acetyl, N-propyl ester derivative of serine and glycine was prepared according to the method of Adam (see Ref. 13) with certain modifications. A Hewlett-Packard model 5985 GC-MS system was utilized. Methane chemical ionization was used, and mass-to-charge ratios (m/z) 232 and 234, representing unlabeled and dilabeled serine, were monitored using the selected ion monitoring software. The gas chromatograph column and oven condition were similar to those for analysis of other amino acids and have been described previously (13). The contribution of serine to the glycine pool was estimated by measuring the tracer enrichment of glycine (m2) in the same plasma sample:m/z 160 and 162 were monitored representing unlabeled and [13C,15N]-labeled glycine. Standard solutions of known enrichments were run along with the unknowns to correct for analytical variations. 13C enrichment of the CO2 in expired air was measured after separation of the CO2 by cryogenic distillation in vacuum, as previously described (1).


The rate of appearance (Ra) of serine in the plasma was calculated using the tracer dilution equation during steady state (44). Ra = I [(Ei/Ep) − 1], where I is the rate of infusion of tracer (μmol · kg−1 · h−1), and Ei and Ep represent the enrichments of infused serine and plasma serine at steady state, respectively. Isotopic steady state was determined by visual inspection of the data. Enrichment data between 120 and 150 min, and between 270 and 300 min, were used to calculate the rates of turnover. The coefficients of variation for the enrichment data for serine and glycine in individual subjects were between 3 and 5%, and the slope was not different from zero.

The fraction (F) of serine turnover oxidized was calculated as follows: F = V˙co 2 × Δ[13C]O2/I, whereV˙co 2 is in μmol · kg−1 · min−1, Δ[13C]O2 is the 13C enrichment of CO2 during steady state, and I is the rate of infusion of [15N,13C]serine in μmol · kg−1 · min−1. The rate of oxidation of serine was calculated by multiplying F with serine turnover (Ra).

The fractional contributions of serine to the glycine pool and to the urea N pool were calculated by the precursor-product relationship.


All data are reported as means ± SD. The data were analyzed using the Statistix statistical package. The Mann-Whitney U-test and Wilcoxon's signed rank test were used for unpaired and paired interval data. Statistical significance was defined a priori as aP value <0.05 (2-tail).


All study subjects were healthy and did not have any intercurrent illness or were receiving any medications. The pregnant subjects were receiving supplemental multiple vitamins and iron. They were also older than the nonpregnant women; however, there was no difference in their prepregnancy body weight or body mass index (Table 1). Although the total calorie intake of pregnant subjects was higher than that of nonpregnant subjects, their weight-specific calorie intake and the fractional intake of carbohydrate and protein were similar. The infants were born at term gestation (39.25 ± 0.97 wk) and had appropriate weight for the gestational age (birth weight 3,303.0 ± 548.1 g).

Serine kinetics during pregnancy.

Plasma serine concentration was significantly lower in pregnant compared with nonpregnant subjects (P < 0.0001, Table2). The (m2) enrichment of serine during the basal study is displayed in Fig. 1. An isotopic steady state was rapidly achieved within 90 min of tracer infusion. The isotopic tracer enrichments between 120 and 150 min and between 270 and 300 min were averaged to calculate the tracer dilution. The individual data during the fasting (12-h) study are displayed in Fig.2. The Ra of serine in the plasma during fasting (12 h) measured by tracer isotope dilution was not significantly lower in the early pregnant compared with the nonpregnant state. However, with advancing gestation, there was a significant decrease in serine Ra during fasting (P < 0.03). With the prolongation of fast from 12 to 15 h, there was no significant change in serine Ra in pregnant subjects examined early in gestation. In five nonpregnant subjects, extension of fast from 12 to 15 h also did not impact the Ra of serine. The effect of intravenous glucose infusion on serine Ra is also displayed in Table 2. Glucose infusion at 3 mg · kg−1 · min−1(16.6 μmol · kg−1 · min−1) resulted in a significant increase in plasma glucose in all subjects (early pregnancy from 74.8 ± 5.8 to 104.2 ± 20.2 mg/dl,P < 0.006; late pregnancy from 76.6 ± 8.4 to 104.8 ± 13.8 mg/dl, P < 0.001). Parenteral infusion of glucose resulted in a significant decrease in the Ra of serine by a similar magnitude in both early and late pregnancy (Table 2). In two nonpregnant subjects studied, infusion of glucose also resulted in a decrease in serine Ra (fasting 121 and 167 μmol · kg−1 · h−1; glucose infusion 119 and 129 μmol · kg−1 · h−1, respectively).

Fig. 1.

Enrichment of serine (m2) during fasting in the plasma of nonpregnant women (●, n = 5), pregnant women early in gestation (□, n = 12), and pregnant women during the third trimester (▵,n = 8). [2-13C15N]serine was administered as prime-constant rate infusion after an overnight fast (∼10 h). Plasma samples were obtained at times shown.

Fig. 2.

Steady-state serine (m2) enrichment in plasma serine during fasting (12 h). Individual data and means ± SD are displayed. [2-13C15N]serine was infused as a prime-constant rate infusion, as described in methods.

Plasma glycine enrichment.

Because serine is rapidly converted into glycine, we measured the m2 enrichment of plasma glycine to examine the effects of fasting, infusion of glucose, and gestation on the contribution of serine to glycine. The m2 enrichment of glycine increased significantly with the duration of fast in both nonpregnant and pregnant subjects (Table 3). During glucose infusion, the m2 enrichment of both serine and glycine increased in pregnant subjects and was of a similar magnitude as that observed during prolongation of fasting. The fraction of glycine derived from serine, calculated from the m2enrichment of glycine and serine after 12 h of fasting, ranged from 0.15 to 0.20. The fraction of glycine derived from serine was highest in nonpregnant subjects (0.20 ± 0.03, n = 5) and lowest in subjects in late gestation (early 0.17 ± 0.01,n = 12; late 0.15 ± 0.02, n = 8;P < 0.01). Prolongation of fast from 12 to 15 h, or glucose infusion, caused a small but significant increase (P < 0.01) (from 0.17 to 0.19) in the fraction of glycine derived from serine.

View this table:
Table 3.

Tracer enrichment in plasma serine and glycine during infusion of [13C15N]serine

Oxidation of serine.

The contribution of serine to urea and respiratory carbon dioxide was estimated from the appearance of 15N in urea (m1) and of 13C in CO2. Serine N represented ∼6% of urea synthesized, with a wide range from 2 to 15% (data not presented). There was no significant difference in pregnant subjects between early and late gestation nor between nonpregnant and pregnant subjects. The wide range in the estimate reflected, in part, the difficulty in the measurement of very low m1 enrichment in urea.

The data on respiratory calorimetry and the contribution of serine C to expired carbon dioxide are displayed in Tables4 and 5. As shown, there was a small but statistically significant increase (P < 0.02) in the rates ofV˙o 2 andV˙co 2 in pregnant subjects (Table 4). As previously reported (1, 13), the respiratory exchange ratio was also higher in late pregnancy, indicating a higher contribution of carbohydrate to energy consumption.

View this table:
Table 4.

Respiratory calorimetry data

View this table:
Table 5.

Contribution of serine C to expired carbon dioxide

During fasting (12 h), 8–10% (10–15 μmol · kg−1 · h−1) of serine appearing in the circulation was oxidized to CO2in pregnant and nonpregnant subjects (Table 5). The rate of oxidation of serine increased as the fast was prolonged from 12 to 15 h in nonpregnant women and in pregnant women in early gestation. During glucose infusion, the fraction of serine C recovered in expired CO2 in pregnant women increased significantly (P < 0.01), suggesting a higher rate of oxidation of serine. The magnitude of this increase was less in late pregnancy compared with early pregnancy.

Plasma amino acids.

The concentration of several dispensable amino acids in the plasma was significantly lower in pregnant women in early gestation than in nonpregnant women, and it remained low in late gestation (Table6). Glucose infusion caused a small decrease in some amino acids, in particular leucine, isoleucine, valine, and glutamine.

View this table:
Table 6.

Plasma amino acid concentrations


There was no statistically significant correlation between serine Ra and the birth weight of infants in early or late pregnancy.


The data from the present study show that the plasma serine concentration and the Ra of serine were lower in healthy pregnant women during late pregnancy compared with nonpregnant women. Additionally, intravenous infusion of glucose with resultant hyperglycemia was associated with a decrease in serine Ra. Studies of serine oxidation showed that 8–10% of serine flux was oxidized to CO2 after a brief fast and that ∼6% of urea N was derived from serine N (range 2–15%). The fraction of the plasma glycine pool derived from serine approximated 15–20%. Interestingly, pregnancy did not appear to have any significant impact on the contribution of serine to glycine, urea, or expired CO2.

The present data are the first to quantify serine kinetics in human pregnancy. Few studies have quantified serine kinetics in humans (12, 19). Using a [5,5,5-2H3]serine tracer, Cuskelly et al. (12) quantified serine Ra in five healthy young subjects. Their estimate of serine Ra (188 ± 44 μmol · kg−1 · h−1) was of the same magnitude as that determined by us in nonpregnant women by use of a [2-15N13C]serine tracer (Table2).

The kidney appears to be the primary site of serine production, both in humans and in the rat. Arteriovenous concentration differences of serine across tissues in vivo show that serine is released into the circulation by the kidney and that there is a net uptake of serine by splanchnic organs and by the skeletal muscle (6, 15, 35, 43, 46,47). In addition, fasting or intravenous infusion of amino acids did not appear to impact the release of serine by the kidney nor its uptake by other tissues (6, 46, 47). Although an enteral protein load resulted in an increased uptake of serine in the splanchnic compartment, it did not result in any change in uptake of serine in the periphery (47). Similar data have been reported for the rat kidney in both in vitro and in vivo preparations (4, 5, 30). No impact of exogenous serine administration, either enteral or parenteral, on serine production by the kidney has been observed (4, 5). Thus the rate of production of serine by the kidney is not influenced by exogenous administration of serine or by other dietary manipulations.

Additionally, there does not appear to be any discernible impact of other substrates and hormones, such as insulin and glucagon, on net release of serine by the kidney in rats or humans. In this context, the present data are of interest in that they show a lower rate of serine turnover in pregnancy. It should be underscored that tracer-determined serine Ra includes multiple miscible pools of serine within tissues and organs and will be higher than the net release measured by arteriovenous differences. The pregnancy-related effect is likely the consequence of the increase in pregnancy-related hormones directly or mediated through other intermediates. Whether the lower rate of serine turnover is part of the overall decrease in α-amino nitrogen turnover remains to be examined.

The major source of serine C has been shown to be glucose via the 3-glycerophosphate pathway (3). Plasma serine levels have been observed to be high in hyperglycemia associated with diabetes mellitus in pregnancy (28). For these reasons, we examined the effect of glucose infusion on serine kinetics. Hyperglycemia induced by intravenous glucose infusion caused a decrease in serine Ra in pregnant subjects. Although we did not measure plasma insulin levels, we anticipate them to be high at these glucose levels. These data suggest that glucose or glucose plus insulin may have an inhibitory effect on serine Ra and that lack of insulin, as in diabetes, particularly in pregnancy, may be associated with higher rates of turnover of serine. Direct measurements of serine kinetics in the absence of insulin will be required to confirm this hypothesis. The lower Ra of serine during glucose infusion could be the consequence of a lower rate of whole body protein turnover or protein breakdown. In addition, as suggested by Fell and Snell (16), the synthesis of serine may be controlled by the demand for serine rather than the supply of the precursor.

Although no net release of serine by the liver has been demonstrated, serine is metabolized in the liver in both the adult and the fetus (8, 42, 48, 49). Tracer isotope dilution methods, as used here, will include the metabolism of serine in both liver and kidney. Therefore, although the renal production of serine has been shown not to be impacted by dietary and hormonal manipulation, the hepatic metabolism of serine may be responsive to dietary and hormonal perturbations and may also explain, in part, the effect of glucose as seen in the present study.

Serine can be oxidized either via pyruvate or by conversion to glycine and oxidation by the glycine cleavage system. In the present study, 8–10% serine flux was oxidized (Table 5). Pregnancy did not have any effect on the contribution of serine C to expired CO2. Infusion of glucose at 3 mg · kg−1 · min−1resulted in a significant increase in the oxidation of serine. Whether the observed increase is a true increase in serine oxidation or is a consequence of higher carbon flux through the tricarboxylic acid intermediates as a result of glucose infusion cannot be separated from these data. The increase in oxidation is unlikely to be due to higher 13C enrichment of infused glucose (25). Exogenous glucose infusion at these low rates has been shown not to cause significant changes in the 13C enrichment of expired CO2.

Serine and glycine have been shown to have a unique metabolism in the fetus and placenta. Studies in human pregnancy have shown a strong correlation between maternal plasma serine levels and neonatal birth weight (r = 0.762, P < 0.001) (28). Such a high correlation has not been documented for any other amino acid. Although no significant transfer of maternal serine to the fetus has been demonstrated in the sheep, the data in humans have been variable (9, 20, 37). Parenteral infusion of an amino acid mixture containing serine did result in an increase in the maternal and umbilical venous concentration of serine in healthy women at term gestation, suggesting placental transfer of serine (37). Because fetal concentrations of serine are markedly higher in sheep than in humans, the fetal hepatic and placental interorgan cycling of serine and glycine in sheep may be unique to that particular species (7, 9). Other studies have also demonstrated unique interspecies differences of serine metabolism (48, 49).

The importance of serine metabolism is underscored by its role as precursor for nucleotide biosynthesis during the rapid phase of growth in rat, as a source for the one-carbon pool for nucleotide synthesis and during cell proliferation, and as a gluconeogenic precursor (11). For these reasons we had hypothesized that pregnancy will result in an increase in the turnover of serine. Therefore, it was surprising that there was a decrease in the rate of turnover of serine in pregnant women with advancing gestation. The decrease may be related to the downregulation of the transaminations to conserve nitrogen, as was observed by us previously in human pregnancy (26).

Brosnan and Hall (5) have speculated, on the basis of indirect evidence, that the source of serine C in rat kidney is probably the small glycine-rich peptides, such as glycyltyrosine, glycylproline, glycylhydroxylproline, and the like, derived from collagen, which are taken up by the kidney. Whether the lower turnover of serine in pregnancy is related to a decrease in precursor supply cannot be ascertained from the present data.

In summary, in association with a decrease in plasma serine concentration, there is a decrease in the Ra of serine in human pregnancy. Exogenous glucose infusion resulted in a decrease in serine Ra and an increase in oxidation. We speculate that the lower serine kinetics in pregnancy may be part of a general downregulation of α-amino nitrogen turnover, probably mediated via lower rates of branched-chain amino acid transamination, aimed at nitrogen conservation and nitrogen accretion.


We thank the nursing staff of the General Clinical Research Center, University Hospitals of Cleveland, for their expert assistance. The secretarial help of Joyce Nolan is gratefully appreciated.


  • These studies were supported by National Institutes of Health Grants HD-11089 and RR-00080.

  • Address for reprint requests and other correspondence: S. Kalhan, M. D. Schwartz Center, Bell Greve Bldg., Rm. G-735, MetroHealth Medical Center, 2500 MetroHealth Dr., Cleveland, OH 44109-1998 (E-mail: sck{at}po.cwru.edu).

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • First published December 17, 2002;10.1152/ajpendo.00167.2002


View Abstract