Cerebrospinal Fluid Levels of Leptin, Proopiomelanocortin, and Agouti-Related Protein in Human Pregnancy: Evidence for Leptin Resistance

Abstract

Context:

Leptin suppresses appetite by modulating the expression of hypothalamic neuropeptides including proopiomelanocortin (POMC) and agouti-related peptide (AgRP). Yet during pregnancy, caloric consumption increases despite elevated plasma leptin levels.

Design and Participants:

To investigate this paradox, we measured leptin and soluble leptin receptor in plasma and leptin, POMC, and AgRP in cerebrospinal fluid (CSF) from 21 fasting pregnant women before delivery by cesarean section at a university hospital and from 14 fasting nonpregnant women.

Results:

Prepregnancy body mass index was 24.6 ± 1.1 (se) vs. 31.3 ± 1.3 at term vs. 26.5 ± 1.6 kg/m² in controls. Plasma leptin (32.9 ± 4.6 vs. 16.7 ± 3.0 ng/ml) and soluble leptin receptor (30.9 ± 2.3 vs. 22.1 ± 1.4 ng/ml) levels were significantly higher in pregnant women. However, mean CSF leptin did not differ between the two groups (283 ± 34 vs. 311 ± 32 pg/ml), consistent with a relative decrease in leptin transport into CSF during pregnancy. Accordingly, the CSF/plasma leptin percentage was 1.0 ± 0.01% in pregnant subjects vs. 2.1 ± 0.2% in controls (P < 0.0001). Mean CSF AgRP was significantly higher in pregnant subjects (32.3 ± 2.7 vs. 23.5 ± 2.5 pg/ml; P = 0.03). Mean CSF POMC was not significantly different in pregnant subjects (200 ± 13.6 vs. 229 ± 17.3 fmol/ml; P = 0.190). However, the mean AgRP/POMC ratio was significantly higher among pregnant women (P = 0.003), consistent with an overall decrease in melanocortin tone favoring increased food intake during pregnancy.

Conclusions:

These data demonstrate that despite peripheral hyperleptinemia, positive energy balance is achieved during pregnancy by a relative decrease in central leptin concentrations and resistance to leptin's effects on target neuropeptides that regulate energy balance.

Pregnancy is characterized by elevated plasma leptin levels (1–7). Although leptin is most well known as an adipocyte-derived hormone that suppresses appetite, the degree of hyperleptinemia in pregnant animals and humans is out of proportion to adiposity. In pregnant rodents, the gestational rise in leptin levels derives from oversecretion by adipocytes beginning at midgestation (8) as well as from the placental production of a soluble leptin receptor (OB-Re) (2, 9). In humans, leptin appears to be synthesized by the placenta because Masuzaki et al. (10) demonstrated ob gene expression in trophoblasts, syncytiotrophoblasts, and amnion cells both in vivo and in vitro. Consistent with placental production, leptin levels peak during the second trimester of human pregnancy and fall precipitously postpartum (3). Leptin normally inhibits food intake by binding to the long form of the receptor (OB-Rb) in the hypothalamus and modulating the activity of key neurons, including proopiomelanocortin (POMC) and agouti-related peptide (AgRP) neurons that regulate appetite and energy expenditure. Leptin stimulates POMC and the POMC-derived MSH peptides that decrease caloric consumption, while inhibiting the release of the orexigenic peptide AgRP that typically stimulates food intake by antagonizing the effects of α-MSH at brain melanocortin receptors (11). This hypothalamic bioactivity is predicated on the successful transport of leptin into the central nervous system. Given that leptin is too large to cross the blood-brain barrier (BBB) via simple diffusion, it is thought to enter the brain via a saturable transport system that may be mediated by the short form of the leptin receptor (12, 13). Leptin may also gain access to brain areas outside the BBB. Paradoxically, despite the hyperleptinemia of pregnancy, appetite and caloric consumption increase during gestation, reflecting the existence of a leptin-resistant state at the level of the hypothalamus that is likely designed to ensure that the metabolic needs of the developing fetus are met (14–16).

Although the central mechanisms underlying this leptin-resistant state have not been studied in human pregnancy, in rodents several authors have suggested that placentally derived OB-Re not only increases circulating leptin concentrations but also, by binding leptin, reduces its availability to hypothalamic target tissues and prevents the appetite suppression that it typically mediates (2, 17, 18). Given that OB-Re concentrations have also been shown to increase during pregnancy in women, in parallel with the rise in leptin levels, OB-Re could play a similar role in the leptin resistance observed during human gestation (7).

Studies in pregnant rats suggest that leptin resistance may also be mediated by a reduction in leptin transport through the BBB and alterations in hypothalamic leptin signaling that favor an increase in the expression of orexigenic neuropeptides AgRP/neuropeptide Y (NPY) and a corresponding decrease in the expression of anorexigenic POMC-derived peptides (19). Until now, evaluating the impact of leptin on the melanocortin pathway in humans has been limited by the dearth of reliable peripheral markers of hypothalamic activity. However, there are animal data to suggest that POMC and AgRP levels in cerebrospinal fluid (CSF) may be reliable surrogates of melanocortin activity in brain. In humans, POMC prohormone is the predominant POMC peptide in CSF, with levels up to 50-fold higher than its peptide products (20). CSF levels of POMC correlate with hypothalamic gene expression in fed and fasted rats, and there is evidence that estrogen-induced changes in CSF AgRP levels in ovariectomized rhesus monkeys may correlate with reported changes in hypothalamic gene expression (21, 22). We have found that POMC and AgRP can be reliably measured in human CSF with sensitive and specific assays. We subsequently sought to investigate, for the first time in human pregnancy, CSF leptin as a surrogate for brain leptin and its relationship to measurements of target melanocortin neuropeptides and to examine differences in OB-Re concentrations to gain insight into the adaptations in leptin physiology that characterize human pregnancy.

Subjects and Methods

Subjects

This prospective study was approved by the Columbia University Institutional Review Board, and written informed consent was obtained from all subjects before participation. Subjects were 21 healthy women with uncomplicated pregnancies at term (age, 32.3 ± 1.3 yr, mean ± se) and 14 healthy nonpregnant reproductive-age women (age, 30.0 ± 1.7 yr; P = 0.30) with regular 25- to 35-d menstrual cycles. Subjects were screened by medical history before participation. All subjects were nonsmokers in good general health who were on no medications other than multivitamins. Subjects were free from past or present major psychiatric disorders, alcoholism, neurological disease, and renal and hepatic disease. Subjects were excluded if they had used weight loss supplements or dieted during the 6 months before the study. Subject characteristics are presented in Table 1.

Table 1.

Subject characteristics

	Nonpregnant (n = 14)	Pregnant (n = 21)	P value
Age (yr)	30.0 ± 1.7	32.3 ± 1.3	0.30
Prepregnancy BMI (kg/m2)	26.5 ± 1.6	24.6 ± 1.1	0.32
BMI (kg/m2)	26.5 ± 1.6	31.3 ± 1.3	0.02
Corrected BMI (kg/m2)	26.5 ± 1.6	30.1 ± 1.2	0.07

Values are expressed as mean ± se.

Protocol

All pregnant subjects were undergoing elective cesarean section the morning after an overnight fast. Two milliliters of CSF were collected at the time of spinal anesthesia/analgesia or combined spinal epidural placement. In patients receiving spinal anesthesia, a 20-gauge introducer needle was inserted into the interspinous ligament before the insertion of a 25-gauge Whitacre needle through the introducer into the subarachnoid space. Two milliliters of CSF were then removed using a 3-ml syringe before injection of anesthetic/analgesic. In subjects receiving a combined spinal-epidural anesthetic, a 17-gauge Tuohy needle was advanced into the epidural space, followed by passage of a 27-gauge Whitacre needle through the Tuohy needle into the CSF. Two milliliters of CSF were then removed using a 3-ml syringe before injection of anesthesia/analgesia and completion of epidural catheter placement. Nonpregnant control subjects were studied in the outpatient clinical research center in the early follicular phase of the menstrual cycle, the morning after an overnight fast. A lumbar puncture was performed using a 20-gauge introducer needle followed by a 25-gauge Whitacre needle, and 2 ml of CSF were collected. In all groups, the first 0.5 ml of CSF obtained was discarded before collection of the sample for study. All CSF samples were immediately transferred to polyethylene tubes and placed on ice. Samples were then centrifuged, aliquoted, and stored at −80 C until assays were performed. A single plasma sample was obtained immediately after CSF collection in all subjects and was similarly centrifuged, aliquoted, and stored at −80 C until assays were performed.

Assays

Plasma leptin levels were measured by RIA (Millipore Corporation, Billerica, MA). CSF leptin levels were measured with a highly sensitive ELISA (R&D Systems, Minneapolis, MN) with an assay sensitivity of 10 pg/ml. The Bland-Altman analysis indicated that the two assays provided similar measures. The bias was −0.62; the 95% limits of agreement between the two assays ranged from −4.0 to 2.7, and there were no measures that fell outside the limits of agreement. Additionally, the correlation coefficient between plasma leptin levels measured by RIA and ELISA was r = 0.98 (P < 0.0001). OB-Re was assayed by ELISA (R&D Systems), with an assay sensitivity of 0.3 ng/ml. CSF POMC was assayed using an in-house two-site ELISA with antibodies provided by Dr. Anne White, with the capture monoclonal antibody directed against ACTH (10–18) and the detection antibody directed against γ-MSH. There is 100% cross-reactivity with 22K pro-ACTH. There is no cross-reactivity with ACTH, α-MSH, γ-MSH or β-EP. Affinity purified human 31K POMC was used for standards (20). Assay sensitivity is 8 fmol/ml. CSF AgRP was assayed by ELISA (R&D Systems) using recombinant full-length human AgRP standard. There is 17% cross-reactivity with AgRP (83–132). Assay sensitivity is 7 pg/ml. An AgRP RIA was also used to characterize the AgRP immunoactivity in CSF after HPLC. The RIA was performed as previously described (23) using an antibody (kindly provided by Dr. Gregory Barsh, Stanford University, Palo Alto, CA), and synthetic human AgRP (83–132) (Phoenix Pharmaceuticals Inc., Burlingame, CA) was used as standard and iodinated for use as tracer. This assay is more sensitive for AgRP (83–132) and has 20% cross-reactivity with full-length AgRP.

Characterization of the AgRP and POMC immunoactivity in CSF

AgRP immunoactivity was characterized by HPLC in two CSF pools from pregnant (7.5 ml) and nonpregnant (10 ml) subjects as previously reported (18). CSF pools were concentrated using Amicon Ultra Centrifugal Filters (Millipore Corporation) with a 3000 molecular weight cutoff. The concentrated CSF (0.5 ml) was then mixed with 0.1% trifluoroacetic acid containing 20% acetonitrile and subjected to reverse phase C18 HPLC at room temperature. Samples were eluted with a gradient of 80% acetonitrile containing 0.1% trifluoroacetic acid. Fractions were collected, evaporated in a Speed Vac Concentrator, and dissolved in buffer for AgRP RIA and AgRP ELISA. The columns were calibrated with 5 ng of synthetic human AgRP (83–132) (Phoenix Pharmaceuticals Inc.) and with 5 ng human full-length AgRP (R&D Systems). POMC immunoactivity was characterized by gel filtration in CSF from a normal subject. CSF (3 ml) was concentrated as described above and chromatographed on a Sephadex G-75 column, and fractions were assayed for POMC by ELISA. The column was calibrated with affinity purified 31K human POMC.

Statistical analysis

Statistically significant differences in the parameters studied were evaluated between pregnant and nonpregnant subjects using unpaired t tests. Data are expressed as mean ± se. The significance level was set at P < 0.05. The associations between plasma leptin levels and OB-Re concentrations in pregnant and nonpregnant subjects were examined by simple linear regression analysis with Pearson's correlation. Analyses were performed with Statview (Abacus Concepts Inc., Berkeley, CA).

Results

Characteristics of subjects

Characteristics of the subjects are presented in Table 1. There was no significant difference in age between nonpregnant and pregnant subjects (P = 0.30). As expected, when the body mass index (BMI) of pregnant subjects was compared with that of healthy reproductive-age nonpregnant controls, a significant difference was noted (P = 0.02); however, it is important to acknowledge that maternal BMI is not likely to be an adequate reflection of adiposity in pregnant subjects, given the weight of the fetus, placenta, uterus, amniotic fluid, and edematous tissues. To provide a more accurate assessment of BMI in the pregnant subjects, we calculated a corrected BMI by subtracting the weight of the fetus and placenta from the maternal term weight. The corrected BMI in pregnant subjects did not differ significantly from the BMI in nonpregnant subjects. Nonetheless, it should be acknowledged that even the corrected BMI likely overestimated the true weight of pregnant subjects at term. Interestingly, there were no differences between the prepregnancy BMI of pregnant subjects and the BMI of nonpregnant subjects at the time of the study (P = 0.32).

Plasma leptin and CSF leptin and plasma OB-Re levels

Plasma leptin levels were almost 2-fold higher in pregnant subjects than in their nonpregnant counterparts (32.9 ± 4.6 vs. 16.7 ± 3.0 ng/ml; P = 0.01), but despite the hyperleptinemia of pregnancy, CSF leptin concentrations did not differ between the two groups (283 ± 34 vs. 311 ± 32 pg/ml; P = 0.58) (Fig. 1A). Consequently, the CSF/plasma leptin percentage was significantly lower in pregnant subjects (1.0 vs. 2.1%; P < 0.0001) (Fig. 1B). As expected, there was a strong positive correlation between plasma leptin and BMI in nonpregnant subjects (r = 0.97; P = 0.0001). Although this positive correlation was preserved in pregnant subjects, the strength of the correlation was not as great (r = 0.74; P = 0.0002). The concentration of OB-Re was significantly higher in pregnant subjects (30.9 ± 2.3 vs. 22.1 ± 1.4 ng/ml; P = 0.007) (Fig. 2A). Simple linear regression analysis of the OB-Re concentration as a function of plasma leptin levels illustrates an inverse relationship between plasma leptin and OB-Re in nonpregnant subjects (β = −0.26; r = −0.57; P = 0.03) that persists in pregnancy (β = −0.31; r = −0.62; P = 0.003) (Fig. 2B).

Fig. 1.

Plasma leptin concentration (A), CSF leptin concentration (B), and CSF/Plasma leptin percentage (C) in pregnant and nonpregnant subjects at term. *, P < 0.0001.

Fig. 2.

A, Soluble leptin receptor (OB-Re) concentrations in nonpregnant and pregnant subjects. B, Simple linear regression of the OB-Re concentration as a function of the plasma leptin concentration in nonpregnant and pregnant subjects.

CSF AgRP and POMC levels

AgRP was measured in CSF and was found to be significantly higher in pregnant subjects (32.3 ± 2.7 vs. 23.5 ± 2.5 pg/ml; P = 0.03) (see Fig. 4A). Concomitantly, there was a nonsignificant trend toward a lower CSF POMC concentration among pregnant subjects (200 ± 13.6 vs. 229 ± 17.3 fmol/ml; P = 0.19). However, overall the CSF AgRP/POMC ratio in the pregnant subjects was 180% of their nonpregnant counterparts (0.18 ± 0.017 vs. 0.10 ± 0.009; P = 0.003) (Fig. 3, B and C).

Fig. 4.

A, HPLC of a CSF pool from nonpregnant subjects with fractions assayed for AgRP by ELISA. Two peaks of immunoactivity were seen corresponding to AgRP (83–132) and full-length AgRP standards (arrows). The solid line represents the actual AgRP ELISA values. The dashed line represents a correction for values eluting in the position of AgRP (83–132) for the 17% crossreactivity of AgRP (83–132) in the AgRP ELISA. B, G-75 Sephadex chromatography of CSF from a nonpregnant subject with fractions assayed for POMC by ELISA. Arrows indicate the void volume and the elution positions of POMC and ACTH standards.

Fig. 3.

CSF AgRP (A), CSF POMC (B), and the CSF AgRP/POMC Ratio (C) in nonpregnant and pregnant subjects. *, P < 0.003.

Characterization of the AgRP and POMC immunoactivity in CSF

AgRP immunoactivity in CSF pools from nonpregnant and pregnant subjects was characterized by HPLC and fractions were assayed by AgRP ELISA and RIA. There were two peaks of AgRP immunoactivity corresponding to the C-terminal and full length peptides noted with both assays. The HPLC profile using the ELISA is shown in Fig. 4A as a solid line. A dashed line is used to depict the values eluting in the position of C-terminal AgRP (83–132) that have been corrected for the 17% cross-reactivity of AgRP (83–132) in the AgRP ELISA. The relative amounts of C-terminal and full length AgRP were the same for pregnant and nonpregnant subjects. The ratio of full length to C-terminal AgRP was 1.2 for the nonpregnant subjects and 1.1 for the pregnant subjects. POMC immunoactivity in CSF from a normal nonpregnant subject was characterized by gel filtration and the fractions were assayed by POMC ELISA. The majority of the immunoactivity eluted in the same position as affinity purified 31K human POMC (Fig. 4B).

Discussion

The energy demands of human pregnancy are significant. To achieve the requisite positive energy balance, mammals markedly increase their caloric intake, maintaining a state of hyperphagia that persists throughout pregnancy. This metabolic adaptation is central to fetal development and occurs in the face of increased leptin levels, which have been repeatedly observed during gestation in both animals and humans. The simultaneous presence of hyperphagia and hyperleptinemia suggests that pregnancy is a state of leptin resistance. This is the first study to describe the potential mechanisms underlying leptin resistance in human pregnancy as reflected by measurements of plasma and CSF leptin and target melanocortin neuropeptides including POMC and AgRP. There is one previous report measuring leptin in CSF during pregnancy, but there was no comparison to nonpregnant subjects (24). The measurement of leptin levels in CSF and the ratio of CSF to plasma leptin have previously been shown to provide a useful index of leptin transport through the BBB (25, 26). The results of this study, using a very sensitive assay that can reliably detect the low levels of leptin present in CSF, demonstrate a significantly lower CSF to plasma leptin ratio in pregnant subjects consistent with decreased leptin transport into the brain. This apparent decline in CSF leptin occurs in parallel with an observed rise in the concentration of circulating OB-Re, supporting the hypothesis that this binding protein prevents leptin transport into CSF. Our results further demonstrate an apparent resistance to the central metabolic effects of leptin on the melanocortin system during pregnancy, as evinced by an increase in CSF AgRP and a trend toward a decrease in CSF POMC at term, despite the presence of CSF leptin levels that are comparable to those observed in nonpregnant individuals. The most significant change was the increase in the CSF AgRP/POMC ratio (P = 0.003) that likely reflects a more integrated measurement of melanocortin peptide activity at the MC4-R. The interpretation of these CSF neuropeptide measurements is predicated on the assumption that they reflect brain melanocortin activity. This assumption is supported by studies in the rat showing that CSF POMC correlates with hypothalamic POMC in fed, fasted, and leptin receptor-deficient rats (21). In humans, the POMC prohormone is the predominant POMC peptide in CSF, with levels up to 50-fold higher than its peptide products (20). There is no correlation of CSF POMC peptide measurements with peripheral blood levels. Furthermore, POMC peptides persist in CSF after hypophysectomy, consistent with brain origin (27). In this study, we have confirmed by gel filtration that the majority of POMC immunoactivity detected in CSF elutes in the position of the 31K POMC standard as has been reported previously (20).

There is little information about CSF AgRP measurements. We have previously measured AgRP levels in lumbar CSF in ovariectomized rhesus monkeys and shown changes after estradiol replacement that parallel what has been reported in brain in ovariectomized rodents (28). When characterized by HPLC, we show that the AgRP immunoactivity detected in human CSF consists of both full-length AgRP and AgRP (83–132). This is similar to the findings in the monkey. Of note, the relative amounts of full-length AgRP and AgRP (83–132) were similar in CSF pools from pregnant and nonpregnant subjects.

The relative decrease in CSF/plasma leptin that we observe during human pregnancy corroborates existing data from rodent models. After midgestation, hyperphagic hyperleptinemic rats demonstrate a similar reduction in CSF/serum leptin levels relative to nonpregnant controls. Impaired transport of leptin across the BBB during pregnancy in these animals is further substantiated by the presence of lower levels of leptin in the CSF of pregnant animals after peripheral leptin administration, compared with nonpregnant controls (19). The relative impairment of leptin transport seen during pregnancy appears similar to what has been observed in both obese animals and humans. The short form of the leptin receptor (OB-Ra), which is densely expressed in the choroid plexus and endothelial cells, has been implicated in leptin transport across the BBB (13). Although saturation of OB-Ra or potentially the down-regulation of the receptor in the hormonal milieu that characterizes pregnancy may be one mechanism impacting leptin transport across the BBB, soluble leptin binding proteins may also mediate leptin resistance during pregnancy.

OB-Re is synthesized via alternative splicing of leptin receptor mRNA and by ectodomain shedding of membrane-spanning receptors (29, 30). In humans, OB-Re levels are inversely related to leptin levels, and we confirm that this relationship persists in pregnancy (31–33). OB-Re functions as a plasma leptin binding protein that reduces leptin clearance, thereby dramatically increasing circulating leptin levels. Yet, increased concentrations of leptin-OB-Re complexes do not appear to promote leptin action. Instead, as with other peptide binding proteins like GH (34) and CRH (35), the bound form of the hormone is sequestered from interactions with its receptor. This is evinced by in vitro studies demonstrating a reduction in leptin binding to OB-Ra as well as to OB-Rb and a corresponding decrease in leptin-induced signal transducer and activator of transcription 3 (STAT3) signaling after incubation of leptin with OB-Re (17, 36).

In vivo, the leptin resistance that characterizes pregnancy has been shown to occur in conjunction with a dramatic rise in plasma leptin binding activity in the final third of gestation in mice (2). Although the pregnancy-associated rise in plasma leptin binding activity is not as great in rats, gestational leptin resistance in these animals also appears to occur secondary to increases in OB-Re as opposed to changes in hypothalamic OB-Rb expression (9). We confirm elevations in OB-Re concentrations during human pregnancy. Although some describe elevations in OB-Re levels only in diabetic pregnancies, our results are consistent with those of Neuteboom et al. (7) who identified a gestational rise in OB-Re that mirrors the rise in leptin in normal pregnant women (3, 7, 37). Given that soluble leptin receptors have been shown to antagonize the central transport of leptin both in vitro and in vivo in animal models, it is reasonable to hypothesize that OB-Re mediates similar effects in human pregnancy, restricting leptin's access to target cells in the brain and hypothalamus.

Additionally, our data suggest that the leptin resistance of pregnancy may be mediated by changes in leptin's impact on target neuropeptides. Although both pregnant and nonpregnant subjects had comparable CSF leptin levels, absolute AgRP levels and AgRP/POMC were significantly increased, suggesting that leptin may be less effective at suppressing AgRP and stimulating POMC during pregnancy. This decrease in hypothalamic responsiveness to the appetite-suppressing effects of leptin has been demonstrated in pregnant rats that exhibit no reduction in appetite after intracerebroventricular (icv) leptin administration (19, 38). Furthermore, there is evidence that the increase in AgRP and AgRP/POMC that we observe in pregnancy reflects alterations in leptin signal transduction. Trujillo et al. (19) demonstrated a loss in the capacity of icv leptin to induce phosphorylation of STAT3 and Akt in the hypothalamus of pregnant rats. This was associated with higher levels of SOCS3 in the hypothalamus, which is known to inhibit leptin signaling. An earlier study reported decreased STAT3 phosphorylation specifically in the arcuate nucleus of pregnant rats after icv leptin (39). These changes in the leptin signaling pathway were accompanied by an increase in AgRP and NPY and a decrease in POMC mRNA levels in the hypothalamus of pregnant rats (19). However, not all studies have reported significant decreases in POMC mRNA in the hypothalamus during pregnancy (16).

Increases in AgRP mRNA have been more consistently reported in the rodent hypothalamus during pregnancy. Rocha et al. (16) observed a selective increase in hypothalamic gene expression of AgRP at d 19 of gestation in Wistar rats. Similarly, using in situ hybridization techniques, Trujillo et al. (19) observed a significant increase in AgRP and NPY in the arcuate nucleus of pregnant rats on d 13 and 18 of gestation, although expression was not different earlier in gestation. In contrast, Ladyman et al. (40) did not observe a change in AgRP mRNA expression during pregnancy, although in the presence of hyperleptinemia the absence of a fall in AgRP is consistent with leptin resistance.

Overall, these data suggest that during pregnancy, women are protected from the suppressive effects of hyperleptinemia on appetite by decreasing leptin transport into CSF and by modifying leptin's effects on the target melanocortin neuropeptides that regulate energy balance. The leptin resistance of pregnancy is not a state of physiological aberrancy, but instead reflects an adaptive physiological response to the metabolic demands of pregnancy that appear to have been evolutionarily conserved from animals to humans.