Evaluation of CSF and plasma biomarkers of brain melanocortin activity in response to caloric restriction in humans

Abstract

The melanocortin neuronal system, which consists of hypothalamic proopiomelanocortin (POMC) and agouti-related protein (AgRP) neurons, is a leptin target that regulates energy balance and metabolism, but studies in humans are limited by a lack of reliable biomarkers to assess brain melanocortin activity. The objective of this study was to measure the POMC prohormone and its processed peptide, β-endorphin (β-EP), in cerebrospinal fluid (CSF) and AgRP in CSF and plasma after calorie restriction to validate their utility as biomarkers of brain melanocortin activity. CSF and plasma were obtained from 10 lean and obese subjects after fasting (40 h) and refeeding (24 h), and from 8 obese subjects before and after 6 wk of dieting (800 kcal/day) to assess changes in neuropeptide and hormone levels. After fasting, plasma leptin decreased to 35%, and AgRP increased to 153% of baseline. During refeeding, AgRP declined as leptin increased; CSF β-EP increased, but POMC did not change. Relative changes in plasma and CSF leptin were blunted in obese subjects. After dieting, plasma and CSF leptin decreased to 46% and 70% of baseline, CSF POMC and β-EP decreased, and plasma AgRP increased. At baseline, AgRP correlated negatively with insulin and homeostasis model assessment (HOMA-IR), and positively with the Matsuda index. Thus, following chronic calorie restriction, POMC and β-EP declined in CSF, whereas acutely, only β-EP changed. Plasma AgRP, however, increased after both acute and chronic calorie restriction. These results support the use of CSF POMC and plasma AgRP as biomarkers of hypothalamic melanocortin activity and provide evidence linking AgRP to insulin sensitivity.

the melanocortin neuronal system plays a key role in regulating energy balance and metabolism (11, 35). This system comprises hypothalamic proopiomelanocortin (POMC) and agouti-related protein (AgRP) neurons whose peptide products interact with downstream melanocortin receptor (MC-R)-expressing neurons (16). The POMC-derived peptide α-melanocyte-stimulating hormone (α-MSH) inhibits food intake and stimulates energy expenditure, whereas AgRP is an MC-R antagonist that stimulates food intake and inhibits energy expenditure. Defects in POMC synthesis, peptide processing, and MC-R signaling cause obesity in rodents and humans (4, 6, 31). POMC and AgRP neurons are responsive to a variety of metabolic signals that regulate energy and glucose homeostasis, including leptin and insulin (17). The physiology of this system has been extensively studied in rodents, but studies in humans are limited by the lack of biomarkers for brain POMC and AgRP. Because levels of intact POMC prohormone in cerebrospinal fluid (CSF) have been shown to correlate with hypothalamic POMC in rodents, we focused on similar measurements in human CSF (23). Although it is the POMC-derived peptide α-MSH that engages brain MC-Rs, CSF α-MSH levels are low and difficult to detect. In rodents, CSF POMC, rather then α-MSH, has been shown to reflect hypothalamic POMC activity (23). We have previously shown that high levels of POMC are present in human CSF, and that concentrations vary as a function of body weight, adiposity, and leptin (19). We found no correlation between CSF POMC and plasma POMC, which is of pituitary origin. We also measured AgRP in human CSF and plasma. In contrast to POMC, there is evidence that plasma and hypothalamic AgRP levels are correlated in rodents (12), and we have demonstrated a correlation between plasma AgRP and adiposity in humans (19). However, previous CSF and plasma measurements were all performed in the basal state, and the effects of feeding and weight loss on these parameters have not yet been studied in humans.

Food restriction induces a host of hormonal and neuronal responses that serve to maintain energy balance (3). Plasma leptin falls after acute and chronic food restriction and is accompanied by a rise in levels of the soluble leptin receptor (sOB-R), which may affect leptin transport into the brain (2, 28). Fasting suppresses POMC and stimulates AgRP in the rodent hypothalamus; these effects can be reversed by leptin (10). Such changes in melanocortin activity stimulate appetite and have been implicated as a cause of recidivism after diet-induced weight loss. Objectives of the current study were to examine hormonal and neuropeptide responses to acute fasting and refeeding in healthy lean vs. obese human subjects as compared with chronic diet-induced weight loss in obese subjects. Accordingly, we measured POMC, AgRP, and leptin in CSF; and AgRP, leptin, and other hormones in plasma to validate POMC and AgRP measurements as biomarkers of melanocortin activity after acute and chronic caloric restriction and to examine related changes in plasma and CSF leptin and sOB-R levels. β-endorphin (β-EP) was also measured in CSF because both β-EP and α-MSH are derived together from POMC, and their levels in the hypothalamus usually change in parallel (9, 13). Effects on insulin sensitivity and glucose tolerance were studied because the melanocortin system can affect glucose metabolism independently of changes in body weight (35).

MATERIALS AND METHODS

Study Participants and Protocols

Study participants were healthy men and women (age, 22–45 yr) who were nonsmokers and not taking medications. Women were studied in the early follicular phase of the menstrual cycle. Subjects with a history of eating disorders, recent weight change ± 5%, or use of weight loss products or dieting within 6 mo of starting the study were excluded. This study was approved by the Columbia University Institutional Review Board, and written informed consent was obtained from all subjects.

Study 1: fasting-refeeding protocol.

Ten subjects (7 men, 3 women) were studied. Six subjects were lean [body mass index (BMI) 23.1 ± 0.9 kg/m²]; and four were obese (BMI 33.0 ± 2.2 kg/m²). Subjects were admitted to the clinical research center at 10:00 A.M. (day 1) after fasting since dinner at 6:00 P.M. the previous day, and continued to fast for a total of 40 h. Subjects had free access to water and received intravenous hydration with 1 liter of normal saline. Lumbar puncture (LP) was performed at the end of the 40-h fast (10:00 A.M., day 2). Subjects were refed 200% of calculated (Harris-Benedict equation) caloric requirements over the next 24 h. Meals (55% carbohydrate, 15% protein, 30% fat) were provided by the Bionutrition Research Core at Columbia University Medical Center: breakfast (10:00 A.M.), lunch (1:00 P.M.), dinner (7:00 P.M.), snacks (4:00 P.M./10:00 P.M.), and breakfast the following day (8:00 A.M.). Twenty percent of calories was provided at each meal and 10% at each snack. A second LP was performed after refeeding (10:00 A.M., day 3). CSF (10 ml) was collected at each LP. Blood was obtained during fasting (F) at 10:00 A.M. (F16), 6:00 P.M. (F24, day 1), 10:00 A.M. (F40, day 2), and refeeding (RF) before lunch (RF3) and dinner (RF8), and before 8:00 A.M. (RF22, day 3) and after breakfast (RF23, 9:00 A.M. day 3), and 10:00 A.M. (RF24, day 3). Subjects consumed an average of 190 ± 8.6% of their caloric requirement. One subject developed a mild headache after the first LP, and thus did not have a second LP but instead was refed and blood was drawn.

Study 2: low-calorie diet protocol.

Nine obese (BMI 33.3 ± 1.6 kg/m²; range, 30 to 41) women were recruited. Eight subjects were studied before and after 6 wk on an 800-kcal/day liquid diet (Optifast). CSF (10 ml) was collected by LP after an overnight fast at baseline and after 6 wk of diet. Blood was obtained concomitantly. A 2-h oral glucose (75 g) tolerance test (OGTT) was performed on a separate day before and at the end of the diet in 7 subjects. Hunger and satiety were assessed by visual analog scale (VAS) before each OGTT. One subject was withdrawn after developing a mild headache after the first LP.

Assays

Leptin and sOB-R were measured in plasma and CSF by ELISA (R&D Systems, Minneapolis, MN) (19). POMC was measured by two-site ELISA (19, 27); there was no cross-reactivity with ACTH, α-MSH, or β-EP. β-EP was measured by RIA as previously described; 3% cross-reactivity with POMC was observed (25). β-EP was also measured with a newly developed two-site ELISA that is specific for β-EP and does not cross-react with POMC. This assay employs the same antibody used in the RIA for capture and a monoclonal antibody (MAB5276; Millipore, Temecula, CA) to met-enkephalin (NH₂-terminal of β-EP) that was biotinylated for detection; sensitivity is 2 pg/ml.

AgRP was measured by ELISA and RIA with relative specificities for full-length AgRP and AgRP_83–132, respectively (19, 34). The ELISA (R&D Systems) uses full-length human AgRP standard; there was 17% cross-reactivity with AgRP_83–132. The RIA uses an antibody provided by Dr. Barsh (Stanford University School of Medicine) and human AgRP_83–132 standard (Phoenix Pharmaceuticals); there was 20% cross-reactivity with full-length AgRP.

Insulin was measured by Immulite1000 (Siemens Healthcare Diagnostics). Glucose was measured by the hexokinase method. Total ghrelin was measured by ELISA (Millipore, Billerica, MA).

Statistical Analysis

Data are expressed as means ± SE. CSF hormone and neuropeptide levels in the fasted and RF states and before and after dieting were analyzed by paired t-test or paired Wilcoxon signed-rank test. Plasma levels of leptin and AgRP measured over time were analyzed by repeated-measures ANOVA. Areas under the hormone response curves (AUC) during the OGTT were calculated by trapezoid analysis and compared using a paired t-test. Correlations were determined by linear regression analysis using Pearson’s correlation. Insulin resistance was calculated using the homeostasis model assessment (HOMA-IR) (15). Insulin sensitivity during the OGTT was calculated using the Matsuda index (MI) (14).

RESULTS

Study 1: Fasting and Refeeding

Changes in leptin and sOB-R in plasma and CSF.

Plasma leptin changed over time during fasting and RF (P < 0.001) (Fig. 1). Levels decreased from a baseline (upon admission) of 15.4 ± 5.1 to 6.1 ± 2.3 ng/ml after fasting (P = 0.002) and then increased to 20 ± 6.2 ng/ml during RF (P = 0.02 vs. baseline). Plasma leptin suppressed to 35.1 ± 3.4% of baseline in the entire group but the degree of suppression was more profound in lean vs. obese subjects (28.7 ± 3.1 vs. 44.8 ± 3.4% of baseline, P = 0.009) (Fig. 1). The peak percent increase from baseline during RF was 173 ± 19 vs. 130 ± 17% in lean vs. obese (P = 0.16) subjects. CSF leptin was 139 ± 41 pg/ml after fasting and increased to 205 ± 55 pg/ml after RF (P = 0.004). The percent increase in CSF leptin was greater in lean vs. obese subjects (275 ± 56 vs. 131 ± 5.7%, P = 0.016) (Fig. 1). Thus, relative changes in plasma and CSF leptin were more pronounced in lean subjects after fasting and RF.

Fig. 1.

Top: plasma leptin over time (means ± SE) during the entire period of fasting and refeeding. Left: leptin levels in the entire group (solid circles); middle: leptin levels in lean subjects (triangles); right: leptin levels in obese subjects (open squares). The degree of leptin suppression during fasting was greater in lean vs. obese subjects. Bottom: mean levels of leptin in plasma and cerebrospinal fluid (CSF) at the time of the two lumbar punctures (LPs) after 40 h of fasting (solid bars) and 24 h of refeeding (RF) (hatched bars). The percent increase in CSF leptin after RF was higher in lean vs. obese subjects (bottom right) (*P < 0.01).

Plasma sOB-R increased over time during fasting and RF (P = 0.004). Baseline sOB-R was 22.4 ± 2.0 ng/ml vs. 23.3 ± 2.3 after fasting; levels continued to increase during RF to 25.7 ± 3.0 ng/ml (P < 0.05). The ratio of CSF to plasma leptin expressed as a percentage was 3.3 ± 0.48% after fasting and decreased to 1.9 ± 0.33% after RF (P = 0.008). Thus, higher plasma sOB-R was associated with a lower CSF to plasma leptin ratio. CSF sOB-R did not change significantly after fasting vs. RF (0.233 ± 0.05 vs. 0.256 ± 0.06 ng/ml, P = 0.14).

Changes in insulin and ghrelin.

Serum insulin decreased from 9.9 ± 2.5 to 5.4 ± 1.3 µIU/ml after 40 h of fasting (P = 0.005). Plasma ghrelin was 658 ± 91 and 657 ± 99 pg/ml after 16 and 24 h of fasting and tended to decrease after 40 h (539 ± 72) of fasting (P = 0.09). Ghrelin then decreased after 3 and 8 h of RF (63 and 51% of baseline, P < 0.001) and returned to 90% of baseline the following morning.

Changes in POMC and β-EP in CSF, and AgRP in CSF and plasma.

The concentration of POMC in CSF was not different after fasting vs. RF (P = 0.49). In contrast, the concentration of β-EP in CSF after fasting was 77.5% of the concentration after RF (P = 0.04) (Fig. 2). The POMC to β-EP ratio tended to be higher during fasting (P = 0.11). CSF levels of AgRP were not different during fasting vs. RF. However, plasma AgRP changed over time during fasting and RF (P < 0.001) (Fig. 2). Plasma AgRP increased from 68 ± 7.8 to 103 ± 16 pg/ml after 40 h of fasting (P = 0.02); levels then decreased toward baseline at the end of RF. Plasma AgRP was higher in lean vs. obese subjects at baseline (80 ± 10 vs. 50 ± 4.8 pg/ml, P = 0.04) and after fasting (131 ± 19 vs. 62 ± 4.5 pg/ml, P = 0.02); AgRP correlated negatively with BMI at baseline (r = −0.719, P = 0.02) and after fasting (r = −0.741, P = 0.01). Plasma AgRP correlated negatively with plasma leptin (r = −0.642), CSF leptin (r = −0.652, P < 0.05) and insulin (r = −0.600. P = 0.07). The relationship between plasma leptin and AgRP throughout the fasting-RF protocol is shown in Fig. 2. The decrease in leptin after fasting is paralleled by an increase in plasma AgRP (153% of baseline) and with RF the increase in leptin is paralleled by a decrease in AgRP. CSF AgRP was measured by ELISA and RIA with relative specificities for the full-length (FL) and C-terminal (CT) peptides, respectively. Although no difference was noted after fasting vs. RF with either assay, the calculated ratio of CT to FL AgRP was 2.24 ± 0.37 after fasting vs. 1.58 ± 0.22 after RF (P = 0.02).

Fig. 2.

Top: proopiomelanocortin (POMC), β-endorphin (β-EP), and agouti-related protein (AgRP) in CSF; and plasma AgRP (left bottom) (means ± SE) after 40 h of fasting (solid bars) and 24 h of RF (hatched bars). Mean concentrations of plasma AgRP (solid circles) and plasma leptin (open squares) are depicted over the entire time (bottom right) (*P < 0.05).

Study 2: Low-Calorie Diet

Changes in leptin and sOB-R in plasma and CSF.

Mean weight loss after 6 wk of dieting was 8.6% (Fig. 3). Plasma leptin decreased to 46% (P = 0.009) and CSF leptin to 70% of baseline (P = 0.004) (Fig. 3). Plasma sOB-R increased to 114% of baseline (P = 0.04). CSF sOB-R did not change. The ratio of CSF to plasma leptin expressed as percent was 1.32 ± 0.19% at baseline vs. 1.76 ± 0.18% after weight loss (P = 0.08).

Fig. 3.

Top left: percent weight loss for the eight subjects in the diet study. Top middle: plasma leptin. Top right: CSF leptin (means ± SE). Bottom: plasma soluble leptin receptor (sOB-R), serum insulin, and plasma ghrelin at baseline before the diet (solid bars) and after 6 wk of dieting (hatched bars) (*P < 0.05).

Changes in insulin, glucose metabolism, and ghrelin.

Fasting serum insulin decreased from 15.0 ± 3.6 to 5.1 ± 0.7 µIU/ml, fasting glucose decreased from 91.3 ± 4.1 to 84.9 mg/dl ± 4.1 (P = 0.04), and HOMA-IR decreased from 3.4 ± 0.9 to 1.0 ± 0.2 (P = 0.02) (Figs. 3 and 5). The AUC for insulin during the OGTT decreased (P = 0.03), but the AUC for glucose was not different. The MI calculated during the OGTT increased from 5.1 ± 2.1 to 9.2 ± 2.5 (P = 0.02) (see Fig. 5). Fasting plasma ghrelin increased after weight loss (Fig. 3), and the AUC for ghrelin during the OGTT increased by 136% (P = 0.04). Fasting plasma ghrelin before weight loss correlated negatively with serum insulin (r = −0.837, P = 0.009) and with percent weight loss after diet (r = −0.766, P = 0.02). Ghrelin did not correlate with VAS scores.

Fig. 5.

Top and bottom left: insulin and ghrelin levels over time during the oral glucose tolerance test (OGTT) at baseline (solid circles) and after weight loss (open squares) (means ± SE). Top and bottom middle: mean calculated homeostasis model assessment (HOMA) and Matsuda index (MI), respectively, at baseline (solid bars) and after weight loss (hatched bars). Top right: correlation of plasma AgRP with fasting insulin. Bottom right: correlation of plasma AgRP with MI calculated during the first OGTT (*P < 0.05).

Changes in POMC and β-EP in CSF and AgRP in CSF and plasma.

The concentrations of POMC and β-EP in CSF decreased significantly following weight loss. Mean and individual changes are shown in Fig. 4. CSF POMC decreased to 86% of baseline (P = 0.003). CSF β-EP, measured by RIA and highly specific ELISA, decreased to 87% and 71% of baseline, respectively (P < 0.05). There was no change in the POMC to β-EP ratio. CSF AgRP did not change after weight loss when measured by either ELISA (20.6 ± 2.1 vs. 19.0 ± 1.5 pg/ml) or RIA (39.4 ± 6.8 vs. 43.8 ± 6.0 pg/ml). However, plasma AgRP increased significantly from 61.7 ± 9.7 to 72.0 ± 11 pg/ml (P = 0.03) (Fig. 4). The CSF POMC to plasma AgRP ratio decreased to 74% of baseline (P = 0.005) and the CSF POMC to CSF AgRP (RIA) ratio decreased to 75% of baseline (P = 0.01), indicating decreased melanocortin activity after weight loss.

Fig. 4.

Top left: POMC and β-EP levels in CSF (means ± SE) at baseline before the diet (solid bars) and after 6 wk of dieting (hatched bars). Top right: graphs of individual CSF POMC and β-EP concentrations Bottom left: CSF and plasma AgRP at baseline before the diet (solid bars) and after the diet (hatched bars). Bottom right: ratios of CSF POMC to plasma AgRP and to CSF AgRP before and after the diet (*P < 0.05).

At baseline, plasma AgRP correlated negatively with serum insulin (r = −0.807, P = 0.008) and HOMA (r = −0.633, P = 0.13) (Fig. 5). There was a positive correlation between plasma AgRP and the MI calculated during the baseline OGTT (r = 0.777, P = 0.04) (Fig. 5). These correlations were no longer evident after weight loss. Subjects tended to report less hunger before the OGTT that was carried out after weight loss (2.27 ± 0.9) than before weight loss (4.5 ± 1.0, P = 0.05). There were no significant correlations between ratings of hunger and satiety with POMC or AgRP at baseline or after weight loss. However, the percent change in plasma AgRP after weight loss correlated positively with VAS hunger scores (r = 0.733, P = 0.06) and the percent change in the ratio of POMC to plasma AgRP correlated negatively with hunger scores (r = −0.836, P = 0.02), suggesting that relative changes in melanocortin activity may contribute to changes in appetite after weight loss. The percent change in plasma AgRP also tended to correlate with the percent change in ghrelin (r = 0.656, P = 0.08).

DISCUSSION

Although previous studies suggest that concentrations of POMC in CSF and of AgRP in plasma may be useful biomarkers of hypothalamic POMC and AgRP activity, the effects of caloric restriction on these parameters and their relationship to leptin and insulin have never been studied in humans. This study demonstrates changes in melanocortin peptides after fasting and RF and after diet-induced weight loss that correspond to the known changes in POMC and AgRP in the hypothalamus. Changes in CSF leptin were also demonstrated for the first time as being related to plasma leptin and sOB-R. Importantly, more evidence is provided to support the use of plasma AgRP as a marker of hypothalamic AgRP activity and links plasma AgRP to insulin sensitivity.

Plasma leptin fell to 35% of baseline after 40 h of fasting and then rapidly rebounded after 24 h of RF. Relative changes in both plasma and CSF leptin were greater in lean vs. obese subjects, suggesting that in lean individuals, the brain receives a more robust signal to indicate energy deficit and surplus. Changes in the CSF to plasma leptin ratio were compared after fasting and RF. We hypothesized that the ratio would be lower during fasting due to increased sOB-R, which can inhibit leptin transport into the brain (20, 28). However, the ratio was higher after fasting vs. RF, which may be because sOB-R levels continued to increase during RF. By comparison, after dieting and achieving 8.6% weight loss, plasma and CSF leptin levels decreased to 46% and 70% of baseline, respectively, but the CSF to plasma leptin ratio did not change despite an increase in sOB-R levels. Weight loss during the fast was minimal compared with the diet, but there was a comparable fall in leptin (3, 33).

The concentration of POMC in CSF did not change after 40 h of fasting but decreased after dieting. However, CSF β-EP declined in both cases. Thus, acute caloric deprivation affects release of processed POMC peptides, whereas chronic restriction leads to a decrease in the POMC prohormone and processed peptides. This is consistent with an initial effect on peptide release and a more delayed effect on POMC synthesis, but it could also reflect changes in POMC processing. Changes in POMC processing enzymes have been reported in the hypothalamus during fasting (22). In contrast, plasma AgRP increased after both acute and chronic caloric restriction. This is consistent with studies showing more rapid changes in hypothalamic AgRP expression compared with POMC during fasting (10, 21). Unfortunately, α-MSH could not be reliably measured in CSF possibly due to degradation or inactivation by prolylcarboxypeptidase (29). Our α-MSH assay is specific for the amidated peptide and does not detect the inactivated peptide. However, CSF β-EP may serve as a marker of both hypothalamic β-EP and α-MSH given that levels of both peptides typically change in parallel (13, 32). As with POMC, CSF β-EP is of brain origin (25). Thus, it is likely that the decline in CSF β-EP during fasting is a reflection of a decline in hypothalamic β-EP and α-MSH.

AgRP was measured in CSF and plasma, but only plasma AgRP showed consistent changes in both studies. The increases in plasma AgRP during fasting and dieting mirror the expected changes in hypothalamic AgRP under those conditions. The reciprocal changes in plasma leptin and AgRP during fasting and RF are consistent with the known inhibitory effect of leptin on AgRP in the hypothalamus. Higher levels of plasma AgRP have been reported in rats during fasting (12, 26) and in humans before versus after breakfast (26). We have previously demonstrated negative correlations between plasma AgRP and BMI and leptin in lean and obese subjects (19). Similar negative correlations are again demonstrated. These results suggest that plasma AgRP is of hypothalamic origin, but how brain AgRP gains access to the circulation remains unclear. Heavy AgRP fiber staining is found in the median eminence, which could be a source of secretion into the blood (7). Although the adrenals may also be a source for circulating AgRP (18), it is notable that plasma AgRP did not change in rats after adrenalectomy (12). In contrast to plasma AgRP, CSF AgRP did not increase significantly after fasting or dieting. The explanation for this is unclear but may relate to anatomical differences in AgRP fiber tracks that gain access to CSF and blood, respectively (1, 7). However, in CSF, there was relatively more AgRP measured by RIA vs. ELISA after fasting. This is consistent with changes in AgRP processing resulting in relatively more AgRP_83–132, which has more biological activity than the full-length peptide (5). We have confirmed by HPLC that both forms of AgRP are present in CSF (19, 34).

AgRP neurons are known to play a role in responding to insulin signaling and regulating glucose metabolism independent of changes in body weight (8, 24). Plasma AgRP correlated negatively with fasting insulin and HOMA in a cohort of lean and obese subjects, and the correlation persisted when adjusted for BMI (19). This was observed again in the present studies. The negative correlation observed at baseline in the diet study is notable given that it involved a more homogeneous group of overweight/obese women. Furthermore, a strong positive correlation was noted with plasma AgRP and the MI calculated during the baseline OGTT, providing more evidence for the use of plasma AgRP as a marker of insulin sensitivity while weight is stable. However, these correlations were no longer evident during weight loss, which is associated with numerous hormonal and metabolic changes and an overall decrease in brain melanocortin activity.

There were no correlations between ratings of hunger and satiety with POMC or AgRP at baseline or after weight loss. However, the percent change in plasma AgRP after weight loss correlated positively with hunger, and the percent change in the ratio of POMC-to-plasma AgRP correlated negatively with hunger, suggesting that changes in melanocortin activity may contribute to changes in appetite after weight loss. Plasma AgRP did not correlate with ghrelin, but the change in plasma AgRP tended to correlate with the change in ghrelin, which is consistent with the known stimulatory effect of ghrelin on AgRP neurons (30).

We have previously shown that AgRP (in plasma and CSF) is positively correlated with CSF POMC in lean and obese subjects (19). This initially appeared paradoxical given the opposite roles that POMC and AgRP play in regulating energy balance. However, we now show that under conditions of caloric restriction, POMC levels fall and AgRP increases as would be predicted from animal studies. The explanation for this may be that the activities of both POMC and AgRP neurons and the entire brain melanocortin circuit are increased in lean vs. obese subjects under basal conditions, but in the setting of a caloric deficit, POMC neuronal activity decreases and AgRP increases to maintain energy balance.

In summary, plasma and CSF leptin decreased substantially after fasting and dieting. The relative changes in both CSF and plasma leptin after fasting and RF were blunted in obese subjects. A significant fall in CSF POMC was observed only after the diet, although CSF β-EP changed in both settings. Plasma AgRP levels increased in both settings, and at baseline, correlated with insulin, HOMA, and MI. This study provides further support for the use of CSF POMC and plasma AgRP measurements as biomarkers of hypothalamic melanocortin activity and provides additional evidence linking plasma AgRP to insulin sensitivity.

GRANTS

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant RO1-DK-093920 to S. L. Wardlaw; National Center for Advancing Translational Sciences Grants UL1 TR-000040 and T32DK-007271 and T32DK-065522 to Columbia University; and by a Naomi Berrie Fellowship from the Russell Berrie Foundation to D. Atalayer.

FINANCIAL DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

S.L.W. conceived and designed research; G.P.-W., K.T.N., D.A., K.M., H.A.B., R.J.G., and R.S. performed experiments; K.T.N., D.A., K.M., R.J.G., S.K.P., and S.L.W. analyzed data; G.P.-W., J.K., and S.L.W. interpreted results of experiments; K.M. and S.K.P. prepared figures; G.P.-W. and S.L.W. drafted manuscript; G.P.-W., J.K., A.W., R.S., and S.L.W. edited and revised manuscript; G.P.-W., K.T.N., D.A., K.M., H.A.B., J.K., R.J.G., S.K.P., A.W., R.S., and S.L.W. approved final version of manuscript.