Acute physiological effects of glucocorticoids on fuel metabolism in humans are permissive but not direct

Background and aims The effects of glucocorticoids on fuel metabolism are complex. Acute glucocorticoid excess promotes lipolysis but chronic glucocorticoid excess causes visceral fat accumulation. We hypothesized that interactions between cortisol and insulin and adrenaline account for these conflicting results. We tested the effect of cortisol on lipolysis and glucose production with and without insulin and adrenaline in humans both in vivo and in vitro. Materials and methods A total of 20 healthy men were randomized to low and high insulin groups (both n = 10). Subjects attended on 3 occasions and received low (c. 150 nM), medium (c. 400 nM) or high (c. 1400 nM) cortisol infusion in a randomized crossover design. Deuterated glucose and glycerol were infused intravenously along with a pancreatic clamp (somatostatin with replacement of glucagon, insulin and growth hormone) and adrenaline. Subcutaneous adipose tissue was obtained for analysis. In parallel, the effect of cortisol on lipolysis was tested in paired primary cultures of human subcutaneous and visceral adipocytes. Results In vivo, high cortisol increased lipolysis only in the presence of high insulin and/or adrenaline but did not alter glucose kinetics. High cortisol increased adipose mRNA levels of ATGL, HSL and CGI‐58 and suppressed G0S2. In vitro, high cortisol increased lipolysis in the presence of insulin in subcutaneous, but not visceral, adipocytes. Conclusions The acute lipolytic effects of cortisol require supraphysiological concentrations, are dependent on insulin and adrenaline and are observed only in subcutaneous adipose tissue. The resistance of visceral adipose tissue to cortisol's lipolytic effects may contribute to the central fat accumulation observed with chronic glucocorticoid excess.

or decreased 6 by glucocorticoids. These discrepancies may be attributed to several factors, including the dose and duration of glucocorticoid treatment, the effect of other hormones in the media and the species being studied.
Results from in vivo studies testing the effect of glucocorticoids on lipolysis have been more consistent, particularly when levels of other counter-regulatory hormones have been clamped by infusing somatostatin and replacing insulin, growth hormone and glucagon (the pancreatic clamp technique), showing that cortisol acutely increases whole body lipolysis. [7][8][9] However, these studies achieved cortisol concentrations between 850 and 1500 nM which are not reflective of those observed physiologically in the absence of acute stress. Two studies have examined the effect of physiological cortisol concentrations on in vivo lipolysis and had conflicting results in the fasted state; 1,10 however, neither study used a pancreatic clamp to control counter-regulatory hormones. Therefore, it is unclear whether changes in cortisol concentrations within the physiological range alter whole body lipolysis. It is also unclear how glucocorticoids enhance lipolysis in humans, as no in vivo study has examined the effect of glucocorticoids on the lipolytic pathway ( Figure S1, Supporting Information).
Over and above prevailing glucocorticoid concentrations, a further critical confounder is the effect of other hormones regulating lipolysis, notably insulin and adrenaline. 5,6 In vivo, the effect of interactions between cortisol and either adrenaline 11 or insulin 12 has been examined only once in humans, although systemic rates of lipolysis were not measured as the appropriate tracers were not infused. We hypothesized that the effects of glucocorticoids on lipolysis in humans are indirect and dependent on the prevailing insulin and/or adrenaline concentrations, with glucocorticoids augmenting the pro-lipolytic effects of adrenaline and antagonizing the suppressive effects of insulin.
In addition, we hypothesized that these interactions may account for the apparently contradictory effects of acute (a state of high adrenaline and low insulin) and chronic (a state of high insulin and low adrenaline) glucocorticoid excess and that the effects may differ between subcutaneous and visceral depots. To test this, we performed a randomized, double-blinded, crossover study to determine the effects of glucocorticoids on whole body lipolysis in the presence of both low and high insulin and adrenaline, respectively. Furthermore, we collected adipose tissue biopsies in vivo to determine how glucocorticoids alter lipolysis and tested in vitro whether glucocorticoids have differential effects on lipolysis in subcutaneous and visceral adipocytes.

| In vivo protocol
A total of 20 healthy men were recruited to a randomized, doubleblind, placebo-controlled crossover study with the following inclusion criteria: age, 18 to 75 years; body mass index, 20 to 25 kg/m 2 ; absence of chronic medical conditions; absence of regular medications; no previous glucocorticoid use in the past year; alcohol intake ≤21 units per week; weight change of <5% over the past 6 months; normal screening blood tests (renal, liver and thyroid function, fasting glucose, full blood count). Local ethical approval was obtained, as was written informed consent from each participant. Research Facility after   overnight fast on 3 occasions, each separated by 3 weeks, and were instructed to avoid alcohol and exercise for 48 hours prior to each assessment. Volunteers were randomized to low, medium or high glucocorticoid (GC) phases ( Figure S2, Supporting Information). The night prior to each assessment (11 PM) subjects received orally the 11β-hydroxylase inhibitor metyrapone (metopirone) 1 g, along with either placebo (low GC phase), hydrocortisone 10 mg (medium GC) or hydrocortisone 20 mg (high GC). At 7 AM the following morning subjects received orally 1 g of metyrapone along with either placebo (low GC), hydrocortisone 5 mg (medium GC) or hydrocortisone 10 mg (high GC). A further 1 g of metyrapone was received orally at 11 AM to maintain inhibition of endogenous cortisol synthesis throughout the protocol. The 3 GC phases aimed to achieve trough and peak cortisol concentrations observed during normal diurnal rhythm (low and medium GC, respectively) and peak cortisol concentrations during stress (high GC).

Subjects attended the Edinburgh Clinical
At each visit, measurements were performed of height, weight, blood pressure, body fat by bioimpedance (using an Omron BF-302 analyser) and core body temperature using a tympanic thermometer.  5 -glycerol (at 0.11 μmol/kg/min following a 1.6 μmol/kg bolus) were commenced for 6.5 hours ( Figure S2, Supporting Information). A "pancreatic clamp" was commenced at t = 0 minutes, comprising intravenous somatostatin (60 ng/kg/min), glucagon (0.5 ng/kg/min) and growth hormone (2 ng/kg/min). At their first visit, subjects were further randomized to receive either low or high insulin replacement (both groups n = 10) at a rate of 0.06 mU/kg/min or 0.2 mU/kg/min (aiming to suppress lipolysis by c. 50%), 13 respectively, and subjects remained in this group for all 3 study visits. At t = 0 minutes, subjects commenced infusion with 0.9% saline (low GC) or hydrocortisone (medium GC at 0.025 mg/kg/h following a 0.04 mg/kg bolus; high GC at 0.12 mg/kg/h following a 0.18 mg/kg bolus) in random order.
At t + 20 minutes, an intravenous infusion of 20% dextrose was commenced and the rate was adjusted every 10 minutes to maintain an arterialized glucose concentration between 7.5 and 8.0 mmol/L. Steady state measurements were taken between t + 180 and t + 240 minutes; following this a subcutaneous abdominal adipose tissue biopsy was obtained by needle aspiration. 14 At t + 285 minutes, an adrenaline infusion was commenced at 0.15 nmol/kg/min for 60 minutes. Blood samples were obtained at regular intervals ( Figure 1).
Samples were stored at −80 C until analysis.

| In vitro protocol
Paired samples of subcutaneous and visceral adipose tissue were obtained from patients undergoing elective abdominal surgery at the Royal Infirmary of Edinburgh (subject characteristics detailed in Table S1, Supporting Information). Adipose tissue from subcutaneous and visceral depots was digested, and the stromal vascular fraction was isolated and differentiated as previously described. 15 In brief, following removal of connective tissue and blood vessels, adipose tissue was digested in collagenase type 1 (615 units/g tissue) for 90 minutes at 37 C. Following plating and overnight incubation in DMEM/ F12 medium containing 33 μM biotin, 17 μM pantothenate and 10% foetal bovine serum, cells were differentiated for 3 days using serumfree medium plus 1 nM triiodothyronine, 10 μg/mL transferrin, 66 nM insulin, 500 μM isobutylmethylxanthine, 1 μM dexamethasone and 10 μM rosiglitazone. From day 4 onwards, cells were maintained in differentiation medium, but without isobutylmethylxanthine, dexamethasone or rosiglitazone.
On day 16, cells were incubated with either 0, 100 or 1000 nM cortisol for 24 hours in the presence or absence of either vehicle, 100 pM insulin or 10 μM adrenaline. Following incubation, cells were used to measure mRNA levels of key genes in the lipolytic pathway or the medium was removed to measure the appearance of glycerol and cells were lysed and stored at −80 C for quantification of total protein. Endogenous and tracer glucose and glycerol concentrations in vivo were measured by GC-MS as previously described. 16 Glycerol in cell medium was measured in duplicate using a colorimetric kit (Sigma, Poole, UK). Protein in cell lysates was measured in duplicate using the DC protein assay (Bio-Rad, Hercules, California) and glycerol appearance corrected for total protein. FIGURE 1 Hormone and metabolite concentrations during infusions. Data are given as mean AE SEM for n = 10 for low glucocorticoid (GC) (squares), medium GC (circles) and high GC (triangles) in low insulin (dotted lines, open shapes) and high insulin (solid lines, filled shapes) groups. A, Plasma cortisol was different between all 3 GC phases (P < .001). B, Serum insulin was increased in the high insulin group (P < .01) but unchanged by GC phase. C, Glucose; D, growth hormone; E, adrenaline; F, glucagon were unchanged by GC phase or between insulin groups. ADR, adrenaline infusion 2.3.2 | Quantitative real time PCR measurements qPCR in whole adipose tissue and cultured adipocytes was performed as previously described. 17 Primer sequences and probe numbers are described in Table S2, Supporting Information. Transcript levels are presented as the ratio of the abundance of the gene of interest: mean of abundance of control genes encoding cyclophilin A and 18S.

| Kinetic analysis
Kinetic analysis for steady state (ss) measurements was performed using the mean of 5 samples obtained from t + 180 to t + 240 minutes. Steele's steady state equation 18 was used to measure the rate of appearance (Ra) of glycerol as shown in Equation 1, where d5-Glycerol TTR ss is the tracer/tracee ratio (eg, d5-Glycerol/Glycerol) during steady state: The rate of disposal (Rd) of glucose was similarly calculated using Ra Glucose ss = Rd Glucose ss − GIR ss ð3Þ Kinetic analysis following commencement of the adrenaline infusion was performed using Steele's modified non-steady state equations. 18 Ra glycerol and Rd glycerol were calculated as follows, where pV is volume of distribution: Non-steady state values for Ra and Rd Glucose were calculated as above, substituting glucose for glycerol, with the addition that the GIR was subtracted from the total Ra glucose to determine the endogenous glucose production as in Equation 3. The effective volume of distribution (pV) used for glycerol was 230 mL/kg. 19,20 For glucose, different values were tested for the pV which comprised 40, 100 and 150 mL/kg. 19 The results were not significantly altered by any of these pV values, probably because d2-glucose enrichment was not significantly altered by adrenaline infusion. The results presented for glucose kinetics are those using 100 mL/kg as the pV.

| Regulation of lipolysis by glucocorticoids in vivo
Subject characteristics are shown in Table 1. Subjects in the low and high insulin groups were of similar age, weight and blood pressure and had similar biochemical measurements.

| Baseline measurements at study visits
Cortisol concentrations were different between GC phases (all P < .01) ( Figure 1A). Fasting insulin, glucose, growth hormone, glucagon and adrenaline (Figure 1), NEFAs and glycerol (data not shown) were similar between insulin groups and unaltered by GC phase.

| Steady state measurements
Cortisol concentrations remained different between phases (all P < .001) and were similar between high and low insulin groups ( Figure 1A). Insulin concentrations were increased in the high insulin group (P < .01) ( Figure 1B). Glucose, growth hormone, glucagon and adrenaline concentrations were similar between GC phases and insulin groups ( Figure 1C-F). Concentrations of these hormones and glucose remained stable during steady state. High GC tended to increase systolic blood pressure (P = .06, Table 1).

| Effects of glucocorticoids on lipolysis and glucose kinetics
Ra glycerol and NEFA concentrations were suppressed by high insulin (Figure 2A,B). The high GC phase increased Ra glycerol and NEFAs only in the high insulin group (Figure 2A,B).
The high insulin group required more intravenous glucose to maintain glucose concentrations over the 345-minute protocol (Table 1). In the low insulin group, the medium and high GC phases reduced the required glucose infusion rate (Table 1). Endogenous glucose production (EGP) and Rd glucose were unchanged by GC phase in either insulin group ( Figure 2C,D).

| Measurements during adrenaline infusion
Adrenaline concentrations achieved during the infusion were similar to those observed during exercise 21 ( Figure 1E). Adrenaline increased heart rate and systolic blood pressure and decreased diastolic blood pressure ( Table 1). The high GC phase increased heart rate during the adrenaline infusion in both insulin groups (Table 1).

| Effects of glucocorticoids on lipolysis and glucose kinetics
Adrenaline increased Ra glycerol and NEFAs (all phases P < .001) (Figure 2A,B). High GC increased Ra glycerol and NEFAs in the high insulin group (Figure 2A,B). In contrast, high GC did not increase Ra glycerol in the low insulin group, although NEFAs were increased (Figure 2A,B). GC phase did not alter Rd glycerol (data not shown).
Adrenaline increased glucose concentrations only in the high insulin group (P < .05) ( Figure 1C) and decreased Rd glucose in both insulin groups (P < .01) ( Figure 2D). GC phase did not alter EGP or Rd glucose in either insulin group ( Figure 2C,D).

| Glucocorticoids increase expression of key genes in the lipolytic pathway
Adipose tissue was analysed in low (n = 10) and high (n = 9) insulin groups. Adipose tissue from one subject in the high insulin group was not obtained because of technical difficulties with the biopsy proce-  Figure 3A,B). Perilipin 1, G0S2 and lipoprotein lipase were increased by high insulin ( Figure 3B).

| Regulation of lipolysis by glucocorticoids in vitro
Cortisol did not increase glycerol appearance (a measure of lipolysis)  Data are given as mean AE SEM. Steady state (SS) data are the mean values obtained from t + 180 to t + 240 minutes of the infusion. Adrenaline (ADR) data are the mean values obtained from t + 285 to t + 345 minutes. Adrenaline increased heart rate and systolic BP and decreased diastolic BP (all P < .01). Total glucose infused during the protocol was increased in the high insulin group (P < .001). NEFAs = non-esterified fatty acids. *P < .05, **P < .01 vs low GC; #P < .05, ##P < .01 vs medium GC.
insulin group during adrenaline infusion, but did increase circulating NEFAs. Therefore, our study provides the first in vivo evidence that glucocorticoids have permissive effects on lipolysis in humans.
Although our data might appear to contrast with some previous work, [7][8][9] in one of those studies replacement glucagon was not infused, while we infused a lower insulin dose than the other 2 studies and, as we have shown, a "threshold" insulin dose is required to mediate the lipolytic effects of cortisol. In addition, unlike the other studies, we used metyrapone to control cortisol levels during all phases and performed longer treatment with hydrocortisone. However, metyrapone does not alter lipolysis independently of its effects on cortisol, 10 while cortisol accumulates slowly in adipose tissue, meaning that a longer duration would be more likely to identify effects on lipolysis. 22 Although previous studies have tested how glucocorticoids regulate the lipolytic pathway in rodents (by increasing ATGL and HSL), 4    , medium GC (grey columns) and high GC (black columns) for A, low insulin (n = 10) and B, high insulin (n = 9) groups. Medium or high GC increased mRNA levels of ATGL, HSL and CGI-58 and suppressed G0S2. High insulin increased perilipin-1, G0S2 and LpL mRNA levels. C,D, Data are given as mean AE SEM for C, subcutaneous and D, visceral adipocytes (both n=11) cultured for 24 hours in 0 nM (white columns), 100 nM (grey columns) or 1000 nM cortisol (black columns). While in vitro regulation of transcripts by cortisol was similar to in vivo data, cortisol also increased perilipin-1, LpL, PEDF, PDE3b and decreased MGL levels in vitro. HSL was increased by cortisol only in the subcutaneous adipocytes. *P < .05, **P < .01 vs low GC/0 nM; #P < .05, ##P < .01 vs medium GC/100 nM. $P < .05 vs low insulin group body lipolysis compared with low cortisol (c.150 nM). This suggests that acute diurnal variation in cortisol concentrations may not substantially alter lipolysis during normal physiology, although the physiological response to "clamped" cortisol levels may differ from ultradian rhythm. 34 In addition, we clamped several other hormones to prevent confounding effects, and the physiological rise in these hormones may be critical for the glucocorticoid-dependent diurnal variation in lipolysis. 10