Glucocorticoids Turn Over Slowly in Human Adipose Tissue in Vivo

Abstract

Context

Lipophilic plasma glucocorticoids are thought to gain rapid access to intracellular compartments in adipose tissue. In other organs, transport can be regulated in a steroid- and tissue-specific manner. Moreover, 11β-hydroxysteroid dehydrogenase type 1 (11βHSD1) generates additional cortisol within adipose.

Aim

The aim was to measure the rate of exchange of cortisol between plasma and adipose for comparison with rates of intracellular cortisol generation by 11βHSD1.

Participants and Interventions

With ethical approval, otherwise healthy females (n = 6) undergoing hysterectomy for benign indications were infused with tracer 9,11,12,12-[²H]₄cortisol (d4-cortisol). Adipose biopsies and peripheral venous samples were obtained during surgery after 3.9–5.5 h of infusion. Glucocorticoids were quantified using liquid chromatography tandem mass spectrometry.

Results

In plasma, d4-cortisol concentrations and appearance rates of cortisol and d3-cortisol (reflecting 11βHSD1 activity) did not change during surgery. In both omental and sc adipose, cortisol concentrations were lower than in plasma, consistent with differences in corticosteroid binding globulin, and enrichment with d4-cortisol was low (sc, 7.2 ± 0.6%; omental, 7.4 ± 0.7%; vs. plasma, 15.5 ± 1.0%). The rate of accumulation of d4-cortisol in adipose depots was 0.5 ± 0.1 (sc) and 0.4 ± 0.1 (omental) nmol/kg · h, and the proportion of intraadipose cortisol replaced each hour only 10.7 ± 1.0 and 10.4 ± 0.7%, respectively. The contribution of 11βHSD1 to this turnover could not be quantified because very little substrate d3-cortisone accumulated in adipose during infusion.

Conclusions

Slow turnover of the adipose glucocorticoid pool suggests that rapid acute fluctuations in circulating cortisol are not reflected in adipose, so that 11βHSD1 activity (previously estimated to generate 9 nmol cortisol/kg · h in sc adipose) may play a relatively important role in modulating activation of glucocorticoid receptors.

Human adipose tissue contains a glucocorticoid pool derived from both the systemic circulation and the local regeneration of cortisol by the intracellular enzyme 11β-hydroxysteroid dehydrogenase type 1 (11βHSD1) (1, 2). Intraadipose cortisol is physiologically important because it interacts with intracellular glucocorticoid receptors (GRs) to regulate gene transcription, e.g. of metabolic enzymes such as hormone-sensitive lipase and phosphoenolpyruvate carboxykinase, and to influence the balance of preadipocyte proliferation and differentiation (3). Acute elevations in plasma cortisol in vivo in humans have been associated with increased lipolysis in adipose tissue (4), although results are inconsistent (5) and vary according to the duration of glucocorticoid exposure (6). Longer-term glucocorticoid excess is associated with adipose lipid accumulation (3).

Concentrations of glucocorticoids in human sc adipose biopsies have been estimated using several techniques with contrasting results (7–10). Using highly specific liquid chromatography tandem mass spectrometry (LC-MS/MS), adipose cortisol has been reported at concentrations of 12.4 ± 0.6 nmol/kg and cortisone at 3.0 ± 0.3 nmol/kg (9). However, these methods do not measure the dynamics of uptake of cortisol by adipose tissue, which may explain discrepancies between effects after different durations of exposure. Traditionally, steroid transport across the cell membrane was assumed to be a passive process following a concentration gradient. However, an increasing number of transporters of steroids have been identified, such as the organic anion transporter-3 in the adrenal (11), the multidrug resistance P-glycoprotein in the central nervous system (12), and the glucocorticoid importer in the liver (13). Moreover, measurement of concentration within adipose does not distinguish cortisol derived from the systemic circulation from that generated locally by 11βHSD1 (1, 2); this enzyme may be important in determining glucocorticoid activity in obesity (10, 14) and is a target for pharmacological inhibition in type 2 diabetes (15).

Dynamic turnover in the intraadipose cortisol pool can be investigated using a stable isotope tracer. We developed 9,11,12,12-[²H]₄cortisol (d4-cortisol) as a tracer, dilution of which by cortisol allows quantification of cortisol production in the plasma (16) and across tissue beds using arteriovenous sampling (2). After removal of the 11-[²H] to form d3-cortisone, d3-cortisol is subsequently regenerated by 11βHSD1 reductase activity; measuring the dilution of d4-cortisol with d3-cortisol allows quantification of 11βHSD1 activity. We have now developed a protocol for efficient extraction of glucocorticoids from human adipose biopsies and studied the incorporation and dilution of d4-cortisol tracer in adipose tissue during iv steady-state infusion.

Subjects and Methods

Chemicals and reagents

Reagents were obtained from Sigma (Poole, UK) or VWR International Ltd. (Lutterworth, UK) unless otherwise specified. Solvents were glass distilled HPLC grade from Fisher Scientific (Loughborough, UK). d4-Cortisol was obtained from Cambridge Isotope Laboratories (Andover, MA), epi-cortisol from Steraloids (Newport, RI), and unlabeled hydrocortisone from Calbiochem (Nottingham, UK).

Clinical protocol

Approval from the local research ethics committee and written informed consent were obtained.

Female volunteers (n = 6) who had not received glucocorticoid treatment by any route for 3 months, who had normal hematological and renal indices, and who were undergoing an elective abdominal hysterectomy for benign gynecological indications attended between 0830 and 1030 h. Body fat was measured by bioimpedance (Body Fat Monitor BF302; OMRON Healthcare (UK) Ltd., Henfield, UK). Venous blood was collected for basal assessment of steroids and background isotopomers. An iv bolus of d4-cortisol was administered (40:60 d4-cortisol:cortisol, 3.5 mg) over 5 min, followed by continuous iv infusion (1.74 mg/h). Venous blood samples were obtained after 180- to 195-min infusion. Participants were transferred to the operating theater, where both sc and omental adipose tissue samples (~5 g) were obtained upon opening the abdomen. Venous blood was sampled at the time of obtaining the sc adipose biopsy from the anterior abdominal wall. Adipose samples were snap-frozen on dry ice and stored at −80 C before analysis. Venous plasma was stored at −20 C. The anesthesiologist avoided using agents that affect adrenal function, such as etomidate, although some participants received benzodiazepines in premedication.

Extraction of glucocorticoids

Extraction and quantitative analysis of plasma endogenous and deuterated glucocorticoids was as described previously (2), with the exception of the use of the HPLC column described below.

After extensive preliminary investigations, a protocol was established for extraction of glucocorticoids from adipose tissue, with an extraction efficiency of 70.1 ± 2.1% for internal standard (see Supplemental Methods, published on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org).

LC-MS/MS analysis

Plasma and adipose extracts were analyzed as previously described (2), but using a Biphenyl Allure column (5 μm, 10 cm, 4.6 mm, 38 C; Restek, Saunderton, UK). Ionization was performed in positive electrospray mode, and mass transitions of protonated ions (cortisone, m/z 361→163; cortisol, m/z 363→121; d4-cortisol, m/z 367→121; d3-cortisol, m/z 366→121; and d3-cortisone, m/z 364→164) determined using tandem MS (spray voltage, 3.25 kV; tube lens, 168 (cortisone), 142 (cortisol), 142 (d4-cortisol), 142 (d3-cortisol), and 168 (d3-cortisone); source temperature, 400 C; and collision gas pressure, 1.5 mTorr).

Data analysis

Plasma concentrations of steroids were determined using Xcalibur Quan Browser software (Thermo Fisher Scientific, Waltham, MA). The peak area of d4-cortisol was corrected for interference from the m+4 isotopomer of cortisol and the m+1 isotopomer of d3-cortisol. The peak area of d3-cortisol was corrected for the m+3 isotopomer of cortisol. Concentrations of cortisol and cortisone were determined from a standard curve. Plasma d3-cortisol and d4-cortisol concentrations were calculated by multiplying the concentrations of cortisol by their respective tracer:tracee ratios. Rate of appearance of cortisol and d3-cortisol in the plasma were calculated as previously described (2). Preoperative data were calculated from the mean results in four samples obtained from each participant between 180 and 195 min of d4-cortisol infusion, and intraoperative data were calculated from the mean results of three samples obtained after a variable period of infusion (234–331 min).

For adipose extracts, the limit of quantitation for the LC-MS/MS was assigned with a signal:noise ratio greater than three and peaks underwent smoothing (×5). Amounts of cortisol and cortisone were determined from a standard curve. Adipose d3-cortisol, d3-cortisone, and d4-cortisol concentrations (nanomoles per kilogram of adipose) were calculated using the calculated amount of the nondeuterated glucocorticoids multiplied by the respective tracer:tracee ratios. All adipose concentrations in mass/kilogram were converted to mass/liter by multiplying by the relative density of fat of 0.9 g/ml (17).

The rate of accumulation of d4-cortisol in adipose tissue was calculated using Equation 1:

$$\text{rate of net}\mathrm{d}4-\text{cortisol uptake into adipose}(\text{nmol}∕\text{kg}\cdot \mathrm{h})={[\mathrm{d}4-\text{cortisol}]}_{\text{adipose}}∕\text{time to biopsy},$$

assuming that no d4-cortisol was present in the adipose before infusion and that the rate of accumulation of tracer in adipose was approximately linear during infusion.

The proportion of the intraadipose cortisol pool that was replaced during d4-cortisol infusion was expressed as percentage per hour and calculated using Equation 2:

$$(\%∕\mathrm{h})=\left(\left(\frac{{(\mathrm{d}4-\text{cortisol}∕(\mathrm{d}4-\text{cortisol}+\text{cortisol}\left)\right)}_{\text{adipose}}}{{(\mathrm{d}4-\text{cortisol}∕(\mathrm{d}4-\text{cortisol}+\text{cortisol}\left)\right)}_{\text{plasma}}}\right)/\text{time to biopsy}\right)\cdot 100$$

Statistics

Data are presented as mean ± sem and were compared using paired Student’s t tests. P < 0.05 was considered statistically significant, using SPSS version 14 (SPSS Inc., Chicago, IL).

Results

Participant characteristics

Participants were aged 48.3 ± 3.8 yr, with body mass index of 30.9 ± 2.9 kg/m² and fat mass of 30.7 ± 3.5 kg. All were undergoing total abdominal hysterectomy under general anesthesia for benign gynecological indications (five for uterine fibroids, one for endometrial polyp). Two were taking progestogens (norethisterone), and one was taking oral estradiol with intrauterine progestogen (levonorgestrel). Other medications included antihypertensives (n = 3), ferrous sulfate (n = 2), mefenamic acid (n = 1), and levothyroxine (n = 1).

Plasma glucocorticoids

Deuterated and endogenous cortisol isotopomers in plasma (Fig. 1A) were in steady state by 3 h of infusion (data not shown) and were unchanged during surgery. Cortisone but not d3-cortisone concentrations fell during surgery (Fig. 1C).

FIG. 1.

Concentrations of deuterated and endogenous cortisol and cortisone in plasma pre- and intraoperatively and in adipose depots intraoperatively. A, Plasma cortisol and its isotopomers (nanomoles per liter). Gray, Cortisol; black, d4-cortisol; white, d3-cortisol. B, Adipose cortisol and its isotopomers (nanomoles per liter adipose tissue). Gray, Cortisol; black, d4-cortisol; white, d3-cortisol. C, Plasma cortisone and d3-cortisone (nanomoles per liter). Gray, Cortisone; white, d3-cortisone. D, Adipose cortisone and d3-cortisone (nanomoles per liter adipose tissue). Gray, Cortisone; white, d3-cortisone. All data are mean ± sem for n = 6. Preoperative plasma values were mean of four samples over 15 min, and intraoperative samples were mean of three samples over 10 min. Adipose values were mean of duplicates; nmol/L, nanomoles per liter. *, P < 0.05, vs. the same steroid isotopomer within each panel. All steroids were statistically significantly different in adipose vs. intraoperative plasma.

Dilution of d4-cortisol by unlabeled cortisol (from 40% in the infusate to ~16% in plasma; Table 1) was used to calculate the rate of appearance of cortisol, which did not differ during surgery (preoperative, 128.3 ± 12.2 nmol/min; intraoperative, 138.4 ± 18.5 nmol/min; P = 0.55). Dilution of d4-cortisol by d3-cortisol was used to calculate the rate of appearance of d3-cortisol, which did not differ during surgery (preoperative, 30.5 ± 2.2 nmol/min; intraoperative, 30.4 ± 3.0 nmol/min; P = 0.92).

TABLE 1.

Plasma and adipose tracer:tracee ratios and enrichments

	Plasma		Adipose
	Preoperative	Intraoperative	Subcutaneous	Omental
d4 Cortisol enrichment (%) d4-cortisol/(cortisol + d4-cortisol)	15.6 ± 0.8	15.5 ± 1.0	7.2 ± 0.6a	7.4 ± 0.7a
d3 Cortisone enrichment d3-cortisone/(cortisone + d3-cortisone)	24.5 ± 0.8	25.4 ± 1.0	3.8 ± 1.2b	5.1 ± 0.9b
Ratio of d4-cortisol:d3-cortisol	1.1 ± 0.1	1.1 ± 0.1	1.0 ± 0.1	1.0 ± 0.1
Ratio of cortisol:d4-cortisol:d3-cortisol	5.7:1:1	6.0:1:1	13.3:1.1:1	12.3:1:1

All data are presented as mean ± sem for n = 6. d, Deuterium.

^aP < 0.001;

^bP < 0.0001, vs. intraoperative plasma values using paired Student’s t tests.

Adipose glucocorticoids

Endogenous and deuterated glucocorticoids were detected in both sc and omental adipose tissue (Fig. 1, B and D). Absolute steroid levels were much lower in both adipose depots than in plasma (intraoperative plasma cortisol, 289.3 ± 49.2 nmol/liter; sc adipose cortisol, 27.4 ± 2.3 nmol/liter, P = 0.003; intraoperative plasma cortisone, 47.2 ± 4.7 nmol/liter; sc adipose cortisone, 8.6 ± 1.0 nmol/liter, P = 0.0005). Cortisone and d3-cortisone concentrations were higher in the omental compared with the sc depot (Fig. 1D).

The proportions of cortisol:d4-cortisol:d3-cortisol in adipose differed from those in plasma (Table 1). Dilution of d4-cortisol with cortisol was substantially greater in both adipose depots than in plasma, but dilution of d4-cortisol with d3-cortisol was not different. Enrichment of the cortisone pool with d3-cortisone was similarly much lower in adipose tissue than in plasma.

The estimated net rate of accumulation of d4-cortisol (Equation 1) was 0.5 ± 0.1 nmol/kg · h for sc and 0.4 ± 0.1 nmol/kg · h for omental depots (P = 0.73). The proportion of the adipose cortisol pool replaced each hour was estimated as 10.7 ± 1.0% for sc and 10.4 ± 0.7% for omental depots (P = 0.82) (Equation 2).

Discussion

These data show that concentrations of deuterated cortisol and its metabolites in human adipose tissue during prolonged steady-state iv infusion are substantially lower than the values anticipated from plasma, whereas concentrations of unlabeled cortisol and cortisone pools in adipose tissue are more than twice as high as previously reported with LC-MS/MS assays in subjects not receiving a tracer infusion (9). We found no differences in tracer accumulation or endogenous steroid levels in omental vs. sc adipose depots. We conclude that in humans, in vivo uptake of glucocorticoids into adipose tissue and turnover of the intraadipose glucocorticoid pool is slow.

This is the first report of d4-cortisol tracer infusion in women and during surgery. We found disproportionately high endogenous cortisol production but similar d3-cortisol production rates as previously reported during 40% tracer infusion (2), consistent with increased adrenal secretion but unaltered 11βHSD1 activity in association with surgery. Interestingly, this occurred in anticipation of surgery as well as during the procedure. These findings are in keeping with higher baseline corticosterone levels and greater responsiveness of the hypothalamic-pituitary-adrenal axis in female compared with male rodents (18). However, in humans, adrenocortical response is more contentious, with no sexual dimorphism after pharmacological or physical stimulation, but a greater response in men in anticipation of and during psychosocial stress (19). Although the women here were overweight/obese, this is unlikely to account for this magnitude of increase in cortisol secretion (1,2). 11βHSD1 activity does not display sexual dimorphism in healthy adults (20), in keeping with similar d3-cortisol production rates reported here in women and previously in men (2). This suggests that any up-regulation of 11βHSD1 by surgical intervention, e.g. in muscle (21), is insufficient to increase whole body cortisol regeneration acutely. We had an insufficient number of participants to test the possible effect of menopausal status on cortisol kinetics (22).

During a primed continuous infusion of d4-cortisol, steady state is achieved in plasma almost immediately for d4-cortisol, within 120 min for d3-cortisone, and within 180 min for d3-cortisol (16, 23). Uptake of steroid into adipose tissue is likely to vary in proportion to the plasma concentration. The net rate of accumulation of d4-cortisol in adipose can therefore be estimated from a single biopsy sample on the assumptions that plasma d4-cortisol concentrations were stable throughout infusion, that d4-cortisol was absent from the adipose tissue before infusion (which we confirmed on archived human sc adipose samples; data not shown), and that steady state was not achieved in adipose tissue in advance of the biopsy. To confirm the lack of steady state and estimate rates of change, it would have been useful to have serial biopsies in each patient, but this was not feasible. However, the disproportionately low concentrations of all deuterated steroids in adipose tissue compared with plasma suggest that steady state has not been reached in adipose. Net accumulation rates cannot be estimated so accurately for d3-cortisone and d3-cortisol because plasma concentrations for these steroids will have varied during the early period of infusion as they are generated from d4-cortisol in tissues.

The net rate of accumulation of d4-cortisol into adipose during infusion was estimated at approximately 0.5 nmol/kg · h (Equation 1). This figure reflects the balance between uptake into adipose tissue and any removal, for example by inactivation by enzymes or by export into plasma. The only relevant clearance enzymes expressed in human adipose tissue are 5α-reductase type 1 (24) and 11βHSD1, which under some circumstances may catalyze 11β-dehydrogenase activity, converting cortisol to cortisone (25, 26). However, a previous study in healthy men undergoing d4-cortisol tracer infusion did not detect removal of d4-cortisol across the sc or visceral adipose using arteriovenous sampling (2), which suggests that the slow rate of accumulation of d4-cortisol in adipose tissue is determined by transport between plasma and intraadipose steroid pools rather than by rapid enzymatic inactivation. This contrasts very markedly, for example, with the liver, for which we have calculated removal of d4-cortisol from previously published arteriovenous sampling data (2) to be greater than 250 nmol/h. It is possible that the slow accumulation reflects relatively low blood flow in adipose tissue [~3 ml/100 g adipose tissue/min (27) vs. ~400 ml/min in liver (2)], although this was not measured in this study, and/or tissue-specific and steroid-specific differences in active transport of glucocorticoids in and out of tissues (11–13). Although, we are not aware of any previous in vivo studies that have studied cortisol transport and turnover in adipose, one ex vivo study reported that adipose uptake of [³H]₂-cortisol was approximately five times slower than that of sex steroids (only 12.7% uptake at 3 h) (28).

Can we infer from slow accumulation of tracer d4-cortisol in adipose tissue that turnover of endogenous cortisol is also slow? There is no evidence that d4-cortisol and cortisol have different pharmacokinetics or metabolism due to a “primary isotope effect” (16), although their affinity for transporters has not been tested. Cortisol production in sc human adipose during a similar tracer infusion protocol has been quantified using arteriovenous sampling in healthy men as 9 nmol/kg · h (2). Because this is substantially higher than the estimated rate of accumulation of d4-cortisol in sc adipose tissue (~0.5 nmol/kg · h), it is likely that a component of the dilution of d4-cortisol by cortisol in adipose is attributable to 11βHSD1 activity. We can estimate (from Equation 2) that the proportion of the adipose cortisol pool that was replaced during d4-cortisol infusion was approximately 10%/h, comprising both uptake of cortisol from the plasma and local regeneration by 11βHSD1.

There is no source for d3-cortisol other than its regeneration from d3-cortisone. Thus, in steady-state conditions, dilution of d4-cortisol by d3-cortisol is the hallmark of 11β-reductase activity by 11βHSD1. We did not detect additional dilution of d4-cortisol by d3-cortisol in adipose tissue compared with plasma. However, we found very low concentrations of d3-cortisone in adipose tissue, consistent with slow uptake of cortisone as well as cortisol from the plasma, and perhaps providing inadequate substrate for 11βHSD1 activity. Moreover, given that d3-cortisol levels rise rather slowly during d4-cortisol infusion, reaching steady state after up to 3 h (16), one might anticipate that accumulation of d3-cortisol in adipose tissue would lag behind accumulation of d4-cortisol, and thus the dilution of d4-cortisol with d3-cortisol would be more marked in plasma than adipose at early time points. The fact that d3-cortisol:d4-cortisol ratios were similar in adipose as in plasma may therefore be an indication of 11βHSD1 activity in adipose tissue. Indeed, a previous arteriovenous sampling study in men has demonstrated d3-cortisol production in sc adipose tissue of approximately 5.2 nmol/kg · h (2). This figure may be an underestimate, given that the d4-cortisol infusion lasted only 210 min in the previous study, and the current data suggest that d3-cortisone substrate concentrations in adipose tissue would still have been low at the time of sampling.

The absolute concentrations of cortisol in adipose were approximately 10% those in plasma, consistent with the approximately 90% binding of cortisol to plasma proteins, making it unavailable for transport into adipose. Concentrations of cortisone were approximately 20% those of plasma, which is somewhat lower than expected given the low plasma protein binding of cortisone. This might reflect conversion of cortisone to cortisol by 11βHSD1, and the lower cortisone and d3-cortisone values in sc vs. omental adipose tissue may reflect higher enzyme activity in the sc depot, as suggested in previous studies (2). Most previous studies of concentrations of steroids in adipose tissue have relied on immunoassays (7, 8, 10) and have paid insufficient attention to matrix effects in crude tissue extracts. Others have used highly sensitive and specific LC-MS/MS (9), reporting lower steroid concentrations than we describe here, although the characteristics of the participants, biopsy method, and reliability of the steroid extraction method are unclear, making comparison with our study difficult.

These results suggest that thinking on the short-term metabolic effects of glucocorticoids needs to be revised. Cortisol accesses intracellular GR to elicit its effects on fatty acid metabolism and cellular differentiation in adipose tissue. It is widely assumed that cortisol gains rapid access to adipose tissue given its lipophilicity and induces lipolysis in the short-term and obesity in the longer term. The current data indicating slow turnover between plasma and tissue pools of cortisol are consistent with the intraadipose cortisol pool integrating the longer-term variations in circulating glucocorticoid concentrations. However, unless there are multiple pools of glucocorticoid within the adipose, with a cytosolic “free” pool that turns over more quickly than a triglyceride-bound pool, then it appears unlikely that intraadipose cortisol concentrations vary widely during acute changes in plasma cortisol. Thus, diurnal variation in intraadipose cortisol concentrations is likely to be small. This contrasts with other tissues, such as hippocampus (29) and liver (30) where, at least in rodents, tissue concentrations vary even with ultradian (hourly) fluctuations in plasma glucocorticoids. Establishing the consequences for intraadipose GR activation and transcription of GR-regulated genes, which may respond to pulsatile variation in steroid concentrations (30), will require more detailed dissection of variation in specific intraadipose glucocorticoid pools. Moreover, the rates of regeneration of cortisol by intraadipose 11βHSD1 measured in previous studies are high compared with the rate of accumulation of plasma-derived cortisol within adipose, emphasizing the importance of 11βHSD1 in human adipose glucocorticoid signaling. Differences in glucocorticoid responses between visceral and sc adipose tissue are not explained by differences in uptake of cortisol. It will be important to establish the rates of exchange between plasma and other intracellular cortisol pools in other organs.

Abbreviations

d4-cortisol

9,11,12,12-[²H]₄-Cortisol

GR

glucocorticoid receptor

11βHSD1

11β-hydroxysteroid dehydrogenase type 1

LC-MS/MS

liquid chromatography tandem mass spectrometry