Evidence That the 11 β-Hydroxysteroid Dehydrogenase (11 β-HSD1) Is Regulated by Pentose Pathway Flux

11 β-hydroxysteroid dehydrogenase type 1 (11 β-HSD1) catalyzes the interconversion of biologically inactive 11 keto derivatives (cortisone, 11-dehydrocorticosterone) to active glucocorticoids (cortisol, corticosterone) in fat, liver, and other tissues. It is located in the intraluminal compartment of the endoplasmic reticulum. Inasmuch as an oxo-reductase requires NADPH, we reasoned that 11 β-HSD1 would be metabolically interconnected with the cytosolic pentose pathway because this pathway is the primary producer of reduced cellular pyridine nucleotides. To test this theory, 11 β-HSD1 activity and pentose pathway were simultaneously measured in isolated intact rodent adipocytes. Established inhibitors of NAPDH production via the pentose pathway (dehydroandrostenedione or norepinephrine) inhibited 11 β-HSD1 oxo-reductase while decreasing cellular NADPH content. Conversely these compounds slightly augmented the reverse, or dehydrogenase, reaction of 11 β-HSD1. Importantly, using isolated intact microsomes, the inhibitors did not directly alter the tandem microsomal 11 β-HSD1 and hexose-6-phosphate dehydrogenase enzyme unit. Metabolites of 11 β-HSD1 (corticosterone or 11-dehydrocorticosterone) inhibited or increased pentose flux, respectively, demonstrating metabolic interconnectivity. Using isolated intact liver or fat microsomes, glucose-6 phosphate stimulated 11 β-HSD1 oxo-reductase, and this effect was blocked by selective inhibitors of glucose-6-phosphate transport. In summary, we have demonstrated a metabolic interconnection between pentose pathway and 11 β-HSD1 oxo-reductase activities that is dependent on cytosolic NADPH production. These observations link cytosolic carbohydrate flux with paracrine glucocorticoid formation. The clinical relevance of these findings may be germane to the regulation of paracrine glucocorticoid formation in disturbed nutritional states such as obesity.

The intracellular peri-receptor availability of glucocorticoids is not determined simply by their circulating concentrations and protein binding interactive kinetics. Ostensibly, the intracellular concentration of the active glucocorticoids (cortisol, corticosterone) is governed more so by 11 ␤-hydroxysteroid dehydrogenase type 1 (11 ␤-HSD1), 2 a bidirectional enzyme that facilitates the equilibrium between the aforesaid active steroids and their biologically inactive 11-keto derivatives (cortisone, 11-dehydrocorticosterone) (1-3). 11 ␤-HSD1 is ubiquitous (4), located in microsomes (5), and has a low substrate affinity (6) (K m in M) involving pyridine nucleotide cofactors, with NADP(H) having a greater affinity than NAD(H) (6,7). Recently, 11 ␤-HSD1 has garnered attention as a potential participant in the pathoetiology of obesity, insulin resistance, and type II diabetes (2,3,8,9). Specifically, the dysregulation of 11 ␤-HSD1 in particular tissues may augment intracellular active glucocorticoid concentrations. In addition, it is certainly plausible that in human obesity, although normal circulating blood cortisol levels are found, intracellular cortisol concentrations may be elevated, conceivably because an alteration in the NADPH/NADP ratio may foster 11 ␤-HSD1 reductase over dehydrogenase. Given the well recognized regulation by glucocorticoids of numerous homeostatic and metabolic processes of intermediary metabolism, the role of 11 ␤-HSD1 in obesity is under question (2, 8, 10 -19). In addition to modulating several key intermediary enzymes, cortisol is pivotal in adipogenesis by promoting the differentiation of stromal cells (preadipocytes to adipocytes) (9 -11, 20 -22).
In vitro studies in intact cells from liver, adipose, and other tissues have consistently found the directionality of 11 ␤-HSD1 is predominantly 11␤ oxo-reduction of inactive cortisone to cortisol (1,9,(23)(24)(25)(26). Interestingly, in human stromal cells from the omentum, upon differentiation to mature adipocytes, the 11 ␤-HSD1 activity converts from a dehydrogenase to an oxo-reductase (9). Antithetically, in disrupted cellular homogenates, the direction is reversed, where the enzyme functions preferentially as a dehydrogenase (6,23,26). The grounds for this paradox are poorly understood, but obviously substrate concentrations, ionic milieu, pH, intracellular location, and membrane binding may explain the kinetic discordance between intact cells versus homogenates (where reactions are often measured under artificial V max conditions). All these factors notwithstanding, the most plausible determinate of the direction of 11 ␤-HSD1 is the mass action effect of pyridine nucleotide cofactors. To date, no studies using intact cells have attempted to correlate ongoing in situ pentose pathway flux, which is recognized as the major intracellular producer of NADPH, with the simultaneous measurement of 11 ␤-HSD1 activity. Furthermore, to what extent the hormonal and metabolic manipulation of adipocyte pentose pathway (PP) may impact in situ microsomal 11 ␤-HSD1 activity has yet to be explored.
Rat Adipocyte Preparation-Isolated adipocytes were prepared from gonadal depots by standard collagenase digestion methods (27).
Rodent Fat Cell Incubation Conditions-Isolated fat cells from 4 -6 rats were pooled and distributed for metabolic studies into incubation tubes (three sets of duplicated tubes; set 1, measure PP flux with [1-14 C] glucose, set 2, measure [6-14 C] glucose oxidation (as below), and set 3, measure 11 ␤-HSD1 activity). All cell groups were distributed, handled, and analyzed in parallel under identical incubation conditions (37°C). Isolated rat fat cells were incubated in 1-2-ml polypropylene tubes (set in 20-ml vials with Teflon/silicone membrane caps for CO 2 collection, PP as below).
Rat Microsome Preparation-Liver microsomes were prepared according to the methods of Raucy and Lasker (28). Adipocyte microsomes were prepared from the pooled fat cells of three to four rats by similar methods (29).
11 ␤-HSD1 Activity-The oxo-reductase direction of 11 ␤-HSD1 in which dehydrocorticosterone (11-DHC) is converted to corticosterone (C) was measured in intact cells and microsomes. Isolated rodent fat cells (0.25 ml) were incubated in 0.1 mM glucose Krebs buffer at 37°C (final assay volume 0.4 ml, final cell concentration 5-6 ϫ 10 5 ml or with microsomes 50 g of protein/ml). The reaction was initiated with 250 -1000 nM 11-DHC (0.15 ml), incubated for 45-90 min, and terminated by freezing on dry ice. The appearance of C in the assay medium (conditioned medium) of isolated intact cells was determined by 3 H radioimmune assay of corticosterone (MP Biomedicals, Irving, CA). Baseline studies examined the effect of substrate concentration (0 -2 M 11-DHC), cell number (0-5-6 ϫ 10 6 fat cells/ml) and time (0 -120 min) on corticosterone production to verify consistent linear rates of 11 ␤-HSD1 activity throughout the assay. The dehydrogenase reaction of 11 ␤-HSD1 (C 3 11-DHC) was measured in intact cells under identical assay conditions by the disappearance of [ 3 H] corticosterone (as above) at a final concentration of non-labeled substrate of 100 nM.
Pentose Pathway Flux-The conversion of [1-14 C]glucose to 14 CO 2 followed previously described methods for 14 CO 2 collection and quantification (30 -32). To correct PP flux for minor 14 CO 2 release from [1-14 C]glucose generated through the glycolytic/citric acid cycle (33)(34)(35), the release of 14 CO 2 from [6-14 C] was also determined in parallel experiments. The incubation medium was a 0.1-mM glucose Krebs buffer, pH 7.4. Metabolic reactions were started by the addition of assay mixture containing substrate ([1-14 C]glucose or [6-14 C]glucose). Methods for this latter assay were identical to the PP assay with the exception that [6-14 C]glucose replaces [1-14 C]glucose.
Glucose Dehydrogenase-Isolated microsomes were tested for glucose dehydrogenase activity in the presence and absence of CHAPSO (a zwitter ionic, non-denaturing detergent) as an index for microsomal membrane integrity. To mimic our experimental conditions, microsomes were first preincubated for 90 min at 37°C in Krebs buffer. Next, they were washed once with 0.15 M Tris-HCL, pH 8.0 and incubated for 30 min at 4°C with 0.8% CHAPSO. Glucose dehydrogenase activity was then monitored over 180 min at 25°C in this same Tris buffer with 50 mM Na 2 SO 4 (36,37).
Statistical Analysis-Data are expressed as a % control (no modifier), mean Ϯ S.D. (n ϭ 3-4). In each case, "n" refers to an individual experiment performed on 1 day with a fresh batch of cells or microsomes. Statistical differences were determined by paired t-test.

RESULTS
Baseline Characteristics of 11 ␤-HSD1 Oxo-reductase Assay-The 11 ␤-HSD1 oxo-reductase assay was measured in both isolated rat adipocytes and liver or fat microsomes. The reaction was linear over time  1D) activities by 54 -60%. Importantly, reaction rates were linear and cell counts as well as lactate dehydrogenase release (cellular integrity) were stable (Ͼ90% baseline) with or without either inhibitor. It is noteworthy that DHEA did not directly affect 11 ␤-HSD1 enzyme activities in ruptured or homogenized fat cells. To determine whether the inhibition of these agents directly affected hexose-6-phosphate dehydrogenase (H6PD), which is the microsomal counterpart of cytosolic G6PD, isolated liver microsomes were incubated with and without DHEA or NE and 11 ␤-HSD1 oxo-reductase measured. As shown in Fig.  2, neither agent altered microsomal 11 ␤-HSD1 oxo-reductase. To determine whether the inhibitory actions of DHEA and NE on 11 ␤-HSD1 (as shown in Fig. 1, B and D) were enzyme specific, the reverse reaction (11-dihydrocorticosterone formation) was measured with and without these compounds. If these agents inhibit 11 ␤-HSD1 oxo-reductase by curtailing the supply of NADPH, then, conversely, they should stimulate the dehydrogenase direction. Both agents stimulated the dehydrogenase reaction by 2.5-3-fold (Fig. 3), although the effect of NE was not statistically significant. Metabolites of the 11 ␤-HSD1 Enzyme Reaction Regulate Pentose Pathway Activity in Isolated Fat Cells-If PP and 11 ␤-HSD1 are metabolically linked, then the substrate/products of the 11 ␤-HSD1 reaction might in turn affect PP flux. To examine this possibility, fat cells were preincubated with either 10 M corticosterone or 11-DHC and then pentose pathway flux measured over 45 min. As predicted from Fig. 10, corticosterone, which would promote dehydrogenase flux (stimulates 11-DHC formation and an increased NAPDH/NADP ratio), inhibited PP by 30% whereas 11-DHC (stimulates corticosterone formation and, hence, reduces the NAPDH/NADP ratio) stimulated PP by 23% (Fig. 4).
Microsomal Membrane Intactness Demonstrated by the Latency of Glucose Dehydrogenase Activity-The intactness of our experimental microsome preparation was tested by measuring glucose dehydrogen-ase activity following preincubation with and without the detergent membrane solubilizer (0.8% CHAPSO for 30 min at 4°C). Glucose dehydrogenase is an NADP-requiring intraluminal microsomal enzyme whose activity is largely undetectable in intact microsomes (36). The results show that untreated microsomes were largely intact (no detectible glucose dehydrogenase activity over a 180-min incubation period) (Fig. 6).
DHEA and Norepinephrine Reduce Adipocyte NAPDH Content-If DHEA and norepinephrine regulate 11 ␤-HSD1 oxo-reductase indirectly by inhibiting PP, then these inhibitors should, consequently, diminish adipocyte NADPH content. Therefore, the cellular content of this pyridine nucleotide phosphate was measured after 45 min of preincubation with either 100 M DHEA or 1 M NE (Fig. 8). The respec-    tive results showed 28 Ϯ 1.4 and 47 Ϯ 15% reductions in total NADPH (p Ͻ0.05 versus control, no additions). As stated previously, both cell count and lactate dehydrogenase release were measured to assess cell integrity; neither compound had any effect.
NADPH Increases Liver Microsomal 11 ␤-HSD1 Activity-Intact liver microsomes were preincubated with various pyridine nucleotides for 60 min and 11 ␤-HSD1 oxo-reductase activity measured over 45 min. NADPH caused significant (Ͼ7-fold) increase in the activity of this intraluminal enzyme, whereas NADH had a lesser activating effect (166 Ϯ 15% of basal). On the other hand, NADP was inhibitory (74 Ϯ 6% basal) (Fig. 9). These data, along with the evidence for membrane intactness (Fig. 6), support that the microsomal membrane is not completely impermeable to NADPH over a prolonged incubation.

DISCUSSION
Numerous studies have commented on the enigmatic directionality of microsomal 11␤-HSD1, seemingly dependent on the enzyme status: in intact cells oxo-reductase predominates, but when the enzyme is studied under cell-free conditions, the dehydrogenase direction prevails.
Indirect evidence for the necessity of a renewable source of NADPH to sustain microsomal 11␤HSD-1 reductase is manifest in patients with cortisol reductase deficiency (39). The latter results in inadequate regeneration of cortisol, subsequent overstimulation of adrenocorticotropin release, and a chronic mild androgen excess that promotes the polycystic ovarian phenotype. These patients have triallelic digenic mutations, ostensibly dosage dependent, involving not only 11␤-HSD1 but also H6PD, the microsomal counterpart of cytosolic glucose-6phosphate dehydrogenase. Notably, H6PD is a bifunctional enzyme that also has 6-phosphogluconolactonase catalytic activity; hence, it contains the oxidative portion of the pentose pathway (40). Not only are these patients deficient in 11 ␤-HSD1 but they have a concomitant decrease in H6PD activity; therefore, impaired microsomal NADPH production nullifies any residual reductase activity. Another recent article buttresses the importance of luminal H6PD in the regulation of 11 ␤-HSD1. In human embryonic kidney 293 cells, in which endogenous expression of theses two enzymes is normally scant, co-expression studies revealed co-localization of these two enzymes in the ER. In addition, enhanced expression of H6PD stimulated 11 ␤-HSD1 reductase activity (41).
In our fat cell experiments, after the targeted inhibition of G6P dehydrogenase with the non-competitive inhibitor DHEA, the resultant 56% reduction in adipocyte pentose flux (Fig. 1A) produced a significant attenuation of the oxo-reductase activity of 11 ␤-HSD1 (Fig. 1B). Cellular NADPH content likewise declined (Fig. 8). In addition, if there is a kinetic linkage between pentose pathway flux 7 NADPH 7 11 ␤-HSD1, this reduction in NADPH by DHEA should promote the dehydrogenase direction for 11 ␤-HSD-1. Indeed, DHEA caused a significant 187% increase (Fig. 3). Further experimental corroboration for this link- FIGURE 6. Effect of CHAPSO detergent on the latency of liver microsomal glucose dehydrogenase activity. Isolated liver microsomes (250 g/ml) were preincubated with Krebs buffer for 90 min at 37°C and then treated without or with CHAPSO before the glucose dehydrogenase assay was performed (see "Experimental Procedures"). Glucose dehydrogenase activity is represented as A 340 over time (mean Ϯ S.D., n ϭ 3). Of note, there was no detectible glucose dehydrogenase activity in the absence of glucose or microsomes. Differences (control versus CHAPSO) were statistically significant by paired t-test at all time points (p Ͻ0.001).   age was borne out in the NE studies. Prior reports (34) have confirmed that this hormone potently attenuates PP flux in adipocytes, most likely by stimulating glycolysis. In our fat cell studies, norepinephrine caused a 54% decrease in PP flux (Fig. 1C), 47% reduction in NADPH content (Fig. 8), and a concomitant 40% reduction in 11 ␤-HSD1 reductase (Fig.  1D). Moreover, as in the case with DHEA, the antithetical dehydrogenase direction was enhanced 141% by NE (Fig. 3). In summary, regardless of the dissimilar mechanism by which PP flux was curtailed, similar directionality changes in 11 ␤-HSD1 were observed.
To further probe the hypothesized metabolic linkage between cytosolic PP and 11 ␤-HSD1 (Fig. 10), we examined whether metabolites of the 11 ␤-HSD1 reaction might alter PP flux in adipocytes. Based on their anticipated effects on cellular NADPH, then corticosterone (increases NADPH) should decrease PP, whereas 11-DHC (reduces NADPH) should increase PP. As predicted, these outcomes were confirmed in isolated fat cell incubations (Fig. 4). These findings again support a kinetic inter-connection between PP and 11 ␤-HSD1.
Despite a sizable (6-fold), and well known (34,42,43) stimulation of PP flux with insulin, no net effect on 11 ␤-HSD1 oxo-reductase was found (data not included). Interestingly, the intracellular NADPH content did not increase but actually declined by 23 Ϯ 6% (n ϭ 4) following the robust increase in PP activity. Indeed, in an earlier study (44) in rat adipose tissue, insulin treatment caused a nearly 40% reduction in the NADPH/NADP ratio despite a 1.5 increase in G6P dehydrogenase activity, the rate-limiting pentose pathway enzyme. Presumably, at least in adipocytes, the simultaneous robust activation of lipogenesis by insulin (NADPH loss) overrides the stimulation of PP (NADPH gain) such that an overall decrease in NADPH content occurs. Another tenable explanation is that the insulin-induced reduction in cellular NADPH, as it is consumed in fat synthesis, may disinhibit the PP (44 -48). Hence, insulin plays an indirect role in enhancing PP flux.
Importantly, the two compounds (DHEA, NE) that attenuated PP flux in the fat cells had no direct effect on 11 ␤-HSD1 in isolated microsomes nor did they alter the stimulation of 11 ␤-HSD1 reductase by added glucose-6-phosphate (Fig. 2). This latter observation infers that neither DHEA nor NE directly affected HGPD. Others have likewise reported no adverse direct effect of DHEA on microsomal H6PD (49,50). Interestingly, an inhibition of 11 ␤-HSD1 reductase by DHEA was previously reported in human myoblasts but no explanation was opined (51); it is quite likely that the PP was inhibited in these cells as in adipocytes. In a just-published study using intact differentiated 3T3-L1 adipocytes, after a 48-h chronic incubation with DHEA a dose-dependent inhibition of 11 ␤-HSD1 reductase gene expression and activity was observed (52). Nevertheless, no acute (30 min) effect of 100 -200 M DHEA on this enzyme in control adipocytes was observed. But similar to our studies, no direct inhibition of 11 ␤-HSD1 reductase activity by DHEA in human embryonic kidney 293 cell lysates was found; thus, it was concluded that the DHEA inhibitory action on 11 ␤-HSD1 was solely due to attenuated gene expression.
Given that the catalytic site of 11 ␤-HSD1 resides in the lumen of the endoplasmic reticulum (ER) (5), one concern about the role of cytosolic PP is that the subcellular aqueous pool of pyridine nucleotides governed by this pathway would not impact endoplasmic 11 ␤-HSD1. However, intact rat liver peroxisomes are permeable to NADϩ (53). Even more relevant, studies exploring which subcellular substrates are available for ER luminal H6PD concluded that the intact untreated rat microsomal membrane was permeable to NADP but not glucose (36,54). An earlier report (55) found minimal ER (ϳ16%) penetrance by NADP, whereas in FIGURE 10. Schematic illustration of proposed metabolic interconnection between cytosolic pentose pathway and 11 ␤-HSD1 in the endoplasmic reticulum (ER). Glucose enters the cell via a plasma membrane transporter (T) and is phosphorylated to glucose-6-phosphate (G-6-phosphate). This hexose phosphate is metabolized via glycolysis or the cytosolic oxidative pentose pathway (wherein NADP is converted to NADPH), or it may also be transported via its specific transporter (T) directly into the ER. NADPH produced in the cytosol accesses the ER via an uncertain mechanism, graphically depicted here as a pore for the sake of simplicity. This cytosolic-derived NADPH is then utilized by the intraluminal 11 ␤-HSD1 oxo-reductase to convert the biologically inert 11 dehydrocorticosterone (or cortisone in humans) to biologically active corticosterone (or cortisol in humans). Also depicted is hexose-6-phosphate dehydrogenase (H6PD), residing proximate to 11 ␤-HSD1, which can locally produce NADPH within the ER space from transported G-6-phosphate.
another study it was found that NADP can slowly enter liver microsomes (56). On the other hand, a recent study demonstrating cooperativity between 11 ␤-HSD1 and H6PD in rat liver microsomes found evidence of meager, if any, NADP trans-membrane influx (57). In this report, the calculated ER latency of H6PD was 90%.
As to the tightness of the ER membrane barrier, this issue is beclouded by manifold methodological approaches including: use of permeabilized cells, differences in microsomal preparations and cell types, salt concentrations, duration of exposure to study compound, etc. To the point, a detailed study (58) in HeLa cells disclosed that the influx across the ER membrane of small (molecular mass 557 Da), nonphysiologic, polar reagents is surprisingly brisk. This report did not study compounds of slightly higher molecular mass such as NADP. Although the ER membrane was quite leaky, both the plasma and lysosomal membranes were impermeable to these same compounds (58). Indeed, the ER translocon pore is estimated to be 40 -60 Å, allegedly the largest diameter hole in a relatively impermeable eukaryotic membrane (59). One difference in some of the aforementioned studies and our series of experiments was the duration of time that the microsomes were exposed to the pyridine nucleotides and the duration of time the fat cells were exposed to the DHEA and NE. For example, in native microsomes, G6P stimulated 11 ␤-HSD1 (by providing luminal NADPH) but NADPH per se failed to do so over a 5-min time course (57). In our intact microsomal studies 60 min of preincubation with NADPH (1 mM) activated 11 ␤-HSD1 during the 45-min assay interval by Ͼ7-fold (Fig. 9). The reason for the discordance with our observations is unknown, but the incubation time may be germane. All this notwithstanding, in whole cells the manipulation of cytosolic PP activity with compounds that had no effect on H6PD or 11 ␤-HSD1 in isolated microsomes caused a significant attenuation in 11 ␤-HSD1 (Fig. 1, B  and D).
In our experiments in intact fat cells the measurement of pentose pathway flux represents both the cytosolic and microsomal (40,60) oxidative enzymes. Radiolabeled glucose-6-phosphate is generated intracellularly by hexokinase, and this compound can be taken up by microsomes. Consequently, our metabolic studies cannot discern to what extent the oxidative pentose flux is transpiring in the cytosol versus ER. However, quantitatively, as reported in rat tissue, the pentose pathway enzymes in the cytosol far exceed those in the ER (only ϳ1.5% total PP) (61). How then to reconcile for 11 ␤-HSD1 the relative roles of cytosolic PP flux versus the G6P-H6PD reaction in ER? Although not addressed in this study, the nugatory impact of changes in the intracellular G6P content on 11 ␤-HSD1 oxo-reductase activity remains puzzling. For example, insulin and vanadate have been shown to cause a profound increase in intracellular adipocyte G6P that, in turn, should enhance microsomal 11 ␤-HSD1 reductase. In our experiments neither vanadate nor insulin altered 11 ␤-HSD1 reductase (data not shown, and Ref. 62). It is unclear why abrupt and marked increases in cellular G6P (after insulin or vanadate) do not change 11 ␤-HSD1 oxo-reductase as would be expected based on the isolated microsome data. Insofar as the microsomal membrane is not completely impermeable to NADP, and given the large difference in nucleotide pool size between cytosol and microsomal aqueous spaces, the pentose pathway activity in the cytosol could conceivably assume a regulatory role. Obviously, segregated cytosolic pools, compartmentalization, substrate channeling, or enzyme clustering may be at play in intact cells (31).
To test the tenet that the cytosolic and ER pools of NADPH are not unconditionally separate, we examined the effect of added substrate and cofactors using intact, isolated microsomes from rat liver and fat. As expected, G6P stimulated 11 ␤-HSD1 activity (Fig. 5). Intactness was established by the latency of glucose dehydrogenase because microsomes are impermeable to the ingress of glucose, i.e. no glucose dehydrogenase activity is detectible unless microsomes are detergent treated (Fig. 6). In our studies, as previously reported, pyridine nucleotide cofactors gained access to intramicrosomal 11 ␤-HSD1 (Fig. 9). Additionally, intactness of the microsomal membrane was further substantiated because there was a near obliteration of the stimulation of 11 ␤-HSD1 oxo-reductase activity by the G6P-specific translocase inhibitors chlorogenic acid (Fig. 7) and phlorizin (63). In terms of 11 ␤-HSD1 reductase activity, both liver and fat microsomes displayed similar kinetic responses to NADPH, NADH, and G6P.
In conclusion, pentose pathway activity modifies 11 ␤-HSD1 activity in adipocytes, thereby linking carbohydrate metabolism to glucocorticoid production. This premise is supported by the following observations. (i) Curtailing adipocyte pentose flux with DHEA or NE reduces 11 ␤-HSD1 oxo-reductase and augments dehydrogenase activity. Neither of these compounds had a direct effect on microsomal 11 ␤-HSD1. (ii) The addition to fat cells of 11 ␤-HSD1 substrates, such as corticosterone (which generates NADPH) or 11-dehydrocorticosterone (which generates NADP), had the predicted directional effect on pentose pathway. (iii) NADPH has some access to the microsomal membrane as evidenced by the marked stimulation of 11 ␤-HSD1 oxo-reductase induced by this compound in isolated microsomes. This study suggests a constitutive metabolic interplay between intermediary carbohydrate metabolism and glucocorticoid production.