Hexose-6-phosphate Dehydrogenase Knock-out Mice Lack 11β-Hydroxysteroid Dehydrogenase Type 1-mediated Glucocorticoid Generation*

The local generation of active glucocorticoid by NADPH-dependent, 11β-hydroxysteroid dehydrogenase type 1 (11β-HSD1) oxoreductase activity, has emerged as an important factor in regulating hepatic glucose output and visceral adiposity. We have proposed that this NADPH is generated within the endoplasmic reticulum by the enzyme hexose-6-phosphate dehydrogenase. To address this hypothesis, we generated mice with a targeted inactivation of the H6PD gene. These mice were unable to convert 11-dehydrocorticosterone (11-DHC) to corticosterone but demonstrated increased corticosterone to 11-DHC conversion consistent with lack of 11β-HSD1 oxoreductase and a concomitant increase in dehydrogenase activity. This increased corticosterone clearance in the knock-out mice resulted in a reduction in circulating corticosterone levels. Our studies define the critical requirement of hexose-6-phosphate dehydrogenase for 11β-HSD1 oxoreductase activity and add a new dimension to the investigation of 11β-HSD1 as a therapeutic target in patients with the metabolic syndrome.

Corticosteroid hormone action is modulated in a tissue-specific fashion by the expression and activity of two isozymes of 11␤-hydroxysteroid dehydrogenase (11␤-HSD) 3 that interconvert hormonally active cortisol (or corticosterone in rodents) to their inactive derivatives, cortisone/11-dehydrocorticosterone (11-DHC) (1). 11␤-HSD2 acts as a NAD-dependent dehydrogenase, inactivating glucocorticoids and protecting the mineralocorticoid receptor from illegitimate activation by cortisol in mineralocorticoid receptor-rich tissues such as the kidney and colon (2).
By contrast, 11␤-HSD1 is a bidirectional enzyme expressed in liver and adipose tissue. It catalyzes both oxidation and oxoreduction of glucocorticoids but acts in vivo predominantly as a NADPH-dependent oxoreductase (1). Independent of circulating glucocorticoid concentrations, the local generation of cortisol/corticosterone by 11␤-HSD1 has emerged as an important factor in regulating hepatic glucose output (by augmenting gluconeogenesis) and visceral adiposity (by increasing adipocyte differentiation). Transgenic mice overexpressing 11␤-HSD1 in liver and adipose tissue recapitulate features of the metabolic syndrome with visceral obesity, hepatic steatosis, glucose intolerance, insulin resistance, hyperlipidemia, and hypertension (3)(4)(5). In contrast, knockout (KO) mice lacking 11␤-HSD1 show improved glucose tolerance, enhanced insulin sensitivity, and reduced weight gain when fed a high fat diet (6,7).
11␤-HSD1 has therefore emerged as a novel therapeutic target to treat patients with obesity and insulin resistance. Indeed, selective 11␤-HSD1 inhibitors improve glucose tolerance in diabetic mice (8 -10).
These in vivo findings are dependent upon the oxoreductase activity of 11␤-HSD1, which is puzzling because activity studies performed on the purified 11␤-HSD1 enzyme or on cell/tissue homogenates indicate almost exclusive unidirectional dehydrogenase activity (11). We hypothesized that the enzyme hexose-6-phosphate dehydrogenase (Hex-6-PDH) is a key factor in conferring oxoreductase activity upon 11␤-HSD1. Hex-6-PDH is a bifunctional enzyme that catalyzes the first two steps of an endoluminal pentose phosphate pathway (12) but is distinct from its cytosolic homologue, glucose-6-phosphate dehydrogenase, in being localized exclusively to the endoplasmic reticulum (ER) lumen. Hex-6-PDH has broad substrate specificity, accepting a variety of hexose 6-phosphates (13,14) and has dual specificity for NAD ϩ and NADP ϩ . However, under physiological conditions, the substrates within the ER are thought to be glucose 6-phosphate (Glc-6-P) and NADP ϩ . Supply of Glc-6-P is ensured by the glucose 6-phosphatetransporter of the ER, but because the ER membrane is relatively impermeable to pyridine nucleotides, supply of NADP ϩ is maintained through a functional cooperation with endoluminal reductases such as 11␤-HSD1.
Rare patients have apparent cortisone reductase deficiency, with decreased urinary excretion of cortisol metabolites relative to those of cortisone. Affected females have increased adrenal androgen levels and a clinical presentation resembling polycystic ovarian syndrome. Although it has been proposed that cortisone reductase deficiency has a digenic etiology, requiring mutations in both the HSD11B1 and H6PD genes (1,15), subsequent studies have demonstrated that polymorphisms in these genes are relatively common and are associated with neither abnormalities in urinary excretion nor an increased risk of polycystic ovarian syndrome (16,17). Subsequently, a series of in vitro studies have demonstrated close cooperativity between 11␤-HSD1 and Hex-6-PDH in preparations of rat liver microsomes with manipulation of Hex-6-PDH expression directly altering the set point of 11␤-HSD1 activity (18 -21). To date, however, little direct in vivo evidence exists as to the effects of Hex-6-PDH upon 11␤-HSD1 activity. Here, through targeted gene inactivation in mice, we demonstrate that Hex-6-PDH inactivation profoundly affects 11␤-HSD1 enzyme activity.
Design-All animal experiments had the approval of the Institutional Animal Care and Use Committee and were performed according to procedures approved by that committee. Mice were housed in standard conditions on a 12-h light/12-h dark cycle with access to standard rodent chow and water ad libitum.
Generation of a Targeted H6PD Gene-The murine H6PD gene spans ϳ15 kb and contains 5 exons. Genomic DNA from 129SvJ embryonic stem cells was used to amplify 5.5-kb 5Ј and 2-kb 3Ј homology arms, which were subsequently cloned into pBluescript SK(ϩ) containing both PGK-Neo and thymidine kinase cassettes. The targeting vector was designed to replace exons 2 and 3. Following verification by DNA sequencing, the construct was linearized with SacII and electroporated into E14TG2a mouse embryonic stem cells. Southern hybridization of NcoI-digested genomic DNA probed with a 500-bp H6PD fragment located at the 5Ј and 3Ј ends of the targeting vector identified cells positive for a recombined H6PD locus after selection in G418 and gancyclovir. Three targeted embryonic stem cell clones were expanded, screened with Neo and 3Ј probes, and karyotyped to ensure correct recombination and chromosomal integrity. Two clones were injected into C57BL/6 blastocysts to produce chimeric mice. Chimeric mice derived from embryonic stem cell clone were mated with C57BL/6 females to achieve germ line transmission of the mutant allele. From these, heterozygote mice were intercrossed to generate WT, heterozygote and KO mice. Genotypes were routinely monitored using PCR and Southern blotting. For blotting, DNA was extracted from tail biopsy, digested with NcoI, and probed with the 3Ј probe as above. For genotyping the following primers were multiplexed in a standard PCR reaction: P1, 5Ј-CTTGTCACTCTGTCTGTCACTGTGG-3Ј; P2, 5Ј-ACCATGTG-GCCTTGTGCCTG-3Ј; P3, 5Ј-CTATGGCTTCTGAGGCGGAAAG-3Ј.
Microsome Preparation-WT and KO liver microsomes were prepared from 13-16-week-old male mice, resuspended in MOPS buffer (100 mM KCl, 20 mM NaCl, 1 mM MgCl 2 and 20 mM MOPS, pH 7.2), aliquoted, and rapidly frozen in liquid nitrogen prior to long term storage. The integrity of the microsomes was determined using the mannose-6-phosphatase assay and was Ͼ90% in all preparations. Protein concentration of the microsomes was established using the Bio-Rad protein assay.
Immunoblotting-SDS-PAGE was performed by the method of Laemmli (23) with 2.5 and 5 g of mouse microsomal liver protein on 10% acrylamide minigels using a Bio-Rad Mini-Protean II apparatus. Following electrophoresis, proteins were transferred to Immobilon-P membrane (Millipore Corp., Bedford, MA). Nonspecific protein binding was blocked by incubating membranes in 20% nonfat milk, 0.1% Tween 20 in phosphate-buffered saline at 25°C for 1 h. Membranes were then incubated with a polyclonal antibody to human Hex-6-PDH at a dilution of 1:500 for 16 h at 4°C. Following three 10-min washes in phosphatebuffered saline, 0.1% Tween 20, membranes were incubated with secondary antibody (goat anti-rabbit IgG peroxidase conjugate (Dako, Glostrup, Denmark) at a dilution of 1:10,000 for 1.5 h at room temperature. Bound peroxidase-conjugated IgG was visualized using an ECL detection kit (Amersham Biosciences) and exposing membranes to x-ray film (Kodak).
Enzyme Assays-Hex-6-PDH enzyme activity was measured by spectrophotometric detection of NADPH using absorbance readings taken at 340 nm using an Ultrospec 2100pro spectrophotometer (Amersham Biosciences). Microsomes were permeabilized with 0.5% Triton X-100 and incubated in MOPS buffer, pH 7.2, at 22°C in the presence of 1 mM of the Hex-6-PDH-specific substrate glucose 6-sulfate and 0.4 mM NADP ϩ . Absorbance readings were taken at 1-min intervals for 5 min.
11␤-HSD1 oxoreductase and dehydrogenase activities were measured using both intact and permeabilized microsomes. Microsomes (200 g) were incubated at 37°C for 20 min in MOPS buffer in the presence of 2 M 11-dehydrocorticosterone/corticosterone added to 20,000 cpm of tritiated 11-dehydrocorticosterone/corticosterone and the Hex-6-PDH substrate Glc-6-P at 1 mM. All experiments were performed in quadruplicate.
Analysis of Urinary Steroid Metabolites-Urine was collected on filter paper spread in the metabolic cages, and samples were pooled from five members of each group (WT female, WT male, KO female, and KO male mice). The filter paper was cut up and eluted with water while being vortexed and sonicated. The water extracts were subjected to steroid analysis employing well described methods (24,25). In brief, the initial steroid recovery was by solid-phase extraction using Waters Sep-Pak cartridges. The dried extracts were subject to conjugate hydrolysis using Helix pomatia (Roman snail) digestive juice. Steroids freed from sulfate and glucuronide conjugation were extracted once again by Sep-Pak. Preliminary experiments showed that the extracts were too complex and contaminated for immediate analysis, so they were purified by Sephadex LH-20 chromatography using 0.5-g columns and a cyclohexane:ethanol (4:1,v/v) solvent system. Three steroid-containing fractions were analyzed for corticosterone metabolites. Following the addition of stigmasterol as an internal standard, methyloxime and trimethylsilyl ether derivatives of the steroid fractions were prepared prior to GC/MS analysis. The samples were analyzed by repetitive scanning on a Themofinnigan Polaris ion trap GC/MS instrument. Corticosterone metabolites were identified through interpretation of their mass spectra. Individual components were quantified by relating the peak areas on the TIC (total-ion-current) chromatograms to the peak area of the stigmasterol internal standard. Because we had no information on urine volume, individual steroids were reported quantitatively as a percentage of the total corticosterone metabolites (22,23).
Blood Analysis-At age 20 weeks both KO and WT mice underwent a terminal bleed, and serum was collected for analysis. Corticosterone levels in serum were measured between 8 and 10 a.m. using a commercial radioimmunoassay-based method (GE Healthcare) (n ϭ 15).
Statistical Analysis-Statistical comparisons were performed using SPSS version 12.0 (SPSS Inc.). Data are expressed as means Ϯ S.E. with statistical significance defined as p Ͻ 0.05. One-way analysis of variance was utilized to compare between KO and WT mice.

RESULTS
Disruption of the Mouse H6PD Gene-A null mutation was created in the murine H6PD gene through replacement of exons 2 and 3 with a neomycin resistance cassette via homologous recombination (Fig. 1A). Recombination was detected in 8 of 110 clones screened by Southern blotting, and correct targeting fidelity was confirmed with further Southern blotting using the 5Ј and neomycin probes. Homozygous KO mice were bred and maintained through intercrosses of 129SvJ and Hexose-6-phosphate Dehydrogenase Knock-out Mouse MARCH 10, 2006 • VOLUME 281 • NUMBER 10 C57BL/6 heterozygote mice and detected by Southern blotting using the 3Ј probe (Fig. 1B). Mice were genotyped by PCR using DNA extracted from tail biopsy. Primers P1, P2, and P3 were used in a multiplex PCR reaction to rapidly characterize WT, KO, and heterozygote mice. Gestation time and litter size were normal in heterozygote crosses, with genotyping of 178 animals revealing no deviation from a Mendelian distribution of alleles. No gross morphological abnormalities were seen in KO mice at birth.
Knock-out Mice Lack Hex-6-PDH Protein and Activity-Immunoblot analysis showed that no Hex-6-PDH protein could be detected in KO liver microsomes (Fig. 1C). Hepatic Hex-6-PDH enzyme activity in WT and KO mice was assessed using glucose 6-sulfate as a specific substrate for murine Hex-6-PDH (26). No Hex-6-PDH activity was measurable in KO mice (Fig. 2) compared with 5.5 Ϯ 0.41 nM NADPH/mg/min in WT mice and 3.2 Ϯ 0.35 nM NADPH/mg/min in heterozygous mice.
Set Point of 11␤-HSD1 Is Altered in Hex-6-PDH KO Mice-11␤-HSD1 enzyme function was assessed by examining the oxoreductase and dehydrogenase activities in liver microsomes from WT and KO mice either in the presence or absence of Glc-6-P, the substrate for Hex-6-PDH (Fig. 3). In the absence of Glc-6-P, intact microsomes from WT mice showed 7.1 Ϯ 1.8% conversion of 11-DHC to corticosterone (oxoreductase activity) and 10.3 Ϯ 1.3% conversion of corticosterone to 11-DHC (dehydrogenase activity). In the presence of Glc-6-P, conversion of 11-DHC to corticosterone increased to 31.8 Ϯ 6.8% in WT mice. By contrast, liver microsomes from KO mice consistently gave Յ5% conversion of 11-DHC to corticosterone, with or without Glc-6-P, indicative of absence 11␤-HSD1 oxoreductase activity (Fig. 3A). This loss in oxoreductase activity in KO animals was mirrored by an increase in dehydrogenase activity (conversion of corticosterone to 11-DHC 29.4 Ϯ 0.5% in KO versus 10.3 Ϯ 1.3% in WT), thus indicating a switch from oxoreductase to dehydrogenase activity in the absence of Hex-6-PDH.
Permeabilization of WT microsomes with detergent resulted in a loss of conversion of 11-DHC to corticosterone regardless of whether substrate for Hex-6-PDH (Glc-6-P) was absent or present (1.6 Ϯ 0.7% without Glc-6-P, 3.9 Ϯ 1.3 with Glc-6-P), indicating the requirement for FIGURE 2. Hex-6-PDH enzyme assay. Hex-6-PDH activity was determined spectrophotometrically by the production of NADPH at 340 nm when the reaction contained the Hex-6-PDH-specific substrate glucose 6-sulfate, with readings taken at 1-min intervals for 5 min. KO microsomes were devoid of activity, whereas activity was attenuated by ϳ50% in heterozygotes (HET) (0 nM NADPH/mg/min for KO mice versus 5.5 Ϯ 0.41 nM NADPH/mg/min for WT and 3.2 Ϯ 0.35 nM NADPH/mg/min for heterozygotes; n ϭ 4).

Hexose-6-phosphate Dehydrogenase Knock-out Mouse
intact microsomes and presumably a close, cooperative environment between Hex-6-PDH and 11␤-HSD1 in the ER to ensure 11-oxoreductase activity.
An assessment of 11␤-HSD1 activity in vivo was inferred through GC/MS analysis of steroid metabolites from pooled urine collections from male and female mice (n ϭ 5 in each group). Tetrahydro-and hexahydrocorticosterone and 11-DHC metabolites were found in mouse urine. These steroids were the only compounds with known mass spectra allowing accurate structures to be assigned. The dominant stereochemistry of the tetrahydro-11-DHC and tetrahydrocorticosterone metabolites was 3␣,5␤, and the hexahydrometabolites were predominately 3␣,5␤,20␣.
WT mice almost exclusively excrete steroids with an 11␤-hydroxyl group, and thus only minor amounts of 11-DHC metabolites were detected. Conversely, the dominant steroids in the KO mice were 11-DHC metabolites. In male WT mice the ratio of 11-DHC metabolites to corticosterone metabolites was 0.06, increasing to 9.3 in the KO animals. In female WT mice the 11-DHC/corticosterone ratio was 0.04, increasing to 1.83 in KO animals. When expressed as a percentage of 11-DHC metabolites in the urine of WT and KO animals, the marked differences confirmed the lack of oxoreductase activity in KO mice (Fig. 4).

DISCUSSION
Our data show that the inactivation of the murine H6PD gene dramatically alters the reaction direction of 11␤-HSD1. In isolated but intact WT microsomal preparations, there was minimal oxoreductase activity and moderate dehydrogenase activity. However, upon addition  Below each graph is indicated whether microsomes were intact or detergent-treated and whether microsomes were incubated with or without Glc-6-P. A, 11␤-HSD1 oxoreductase activity. The greatest activity was seen from intact WT microsomes incubated with Glc-6-P compared with KO. B, 11␤-HSD1 dehydrogenase activity. Intact KO microsomes with or without Glc-6-P had significantly higher levels of activity compared with WT. *, p Ͻ 0.01; **, p Ͻ 0.001.

Hexose-6-phosphate Dehydrogenase Knock-out Mouse
of Glc-6-P (the intracellular substrate for Hex-6-PDH), oxoreductase activity was stimulated ϳ6-fold, whereas dehydrogenase activity was unaffected. Upon permeabilization of the microsomes, this stimulation of reductase activity by Glc-6-P was lost, presumably because of the disruption of an intimate interaction between 11␤-HSD1 and Hex-6-PDH proteins. 11␤-HSD1 is bound to the inner ER membrane (27), whereas Hex-6-PDH is free-floating within the ER lumen (28). Further studies are warranted to define the interaction between these two proteins within the ER lumen. By contrast, intact hepatic microsomes from Hex-6-PDH KO mice show minimal basal oxoreductase that cannot be stimulated with the addition of Glc-6-P. Additionally, they have elevated dehydrogenase activity, indicating that in vivo these mice have a switch in hepatic 11␤-HSD1 enzyme activity in the liver from oxoreductase to dehydrogenase. An in vivo analysis of corticosterone metabolism obtained through urinary steroid GC/MS analysis endorsed these in vitro data. In man, the ratio of 11-hydroxyl C 21 steroids to 11-oxo C 21 derivatives (cortisol/cortisone) is an accurate reflection of 11␤-HSD activity (1,2). Here, using a similar principle, the ratio of 11-dehydrocorticosterone to corticosterone metabolites differed markedly between WT and KO mice. Based on the urinary 11-DHC metabolite/corticosterone metabolite ratio, the set point of 11␤-HSD1 was strongly in favor of corticosterone in WT mice (i.e. oxoreductase Ͼ Ͼ dehydrogenase), but this was reversed in KO mice. Collectively, these data are consistent with previous observations (18 -20, 29) and demonstrate directly for the first time that Glc-6-P within the ER provides the substrate for Hex-6-PDH and production of NADPH for the 11␤-HSD1-mediated generation of active glucocorticoid (in this case corticosterone) within hepatocytes (Fig. 6).
In the Hex-6-PDH KO mice, circulating corticosterone levels were significantly lower compared with the WT controls, despite adrenal hyperplasia. This is in contrast to the data obtained from the 11␤-HSD1 KO mice (6) where plasma corticosterone levels were higher in KO than in WT mice when measured at a single time point (the nadir). Unlike the 11␤-HSD1 KO mice, however, the activity of 11␤-HSD1 in the Hex-6-PDH mice reflects not only the removal of oxoreductase activity but also a highly significant increase in its dehydrogenase activity. It remains to be seen whether this alteration in 11␤-HSD activity accounts for differences in the function of the hypothalamus-pituitary-adrenal axes between these mice.
Our study has focused upon what appears to be a crucial interaction between Hex-6-PDH and 11␤-HSD1, largely in hepatocytes. However, Hex-6-PDH is ubiquitously expressed in mammalian tissues, including the liver, adrenal glands, spleen, kidney, heart, lungs, muscle, testes, ovaries, prostate, uterus, intestine, and placenta (30,31). Despite an extensive search we have been unable to identify other endoluminal NAPDH-dependent enzymes, but it is possible that these exist and will also be compromised. A detailed phenotypic analysis of these mice is now indicated, not only to define the effect of lack of glucocorticoid generation in liver and adipose tissue upon glucose tolerance and adiposity but also to investigate the consequences of disruption of ER redox potential in other key tissues.
In summary, these data further our understanding of the regulation and function of 11␤-HSD1, demonstrating that NADPH-dependent oxoreductase activity is critically dependent on the activity of the enzyme Hex-6-PDH. Mice lacking Hex-6-PDH have a profound switch in 11␤-HSD1 activity from oxoreductase to dehydrogenase, i.e. activation to inactivation. There is great interest in the role of 11-oxoreductase activity in the treatment of the metabolic syndrome, with selective 11␤-HSD1 inhibitors demonstrating improved glucose tolerance and weight reduction in mouse models of obesity (8,9). Our findings add a new complexity to this exciting pathway but also define a completely novel biochemical pathway that regulates the ER NADPH:NADP ϩ ratio and redox potential.