Mice Deficient in the Insulin-regulated Membrane Aminopeptidase Show Substantial Decreases in Glucose Transporter GLUT4 Levels but Maintain Normal Glucose Homeostasis*

The insulin-regulated aminopeptidase (IRAP) is a zinc-dependent membrane aminopeptidase. It is the homologue of the human placental leucine aminopeptidase. In fat and muscle cells, IRAP colocalizes with the insu-lin-responsive glucose transporter GLUT4 in intracellular vesicles and redistributes to the cell surface in response to insulin, as GLUT4 does. To address the question of the physiological function of IRAP, we generated mice with a targeted disruption of the IRAP gene (IRAP (cid:1) / (cid:1) ). Herein, we describe the characterization of these mice with regard to glucose homeostasis and regulation of GLUT4. Fed and fasted blood glucose and insulin levels in the IRAP (cid:1) / (cid:1) mice were normal. Whereas IRAP (cid:1) / (cid:1) mice responded to glucose administration like control mice, they exhibited an impaired response to insulin. Basal and insulin-stimulated glucose uptake in extensor digitorum longus muscle, and adipocytes isolated from IRAP (cid:1) / (cid:1) mice were decreased by 30–60% but were normal for soleus muscle from male IRAP (cid:1) / (cid:1) mice. Total GLUT4 levels were diminished by 40–85% in the IRAP (cid:1) / (cid:1) mice

fat cells with regard to its subcellular localization and the regulation of its trafficking by insulin (1, 2, 4 -8). The results from these studies showed that IRAP behaves almost identically to GLUT4. Under basal conditions, IRAP and GLUT4 are efficiently sequestered in intracellular membrane compartments. When cells are treated with insulin, IRAP and GLUT4 translocate to the plasma membrane. In adipocytes, this results in an 8-and 10 -20-fold increase of IRAP and GLUT4, respectively, at the cell surface (4,9,10). GLUT4 and IRAP are the only known molecules in fat and muscle cells that exhibit such marked translocation in response to insulin.
The physiological role of GLUT4 and the significance of its differential subcellular distribution under basal and insulinstimulated conditions are well established. The number of GLUT4 at the cell surface is the major determinant for the amount of glucose transported into muscle and fat tissues (10,11). Since skeletal muscle accounts for 80% of glucose disposal after feeding (12), and glucose uptake is the rate-limiting step in glucose metabolism in skeletal muscle (13), the regulation of the subcellular distribution of GLUT4 is key to the maintenance of glucose homeostasis. The physiological function of IRAP and the implication of its differential subcellular distribution under basal and insulin-stimulated conditions, however, are unknown.
The cloning of IRAP revealed that it is a member of the family of zinc-dependent aminopeptidases (14). IRAP's closest relatives in this family, the mammalian membrane aminopeptidases A and N, were known to process peptide hormones in vitro (15,16). Recent studies in mice using specific inhibitors for the two aminopeptidases support their role in the processing of angiotensin II and III, respectively, in vivo (17). We have previously shown that IRAP removes the N-terminal amino acids from vasopressin, Lys-bradykinin, and angiotensins III and IV efficiently (18). Insulin itself is not a substrate. Angiotensins III and IV are also well cleaved by aminopeptidase N (17,19). However, vasopressin, with its N-terminal cystine, was known to be cleaved only by the human placental leucine aminopeptidase/oxytocinase (P-LAP) (19). The cloning of P-LAP revealed that it is 87% identical to IRAP at the amino acid level (20). In agreement with our results, P-LAP cleaves vasopressin and angiotensin IV, as well as oxytocin (21). Whether the peptides cleaved by IRAP and P-LAP in vitro are also in vivo substrates has not been established.
The aminopeptidases homologous to IRAP are constitutively expressed at the cell surface and so have continued access to their extracellular substrates (22). IRAP, however, was expected to process its extracellular substrates only after translocation to the cell surface. In support of this, we have found that insulin, concomitant with the translocation of IRAP to the plasma membrane, increases aminopeptidase activity toward extracellular vasopressin in isolated adipocytes (18). The cloning of IRAP and characterization of it as an aminopeptidase that can process regulatory peptide hormones revealed thus a hitherto unrecognized possible action of insulin, the potential to modify, through IRAP, the action of other peptide hormones and thus broaden its own spectrum of action. This mode of action may not be restricted to insulin. It has been shown that endothelin-1 recruits IRAP to the plasma membrane in adipocytes and cardiomyocytes (23,24). Furthermore, IRAP may also play a role in processing peptide hormones in other tissues than fat and muscle. IRAP is well expressed in all major tissues with the exception of liver, where its expression levels are low (14). Recently, it has been reported that stimulation of human umbilical vascular endothelial cells with oxytocin leads to the translocation of IRAP to the cell surface (25).
To address the question of the function of IRAP in a physiological context, we have generated mice with a targeted disruption of the IRAP gene (referred to as IRAPϪ/Ϫ mice). We first tested these mice for defects in the regulation of glucose homeostasis, a major insulin action that is dependent on the proper regulation of GLUT4.

EXPERIMENTAL PROCEDURES
Generation of IRAPϪ/Ϫ Mice-Genomic clones containing fragments of the IRAP gene were isolated from a mouse embryonic stem cell genomic P1 library (mouse strain 129) by PCR screening with two primers derived from the rat IRAP cDNA (5Ј-GTCTTGGTGAGCAT-GAGATGG-3Ј and 5Ј-CTAAGGTCCTGGCAGAGGGTA-3Ј, corresponding to nucleotides 241-261 and 385-405, respectively (14)) by Genome Systems Inc. The genomic clones were analyzed by restriction digestions, Southern blotting, and sequencing, and a detailed exon-intron map of the IRAP gene was obtained. 2 Based on this information, a targeting vector was constructed as follows (Fig. 1A). An XhoI (bluntended)-HindIII genomic IRAP fragment corresponding to 3.8 kb of intron sequence with its 3Ј-end 0.4 kb upstream of exon 2 was cloned into the ClaI (blunt-ended)-HindIII restriction sites in the vector pPol II long neo bpA 5Ј to the neomycin resistance cassette. The neomycin resistance gene in this cassette is oriented such that transcription proceeds in the opposite direction to the introduced IRAP sequences. A 5.6-kb XhoI-ClaI (blunt-ended) restriction fragment of IRAP genomic DNA with its 5Ј-end 0.5 kb downstream of exon 6 and its 3Ј-end 2 kb downstream of exon 8 was inserted into the XhoI and EcoRI (bluntended) restriction sites in the vector downstream of the neomycin resistance cassette and upstream of the thymidine kinase gene. The thymidine kinase gene was derived from the pMC1TK vector. Correct insertion of the fragments in the targeting construct was verified by restriction analysis and sequencing of the junctions. The targeting vector was linearized with SrfI and introduced into mouse embryonic stem cells (from strain 129 SvEv) by electroporation. Embryonic stem cell clones were selected for resistance to G418. Genomic DNA was isolated from resistant clones, digested with XbaI, and screened for homologous recombination by Southern blotting. The 1032-bp probe used for screening was derived from exon 9 and 0.5 kb of intron sequence upstream and 0.35 kb of intron sequence downstream of exon 9 and was labeled with dUTP-digoxigenin by PCR (primers 5Ј-CAAGCT-CAGTCTGGAGTCTTAGTG-3Ј and 5Ј-GATTCACAGGGCTTCATAGA-GAC-3Ј). The hybridized probe was detected with an anti-digoxigenin antibody conjugated to alkaline phosphatase and chemiluminescence according to the manufacturer's instructions (Roche Molecular Biochemicals). Of the 384 clones analyzed, three showed the expected fragments of 12 and 8.5 kb characteristic for the endogenous intact and the disrupted IRAP allele, respectively (see Fig. 1B). The XbaI-digested genomic DNA of these three clones was also hybridized with a 783-bp dUTP-digoxigenin-labeled probe that was obtained by PCR with primers derived from the neomycin resistance gene (5Ј-GGGATCGGCCAT-TGAACAAG-3Ј and 5Ј-AAGGCGATAGAAGGCGATGC-3Ј) to test for correct recombination between the 5Ј homologous IRAP region in the targeting vector and the endogenous IRAP. All three clones showed the expected 8-kb XbaI fragment ( Fig. 1A and data not shown).
One of these clones was injected into blastocysts derived from Black Swiss mice, and the blastocysts were implanted into pseudopregnant recipients. Male chimeric offspring were bred with wild type Black Swiss female mice (Taconic). In the agouti offspring, germ line transmission of the mutant IRAP allele was identified by Southern blotting of genomic DNA obtained from tail snips. The dUTP-digoxigenin-labeled IRAP probe was used as described above and marked in Fig. 1A. Six male and female mice heterozygous for the mutant IRAP allele (IRAPϩ/Ϫ) were bred to obtain the IRAP null (IRAPϪ/Ϫ) together with wild type (IRAPϩ/ϩ) and heterozygous (IRAPϩ/Ϫ) mice. The mice were genotyped by Southern blotting as described above. Representative results from Southern blots are shown in Fig. 1B. Nine male and female IRAPϪ/Ϫ mice as well as eight male and female IRAPϩ/ϩ mice were paired, and the offspring from these breeding pairs were used for all of the experiments described below unless stated otherwise. Offspring from different breeding pairs for each genotype were used in each experiment to assure equal representation of the different mixed genetic pools obtained through the breeding of the outbred Black Swiss with the inbred 129 SvEv mice. The mice were housed under a constant light (6 a.m. to 6 p.m.) and dark cycle and fed Teklad LM 485 mouse/rat diet (Harlan Teklad) with free access to water.
Analysis of Blood Glucose and Plasma Insulin Concentrations and Glucose and Insulin Tolerance Tests-The determination of blood glucose and plasma insulin concentrations as well as the oral glucose tolerance test were carried out as described in Ref. 26 with the exception that the blood samples for the determination of fed glucose and insulin were taken between 8 and 9 a.m. The insulin tolerance test was performed as described in Ref. 27.
Determination of Glucose Uptake in Isolated Muscles-Mice were anesthetized with sodium pentobarbital (Nembutal) (0.1 mg/g body weight), and soleus and EDL were isolated. Glucose transport into the isolated muscles was assayed by the uptake of 3 H-labeled 3-O-methylglucose after the method described in Ref. 28  C]mannitol/mmol (PerkinElmer Life Sciences)), 2 mM pyruvate, and either 0 or 13 nM insulin. After 10 min of incubation at 29°C, glucose uptake was terminated, and muscle pieces were processed as described in Ref. 28. Glucose uptake was determined from the intracellular 3 Hlabeled 3-O-methylglucose by subtracting extracellular 3 H-labeled 3-Omethylglucose from the total muscle-associated 3 H-labeled 3-O-methylglucose. The extracellular 3 H-labeled 3-O-methylglucose was estimated from the content of the extracellular marker [ 14 C]mannitol in each muscle piece. Data are expressed as mol of 3-O-methylglucose/g of muscle/10 min.
Determination of Glucose Uptake In Isolated Adipocytes-Adipocytes were isolated from the epididymal and parametrial fat pads from male and female mice, and the uptake of [U-14 C]D-glucose (ICN) was measured as described in Ref. 26 with the exception that glucose uptake was terminated after 15 min of incubation. Aliquots of the adipocyte preparation used for the glucose uptake assays were also subjected to lipid extraction to determine lipid weight as described in Ref. 26. Glucose uptake was normalized to the lipid content of each sample and expressed as fmol of glucose/min/mg of lipid.
Analysis of Tissue Homogenates-Mice were euthanized with 100% CO 2 . Tissues were isolated, rinsed in ice-cold phosphate-buffered saline, blotted dry, immediately frozen in liquid nitrogen, and stored at Ϫ80°C until further processing. The frozen tissues were weighed and homogenized in a buffer containing 25 mM Hepes, pH 7.4, 100 mM NaCl, 1 mM Na 3 VO 4 , 20 mM sodium pyrophosphate, 3 mM N-ethylmaleimide, 1 g/ml pepstatin, 10 M EP475, 10 g/ml aprotinin, and 0.2 mM phenylmethylsulfonyl fluoride. Experimentally determined protein content per mg of tissue was used as the basis to calculate the amount of homogenization buffer to be added to each tissue such that the protein concentrations of the final homogenates were between 10 and 20 mg/ml. Aliquots of the homogenates were mixed with 4ϫ SDS sample buffer (370 mM Tris, pH 6.8, 4 mM EDTA, 40% (v/v) glycerol, 0.016% (w/v) bromphenol blue, 16% SDS, 80 mM dithiothreitol, 4 g/ml pepstatin, 40 M EP475, 4 mM phenylmethylsulfonyl fluoride, 40 M leupeptin, and 40 g/ml aprotinin). 2 S. R. Keller, manuscript in preparation.
For the preparation of adipocyte homogenates, adipocytes were isolated from the epididymal and parametrial fat pads as described in Ref. 26 in Krebs-Ringer-Hepes-BSA buffer (120 mM NaCl, 4.8 mM KCl, 1.3 mM CaCl 2 , 1.2 mM KH 2 PO 4 , 1.2 mM MgSO 4 , 20 mM HEPES, pH 7.4, 1% BSA, 3 mM glucose, and 200 nM adenosine). To remove the BSA and thus avoid complications of interference in immunoblotting for GLUT4, the cells were washed twice in KRBH without BSA, and a 5% fat cell suspension in KRBH was prepared. Aliquots of the suspension were either processed for the determination of lipid weights as described in Ref. 26 or were used to prepare adipocyte lysates as follows. Cells were allowed to float, the infranatant was removed, and 1.2ϫ SDS sample buffer was added to the cells. The protein concentrations in all of the SDS samples were determined by a precipitating Lowry assay with BSA as the standard (29).
Preparation of Subcellular Fractions from Isolated Adipocytes-Mice were euthanized with 100% CO 2 . Adipocytes were isolated from the epididymal and parametrial fat pads from male and female mice as described in Ref. 26 in KRBH-BSA buffer (120 mM NaCl, 4 mM KH 2 PO 4 , 1 mM CaCl 2 , 1 mM MgSO 4 , 10 mM NaHCO 3 , 30 mM HEPES, pH 7.4, 1% BSA, 3 mM glucose, and 200 nM adenosine). After the digestion with collagenase, the cells were washed twice in KRBH-BSA. A 10% cell suspension was prepared in KRBH-BSA and split into two equal aliquots. To one aliquot insulin was added to 10 nM. All cells were incubated for 30 -45 min at 37°C with gentle shaking (40 rpm). The cells were then washed once in homogenization buffer (255 mM sucrose, 20 mM Hepes, pH 7.4, with or without insulin) at room temperature and subsequently homogenized with 10 up and down strokes in a Teflon homogenizer rotating at 1250 rpm in the homogenization buffer containing 1 g/ml pepstatin, 10 M EP475, 10 g/ml aprotinin, 10 M leupeptin, and 1 mM phenylmethylsulfonyl fluoride with or without insulin. Subcellular fractions were obtained from the homogenates as described in Ref. 30. The fractions were resuspended in homogenization buffer containing 1 g/ml pepstatin, 10 g/ml aprotinin, 10 M leupeptin. The protein concentrations in each fraction were determined by the precipitating Lowry assay (29). Aliquots from each of the fractions were used for the measurement of aminopeptidase activities as described below. SDS samples containing equal amounts of protein for each fraction from the IRAPϪ/Ϫ and IRAPϩ/ϩ mice were prepared from aliquots from the fractions and were analyzed by immunoblotting.
Immunoblotting of Tissue Homogenates and Subcellular Fractions-SDS samples of tissue homogenates containing equal amounts of protein were immunoblotted for IRAP with an antibody raised against the N-terminal cytoplasmic domain of IRAP as described in Ref. 4. Equal amounts of protein of muscle tissue homogenates in SDS sample buffer were separated on 16 ϫ 16 ϫ 0.15-cm 10% SDS-polyacrylamide gels and transferred for 4 h or overnight to Immobilon-P membranes (Millipore Corp.) at 100 mA. SDS samples of adipocyte homogenates and subcellular fractions were separated on 9 ϫ 13 ϫ 0.1-cm 10% SDS-polyacrylamide gels and transferred for 2 h to Immobilon-P membranes (Millipore) at 100 mA. Membranes were probed for GLUT4 with antibodies raised against the C-terminal peptide of GLUT4 as described in Ref. 4. Immunoblots for GLUT1 in heart and adipocytes were carried out with polyclonal antibodies raised against the C-terminal peptide of GLUT1 (Research Diagnostics Incorporation; catalog no. RDI-MGLUT1Cabr) following the same protocol as described for GLUT4 in Ref. 4 except that the antibody was used at a concentration of 2.5 g/ml. For the detection of GLUT1 in soleus and EDL, a polyclonal antibody raised against the entire purified mouse GLUT1 (a gift from G. E. Lienhard) was used under the same conditions as described for GLUT4 in Ref. 4 except that BSA was used as the blocking reagent, and Tween 20 was added to the antibody solution to a final concentration of 0.3%. Relative expression levels were determined by analyzing different dilutions of the samples and quantitating the signals by densitometry using ImageQuant software (Molecular Dynamics, Inc., Sunnyvale, CA).
Aminopeptidase Assays-Aliquots of plasma membrane and low density microsomal fractions containing 5 g of total protein in 100 l of homogenization buffer were solubilized with the nonionic detergent C 12 E 9 (0.5%) by incubation for 10 min at room temperature. The aliquots were then added to 400 l of substrate in 120 mM NaCl, 4 mM KH 2 PO 4 , 1 mM CaCl 2 , 1 mM MgSO 4 , 30 mM HEPES, pH 7.4, at 35°C. The substrate used was L-lysine ␤-naphthylamide (Sigma), and it was present in the assay at a final concentration of 500 M. The fluorescence of the liberated ␤-naphthylamine was recorded at excitation of 340 nm and emission of 410 nm over a 6-min incubation period at 35°C (Hitachi Instruments F-2500). The rates of cleavage were determined from the linear slope of the graph.
Statistical Analysis-For each experiment, unpaired two-tailed t tests were used to compare the mean values for the wild type and the IRAPϩ/Ϫ and IRAPϪ/Ϫ mice. Differences were taken to be not significant for p values greater than 0.05. Data are expressed as means Ϯ S.E.

Growth and Development of IRAPϪ/Ϫ Mice-
The IRAP gene was disrupted in 129 SvEv embryonic stem cells by replacing exons 2-6 and the intervening introns (in total 10 kb of IRAP genomic sequence) with the neomycin resistance gene through homologous recombination with the targeting vector (Fig. 1A). Exons encoding domains that are crucial for the function of IRAP were thus eliminated. Exon 2 encodes the cytoplasmic tail that confers to IRAP its specific subcellular localization and regulation by insulin (31) as well as the transmembrane domain. The zinc-binding site that is encoded in exon 6 is essential for IRAP's catalytic activity (14,32). Mice heterozygous for the mutant allele were obtained on a mixed Black Swiss/129SvEv genetic background. They were bred to yield IRAPϪ/Ϫ mice together with IRAPϩ/ϩ and IRAPϩ/Ϫ mice. The offspring from six heterozygous breeding pairs, a total of 280 mice (126 female and 154 male), were genotyped by Southern blotting (Fig. 1B). IRAPϩ/ϩ, IRAPϩ/Ϫ, and IRAPϪ/Ϫ were obtained at a ratio of 26:53:21 for the male and 22.2:52.4:25.4 for the female mice. The average litter size was 6.4 pups. The ratios for the different genotypes reflect the expected Mendelian frequencies for the different genotypes. The lack of IRAP thus does not affect embryonic viability.
The absence of the IRAP protein was confirmed by immunoblotting tissue homogenates derived from IRAPϪ/Ϫ mice with an antibody against IRAP. Immunoblots for muscle and fat tissues are shown in Fig. 1C. No IRAP was detected in these or in any other of the tissues where IRAP is normally expressed. An additional protein of slightly higher molecular weight than IRAP is detected by the antibody in skeletal muscle and diaphragm of both IRAPϩ/ϩ and IRAPϪ/Ϫ mice (Fig. 1C). In contrast to IRAP itself, this protein is not immunoprecipitated with the antibody used (14), and we conclude that it is a protein that is not related to IRAP. Subcellular fractions that were obtained from isolated adipocytes were tested for aminopeptidase activity toward L-lysine ␤-naphthylamide. L-Lysine ␤-naphthylamide is a good synthetic substrate for IRAP (14) and is not cleaved or is poorly cleaved by the membrane aminopeptidases A and N (33). The aminopeptidase activities toward L-lysine ␤-naphthylamide we measured in the plasma membrane and in the low density microsomal membranes derived from the IRAPϪ/Ϫ cells were 37 Ϯ 3 and 14 Ϯ 1%, respectively, of the activities we determined in the corresponding fractions from wild type cells (Fig.  2). We have previously determined that the aminopeptidase activities toward the L-lysine ␤-naphthylamide that cannot be immunoprecipitated with the antibody against IRAP from the plasma and low density microsomal fractions prepared from isolated rat adipocytes were 30 Ϯ 4 and 7 Ϯ 4% of the total activities (n ϭ 4 individual preparations and immunoprecipitations), respectively (18). 3 The remaining aminopeptidase activities in the fractions after depletion of IRAP are thus similar to the activities remaining in each of the corresponding fractions from IRAPϪ/Ϫ cells. This suggests that there is no increase in an alternative aminopeptidase that compensates for the lack of IRAP in the IRAPϪ/Ϫ cells. The remaining activity is most likely due to aminopeptidase N, an enzyme that is abundant in the plasma membrane (22).
Growth of the offspring from the heterozygous breeding pairs was assessed by measuring the body weights of male IRAPϩ/ϩ, IRAPϩ/Ϫ, and IRAPϪ/Ϫ mice at 30-day intervals from 30 to 150 days of age (Fig. 3). We found that the weights were not significantly different between the three genotypes at any time point. We also could not detect any differences in weights between the three genotypes in the same mice at 1 year of age (Table I and data not shown). From these results, we conclude that in the absence of IRAP, postnatal growth and weight gain during adult life is normal.
Eight pairs of male and female wild type and nine pairs of male and female IRAPϪ/Ϫ mice were bred. The ages of the mice used for breeding were matched for the two genotypes. The IRAPϩ/ϩ pairs yielded an average of 4.5 Ϯ 0.3 litters with 6.5 Ϯ 0.4 pups/litter. The average interval between the litters was 27.6 Ϯ 2 days. Within the same time period, the IRAPϪ/Ϫ pairs produced an average of 4.9 Ϯ 0.4 litters with 5.9 Ϯ 0.4 pups/litter. The average interval between the litters was 28.6 Ϯ 2.3 days. The values obtained for the IRAPϪ/Ϫ and IRAPϩ/ϩ breeding pairs were not significantly different. Thus, reproduction is not impaired in the IRAPϪ/Ϫ mice. We also found that B, genotyping of wild type (ϩ/ϩ), IRAP-heterozygous (ϩ/Ϫ), and IRAP null (Ϫ/Ϫ) mice by Southern blot analysis. XbaI-digested genomic DNA was hybridized with the probe derived from a sequence downstream of the 3Ј homologous recombination region. The location of the probe is shown in A. The XbaI fragments derived from the wild type and IRAP recombinant alleles (indicated in A) were 12 and 8.5 kb, respectively, as judged by comparison with size markers and are marked with arrows on the right of the blot. C, immunoblots of tissue homogenates from wild type (ϩ/ϩ) and IRAP null (Ϫ/Ϫ) mice. SDS samples containing 100 g of total protein from each tissue were separated on a 6% SDS-polyacrylamide gel and immunoblotted for IRAP as described in Ref. 4. Lanes 1 and 4, skeletal muscle; lanes 2 and 5, diaphragm; lanes 3 and 6, heart; lanes 7 and 9, brown adipose tissue; lanes 8 and 10, white adipose tissue from female wild type and IRAPϪ/Ϫ mice, respectively. The position of IRAP is indicated with an arrow on the left of the blot. The signal above the position of IRAP at slightly higher molecular weight in skeletal muscle and diaphragm is due to nonspecific reaction of the antibody with a protein not related to IRAP.

FIG. 2. Aminopeptidase activities in subcellular fractions from isolated adipocytes.
Aminopeptidase activities toward L-lysine-␤naphthylamide were measured in plasma membrane (PM) and low density microsomal fractions (LDM) as described under "Experimental Procedures." The values obtained for each individual fraction for the IRAPϪ/Ϫ cells (Ϫ/Ϫ) were compared with the respective fractions obtained from IRAPϩ/ϩ cells (ϩ/ϩ) in the preparation carried out simultaneously and expressed as a percentage. Three separate subcellular fractionations with 20 mice for each genotype and with basal and insulin-stimulated cells were carried out. Two preparations were done for male mice, and one was done for female mice. The relative values for the respective male and female IRAPϪ/Ϫ fractions and the respective fractions from basal and insulin-stimulated cells were the same, and the values were thus pooled (n ϭ 6). The data in the graph represent the means Ϯ S.E.

FIG. 3. Growth curves.
Body weights for male IRAPϩ/ϩ (rectangles) (n ϭ 7-9), IRAPϩ/Ϫ (n ϭ 10) (reversed triangles), and IRAPϪ/Ϫ (triangles) (n ϭ 6 -7) mice at 30-day intervals from 30 to 150 days of age were measured. The means Ϯ S.E. of the weights for each genotype for each age were plotted. Littermates from the original heterozygous breeding pairs were used. the survival of the pups from the IRAPϪ/Ϫ mice was the same as for the wild type mice, indicating that the IRAPϪ/Ϫ mothers nursed their pups well.
Glucose Homeostasis in IRAPϪ/Ϫ Mice-To determine whether glucose homeostasis was affected by the lack of IRAP, blood glucose and plasma insulin levels were measured in fasted and randomly fed male and female mice at 6 months of age (Table I). The values that were obtained for the IRAPϪ/Ϫ mice for both sexes were not significantly different from the IRAPϩ/ϩ mice. To determine whether disturbances in glucose homeostasis would become apparent with advanced age, fasted and fed blood glucose and plasma insulin levels were also determined in 1-year-old male mice (Table I). As in the 6-month-old mice, no differences were found between the IRAPϪ/Ϫ and the wild type mice. Next, glucose and insulin tolerance tests were performed on male and female IRAPϪ/Ϫ and wild type mice (Fig. 4). In the oral glucose tolerance test, the blood glucose values for the male IRAPϪ/Ϫ mice were slightly higher than the ones for the control mice at each time point after the administration of glucose (Fig. 4A). However, the values were not statistically significantly different from the values obtained for the wild type mice. The blood glucose values for the female mice were identical for IRAPϩ/ϩ and IRAPϪ/Ϫ mice (Fig. 4B). In the insulin tolerance test, the blood glucose levels in the male IRAPϪ/Ϫ mice did not decline to the same degree as in the IRAPϩ/ϩ mice (Fig. 4C). Minimal glucose values obtained for IRAPϪ/Ϫ and IRAPϩ/ϩ mice were 76.4 Ϯ 5 and 53 Ϯ 3.4% of basal levels before the injection of insulin, respectively. In the female IRAPϪ/Ϫ mice, the blood glucose levels at each time point were higher than in the IRAPϩ/ϩ mice, but the values were not significantly different (Fig. 4D). The results from these tests show that the IRAPϪ/Ϫ mice maintain normal glucose homeostasis under randomly fed conditions and handle an oral glucose challenge similar to wild type mice. The effectiveness of a bolus of insulin to lower blood glucose is impaired in the male but not in the female IRAPϪ/Ϫ mice.
Glucose Uptake into Isolated Muscles and Adipocytes-To determine whether the impaired response to the glucose-lowering effect of insulin in the IRAPϪ/Ϫ mice could be due to a decrease in glucose uptake into the tissues responsible for the major fraction of insulin-mediated glucose disposal, we measured glucose transport into isolated muscles and adipocytes. 3-O-Methylglucose uptake into soleus and EDL was measured in the absence and presence of a maximally stimulating dose of insulin (13 nM) as described in Ref. 28. Glucose uptakes in soleus were the same for male wild type and IRAPϪ/Ϫ mice under basal (0.53 Ϯ 0.07 and 0.52 Ϯ 0.03 mol/g of muscle/10 min) as well as under insulin-stimulated conditions (1.22 Ϯ 0.08 and 1.13 Ϯ 0.13 mol/g of muscle/10 min) (Fig. 5A). For the soleus isolated from female wild type and IRAPϪ/Ϫ mice, glu-cose uptake was not different under basal conditions (0.59 Ϯ 0.07 and 0.56 Ϯ 0.06 mol/g of muscle/10 min). The insulinstimulated glucose uptake, however, was decreased by 23% in the IRAPϪ/Ϫ when compared with the wild type mice (1.25 Ϯ 0.09 compared with 1.62 Ϯ 0.07 mol/g of muscle/10 min) (Fig.  5A). The insulin-stimulated increases in glucose uptake over basal level were 2.3-and 2.2-fold for the male IRAPϩ/ϩ and IRAPϪ/Ϫ mice, respectively, and 2.7-and 2.2-fold for the female IRAPϩ/ϩ and IRAPϪ/Ϫ mice, respectively. In the EDL isolated from IRAPϪ/Ϫ mice, glucose uptake was decreased by 30% for both male and female when compared with wild type muscles under basal (for male 0.40 Ϯ 0.02 compared with 0.61 Ϯ 0.07 mol/g of muscle/10 min; for female 0.39 Ϯ 0.05 compared with 0.56 Ϯ 0.09 mol/g of muscle/10 min) and in-TABLE I Blood glucose and plasma insulin levels for wild type (ϩ/ϩ) and IRAP null (Ϫ/Ϫ) mice in the fed and fasted state Measurements were made for 6-month-old male and female mice and for 12-month-old male mice. Values are displayed as mean Ϯ S.E. Data are from eight mice for each genotype and sex for the 6-month-old mice and from six male mice for each genotype for the 12-month-old mice (except insulin-fed, nϭ5 for each genotype). For the measurements of blood glucose and insulin on 1-year-old mice, IRAPϩ/ϩ and IRAPϪ/Ϫ offspring from the original heterozygous breeding pairs were used. Blood samples were obtained from the cut tail vein between 8 and 9 a.m. for the determination of fed values and at 9 -10 a.m. after a 16-h overnight fast for the fasted values. Glucose values were determined with the Precision-G blood glucose testing system (MediSense). Insulin values were determined in plasma using an insulin radioiummune assay (Linco Research Inc.

FIG. 4. Oral glucose and insulin tolerance tests.
A and B, oral glucose tolerance tests with fasted 10-month-old male (A) and 6-monthold female (B) wild type (rectangles) and IRAPϪ/Ϫ mice (triangles) (n ϭ 8 for each genotype and sex). 1 mg of D-glucose/g of body weight was given with a feeding needle to mice fasted for 16 h, and glucose concentrations were determined in blood drops obtained from the cut tail vein at the indicated times. C and D, insulin tolerance tests with randomly fed 8 -10-month-old male (C) (n ϭ 15 for each genotype) and female (D) (n ϭ 16 for each genotype) wild type (rectangles) and IRAPϪ/Ϫ mice (triangles). Insulin (Humulin R; 0.75 units/kg of body weight) was given by intraperitoneal injection at 2 p.m. to randomly fed mice, and blood glucose measurements were taken at the indicated times as described under oral glucose tolerance tests. Data points and error bars in the graphs represent means Ϯ S.E. Only the values in the insulin tolerance test for the male (C) were statistically significantly different between the IRAPϪ/Ϫ and wild type mice (p Ͻ 0.05). The average body weights for the IRAPϪ/Ϫ and wild type mice used in the tests were not statistically significantly different.
Glucose uptake was measured into adipocytes isolated from the epididymal and parametrial fat depots of male and female mice, respectively, in the absence and presence of maximally stimulating insulin concentrations (10 and 3 nM for male and female, respectively) as described in Ref. 26 (Fig. 5C). The basal glucose uptake for the adipocytes from male IRAPϪ/Ϫ mice was decreased by 30% when compared with wild type cells (17.5 Ϯ 3.3 compared with 27.3 Ϯ 2.8 fmol/min/mg of lipid). For the female IRAPϪ/Ϫ adipocytes, basal glucose uptake was decreased by 45% versus wild type cells (46.3 Ϯ 11.2 versus 84.9 Ϯ 8.4 fmol/min/mg of lipid). The insulin-stimulated glucose uptake into isolated adipocytes from male and female IRAPϪ/Ϫ mice was decreased by 52 and 57%, respectively, when compared with the glucose uptake by wild type cells (for male 46.9 Ϯ 8.7 compared with 103.9 Ϯ 10.3 fmol/min/mg; for female 206.5 Ϯ 42.9 compared with 479.2 Ϯ 35.6 fmol/min/mg). The -fold increases in glucose uptake induced by insulin over basal were 3.8 and 2.9 for male IRAPϪ/Ϫ and IRAPϩ/ϩ mice, respectively, and 6 and 4.8 for female IRAPϪ/Ϫ and IRAPϩ/ϩ mice, respectively. The -fold increases were not statistically significantly different between the two genotypes. The insulin dose that elicited a half-maximal response (ED 50 ) was determined from four independent experiments in which glucose uptake was measured at 0, 0.1, 1, and 10 nM insulin for the epididymal adipocytes and at 0, 0.03, 0.3, and 3 nM for parametrial adipocytes. The ED 50 values were 0.11, 0.19, 0.05, and 0.06 for adipocytes from wild type and IRAPϪ/Ϫ male and wild type and IRAPϪ/Ϫ female mice, respectively. The epididymal and parametrial fat pads from which the adipocytes for the glucose uptake assays were isolated weighed 0.44 Ϯ 0.09 and 0.48 Ϯ 0.1 g for male IRAPϩ/ϩ and IRAPϪ/Ϫ mice, respectively, and 0.35 Ϯ 0.05 and 0.44 Ϯ 0.08 g for the female IRAPϩ/ϩ and IRAPϪ/Ϫ mice, respectively. These weights were not statistically significantly different. We conclude that basal as well as insulin-stimulated glucose uptake into isolated EDL and adipocytes is impaired in the IRAPϪ/Ϫ mice. Glucose uptake into isolated soleus is normal for male IRAPϪ/Ϫ mice under basal and insulin-stimulated conditions. It is normal under basal conditions but impaired after insulin stimulation for female IRAPϪ/Ϫ mice. Since the insulin dose required for half-maximal stimulation of glucose uptake in isolated adipocytes is the same for wild type and IRAPϪ/Ϫ mice, we infer that insulin sensitivity is normal in the IRAPϪ/Ϫ mice.
Determination of Glucose Transporter Amounts in Muscle and Fat Tissues-To determine whether a decrease in the expression of GLUT4 was responsible for the decreases in glucose uptake observed in EDL and adipocytes, immunoblots for GLUT4 were performed and analyzed by densitometry (Fig. 6). GLUT4 was barely detectable in EDL from male and female IRAPϪ/Ϫ and was determined to be 15 Ϯ 2 and 20 Ϯ 2%, respectively, of wild type levels (Fig. 6B). In isolated adipocytes, GLUT4 was decreased to 53 Ϯ 2 and 57 Ϯ 5% for male and female IRAPϪ/Ϫ mice, respectively, when compared with wild type cells (Fig. 6C). Although glucose uptake was normal in soleus (male) or slightly impaired (female), GLUT4 expression was decreased to 43 Ϯ 3% in male and to 35 Ϯ 7% in female IRAPϪ/Ϫ mice when compared with wild type soleus (Fig. 6A). GLUT4 protein levels were also assessed in heart. In hearts from male and female IRAPϪ/Ϫ mice GLUT4 was decreased to 15 Ϯ 2 and 35 Ϯ 6% (Fig. 6D), respectively, when compared with wild type male and female hearts. The decreases in GLUT4 are expressed relative to the same amount of total protein analyzed for each tissue for the different genotypes. Since the amounts of total protein obtained per mg of tissue were the same for the IRAPϪ/Ϫ mice and the wild type mice, the relative decreases in GLUT4 reflect absolute decreases in GLUT4 amounts in the cells. To correlate the relative decreases in GLUT4 to glucose uptake in isolated adipocytes, we also determined the amount of total protein per mg of lipid. We found that the numbers were similar for adipocytes isolated from IRAPϪ/Ϫ and IRAPϩ/ϩ mice. The relative decreases of GLUT4 per the same amount of protein therefore are equivalent to the relative decreases in glucose uptake expressed per mg of lipid. The decrease in GLUT4 expression could thus explain the decrease in glucose uptake in IRAPϪ/Ϫ adipocytes. However, in soleus and EDL from IRAPϪ/Ϫ mice, the relative level of glucose uptake was higher than what could be expected from the relative decrease in GLUT4.
In fat and muscle cells, the glucose transporter isoform GLUT1 is also expressed. It is much less abundant than GLUT4 and comprises only about 10% of the total glucose transporters in adipocytes and skeletal muscle (34,35). To determine whether changes in the expression of GLUT1 could compensate for the decreases in GLUT4, immunoblots for  A and B) and quadruplicates (C). An asterisk above the bars for the IRAPϪ/Ϫ mice indicates that the value was statistically significantly different from the respective value obtained for the IRAPϩ/ϩ mice (p Ͻ 0.05, except for glucose uptake into male adipocytes in the absence of insulin, where p ϭ 0.06). Glucose uptake assays into adipocytes were also performed in 6-month-old male mice, and glucose uptake was expressed per cell. The relative changes in glucose uptake between the wild type and the IRAPϪ/Ϫ mice were the same as for the assays shown in the figure.
GLUT1 were performed on soleus and EDL, adipocyte, and heart homogenates (Fig. 7). No increase in the expression of GLUT1 was detected in any of these tissues when compared with the respective tissue homogenates from wild type mice. The relative expression levels in the IRAPϪ/Ϫ tissues were 93 Ϯ 9 for soleus (n ϭ 3), 95 Ϯ 18% for EDL (n ϭ 3), 105% for adipocytes (pooled from five mice), and 105 Ϯ 30% for heart (n ϭ 3). This finding rules out the possibility that GLUT1 compensates for the decrease in GLUT4.
Subcellular Distribution of GLUT4 in Adipocytes-To evaluate whether the absence of IRAP led to changes in the subcellular distribution of GLUT4, subcellular fractions were prepared from basal and insulin-stimulated adipocytes from male and female 6-month-old IRAPϪ/Ϫ and IRAPϩ/ϩ mice. Plasma membranes, low density microsomal membranes, and homogenates were immunoblotted for GLUT4. Fig. 8 shows representative immunoblots of different dilutions of plasma and low density microsomal membranes isolated from male IRAPϪ/Ϫ and IRAPϩ/ϩ adipocytes. We found that GLUT4 was decreased to 48 Ϯ 7 and 49 Ϯ 2%, respectively, in the plasma membrane fraction from basal and insulin-stimulated IRAPϪ/Ϫ adipocytes when compared with the respective fractions from wild type cells. In the low density microsomes, the decreases were similar to 46 Ϯ 6 and 50 Ϯ 2%, respectively, in the fractions from basal and insulin-stimulated cells when compared with the respective fractions from wild type cells. The relative decreases in GLUT4 in each of the fractions from the IRAPϪ/Ϫ mice were the same as in the homogenates prepared from the same cells. The relative level in these was 53 Ϯ 4%. The -fold increase of GLUT4 in the plasma membrane in response to insulin was 2 and 1.6 for IRAPϪ/Ϫ and IRAPϩ/ϩ mice. From these results, we conclude that in the absence of IRAP the relative subcellular distribution of GLUT4 is unchanged at steady state. The glucose uptake assays for which the results are shown in Fig. 5C were performed with adipocytes isolated from 1-year-old mice. However, we obtained similar results for glucose uptake measurements in adipocytes from male IRAPϪ/Ϫ mice at 6 months of age. Thus, the relative decrease in glucose uptake in adipocytes correlates exactly with the relative decrease of GLUT4 in the plasma membrane fraction.
Tissue Weights-The weights of adipose tissue and skeletal muscles were not significantly different between the wild type and the IRAPϪ/Ϫ mice. However, the weights of the hearts isolated from male IRAPϪ/Ϫ mice were on average 20% larger than the hearts obtained from the IRAPϩ/ϩ mice (203.7 Ϯ 13.4 mg versus 169.9 Ϯ 8.6 mg, p ϭ 0.048, n ϭ 7-8, 11-month-old mice). The body weights were identical with 42.3 Ϯ 3.1 g for the IRAPϩ/ϩ and 42.5 Ϯ 2.4 g for the IRAPϪ/Ϫ mice. The ratios of heart weights over body weights were 0.0041 Ϯ 0.0002 and 0.0048 Ϯ 0.0004 (p ϭ 0.0697).

DISCUSSION
One of the most important results obtained in this initial characterization of the IRAPϪ/Ϫ mice is the finding of the substantial reduction in GLUT4 protein in all of the tissues where GLUT4 is predominantly expressed (36): in skeletal muscle, heart, and adipose tissue. The decreases for GLUT4 were similar in the tissues from male and female IRAPϪ/Ϫ mice. These findings suggest that the presence of IRAP is required for the maintenance of normal GLUT4 levels independent of the tissue type and sex. Preliminary data that were obtained for skeletal muscle and heart showed that the defect was present as early as postnatal day 18 (data not shown), thus indicating that it is also independent of the age of the mice. Reduced expression of GLUT4 has been found in various insulin-resistant states in humans and animals (36). In most of FIG. 6. GLUT4 expression in muscle and adipose tissues. A and B, quantitation of GLUT4 in skeletal muscle homogenates, soleus (A), and EDL (B). Homogenates containing 150, 75, 37.5, and 18.8 g of total protein for IRAPϩ/ϩ samples and 150 and 75 g for IRAPϪ/Ϫ samples were subjected to SDS-polyacrylamide gel electrophoresis and immunoblotted with affinity-purified antibodies raised against the C-terminal peptide of GLUT4 (4). The signals were quantitated by densitometry, and the values obtained for the IRAPϪ/Ϫ samples were expressed as the percentage of the values for those IRAPϩ/ϩ samples for which the signals in the respective dilutions were linear and in the range of the values for the IRAPϪ/Ϫ samples. The data in the graphs represent the means Ϯ S.E. from five male wild type (ϩ/ϩ) and five male IRAPϪ/Ϫ (Ϫ/Ϫ) mice (8 months old). The insets show representative results for two IRAPϩ/ϩ and two IRAPϪ/Ϫ samples (75 g of total protein loaded in each lane). Immunoblots performed on muscle homogenates derived from female mice (n ϭ 3 for each genotype) yielded similar results. C, quantitation of GLUT4 in white adipocyte homogenates. Adipocytes were isolated from five male 12-month-old IRAPϩ/ϩ and IRAPϪ/Ϫ mice as described under "Experimental Procedures." From the resulting suspensions of the cells pooled for each genotype, SDS samples were prepared, and aliquots containing 15 g (2ϫ), 7.5 g (1ϫ), and 3.75 g ( 1 ⁄2ϫ) g of total protein were immunoblotted with antibodies against GLUT4 (shown in the inset). Signals were quantitated as described for A and B. The graph shows the means Ϯ S.E. from three different adipocyte preparations. Immunoblots on two separate preparations from female mice for each genotype yielded similar results. D, quantitation of GLUT4 in heart. Homogenates containing 50, 25, 12.5, and 6.25 g of total protein for IRAPϩ/ϩ samples and 50 and 25 g for IRAPϪ/Ϫ samples were immunoblotted for GLUT4, and the signals were quantitated as described for A and B. The graphs show the means Ϯ S.E. from six 6 -8-month-old male mice for each genotype. The inset shows a representative immunoblot for two IRAPϩ/ϩ and IRAPϪ/Ϫ male mice with 50 g of total protein loaded in each lane. Immunoblots were also performed for female mice. The relative expression of GLUT4 in the female IRAPϪ/Ϫ hearts was 35 Ϯ 6% of that in female IRAPϩ/ϩ hearts (n ϭ 3 for each genotype) (data not shown). these, the decreased expression of GLUT4 is limited to fat. Only in a few animal models was the defect observed in fat and muscle tissues: the Zucker diabetic fatty rat, the diabetic (KK/ A y ) mouse, the viable yellow (A vy /a) mouse, and the obese SHR/N-cp rat (37)(38)(39). For these models evidence was provided that the decrease in the expression of GLUT4 was a consequence of hyperglycemia. IRAPϪ/Ϫ mice do not have elevated blood glucose levels, and thus hyperglycemia cannot be the cause for the decrease in GLUT4 in the IRAPϪ/Ϫ mice. Downregulation of GLUT4 in skeletal muscle, heart, and fat tissues has also been observed in insulin-deficient states, such as fasting and streptozotocin-induced diabetes (40). The decrease in GLUT4 levels in the streptozotocin-induced diabetic animals is due to insulin deficiency per se. Only insulin therapy restores the expression of GLUT4; the normalization of glycemia alone is not sufficient (41). The molecular mechanisms by which insulin regulates GLUT4 levels in vivo are not known. In the IRAPϪ/Ϫ mice, insulin levels are normal (Table I), and our data from the insulin dose response for glucose uptake in adipocytes suggest that insulin sensitivity for the IRAPϪ/Ϫ cells is comparable with wild type cells. It therefore seems unlikely that the decrease in GLUT4 levels in the IRAPϪ/Ϫ mice is due to an absolute or relative insulin deficiency.
Common to insulin-deficient animals and the IRAPϪ/Ϫ mice, however, is the lack of IRAP function at the cell surface. It would be expected that in the former insulin-triggered translocation of IRAP to the cell surface is severely diminished, and in the latter IRAP is lacking entirely. We have previously shown that insulin increases the processing of extracellular substrates concomitant with the translocation of IRAP to the cell surface (18). Thus, when IRAP is absent from the cell surface extracellular peptide hormones that are substrates for IRAP may not be processed efficiently. We thus speculate that the altered action of a peptide hormone that is a substrate for IRAP or is under the control of a substrate for IRAP leads to the down-regulation of GLUT4. We presently do not know what the in vivo substrate for IRAP is, but we have previously shown that IRAP cleaves the N-terminal amino acid from vasopressin, Lys-bradykinin, angiotensin III, and angiotensin IV in vitro (18). The removal of the N-terminal amino acid inactivates vasopressin (42) and converts angiotensin III and angiotensin IV into products with different receptor specificity (43). There is evidence that angiotensins may be involved in the control of GLUT4 levels. An angiotensin type-1 receptor antagonist prevents the decrease in GLUT4 levels in the hearts of streptozotocin-diabetic rats (44). Both angiotensin II and angiotensin III bind to the angiotensin-1 receptor and exert similar effects through the interaction (45). It is possible that in the IRAPϪ/Ϫ mice the half-life of angiotensin III is increased and its prolonged action may lead to the down-regulation of GLUT4.
Another mechanism that could be responsible for the decrease in GLUT4 is that the absence of IRAP leads to a trafficking defect and mistargeting of GLUT4 and subsequently to increased degradation of GLUT4. This defect would most likely be due to the physical absence of IRAP and would be independent of its catalytic activity. The specific intracellular compartment where IRAP and GLUT4 predominantly reside has not been established. One model is that they are in specialized secretory vesicles that fuse with the cell surface when cells are stimulated with insulin (46). Another model is that they are in the endosomal recycling compartment from which they are sorted into vesicles going to the plasma membrane upon stimulation with insulin (46). The sorting mechanism that has to occur in either of the models is unknown. Also, the molecular mechanisms responsible for the sequestration of IRAP and GLUT4 in the intracellular compartment and the insulin-triggered translocation to the cell surface are unknown. The cytoplasmic domain of IRAP contains several potential trafficking motifs (two dileucine motifs and two clusters of acidic amino acids) similar to those found in the C-terminal cytoplasmic tail of GLUT4 (14). For IRAP, we have been able to show that the cytoplasmic tail carries all of the information for its specific intracellular localization and insulin-regulated trafficking (31). Extensive studies with the three major cytoplasmic domains in GLUT4 (the N terminus, the cytoplasmic loop between membrane-spanning domains 6 and 7, and the C terminus) have identified roles for each of these in GLUT4 trafficking (47). Injection of either a GST fusion protein with the cytoplasmic tail of IRAP or a peptide derived from the C-terminal end of GLUT4 triggers the redistribution of both IRAP and GLUT4 to the cell surface (48,49). These findings suggest that the same mechanism may be responsible for their sequestration within the cells under basal conditions. However, the protein(s) interacting with the cytoplasmic domains of IRAP and/or GLUT4 have not been identified. A strong physical interaction between IRAP and GLUT4 has been ruled out; no IRAP is found in immunoprecipitates derived from detergent cell lysates using an antibody against GLUT4 and vice versa (4). Our analysis of subcellular fractions obtained from isolated adipocytes from IRAPϪ/Ϫ and IRAPϩ/ϩ mice does not support a role for IRAP in the sequestration and insulin-stimulated trafficking of GLUT4. Under steady state conditions in basal and insulintreated cells, the relative distribution of GLUT4 between the plasma membrane and the low density microsomes is the same in the IRAPϪ/Ϫ and IRAPϩ/ϩ adipocytes. The subcellular fractionation method we used in this study does not differentiate between specific compartments within the low density microsomes. However, we find that in the IRAPϪ/Ϫ adipocytes, GLUT4 is targeted to an intracellular compartment from which it can be recruited to the cell surface by insulin to the same extent as in wild type cells. The finding of the normal relative subcellular distribution of GLUT4 in the IRAPϪ/Ϫ adipocytes contrasts with the finding for the subcellular distribution for IRAP described in GLUT4 Ϫ/Ϫ adipocytes. In the latter, IRAP is redistributed to the plasma membrane under basal condi- FIG. 8. Subcellular distribution of GLUT4 in basal and insulinstimulated adipocytes. Adipocytes were isolated from twenty 6-month-old male IRAPϪ/Ϫ (Ϫ/Ϫ) and wild type (ϩ/ϩ) mice and incubated with (10 nM) (lanes 6 -10) or without insulin (lanes 1-5). Plasma (PM) and low density microsomal membrane (LDM) fractions were prepared from the adipocytes by differential centrifugation. SDS samples from each of the fractions were immunoblotted for GLUT4. Lanes 1-3 and 8 -10 show fractions derived from wild type adipocytes, and lanes 4 -7 show fractions derived from IRAPϪ/Ϫ cells. Total protein analyzed were for plasma membrane (10 g (1ϫ), 5 g ( 1 ⁄2ϫ), and 2.5 g ( 1 ⁄4ϫ)) and for low density microsomal membranes (5 g (1ϫ), 2.5 g ( 1 ⁄2ϫ), and 1.25 g ( 1 ⁄4ϫ)). The subcellular fractionation and subsequent immunoblotting of the fractions for GLUT4 was repeated for another set of male mice as well as for female mice. Similar results to the ones shown were obtained. Immunoblots for GLUT4 were also carried out on the homogenates from which the plasma and low density microsomes were derived. The results were identical to the ones for the adipocyte cell lysates shown in Fig. 6C. tions, and there is no further increase in IRAP in this fraction in response to insulin (50).
Yet a third explanation for the decrease in GLUT4 in the absence of IRAP is that GLUT4 and IRAP may be routed through a specific common degradation pathway. At present, we do not know what the mechanisms are that guide the degradation of either GLUT4 or IRAP. We have shown that IRAP and GLUT4 are approximately equally abundant in adipocytes (1), and it is therefore conceivable that in the absence of IRAP GLUT4 would be targeted for degradation more efficiently because of a lack of competition. In this instance, it would be expected that in the absence of GLUT4, IRAP levels would be decreased. A recent study reports that IRAP is indeed decreased by 40 and 60% in skeletal muscle and heart, respectively, in GLUT4Ϫ/Ϫ mice (50).
Transgenic mouse models with decreased or lack of expression of GLUT4 in muscle and/or adipose tissues have confirmed the importance of GLUT4 in the maintenance of glucose homeostasis (51)(52)(53)(54). The mice with GLUT4 levels most similar to the ones found in IRAPϪ/Ϫ mice are mice heterozygous for the deletion of GLUT4 in all tissues (GLUT4ϩ/Ϫ) (51). The GLUT4 ϩ/Ϫ mice show 46 and 26% decreases in GLUT4 content in soleus and EDL, respectively, a 75% decrease of GLUT4 in white adipocytes, and no decrease of GLUT4 in heart. Many of the male GLUT4ϩ/Ϫ mice develop hyperinsulinemia with advancing age, and a high proportion of these subsequently become diabetic. It is surprising that in the IRAPϪ/Ϫ mice, similar and even more marked decreases in GLUT4 expression in skeletal muscle, the major site of glucose disposal, were not accompanied by changes in glucose homeostasis. Fed and fasted blood insulin and glucose levels in IRAPϪ/Ϫ mice were comparable with IRAPϩ/ϩ mice at 6 months as well as at 1 year of age. The male IRAPϪ/Ϫ mice only showed an impaired response to the glucose-lowering effect of insulin in the insulin tolerance test. The minimal effect of the 57-85% decrease of GLUT4 in skeletal muscle on glucose homeostasis in the IRAPϪ/Ϫ mice may be explained by the fact that the actual glucose uptake in the different isolated muscle types was not as severely impaired as was expected from the decreases in GLUT4. In soleus, glucose uptake is normal in male and only reduced by 23% in the female despite a decrease in GLUT4 of 57 and 65%, respectively. In EDL, glucose uptake is 60% of wild type, whereas GLUT4 levels are only 15% of wild type levels. For the male GLUT4ϩ/Ϫ mice, glucose uptake into EDL was reduced to a similar extent as in the IRAPϪ/Ϫ mice, but glucose uptake into soleus was also reduced by 38% (51). Euglycemic insulin clamp studies in rats have shown that the red and white fiber type muscles have different sensitivities to insulin to increase glucose uptake (55). In the presence of physiological concentrations of insulin, a maximal increase in glucose uptake in soleus (a predominantly red fiber type muscle), but only a half-maximal increase into EDL (a predominantly white fiber type muscle) is achieved. Only supraphysiological concentrations of insulin led to maximal glucose uptake in EDL. Since glucose uptake into soleus in the IRAPϪ/Ϫ mice is normal or close to normal, glucose homeostasis may be maintained under randomly fed conditions and in response to a glucose load when physiological insulin levels are present. In the insulin tolerance test, when insulin concentrations are raised to supraphysiological levels, the 40% decrease in glucose uptake into the white fiber type muscles in the IRAPϪ/Ϫ mice will cause the apparent decreased clearance of glucose in the IRAPϪ/Ϫ when compared with wild type mice. The situation in IRAPϪ/Ϫ mice may be more comparable with the GLUT4Ϫ/Ϫ mice (56). These mice, in the complete absence of GLUT4, maintain normal glucose homeostasis, and the fe-male mice exhibit close to normal insulin-responsive glucose uptake into isolated soleus (56,57).
The discrepancy in the expression of GLUT4 and the actual glucose uptake measured into the different muscles in IRAPϪ/Ϫ mice may be explained by the increased expression of an alternative insulin-responsive glucose transporter isoform. Such an explanation was proposed in the case of the soleus in the GLUT4Ϫ/Ϫ mice (57). GLUT1, the other well characterized glucose transporter isoform that is somewhat responsive to insulin and is expressed in skeletal muscle tissues, was not altered in the GLUT4Ϫ/Ϫ mice (57). We also could not detect any increase in the levels of GLUT1 in the IRAPϪ/Ϫ tissues. Whether there is another novel glucose transporter isoform expressed in the IRAPϪ/Ϫ mice that compensates for the reduction of GLUT4 remains to be determined. Good candidates are the recently cloned GLUT8, a transporter isoform that has been shown to be insulin-responsive in blastocysts (58), and GLUT11, a transporter isoform that is specifically expressed in skeletal muscle and heart (59).
Alternative explanations for the discrepancy between glucose uptake and GLUT4 levels in the skeletal muscles of the IRAPϪ/Ϫ mice are that in the absence of IRAP more GLUT4 may be at the cell surface, or GLUT4 at the cell surface may have a higher intrinsic activity. Due to technical limitations, the subcellular distribution of GLUT4 has not been determined in muscles. However, we determined the subcellular distribution of GLUT4 in adipocytes. As described above, we could not find any differences in the relative distribution of GLUT4 between the IRAPϪ/Ϫ and IRAPϩ/ϩ adipocytes. Indeed, in adipocytes there is a direct correlation of glucose uptake with the relative levels of total GLUT4 and the relative levels of GLUT4 protein at the cell surface. This confirms earlier reports that have shown that the relative levels of GLUT4 at the cell surface correlate directly to the relative glucose uptake measured in basal and insulin-stimulated adipocytes (10). The same has been found to be the case for skeletal muscle, indicating a similar dependence on the presence of GLUT4 in skeletal muscle (11). Our data in adipocytes also indicate that in the absence of IRAP the intrinsic activity of GLUT4 is the same as in wild type cells.
In all of the mouse models with decreased or lack of GLUT4 expression, cardiac hypertrophies have been observed (52,53,56,57). The nature of the enlargement of the heart in the IRAPϪ/Ϫ mice needs to be further evaluated. Whether it is due to the decrease in GLUT4 expression or whether it is a consequence of the impaired processing of a peptide hormone that is a substrate for IRAP remains to be determined. It is noteworthy that all of the peptide hormones found to be substrates for IRAP in vitro are vasoactive peptide hormones (18). If one of these is an in vivo substrate for IRAP and it is not processed properly in the IRAPϪ/Ϫ mice, cardiovascular abnormalities may develop.
IRAP is the homologue of the human P-LAP, also known as human oxytocinase. Expression levels of the human isoform are high in placenta (20). It has been shown that the expression of P-LAP in placenta increases during pregnancy (60,61). Concomitantly, increased P-LAP activity can be detected in maternal serum in both humans and mice (60,61). Furthermore, it has been shown that decreased P-LAP activity in maternal serum correlates with spontaneous preterm delivery (62). Our observations on the breeding pattern of the IRAPϪ/Ϫ mice indicate that the IRAPϪ/Ϫ females do not have any problems in maintaining pregnancies and do not deliver their pups prematurely. IRAP thus does not play a critical role in mouse pregnancy.
The recruitment of IRAP to the cell surface is impaired in type 2 diabetics (63), and consequently it would be expected that IRAP action at the cell surface is diminished in these individuals. Type 2 diabetics show decreased expression of GLUT4 in adipose tissues (36), and the development of cardiovascular complications in diabetics is common (64). The changes we observed in the initial characterization of the IRAPϪ/Ϫ mice, decreased expression of GLUT4 and the enlargement of the heart, suggest that impaired function of IRAP at the cell surface may play a role in the development of complications in insulin-resistant individuals.
The initial characterization of the IRAPϪ/Ϫ mice reported herein provides strong evidence for a role of IRAP in insulin action. Further characterization of the molecular mechanisms that are responsible for the changes observed in the IRAPϪ/Ϫ mice will provide novel insights into the physiological function of IRAP.