Increased β-Oxidation but No Insulin Resistance or Glucose Intolerance in Mice Lacking Adiponectin*

Previous reports showed that recombinant fragments of adiponectin (adipo) displayed pharmacological effects when injected into rodents, but the relevance of these observations to the physiological function of adipo is unclear. We generatedAdipo −/− mice by gene targeting.Adipo −/− mice are fertile with normal body and fat pad weights. Plasma glucose and insulin levels ofAdipo −/− and Adipo +/+mice are similar under fasting conditions and during an intraperitoneal glucose tolerance test (GTT). Insulin tolerance test (ITT) also produces similar plasma glucose and insulin levels in the two groups of mice. Hyperinsulinemic-euglycemic clamp analysis showed thatAdipo −/− and Adipo +/+mice have similar glucose infusion rates to maintain a similar serum glucose. High-fat diet feeding for 7 months led to similar weight gain and similar GTT and ITT responses. We next measured β-oxidation and found it to be significantly increased in muscle and liver ofAdipo −/− mice. In conclusion, our study indicates that absence of adipo causes increased β-oxidation but does not cause glucose intolerance or insulin resistance in mice.

Previous reports showed that recombinant fragments of adiponectin (adipo) displayed pharmacological effects when injected into rodents, but the relevance of these observations to the physiological function of adipo is unclear. We generated Adipo ؊/؊ mice by gene targeting. Adipo ؊/؊ mice are fertile with normal body and fat pad weights. Plasma glucose and insulin levels of Adipo ؊/؊ and Adipo ؉/؉ mice are similar under fasting conditions and during an intraperitoneal glucose tolerance test (GTT). Insulin tolerance test (ITT) also produces similar plasma glucose and insulin levels in the two groups of mice. Hyperinsulinemic-euglycemic clamp analysis showed that Adipo ؊/؊ and Adipo ؉/؉ mice have similar glucose infusion rates to maintain a similar serum glucose. High-fat diet feeding for 7 months led to similar weight gain and similar GTT and ITT responses. We next measured ␤-oxidation and found it to be significantly increased in muscle and liver of Adipo ؊/؊ mice. In conclusion, our study indicates that absence of adipo causes increased ␤-oxidation but does not cause glucose intolerance or insulin resistance in mice.
Adiponectin (adipo, also known as Acrp30, adipoQ, GBP28, and apM1) 1 is a major adipocyte secretory protein of unknown function (1)(2)(3). Injection of full-length and partial-length fragments of recombinant adipo in rodents was shown to suppress hepatic glucose production (4,5) and increase fatty acid oxidation in muscle (6,7). The plasma concentration of adipo is reduced in obesity and type 2 diabetes (8). These observations led to the proposal that adipo may play important roles in glucose homeostasis and insulin resistance.
To elucidate the in vivo role of adipo, we generated Adipo Ϫ/Ϫ mice by gene targeting. We found that these mice had no glucose intolerance or insulin resistance under basal conditions or even after they were fed a high-fat diet for 7 months. Unexpectedly, we found that Adipo Ϫ/Ϫ mice had increased ␤-oxidation in muscle and liver. This study sheds light on the function of native adipo in vivo.

EXPERIMENTAL PROCEDURES
Generation of Targeted Mice-We used a replacement-type targeting vector constructed from a mouse 129/Sv strain bacterial artificial chromosome genomic clone and R1 ES cell line for gene targeting (Fig. 1A). Transfection and ES cell clone selection were as described (9). Nine positive ES clones were injected into blastocysts of C57BL/6J, and chimeric mice obtained were mated with C57BL/6J mice. Germ line transmission was confirmed by Southern blots using tail DNA.
Northern Blot and Immunoblot Analysis-Northern blots were performed on total RNA isolated from white and brown adipose tissue using a 32 P-labeled mouse full-length cDNA probe as described (9,10). Glyceraldehyde 3-phosphate dehydrogenase cDNA was used as a control. Polyclonal rabbit antibody was raised against a C-terminal fragment of mouse adipo (residues 110 -247) fused to glutathione S-transferase. Immunoblotting were performed as described previously using 3 l of plasma (9).
Glucose Tolerance Test and Insulin Tolerance Test-Intraperitoneal glucose tolerance test (GTT), using 1.5 g of glucose/Kg and intraperitoneal insulin tolerance test (ITT) using 1 unit of insulin/Kg were performed in mice fasted for 16 h as described (9,11).
Hyperinsulinemic-Euglycemic Clamp-We measured in vivo glucose utilization by the hyperinsulinemic-euglycemic clamp method as described previously (12) with slight modification. Mice received an insulin infusion (10 milliunits/kg/min) for 90 min. The infusion rate of glucose solution (4%) was adjusted to a target plasma glucose level of 100 mg/dl. Plasma glucose was monitored every 3 min for 90 min. Total body glucose infusion rate was calculated as described (12).
Adiponectin-deficient Mice Have No Glucose Intolerance or Insulin Resistance-Intraperitoneal GTT performed on F2 and F3 mice revealed no difference between Adipo ϩ/ϩ mice and Adipo Ϫ/Ϫ littermates (data not shown). To exclude the possibility that the uneven mixed genetic background of these earlygeneration mice might have masked subtle changes in glucose homeostasis, we backcrossed the mice into C57BL/6J and performed the experimental analyses in F6 mice. Like their F6 wild-type littermates, both male and female F6 Adipo Ϫ/Ϫ mice (with Ͼ99% C57BL/6J background) had normal fasting plasma glucose and insulin levels; furthermore, there was no difference in these parameters during a 2-h GTT (Fig. 2, A and B). We next performed an ITT, again detecting no difference in the plasma glucose response to insulin between F6 Adipo Ϫ/Ϫ mice and Adipo ϩ/ϩ littermates (Fig. 2C), which suggests that there was no impairment in insulin sensitivity.
Hyperinsulinemic-Euglycemic Clamp Analysis Indicates No Insulin Resistance in Adiponectin-deficient Mice-The GTT and ITT suggest that Adipo Ϫ/Ϫ mice have no significant glucose intolerance or insulin resistance. To assess insulin resistance with a more stringent test, we studied Adipo Ϫ/Ϫ and wild-type mice by the hyperinsulinemic-euglycemic clamp technique. As shown in Table IB, Adipo Ϫ/Ϫ and Adipo ϩ/ϩ had similar serum steady-state insulin concentration. To maintain a similar serum glucose, the glucose infusion rate tended to be higher in Adipo Ϫ/Ϫ mice (suggesting that they tended to be more sensitive to insulin), although the difference between Adipo Ϫ/Ϫ and Adipo ϩ/ϩ mice was not significant. Therefore, compared with Adipo ϩ/ϩ mice, Adipo Ϫ/Ϫ mice displayed no insulin resistance.
Adiponectin-deficient Mice Fed a HF/HS Diet for 7 Months Do Not Develop More Insulin Resistance than Wild-type Controls-To bring out potential subtle changes in Adipo Ϫ/Ϫ mice that were not evident under basal conditions, we fed these mice a HF/HS diet for 7 months. Plasma lipids of Adipo Ϫ/Ϫ and Adipo ϩ/ϩ mice were similar (mean Ϯ S.E., triglyceride: Ϫ/Ϫ, 73.08 Ϯ 10.81 mg/dl; ϩ/ϩ, 79.60 Ϯ 9.23 mg/dl; cholesterol: Ϫ/Ϫ, 238.88 Ϯ 28.84 mg/dl; ϩ/ϩ, 261.71 Ϯ 24.94 mg/dl; free fatty acids: 1.67 Ϯ 0.53 mM; ϩ/ϩ, 1.69 Ϯ 0.24 mM) while they were maintained on this diet. They also had similar body weights throughout the 7-month feeding period (data not shown). Both wild-type and knock-out mice developed mild fasting hyperglycemia and fasting hyperinsulinemia (compare Figs. 2 and 3), indicating that the diet-induced weight gain produced similar degrees of mild insulin resistance in mice that produced adipo as well as those that lacked the protein. The plasma glucose and insulin response of both types of mice to GTT was also similar (Fig. 3, A and B). Although the plasma insulin in wild-type mice was higher than that of Adipo Ϫ/Ϫ mice during the GTT, the difference was not significant (Fig. 3B). We next performed an ITT, which revealed that Adipo Ϫ/Ϫ mice and their Adipo ϩ/ϩ littermates had a similar plasma glucose response to insulin (Fig. 3C). Therefore, there was no difference in insulin sensitivity (or resistance) in Adipo ϩ/ϩ or Adipo Ϫ/Ϫ mice. This was true irrespective of whether the comparison was made while the animals were on regular chow or after they were fed a HF/HS diet for 7 months.
Absence of Adiponectin Stimulates ␤-Oxidation-As recombinant fragments of adipo were reported to stimulate ␤-oxidation in muscle of rodents (6, 7), we measured the ␤-oxidation activity of soleus muscle isolated from Adipo Ϫ/Ϫ and Adipo ϩ/ϩ mice. Unexpectedly, we found that ␤-oxidation rate was significantly increased (by ϳ47%) in the muscle of Adipo Ϫ/Ϫ mice compared with wild-type littermate controls (Table IC). The ␤-oxidation rate in liver homogenates of these animals was also determined and was found to be ϳ30% higher in Adipo Ϫ/Ϫ mice than that in wild-type littermate controls. Therefore, the absence of adipo stimulates ␤-oxidation.

DISCUSSION
While we were preparing this manuscript for publication, Kubota et al. (15) reported that F1 and F2 Adipo knock-out mice displayed insulin resistance; they did not examine ␤-oxidation in their animals. The reason for the difference between our study and that of Kubota et al. (15) is unclear, but may be related to environmental factors or genetic background. We did not detect any significant difference in glucose homeostasis in F2 and F3 mice produced in our laboratory. The variation among F2 and F3 mice was higher than that among F6 C57BL/6J mice, presumably because of the inhomogeneous genetic background in the former groups. None of these differences were found to be significant in large groups. To rule out the mixed genetic background as a factor that might have masked differences between knock-out and wild-type mice, we bred them into C57BL/6J background and re-examined the mice at the F6 generation. Again, there was no difference in glucose tolerance or insulin sensitivity between Adipo Ϫ/Ϫ and Adipo ϩ/ϩ mice (Figs. 2 and 3). This lack of a difference was evident in regular chow-fed mice as well as in mice fed a HF/HS diet for 7 months. The high-fat diet did induce a mild, but similar, degree of glucose intolerance and insulin resistance in Adipo Ϫ/Ϫ and Adipo ϩ/ϩ mice (Fig. 3). Interestingly, both Kubota et al. (15) and we observed no difference between knock-out and wild-type mice in their plasma insulin level in the postabsorptive, i.e. fasting state, or after a glucose load. In both studies the plasma insulin during a GTT tended to be (insig- PGKneobpA cassette (Neo) was inserted between the BglII and EcoRI site, replacing exon 2 that contains the start codon. A pMC1-TK-poly(A) cassette was attached to the 5Ј end of the targeting construct. A genomic fragment 5Ј to the targeting construct (horizontal bar marked Probe) was used for genotyping. The expected sizes of the wild-type (12 kb) and the targeted allele (6 kb) were as indicated. B, Southern blot of the positive (ϩ/Ϫ) and negative (ϩ/ϩ) ES cell clones after selection. C, Southern blot of targeted allele of tail DNA of F2 mice. D, Northern blot of white (WAT) and brown (BAT) adipose tissue adipo and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNAs. E, immunoblot analysis of wild-type and Adipo Ϫ/Ϫ mouse plasma. Position of plasma adipo is indicated by an arrow. The anti-adipo antibody used was directed against the C-terminal 138-amino acid fragment. nificantly) higher in the wild-type mice than in Adipo Ϫ/Ϫ mice (Fig. 2J in Kubota et al. (15) and Figs. 2B and 3B in this study). These are unlikely scenarios if significant insulin resistance existed. The absence of insulin resistance in Adipo Ϫ/Ϫ mice was also confirmed by the hyperinsulinemic-euglycemic clamp method in the current study (Table IB).
How can we reconcile the absence of insulin resistance in Adipo Ϫ/Ϫ mice and the previously reported effects of full-length and partial-length adipo fragments injected into rodents (6, 7)?
The overall phenotype of Adipo Ϫ/Ϫ mice could be the result of the activation or overexpression of compensatory biochemical pathways or molecules in Adipo Ϫ/Ϫ mice that reversed the insulin resistance, if indeed adipo is a natural insulin-sensitizing hormone in vivo. The expression of leptin is largely unchanged in these mice, but there are potentially many other molecules involved in insulin resistance that could be activated in the absence of adipo (16). Another consideration is the fact that the data obtained in many of the experiments involving the injection of recombinant adipo into rodents may not reflect the normal action of the native protein. Adipo normally exists in plasma as trimers that associate noncovalently to form high molecular complexes, and most of the injected fragments were not competent to reassemble into the native multimeric complexes in vivo (17). Furthermore, adipo undergoes posttranslational modification (17,18), which does not occur in bacterially derived recombinant adipo. Adipo contains tissue-specific ␣2,8-linked di/oligosialic acid chains that appear to be adipocyte-specific (18); recombinant adipo produced in non-adipocyte mammalian cells may not contain this specific modification. Such ␣2,8-linked di/oligosialic acid chains have been shown to be involved in signal transduction as well as other biological activities (19,20). The fact that increased ␤-oxidation was observed in mice that received recombinant adipo fragments (6, 7), but also occurred in Adipo Ϫ/Ϫ mice that lacked native adipo, suggests that the injected fragments might have acted as dominant negative molecules that blocked the normal action of native adipo in vivo. Additional experiments will be needed to address this important issue.
While this paper was under review, Maeda et al. (21) reported the production of Adipo Ϫ/Ϫ mice in their laboratory. In agreement with our study, they observed no evidence of insulin resistance when Adipo Ϫ/Ϫ mice were fed a regular chow. However, they found that feeding the Adipo Ϫ/Ϫ mice a HF/HS diet for 2 weeks induced insulin resistance in these animals. The reason for the difference between our study, which involved a 7-month HF/HS diet feeding period, and that of Maeda et al. (21) is unclear.