Macrophage-derived human resistin exacerbates adipose tissue inflammation and insulin resistance in mice (original) (raw)
Derivation and initial characterization of humanized resistin mice. Prior to generating the humanized transgenic mice, we showed that human resistin is able to activate the mouse resistin signaling pathway by inducing Socs3 gene expression in 3T3L1 adipocytes (Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI37273DS1) (19). To better understand the role of human resistin in the pathophysiology of insulin resistance and diabetes in humans, we derived transgenic mice on the C57BL/6 background with macrophage-specific expression of human resistin using the human CD68 promoter (Figure 1A) (36). Several independent lines were generated, 2 of which were established and bred to C57BL/6 Retn–/– mice to generate mice that express human resistin but lack any expression of murine resistin (Retn) (20). The line we have studied the most, referred to hereafter as “humanized resistin mice,” had circulating human resistin levels that are comparable to pathophysiological levels seen in humans (48.75 ± 7.50 ng/ml) and were compared with their Retn–/– littermates (“controls,” hereafter) (37, 38). The second line had higher human resistin levels (141.72 ± 11.45 ng/ml) and were designated “the second line of humanized resistin mice.” The humanized resistin mice were viable, were born in the expected Mendelian ratios (Supplemental Table 1), and appeared normal. As in humans, the highest level of tissue expression of human resistin was seen in macrophages, followed by high expression in bone marrow as well as peripheral blood mononuclear cells (Figure 2B). There was also significant expression in tissues that harbor large numbers of macrophages, such as spleen, lung, and fat. The second line of humanized resistin mice had similar tissue expression patterns (Supplemental Figure 2A). The expression of human resistin in fat was predominantly in the stromal vascular fraction (SVF) compared with adipocytes, analogous to the situation in humans (Figure 1B) (39).
Derivation and initial characterization of humanized resistin mice. (A) CD68 promoter construct used to generate the mice. The CD68/human resistin cassette was excised using SalI and BsaAI, and the construct was injected into C57BL/6 fertilized eggs. Arrows represent the locations of the primers used to genotype the transgenic mice. (B) Expression profile of human resistin in different tissues in the humanized resistin mice as well as its expression profile in the different fractions of WAT (inset). Real-time RT-PCR was used to measure mRNA levels. Data are expressed as mean ± SEM; n = 5. S. int, small intestine; L. int, large intestine.
Glucose homeostasis in humanized resistin mice. (A) Humanized resistin mice have decreased glucose tolerance and (B) decreased insulin sensitivity. *P < 0.05. Data are expressed as mean ± SEM; n = 16.
Lipid dysregulation in humanized resistin mice. The humanized resistin mice were studied under conditions of normal-chow feeding as well as on high-fat–diet feeding. There was no difference in weight gain or adiposity between the humanized resistin mice and the controls during the time frame of our experiments (Supplemental Figure 3). In the normal-chow–fed mice, there were also no appreciable differences in serum metabolic parameters, including serum triglycerides, FFAs, glycerol, and blood glucose and insulin levels (data not shown). However, with high-fat feeding, serum insulin levels were significantly higher in the humanized resistin mice, while serum glucose levels were higher, but the difference did not reach statistical significance (Table 1). Moreover, the humanized resistin mice had significantly higher serum levels of FFAs and glycerol, suggestive of increased lipolysis in these animals. Similar results were obtained with the second line of humanized resistin mice (Supplemental Figure 2B). The humanized resistin mice also had significantly lower triglycerides compared with the control animals, which may be explained by increased skeletal muscle lipoprotein lipase (Lpl) activity (see below).
Plasma metabolic profile of humanized resistin mice
Insulin resistance and decreased glucose disposal rate in the humanized resistin mice. There was no difference in glucose homeostasis between genotypes in the normal-chow–fed mice (Supplemental Figure 4). However, with high-fat feeding, glucose-tolerance test (GTT) revealed that the humanized resistin mice had impaired glucose metabolism compared with mice not expressing human resistin (Figure 2A). These mice also showed a general decrease in insulin sensitivity in the insulin tolerance tests (Figure 2B). The animals also showed higher levels of serum insulin during GTT, which reflects the higher glucose levels in these animals under GTT conditions and is consistent with their prediabetic state (Supplemental Figure 5).
In order to ascertain the specific tissue defect responsible for the insulin resistance seen in the humanized resistin mice, we performed hyperinsulinemic-euglycemic clamp studies (Figure 3). The glucose-infusion rate (GIR) required to maintain euglycemia in the humanized resistin mice was 39% lower than in control littermates not expressing human resistin, indicating marked whole-body insulin resistance in this group. The exacerbated insulin resistance was due to decreased rate of disposal in the humanized resistin mice compared with controls (Figure 3A). Similar results were obtained with the second line of humanized resistin mice (Supplemental Figure 2, C and D). Given the defect in the rate of glucose disposal, we investigated the rate of glucose uptake in muscle and WAT, the main tissues responsible for glucose uptake, as well as insulin signaling in these tissues. Indeed, insulin-stimulated glucose uptake in skeletal muscle and WAT were significantly decreased (56% and 22%, respectively; Figure 3B). This was accompanied by decreased tyrosine phosphorylation of Irs-1 (Figure 3C). Insulin-stimulated Akt phosphorylation was also attenuated in the muscle of the humanized resistin mice (Supplemental Figure 6). The resistance to insulin action in WAT was also evident by the decreased suppression of FFA release in response to insulin in the hyperinsulinemic clamp setting, as humanized resistin mice had serum FFA levels significantly higher than those of control mice not expressing human resistin (0.304 ± 0.033 versus 0.208 ± 0.009 mEq/l).
Humanized resistin mice are insulin resistant. (A) Hyperinsulinemic-euglycemic clamp analysis of humanized resistin mice versus Retn–/– controls. (B) Rate of insulin-stimulated glucose uptake in muscle and WAT. (C) Insulin-stimulated tyrosine phosphorylation of IRS-1 in muscle and WAT normalized to controls. Rd, rate of glucose disposal; HGP, hepatic glucose production; pan-Tyr, pan tyrosine. *P < 0.05; **P < 0.01. Data are expressed as mean ± SEM; n = 8.
Activation of the Pkcθ pathway in skeletal muscle in the humanized resistin mice. To understand the mechanism of the exacerbated skeletal muscle insulin resistance, we next evaluated gene-expression changes in the insulin-resistant muscles of the humanized resistin mice. No differences were observed in the expression of several inflammatory cytokines (Figure 4A). In contrast, Lpl expression was approximately 2-fold higher in the skeletal muscles of the humanized resistin mice (Figure 4A), which was accompanied by a significant increase in Lpl enzyme activity (Figure 4B). Lpl is the rate-limiting enzyme for the uptake of triglyceride-derived fatty acids from the serum (40), which may explain why serum triglycerides were reduced in the humanized resistin mice (Table 1 and Supplemental Figure 2B). Increased skeletal muscle Lpl causes insulin resistance due to increased accumulation of intramyocellular lipids and lipid metabolites, including diacylglycerols (DAG) (41). Indeed, triglyceride levels were approximately 52% higher in the muscle of humanized resistin mice (Figure 4C). Consistent with this, skeletal muscle DAG in humanized resistin mice on high-fat diets was approximately 85% higher than in controls (Figure 4D). DAG activates Pkcθ, which impairs insulin signaling by increasing the inhibitory serine phosphorylation of Irs-1 in muscle (42), and Pkcθ activity was found to be increased in the skeletal muscle of the humanized resistin mice (Figure 4E). No difference was seen in intrahepatic triglyceride levels or Lpl expression in the livers of these animals (Supplemental Figure 7).
Increased lipid accumulation in the humanized resistin mice. (A) Gene-expression analysis in muscle of humanized resistin mice compared with control mice. (B) Heparin-released muscle Lpl activity per mg protein. (C) Intramyocellular triglycerides concentration in skeletal muscle. (D) Intracellular muscle DAG levels. (E) Muscle Pkcθ activity expressed as membrane/cytosol ratio of muscle Pkcθ protein levels. TGs, triglycerides. *P < 0.05; **P < 0.01. Data are expressed as mean ± SEM; n = 9.
Increased WAT inflammation and macrophage infiltration in the humanized resistin mice. We also performed gene-expression analysis in WAT after high-fat–diet feeding. The humanized resistin mice had increased expression of several inflammatory markers in WAT, including Tnf-α, Il-6, Il-1β, Cxcl5, monocyte chemotactic protein-1 (Mcp-1), and Cd68 (Figure 5A). This is reminiscent of the inflammation seen in humans during obesity, which has been shown to contribute to the insulin resistance that accompanies this condition (43–45). In addition, there was a modest but significant decrease in the expression of phosphodiesterase 3b (Pde3b), the major enzyme responsible for the breakdown of cAMP and hence attenuation of hormone-sensitive lipase (Hsl) activity in fat cells in these animals. This might be secondary to the increase in Tnf-α in the WAT of the humanized resistin mice and is likely to have contributed to their increased lipolysis (Table 1) (46). The increase in Cd68 expression in the WAT of the humanized resistin mice indicated an increase in the accumulation of macrophages in this tissue. Indeed, immunohistochemical analysis demonstrated that humanized resistin mice had higher accumulation of WAT macrophages (Figure 5B). This is likely related to the increased expression of Mcp-1, which contributes to macrophage infiltration into WAT (47, 48) and has been shown to be induced in endothelial cells treated with human resistin (49). One of the mechanisms by which inflammation leads to impairment in insulin signaling is through the inhibitory phosphorylation of Irs-1 on serine 307 (ser307) (50, 51). Consistent with the increased inflammation, ser307 phosphorylation of Irs-1 was increased in the WAT of humanized resistin mice (Figure 5C).
Inflamed WAT in humanized resistin mice. (A) Gene-expression analysis in WAT of humanized resistin and control mice. (B) Immunohistochemical detection of the macrophage-specific antigen F4/80 (white arrows) in epididymal adipose tissue from humanized resistin and control mice. Original magnification, ×200 (upper panels); ×400 (lower panels). (C) Ser307 phosphorylation levels of Irs-1 in WAT normalized to controls. *P < 0.05; **P < 0.01. Data are expressed as mean ± SEM; n = 9.
Adipose tissue inflammation and increased lipolysis precede the skeletal muscle changes in humanized resistin mice. Thus far, we have shown that after 2–3 weeks on the high-fat diet, the humanized resistin mice had increased inflammation in WAT as well as skeletal muscle insulin resistance. To address the question of which of these metabolically deleterious effects of human resistin are primary, humanized resistin mice were evaluated after only 4 days on the high-fat diet. Remarkably, even at this early time, several inflammation markers were already increased in the WAT of humanized resistin mice, including Tnf-α, Mcp-1, and the macrophage marker Cd68 (Figure 6A). Also at this time, serum FFAs and glycerol were higher in the humanized resistin mice (Table 2), suggestive of increased lipolysis in these animals. Consistently, ex vivo analysis of WAT from the humanized resistin mice revealed increased basal lipolysis with reduced suppression by insulin (Figure 6B). This insulin resistance was paralleled by reduced tyrosine phosphorylation of Irs-1 and increased inhibitory serine phosphorylation of Irs-1 in the WAT of the humanized resistin mice (Figure 6C). Among the cytokines increased in the inflamed WAT of humanized resistin mice, Tnf-α in particular is known to induce lipolysis by suppressing expression of Pde3b, thereby activating Hsl by increasing its phosphorylation (52, 53). Indeed, Hsl phosphorylation on ser563, an index of HSL activity, was increased in humanized resistin mice (Figure 6, D and E, respectively). In contrast, at these early times, no difference was observed in serum triglyceride levels, muscle Lpl expression, muscle triglyceride levels, or muscle DAG levels (Table 2 and Supplemental Figure 8, respectively). In hyperinsulinemic-euglycemic clamp studies, the humanized resistin mice showed a tendency toward lower GIR and lower rate of disposal, but this difference did not reach statistical significance (Supplemental Figure 9). Thus, increased inflammation and lipolysis in WAT precedes the insulin resistance and the accumulation of lipid in skeletal muscle of the humanized resistin mice.
WAT inflammation and lipolysis precede the muscle changes in humanized resistin mice. (A) Gene-expression analysis in WAT of humanized resistin and control mice after short-term (4-day) high-fat–diet feeding. (B) Ex vivo lipolysis measured by glycerol release normalized to PBS-treated samples from control mice. (C) Pan tyrosine and ser307 phosphorylation of WAT IRS-1 normalized to controls. (D) Western blot analysis of Hsl ser563 phosphorylation in WAT of humanized resistin mice compared with control mice. (E) Quantitation of the blot in D normalized to controls. p-ser563, phosphor-ser563. *P < 0.05. Data are expressed as mean ± SEM; n = 6–8.
Plasma metabolic profile of humanized resistin mice on high-fat diet for 4 days
High-fat diet increases human resistin in WAT. We next assessed the mechanism by which high-fat feeding exacerbates the insulin resistance phenotype in the humanized resistin mice. In control mice, the expression of a macrophage-specific marker, Cd68, was increased during both short-term (4 days) and longer-term (3 weeks) high-fat feeding (Figure 7A). In the humanized resistin mice, high-fat feeding increased WAT Cd68 expression much more dramatically (Figure 7A). As expected, this increase in WAT macrophage infiltration led to a significant increase in the expression of human resistin in the WAT of the humanized resistin mice on short-term (5-fold) and longer-term (11.7-fold) high-fat feeding compared with normal-chow–fed animals (Figure 7B). These data are consistent with the conclusion that high-fat feeding increases WAT macrophage infiltration, and this is markedly increased when the macrophages express human resistin.
High-fat feeding increases WAT macrophage infiltration and human resistin expression in humanized resistin mice. (A) Gene-expression analysis of Cd68 in WAT of humanized resistin and control mice on high-fat diet normalized to the expression in mice on normal chow. (B) Expression of human RETN in WAT of humanized resistin mice on high-fat diet normalized to its expression in WAT of mice on normal chow. NC, normal-chow diet; HF, high-fat diet. *P < 0.05; **P < 0.01. Data are expressed as mean ± SEM; n = 5–9.