Hyperglycemic Ins2AkitaLdlr⁻/⁻ mice show severely elevated lipid levels and increased atherosclerosis: a model of type 1 diabetic macrovascular disease.

Accelerated atherosclerosis is the leading cause of death in type 1 diabetes, but the mechanism of type 1 diabetes-accelerated atherosclerosis is not well understood, in part due to the lack of a good animal model for the long-term studies required. In an attempt to create a model for studying diabetic macrovascular disease, we have generated type 1 diabetic Akita mice lacking the low density lipoprotein receptor (Ins2AkitaLdlr−/−). Ins2AkitaLdlr−/− mice were severely hyperglycemic with impaired glucose tolerance. Compared with Ldlr−/− mice, 20-week-old Ins2AkitaLdlr−/− mice fed a 0.02% cholesterol AIN76a diet showed increased plasma triglyceride and cholesterol levels, and increased aortic root cross-sectional atherosclerotic lesion area [224% (P < 0.001) in males and 30% (P < 0.05) in females]. Microarray and quantitative PCR analyses of livers from Ins2AkitaLdlr−/− mice revealed altered expression of lipid homeostatic genes, including sterol-regulatory element binding protein (Srebp)1, liver X receptor (Lxr)α, Abca1, Cyp7b1, Cyp27a1, and Lpl, along with increased expression of pro-inflammatory cytokine genes, including interleukin (Il)1α, Il1β, Il2, tumor necrosis factor (Tnf)α, and Mcp1. Immunofluorescence staining showed that the expression levels of Mcp1, Tnfα, and Il1β were also increased in the atherosclerotic lesions and artery walls of Ins2AkitaLdlr−/− mice. Thus, the Ins2AkitaLdlr−/− mouse appears to be a promising model for mechanistic studies of type 1 diabetes-accelerated atherosclerosis.

under a protocol approved by the Institutional Animal Care and Use Committee in a specifi c pathogen-free environment in rooms with a light-dark cycle. Experimental mice were weaned and separated by gender at 28 days of age and fed a semisynthetic, modifi ed AIN76 diet containing 0.02% cholesterol (Research Diets, New Brunswick, NJ) for 16 weeks until euthanasia at 20 weeks of age.

Intraperitoneal glucose tolerance test (IPGTT)
Three days prior to sacrifi ce mice were fasted for 6 h following the dark (feeding) cycle with free access to water. After fasting, mice were injected intraperitoneally with 2 mg glucose/g body weight. Plasma glucose values were obtained from venous blood from a small tail clip at 0, 15, 30, 60, and 120 min using the Ascensia Elite XL (Bayer HealthCare, Tarrytown, NY) handheld blood glucometer.

Tissue preparation
On the day of euthanasia, mice were fasted for 6 h following the dark (feeding) cycle. Immediately prior to euthanasia, the fasting plasma glucose was measured, and then mice were anesthetized by intraperitoneal injection with sodium pentobarbital (Henry Schein, Melville, NY). Mice were exsanguinated by left-ventricular puncture, and blood was collected into EDTAcontaining syringes. Plasma was prepared by spinning at 16,000 g for 10 min. The circulation was fl ushed with PBS, and the heart was removed and stored frozen in Tissue-Tek OCT compound as we described before ( 19 ). Liver and other tissues were collected and stored in RNAlater solution (Life Technologies, Carlsbad, CA).

Blood analysis
Total cholesterol concentrations were determined enzymatically by a colorimetric method (Roche, Indianapolis, IN). Lipoproteins fractions were isolated by spinning 60 l of plasma in a TL-100 ultracentrifuge (Beckman Coulter, Brea, CA) at its own density (1.006 g/ml) at 70,000 RPM for 3 h to harvest the supernatant and after adjusting the infranatant with solid KBr to a density of 1.063 g/ml, then spinning it for 70,000 RPM for 18 h to harvest the supernatant. The cholesterol content of each supernatant and the fi nal infranatant were measured and taken to be VLDL (<1.006 g/ml), LDL (1.006 р d р 1.063 g/ml), and HDL (d >1.063 g/ml) cholesterol. Cholesterol concentrations in all three fractions were then determined enzymatically by a colorimetric method (Roche, Indianapolis, IN). Plasma triglyceride levels were determined enzymatically in the original plasma sample. Plasma insulin was measured using a rat/mouse insulin ELISA kit (EZRMI-13K, Linco Research, St. Charles, MO).

Quantifi cation of atherosclerosis
To quantify atherosclerosis at the aortic root, OCT-embedded hearts were sectioned and stained with oil red O as described ( 19 ). The heart was oriented so that the three valves of the aortic root were in the same plane, and 12 m sections were saved onto glass slides. Sections were stained with oil red O. The lesion area was quantifi ed in every fourth section, and the average was reported for fi ve measurements.

RNA isolation and quantitative real-time PCR analysis
Total RNA was isolated from the livers of male Ins2 Akita Ldlr Ϫ / Ϫ

and Ldlr
Ϫ / Ϫ mice using the RNeasy mini kit (Qiagen USA, Valencia, CA) according to the manufacturer-supplied protocol. Quantitative real-time PCR (QPCR) was performed using gene-specifi c primers and the SYBR green PCR kit (Life Technologies, Carlsbad, CA) in an ABI 7900 system (Life Technologies) as described before ( 20 ). All samples were quantifi ed using the comparative The Akita mouse (Ins2 Akita ) carries a single copy of a dominant mutation in the Ins2 gene (Cys96Tyr). This mutation disrupts intramolecular disulfi de bond formation causing improper folding of proinsulin. Proinsulin accumulates intracellularly and, by engorging the endoplasmic reticulum (ER) and triggering the ER stress response, leads to apoptosis of pancreatic ␤ -cells ( 13 ). Despite the co-expression of a normal insulin gene allele, by 3 to 4 weeks of age, Ins2 Akita mice exhibit hypoinsulinemia, hyperglycemia, polydipsia, and polyuria in the absence of obesity (14)(15)(16). On the C57BL/6J background on a chow diet, Ins2 Akita mice have persistent hyperglycemia with fasting blood glucose levels of greater than 400 mg/dl (14)(15)(16). The Ins2 Akita model has been used to study diabetic microvascular complications, such as retinopathy, neuropathy, and nephropathy ( 9 ). However, macrovascular diabetic complications, such as atherosclerotic cardiovascular disease, have not been examined.
In our initial studies we found Ins2 Akita mice on the C57BL/6J background fed the 0.02% cholesterol AIN76a diet (low cholesterol, low fat) from weaning to 20 weeks of age averaged total cholesterol levels of ‫ف‬ 112 mg/dl and triglycerides of 52 mg/dl and had no signs of atherosclerotic lesions at the aortic root (data not shown). The latter was not surprising since mice are normally atherosclerosisresistant and much higher lipid levels are required to foster lesion development ( 17 ). Therefore, to enable studies of the effect of hyperglycemia on atherosclerosis, we bred the Ins2 Akita trait onto the atherosclerosis-susceptible Ldlr Ϫ / Ϫ background and compared Ins2 Akita Ldlr Ϫ / Ϫ to Ldlr Ϫ / Ϫ controls. We chose the Ldlr Ϫ / Ϫ background over the apoE Ϫ / Ϫ background because its plasma lipid profi le more closely resembles that of most atherosclerosis-prone humans. We also chose the 0.02% cholesterol AIN76a diet to avoid the additional stresses of obesity and insulin resistance apart from hyperglycemia present in other models ( 18 ). On the 0.02% cholesterol AIN76a diet at 20 weeks of age, Ins2 Akita Ldlr Ϫ / Ϫ mice had higher levels of total, VLDL, and LDL cholesterol and triglycerides, as well as increased aortic root cross-sectional lesion areas. Liver gene expression revealed alteration in lipid homeostasis genes and increased expression of pro-infl ammatory cytokine genes. Immunofl uorescence staining showed that the expression levels of several pro-infl ammatory cytokines were also increased in the atherosclerotic lesions and artery walls of Ins2 Akita Ldlr Ϫ / Ϫ mice. These data suggest the Ins2 Akita Ldlr Ϫ / Ϫ mouse is a promising model for mechanistic studies of accelerated macrovascular disease associated with type 1 diabetes. Ϫ / Ϫ and Ldlr Ϫ / Ϫ mice, global functions, networks, and canonical pathways were also analyzed by Ingenuity Pathway Analysis (IPA) (http://www.ingenuity.com) (Ingenuity Systems, Redwood City, CA). IPA is web-based software designed to organize biological information in a way that allows one to gain a high-level overview of the general biology associated with microarray data (24)(25)(26). In this study, the gene list from Gene-Spring GX 10.0 was generated with gene identifi ers, and the difference in gene expression was entered into IPA. The IPA Canonical Pathway Analysis tool was used to identify the signaling pathways associated with the database. The functional analysis identifi ed the molecular and cellular function that was most signifi cant to the data set as a whole and generated functional interpretation of microarray data ( 26 ).

Immunohistochemistry
Immunohistochemistry were performed on 12 m sections of aortic roots freshly embedded in OCT. Sections were fi rst fi xed in 100% ice-cold acetone for 15 min and then washed with PBS for 20 min. Sections were permeabilized with PBS + 0.1% Triton ×100 (PBST) for 10 min. Nonspecifi c binding was reduced by incubating slides in 10% rabbit sera diluted in PBST for 20 min at room temperature. Sections were then incubated with rabbit antibodies against mouse Mcp1 (1:50; Abcam, Cambridge, MA), tumor necrosis factor (Tnf) ␣ (1:100; Abcam,) or interleukin Ct method for relative quantifi cation of gene expression, normalized to GAPDH ( 21 ). The primer sets utilized in this study are shown in supplementary Table I.

Microarray and gene set analyses
Total RNA samples were prepared from the livers of Ins2 Akita Ldlr Ϫ / Ϫ and Ldlr Ϫ / Ϫ mice using the RNeasy mini kit (Qiagen USA, Valencia, CA). cRNA was prepared and hybridized to the MouseRef-8 v2.0 Illumina Genome-Wide Expression Bead-Chip Array (Illumina, San Diego, CA). The hybridized BeadChip was washed and labeled with streptavidin-Cy3 and scanned with Illumina BeadScan by the Rockefeller University Genomics Resource Center according to the manufacturer's protocol (Illumina). BeadStudio 3.2 software was used for background correction of the imported scanned image. Gene expression profi les were analyzed using the GeneSpring GX 10.0 software (Agilent Technologies, Santa Clara, CA). In addition to microarray analysis, gene set enrichment analysis (GSEA) was conducted using the GSEA desktop application (http://www.broad.mit.edu/GSEA) (Broad Institute, Cambridge, MA) ( 22,23 ). This application ranks the expression of members of a gene set in Ins2 . To determine whether a gene set is signifi cantly enriched, GSEA identifi es signifi cant changes in gene sets by assigning each a calculated enrichment score (ES). For each gene set, ES was calculated by using weighted Kolmogorov-Smirnov statistics to measure the proximity of the gene set to the top of the Ins2 Akita Ldlr Ϫ / Ϫ -effect ranked list. A highly positive ES indicated that the gene set or pathway was collectively upregulated by the Akita mutation intervention, while a highly negative ES indicated downregulation. To account for differences in numbers of genes in each set, the normalized ES (NES) was used to compare analy- Akita mutation also increased mortality of Ldlr Ϫ / Ϫ mice in both males and females (supplementary Fig. IB). Ins2 Akita Ldlr Ϫ / Ϫ mice were severely hyperglycemic with fasting blood glucose levels of 423 ± 162 mg/dl in males and 392 ± 130 mg/dl in females, compared with Ldlr Ϫ / Ϫ mice with glucose levels of 153 ± 29 mg/dl in males and 129 ± 17 mg/dl in females ( Fig. 1A ). In addition, Ins2 Akita Ldlr Ϫ / Ϫ mice had low fasting plasma insulin levels of 0.31 ± 0.05 ng/ml in males and 0.28 ± 0.07 ng/ml in females, compared with Ldlr Ϫ / Ϫ mice with insulin levels of 0.89 ± 0.32 ng/ml in males and 0.61 ± 0.22 ng/ml in females ( Fig. 1B ). Finally, Ins2 Akita Ldlr Ϫ / Ϫ mice of both genders showed impaired glucose tolerance compared with Ldlr Ϫ / Ϫ mice as measured by the IPGTT ( Fig. 1C, D ).

Statistical analysis
All data are expressed as mean ± SD unless indicated otherwise. Statistically signifi cant differences between two groups were analyzed by t -test for data normally distributed and by the Mann-Whitney test for data not normally distributed using Prism version 4.0 (GraphPad Prism Software, San Diego, CA). Grubbs's test was also performed to detect signifi cant outliers ( P < 0.05). areas of 224% in males and 30% in females ( P < 0.001and P < 0.05, respectively).

Akita mutation affects hepatic genes involved in lipid metabolism and infl ammation
To elucidate possible molecular mechanisms through which the Akita mutation might augment hyperlipidemia and atherosclerosis in Ldlr Ϫ / Ϫ mice, gene expression microarray studies were performed comparing livers from Ins2 Akita Ldlr Ϫ / Ϫ and Ldlr Ϫ / Ϫ mice. Genes with more than 1.5-fold expression changes are listed in supplementary  Table II.
To confi rm the results of the microarray analysis and to investigate other genes involved in lipid homeostasis that were not revealed by the microarray analysis, hepatic gene expression in Ins2 Akita Ldlr Ϫ / Ϫ and Ldlr Ϫ / Ϫ mice were evaluated by QPCR ( Fig. 4 ). Expression of several key genes involved in lipid homeostasis, including sterol-regulatory element binding protein (Srebp)1c, liver X receptor males and 688 ± 173 mg/dl in females. Similarly, Ins2 Akita Ldlr Ϫ / Ϫ mice had increased fasting triglyceride levels of 467 ± 239 mg/dl in males and 368 ± 181 mg/dl in females, compared with Ldlr Ϫ / Ϫ mice with triglyceride levels of 65 ± 26 mg/dl in males and 148 ± 64 mg/dl in females ( Fig. 2A ). Comparison of lipoprotein cholesterol levels between Ins2 Akita Ldlr Ϫ / Ϫ and Ldlr Ϫ / Ϫ mice revealed VLDL cholesterol levels increased 7-fold in males and 1.8fold in females; LDL cholesterol levels increased 2-fold in males but only 24% in females; and HDL cholesterol levels were unchanged in males but slightly and signifi cantly increased in females ( Fig. 2B ).

Atherosclerosis is accelerated in diabetic hyperlipidemic
The Ins2 Akita Ldlr Ϫ / Ϫ and Ldlr Ϫ / Ϫ mice were sacrifi ced at 20 weeks of age, and aortic root cross-sectional lesion areas were determined ( Fig. 3 ). Compared with Ldlr Ϫ / Ϫ mice, Ins2 Akita Ldlr Ϫ / Ϫ mice had increased cross-sectional lesion acid metabolism, and bile acid biosynthesis. The upregulated gene sets include pathways involved in carbon fi xation, glutathione metabolism, oxidative phosphorylation, pentose phosphate pathway, metabolism of xenobiotics by cytochrome p450, citrate cycle, and hematopoietic cell lineage.
In addition to GSEA, the microarray results were analyzed by IPA Canonical Pathway Analysis. The genes from microarray results were overlaid onto the global signaling pathways developed from information contained in the Ingenuity Pathways Knowledge Base ( 26 ). The IPA revealed differential expression of genes in lipid and carbohydrate metabolism and cell death ( Table 2 ). Both analyses indentifi ed signifi cantly changed gene sets or pathways related to lipid homeostasis in the livers of Ins2 Akita Ldlr Ϫ / Ϫ mice.

GSEA and IPA of genes differentially expressed between Ins2 Akita Ldlr ؊ / ؊ and Ldlr ؊ / ؊ mice
Microarray results were further analyzed by GSEA and IPA to identify gene sets and pathways affected by breeding the Ins2 Akita mutation onto the Ldlr Ϫ / Ϫ background.
GSEA is a computational method that determines whether an a priori defi ned set of genes shows statistically significance between two biological states ( 22 ). Comparing Ins2 Akita Ldlr Ϫ / Ϫ to Ldlr Ϫ / Ϫ mice, GSEA analysis identifi ed seven downregulated and seven upregulated gene sets ( Table 1 ). The downregulated gene sets in Ins2 Akita Ldlr  The Srebp and Lxr transcription factor genes are known to be particularly important for regulation of lipid metabolism. In the liver of Ins2 Akita Ldlr Ϫ / Ϫ mice, Srebp1a and Srebp1c mRNA levels were signifi cantly downregulated, but Srebp2 mRNA levels remained the same. Insulin is a known direct stimulator of Srebp1 transcription ( 27 ), so a likely cause of decreased Srebp1 mRNA levels in the liver is the hypoinsulinemia of the Ins2 Akita Ldlr Ϫ / Ϫ mice. However, since Srebp1 stimulates fatty acid and triglyceride synthesis, it is unlikely that decreased Srebp1 mRNA levels are the primary explanation for the Akita mutationmediated hyperlipidemia. Lxr ␣ is a nuclear hormone receptor that acts as a cholesterol sensor. It responds to excess cellular cholesterol by upregulating genes involved in promoting reverse cholesterol transport, increasing cellular cholesterol export, and promoting decreased intestinal cholesterol absorption and increased hepatic cholesterol excretion ( 28 ). Breeding the Lxr ␣ knockout trait to the apoE Ϫ / Ϫ background resulted in accelerated atherosclerosis ( 29,30 ). Thus, the decreased Lxr ␣ mRNA level in the livers of Ins2 Akita Ldlr Ϫ / Ϫ mice likely contributed to the increase in atherosclerosis observed. It has also been reported that In-s2 Akita mice have reduced levels of Lxr ␣ mRNA in their kidneys ( 31 ). Thus the Akita mutation might result in reduced Lxr ␣ mRNA in multiple tissues, but in no case is the mechanism known.
Cyp7b1 and Cyp27a1 are two enzymes involved in cholesterol catabolism and bile acid biosynthesis, and both were significantly decreased in Ins2 Akita Ldlr Ϫ / Ϫ mice.
to investigate the protein content of these cytokines in the atherosclerotic lesions. Immunofl uorescence staining showed that the expression levels of several key infl ammatory cytokines, including Mcp1, Tnf ␣ , and Il1 ␤ , were increased in the atherosclerotic lesions and artery walls of Ins2 Akita Ldlr Ϫ / Ϫ mice ( Fig. 6 ). Therefore, the enhanced infl ammation in Ins2 Akita Ldlr Ϫ / Ϫ mice may contribute, at least in part, to the increased atherosclerosis development in these mice.

DISCUSSION
There is a major need for a good animal model to study the mechanism of type 1 diabetes-accelerated atherosclerosis. To address this need, we bred the Ins2 Akita mice onto the atherosclerosis-prone Ldlr Ϫ / Ϫ background, and we showed augmented levels of VLDL and LDL cholesterol as well as increased aortic root cross-sectional lesion areas. Since the liver plays a central role in lipid metabolism and controls aspects of infl ammation relevant to atherosclerosis, we also assessed the alteration of hepatic gene expression by microarray and QPCR determinations. We found the Ins2 Akita mutation altered expression of hepatic lipid metabolism genes, including Srebp1, Lxr ␣ , Abca1, Lpl, Cyp7b1, and Cyp27a1, and increased expression of pro-infl ammatory cytokine genes. In further analysis, both GSEA and IPA indicated alterations in lipid and carbohydrate metabolism genes, whereas IPA alone indicated altered expression of celldeath genes. overexpression of Lpl have increased liver lipid droplets, VLDL production, and plasma ketone body levels ( 39 ), and they are also insulin resistant ( 40 ). Liver Lpl expression may increase at times of metabolic stress and shunt circulating triglycerides to the liver. The increased liver Lpl expression in Ins2 Akita Ldlr Ϫ / Ϫ mice might be due to the defect in liver insulin signaling and impaired ability of insulin to suppress endogenous glucose production in these mice. The increased hepatic Lpl can increase VLDL production and might contribute to the Akita mutation-induced hyperlipidemia in Ins2 Akita Ldlr Ϫ / Ϫ mice. The precise mechanisms through which the Akita mutation modulates lipid metabolism and cholesterol levels in Ldlr Ϫ / Ϫ mice remain to be determined. Nevertheless, the altered expression levels of genes suggest that the Akita mutation can affect multiple genes involved in lipid homeostasis in Ldlr Ϫ / Ϫ mice and might provide us some insight into how type 1 diabetes induces abnormal lipid metabolism in patients.
In addition to hyperlipidemia, infl ammation plays a critical role in the initiation and progression of atherosclerosis. We found that gene sets related to infl ammation were also signifi cantly altered in the livers of Ins2 Akita Ldlr Ϫ / Ϫ mice, with many pro-infl ammatory cytokine genes upregulated ( Fig. 5 ). Furthermore, the protein levels of several pro-infl ammatory cytokines, including Mcp1, Tnf ␣ , and Il1 ␤ , were increased in the atherosclerotic lesions and Cyp7b1 catalyzes the 7 ␣ -hydroxylation of oxysterols, an early step in the alternative bile acid biosynthesis pathway, and Cyp7b1-defi cient mice had elevated plasma and tissue levels of 25-and 27-hydroxycholesterol ( 32 ). Cyp27a1 catalyzes the oxidative cleavage of the sterol side chain in the bile acid biosynthesis pathway and 27-hydroxylation of cholesterol ( 33 ). Cyp27a1 plays an important role in cholesterol catabolism and atherosclerosis protection, and mutations in the human Cyp27a1 gene have been linked to the accumulation of cholesterol and premature atherosclerosis in cerebrotendinous xanthomatosis patients ( 33,34 ). Therefore, the decreased expression of both of these cholesterol catabolism genes might contribute to the increased lipid levels and accelerated atherosclerosis of Ins2 Akita Ldlr Ϫ / Ϫ mice.
Lpl mRNA levels were increased more than 4-fold in the liver of Ins2 Akita Ldlr Ϫ / Ϫ mice. LPL is the ratecontrolling enzyme involved in plasma lipoprotein triglyceride hydrolysis and is responsible for the generation of fatty acids from circulating triglyceride-rich lipoproteins ( 35 ). The Lpl gene is normally expressed mainly in adipose tissue and muscle, not in liver. We have previously generated Lpl-defi cient mice ( 36 ) as well as transgenic mice expressing Lpl exclusively in muscle ( 37 ), adipose tissue ( 38 ), or liver ( 39 ). Lpl-defi cient mice develop lethal hypertriglyceridemia within the fi rst day of life, at which point they have markedly reduced intracellular lipid stores ( 36,38 ). Interestingly, mice with liver-specifi c In summary, we generated a genetically induced type 1 diabetic mouse model (Ins2 Akita Ldlr Ϫ / Ϫ ) to study the mechanism of diabetes-accelerated atherosclerosis.
artery walls of Ins2 Akita Ldlr Ϫ / Ϫ mice. These cytokines play an important role in atherosclerosis initiation and progression. For example, monocytes are attracted by Mcp1 to lesion-prone areas; this is considered a critical step in atherosclerosis initiation ( 41,42 ). Defi ciency of Mcp1 in mice signifi cantly reduces atherosclerotic lesions in Ldlr Ϫ / Ϫ mice ( 43 ). Tnf ␣ and Il1 ␤ can be produced by immune cells (e.g., macrophages, T cells) within the lesion and contribute to atherosclerosis progression ( 44 ). Lxr might be important for regulation of infl ammatory cytokines in Ins2 Akita Ldlr Ϫ / Ϫ mice, since Lxr can inhibit immune and infl ammatory responses by repressing Nf-b-dependent induction of infl ammatory gene expression, including Il-1 ␤ , Tnf ␣ , and Mcp1 ( 28 ). Furthermore, one of the Lxr target genes, Abca1, can also modulate infl ammatory response by promoting cellular cholesterol effl ux and inhibiting Toll-like receptor (TLR) signaling ( 45 ). Abca1defi cient macrophages have increased lipid rafts and enhanced infl ammatory responses ( 46 ). Taken together, reduced Lxr and Abca1 expression levels might, at least in part, account for the increased infl ammatory cytokine gene expression in livers of Ins2 Akita Lldr Ϫ / Ϫ mice.
GSEA and IPA have also identifi ed signifi cantly altered genes or gene networks related to other pathways, including cell death, complement and coagulation cascades, SNARE interactions in vesicular transport, methylnaphthalene degradation, and linoleic acid metabolism. It is  Ϫ / Ϫ mice were stained with rabbit antibodies against mouse Mcp1, Tnf ␣ , or Il1 ␤ followed by fl uoresceinlabeled goat anti-rabbit secondary antibody (red). The nuclei were stained with DAPI (blue).