Identification of Differentially Expressed Genes and Processes in Myocardial Tissue of Liver-Specific Gck Knockout Mice

Background: Diabetic cardiomyopathy is a ventricular disease caused by diabetes mellitus. Abnormalities in the function of the glucokinase (GCK) play an important role in the development of diabetes. The present study is aimed at exploring changes in gene expression and related molecular mechanisms of diabetic myocardial injury in Gck knockout mice. Methods: Liver-specific glucokinase gene knockout mice( Gck w/- ) and wide type ( Gck w/w )mice generated using the Cre-loxP gene targeting strategy of 30- and 60-weeks of age were used in these studies. Determination of liver glucokinase enzyme activity, liver glycogen content and serum biochemistry parameters reflect the metabolic disorder in these mice. Echocardiography and surface electrocardiographs were used to evaluate cardiac function. Superoxide dismutase activity and malondialdehyde levels reflect oxidative stress in the myocardium. RNAseq, GO enrichment analysis and qPCR were used to detect differences in the myocardial gene expression profiles of Gck w/- and Gck w/w mice. Gck knockout mice was decreased. These results suggest that the down regulation of the insulin signaling pathway and the up regulation of NADH dehydrogenase, combined with changes in blood glucose and insulin levels, leads to abnormal metabolism in the myocardium, which is mainly seen as a decrease in glucose uptake and increase in fatty acid synthesis, leading to myocardial toxicity and cardiac dysfunction in 60-week-old Gck knockout mice. These results indicate that the abnormalities of

genes and networks among the transcripts influenced by Gck knock out. In addition, by profiling selected important genes, and their functions, we identified specific cardiac gene expression pattern features of the Gck w/mice. Overall, these approaches uncover a new gene expression pattern that is induced in the liver-specific Gck knockout, as well as its influence on DCM, which should provide theoretical support for early screening for the diagnosis of diabetic cardiomyopathy, as well as a new target for drug therapy.

Animals
Liver-specific glucokinase gene knockout mice were previously generated using a Cre-loxP gene targeting strategy by our lab [10]. Liver-specific glucokinase gene knockout (Gck w/-) mice were obtained by cross breeding mice, whose glucokinase gene was flanked by loxP sites, with the Alb-Cre transgenic mice. Gck w/knockout and wide type (Gck w/w ) mice at 30 and 60 weeks of age were used in these studies (n = 3-6 as stated in the figure legends). A protocol for these experiments, following the "Guidelines for Animal Experiments", was approved by the Peking University Health Science Center.

Glucokinase enzyme activity
100 mg of mouse liver tissue was homogenized in pre-cooled buffer solution containing 100 mM KCl, 25 mM HEPES, 7.5mM MgCl2, and 4 mM dithiothreitol (pH 7.4)and the enzyme was released from the extract after being left at 4 °C for 5 hours. After centrifugation at 4 °C for 5 min at 3,000 rpm, the supernatant was carefully extracted to form the crude extract. A equal volume of 100 or 0.5 mM glucose reaction buffer was then added and preheated at 30 °C for 15 minutes. To measure GCK enzyme activity 0.2 unit of glucose-6-phosphate 1-dehydrogenase was immediately mixed and the absorbance of the extract was measured at 340 nm for 10 minutes. Enzyme activity is determined by measuring the increase of absorbance at 340nm in a unit of time, and is expressed as mU/mg protein. 5

2.3Liver glycogen content
The anthrone method was used to determine liver glycogen content [11]. Liver tissue was prepared and washed with 0.9% NaCl. After weighing, about 100 mg liver tissue was treated by boiling for 30 min in alkaline solution and cooled on ice. A detection buffer containing concentrated sulfuric acid and anthrone was then added followed by boiling for 5 min. Blue compounds generated were then assayed with aspectrophotometer at 620 nm.

2.4Serum biochemistry parameters
Before measurement of serum biochemistry, mice were fasted for 8h. Blood samples, 100μl, were collected from the eyes. Fasting blood glucose levels was measured with a Roche blood glucose monitor (Glucotrend 2, Roche, Germany), while plasma triglyceride(TG) and total cholesterol (TC) (NanjingJiancheng Bioengineering Institute, CHN) levels were assayed by enzymatic methods.
To determine intraperitoneal glucose tolerance(IPGTT),blood was taken from the tail vein of mice 8 hours after fasting, and blood glucose levels were measured by a Roche blood glucose monitor (Glucotrend 2, Roche, Germany). Mice were then injected with 2 g/kg glucose intraperitoneally. Blood glucose levels were measured at 30, 60 and 120 min after injection, and the IPGTT blood glucose values were plotted and the area under the curve (AUC) was calculated.
Fasting insulin levels were quantified using a commercially available radio immune assay kit (China Institute of Atomic Energy, CHN). Fasting blood glucose and serum insulin were measured, and the homeostasis model assessment (HOMA) index [12] was calculated to evaluate peripheral insulin resistance and the function of β-cells.

Echocardiography
Transthoracic echocardiography was performed on anesthetized mice using a high-resolution 6 imaging system for small animals (Vevo 770, VisualSonics, CAN), equipped with a high-frequent ultrasound probe (RMV-707B). Under the guidance of long-axis images of the left ventricle, the maximum left ventricular diameter (the level of papillary muscle) displayed an M-mode image, the left ventricular diameter and wall thickness were measured, and left ventricular function was analyzed. Ejection fraction(EF) and fractional shortening (FS) were calculated using the Vevo770 software. The final data represent the averaged values of 3-6 cardiac cycles.

Electrocardiographic recordings
Surface electrocardiographic (ECG) recordings were obtained from conscious mice. The onset and the offset times of P, Q, R, S, and T waves were measured, and the ECG parameters were analyzed from lead II. Since the length of the QT interval is affected by the heart rate, corrected QT (QTc) intervals were calculated using the following formula [13].

Cardiac histological analysis
Hematoxylin and eosin (HE), trichrome Masson, and periodic acid-schiff (PAS)stains were used to evaluate the heart pathology of the mice. After euthanasia, the heart was immediately removed, and a portion of each heart was fixed in 4% paraformaldehyde, embedded in paraffin and sectioned. These sections were stained with HE, Masson and PAS to visualize tissue fibrosis. The extent of fibrosis in the myocardial tissue sections was quantified using Image-Pro Plus 6 (Media Cybernetics, USA)as the relative area of positive stained area (PAS: purple red-stained glycogen; Masson: blue-green fibrosis) normalized to the total tissue area [14].

2.8RNAseq
RNAseq was used to determine whether there were differences in the levels of gene expression in different samples. mRNA was purified with poly-T oligo-attached magnetic beads from 3μg total RNA per sample (60-week-old Gck w/and Gck w/w mice). Single-stranded cDNA was synthesized using random hexamers as primers with reverse-transcriptase, and then double-stranded cDNA was synthesized using DNA polymerase I. AMPure XP beads were used to 7 select the fragment size, and the cDNA library was enriched by PCR. RNAseq was performed on an IlluminaHiseq 2000 platform. Clean reads were obtained after removing reads with adapters, removing reads with an N ratio greater than 10%and removing low quality reads. The threshold for differential gene screening was set to|log2(Fold change)|>0.58with a corrected P value<0.001.

GO enrichment and PPI analysis of differentially expressed genes
Gene Ontology(GO) terms of cellular component, molecular function and biological process in DAVID[http://david.abcc.ncifcrf.gov/home.jsp] were employed to categorize the enriched biological themes of the differential expression genes. GO terms with corrected P value less than 0.05 were considered significantly enriched in the differential expressed genes.PPI analysis of the differentially expressed genes was based on the STRING database(http://string-db.org/), which contains known and predicted Protein-Protein Interactions(PPI). The networks were imported into Cytoscape for further analysis and visualization.

2.10Quantitative RTPCR
Total RNA was extracted from mouse myocardium tissues using Trizol Reagent (Invitrogen, USA) and reverse-transcribed into cDNA by using Prime Script 1st Strand cDNA Synthesis Kit (Takara, CHN). Quantitative RT PCR was performed with the StepOne™ System with PowerUpTM SYBR® Green Master Mix (ABI, USA). Amplifications were performed using the following conditions with the primer list in Table1: initial denaturation at 95°C for 10 min followed by 39 cycles performed at 95°C for 15 s and 54°C, 57°C or 60°C for 1 min.
Transcription levels were normalized to those of β actin.

Myocardial tissue SOD activity and MDA levels
Frozen left ventricular myocardium samples were weighed and homogenized (1:10, w/v) in 50 mmol/l phosphate buffer (pH 7.4) and kept in an ice-bath. Superoxide dismutase (SOD) activities and malondialdehyde(MDA) levels in the myocardial tissues were determined using commercially available kits (NanjingJiancheng Bioengineering Institute, China).

2.12Statistical analysis
Differences between the Gck w/and Gck w/w groups were determined by analyses with one-way ANOVA with SPSS 13.0 software. P values less than 0.05 were considered to be statistically significant. Experimental data are expressed as means ± SD.

GCK activity and glycogen content is decreased in the liver of Gck w/mice
To confirm the effect of the liver-specific deletion of glucokinase, we measured liver glucokinase protein expression, glucokinase activity and glycogen content, which is an approximate measure of glucokinase function. Results of our western blots show decreased expression(Figure1a and b) and activity of glucokinase (Figure1c) in the liver of Gck w/mice and indicate that the Gck knockout mice were heterozygous and the glucokinase alleles in the liver were only partially deleted.GCK catalyzes the conversion of glucose to glucose-6-phosphate, which is then the substrate for glycogen synthesis. When liver tissue lysates were assayed for glycogen content, the Gck gene knockout resulted in 43% (30weeks) and 33% (60weeks) decreases inn glycogen levels(Figure1d).This result can be explained by the reduction inglucose-6-phosphate levels caused by the Gck knockout.
Compared with Gck w/mice at 30 weeks of age, the liver GCK protein expression, GCK activity and the content of glycogen of the livers of 60-week old Gck w/mice were not significantly different, but were significantly lower than those of wild-type (Gck w/w ) mice of the same age. This shows that the effect of the Gck gene knockout in the liver of mice persists with an increase in age.

Metabolic disorders and insulin resistance inGck w/mice
To determine the physiological consequences of liver-specific Gck knockout, glucose metabolism, body weight, serum glucose, insulin, total cholesterol and triglyceride were investigated. To investigate long-term changes due to Gck gene knockout on body weight and serum glucose, we compared the weights and fasting glucose levels of Gck w/mice to Gck w/w maintained on normal chow diets. In the Gck w/mice, fasting glucose was elevated at 6 weeks compared to control littermate wild-type (Gck w/w ) mice, and this was maintained until 70weeks (Figure2a). Despite this, changes in body weights were not statistically different in the Gck w/mice, compared to age matched gck w/w mice, from 2 to 70weeks of age (Figure2b).
Glucose tolerance in the 60-week-old mice was also significantly worse than in the Gck w/mice compared to their Gck w/w littermate controls. Fasting blood glucose levels and the 30, 60, and 120 minute blood glucose levels after a glucose load were significantly higher in the Gck w/mice than in Gck w/w littermate controls, however, no significant difference in blood glucose levels was seen among the wild-type mice of different ages, or among the Gck w/mice of different ages(Figure2c and d).With the results of the IPGTTs, we calculated the area under the blood glucose curves. A significant increase was seen in the Gck knock-out group compared with their littermate control group in the 30-and 60-week-old mice (Figure2e).
Insulin is an anabolic hormone and its secretion is stimulated by hyperglycemia. To determine whether long-term changes in glucose metabolism affect insulin secretion and induce insulin resistance, fasting insulin levels were measured and HOMA-IR and HOMA-β-cell values were calculated. There was no significant change in serum insulin levels in the mice of different ages or different genotypes (Figure2f). In the 60-week-old mice, HOMA-IR levels were significantly higher and HOMA-β-cell levels significantly lower in the Gck w/compared to Gck w/w mice. In the 30-week-old mice, only the HOMA-β-cell levels were significantly lower in the Gck w/mice (Figure2g-h). These results indicated that the long-term Gck gene deletion and hyperglycemia gradually cause insulin resistance in peripheral tissues.
Taking this into account, we conclude that glucose homeostasis and insulin sensitivity were affected by the Gck gene deletion, while total cholesterol and triglyceride display no significant changes (Figure2i,j). These data demonstrate that the Gck w/mice had a glucose metabolism disorder and insulin resistance.

3.3Age-dependent cardiac hypertrophy occurs in Gck w/mice
To determine the effect of Gck knockout on the heart in mice, we examined Gck liver-specific knockout (Gck w/-) and wild-type (Gck w/w ) mice by echocardiography, electrocardiogram and cardiac histological analysis. At 30 weeks of age, only the left ventricle (LV) internal dimension during systole (LVID;s) was significantly decreased in the Gck w/mice, compared to Gck w/w mice.
At an age of 60weeks, the LV posterior wall thickness during systole (LVPW;s) has also significantly increased in Gck w/mice, compared to wild-type littermates (P<0.001). Similarly, significant reduction of LVID and an increase of LVPW during diastole were observed (Figure3a-g).Myocardial hypertrophy was followed by electrophysiological disorders at the age of 60weeks, with the Gck w/mice developing remarkably longer PR and QRS intervals compared to Gck w/w mice (Figure3h-k). In Gck w/mice, significantly increased levels of collagen (Masson positive material) and glycoproteins (PAS positive material) were detectable, as assessed by tissue structural analyses in60-week-old mice. In contrast, only increased levels of glycoproteins could be detected in the30-week-oldGck w/mice (Figure3l-n).Thus, we conclude that Gck deletion in the liver causes age-dependent heart dysfunction with a disease course going from glycoprotein aggregation and left ventricular cavity reduction to significant hypertrophy and fibrosis.

Analysis of the differences in gene expression in the myocardium of Gck w/and Gck w/w mice by RNAseq
To elucidate the cellular mechanisms underlying age-dependent heart dysfunction in Gck w/-mice, we applied high-throughput RNAseq to identify the myocardium gene expression profiles in60-week-oldGck w/w and Gck w/mice. Expression was detected for 12,680 unique genes, with fold change in expression defined as the ratio of Gck w/-FPKM to Gck w/w FPKM. Based on a threshold of total fold change of>1. 5  To analyze the functional significance of the proteins with changed expression corresponding to the Gck knockout, we used the STRING programs to identify functional networks, which were visualized with Cytoscape. Figure 4b provides a graphical overview of the STRING results and indicates that the changed behavior in response to Gck knockout is observed for different groups of proteins. Using the GeneMANIA Cytoscape plugin gene function prediction, the protein interaction network can be seen as mainly divided into four subgroups, which are related to (1) myocardial cell development and function, (2) material metabolism,(3) mitochondria oxidative phosphorylation and (4) ribosomal function (Figure4b). This phenotype is supported by the GO analysis.
To further categorize biological processes, we classified the differentially expressed genes using the Functional Annotation Cluster (FAC) tool available in DAVID. GO analysis of the 68 up-regulated genes revealed a focus on 4 biological processes, 1 molecular function and 7 pathways as defined by DAVID bioinformatics (P values<0.05). From these results, it can be seen that the GO terms are enriched in genes with functions necessary for mitochondria oxidative phosphorylation such as electron transport chain, cytochrome-c oxidase activity, oxidative phosphorylation and respiratory electron transport ATP synthesis, all of which were up-regulated (Figure4c). Similarly, 18 biological processes, 7 molecular functions and 6 pathways were enriched by the 141 down-regulated genes. The major processes enriched by lower levels of gene expression in the Gck w/mice include GO terms related to myocardial cell development and function (such as muscle structure development , muscle contraction, and cardiac muscle contraction) and material metabolism(such as insulin receptor substrate binding, hexokinase 12 activity, insulin signaling pathway, and type II diabetes mellitus)(Figure4d-e). Western blot analysis showed that the expression of MYC2 was significantly increased in 60-week-old Gck w/mice compared to the control group, and that MYC2 was increased in 30-week-old Gck w/mice, however, there was no significant difference with Gck w/w mice of the same age (Figure5c and d). The trends in MYC2 protein gene changes are consistent.

Metabolic and insulin pathways are impaired in Gck w/mice
Genes for confirmation were selected based on their perceived importance and likely functions for material metabolism. qPCR results show that the gene expression levels of Snca expression was robustly decreased in the Gck w/mice compared to wild-type mice, which was paralleled by significant decreases in ACC and AMPKβ1 phosphorylation. In contrast, no significant changes in the levels of PI3K, mTOR or AMPK α phosphorylated proteins were observed. Interestingly, in the 30week-old mice, the above-mentioned proteins of the insulin signaling pathway were found at comparable levels in both Gck w/and wild-type mice, suggesting that normal insulin signaling existed in 30-week-old Gck w/mouse myocardial tissue. (Figure6 c-e).

4.1Gck knockout induced metabolic disturbances and cardiac dysfunction
Glucokinase is mainly expressed in β cells of pancreatic islets and hepatocytes and catalyzes the phosphorylation of glucose to glucose 6 phosphate, the initial and rate-limiting enzymatic step in glucose metabolism. In mammals, it is a key component of the glucose sensing system and plays a very important role in maintaining the stability of blood glucose levels [15]. The primary role of glucokinase in the liver is to increase the use of glucose by this tissue, especially in postprandial and hyperglycemic conditions, to reduce blood glucose levels and increase glycogen synthesis [16].
We used a Cre-loxP gene targeting strategy to generate liver-specific Gck knockout mouse [10] where liver GCK activation is profoundly impaired (30-50%) and blood glucose homeostasis isdisrupted. Western blot analysis showed that GCK expression in the liver is reduced in Gck w/mice. At the same time, enzymatic activity of GCK is also decreased in these Gck gene knockout mice. Results from our Gck knockout mice showed that heterozygous disruption of Gck results in a 42% and 31% reduction in liver GCK activity at 30 and 60weeks, respectively. Additionally, it is interesting that the reduction in hepatic GCK activity is substantially greater than the reduction in the GCK protein levels. This suggests that some of the synthesized GCK proteins, despite retaining immunogenicity, exhibit abnormal function with loss of enzymatic activity in the Gck w/mice. Glucokinase, a rate-limiting enzyme in glycogenesis, plays a key role in glucose metabolism. Hepatic glycogen assay results showed that the glycogen content of livers is significantly decreased in Gck w/mice due to the glucokinase deficiencies. This indicates that glucokinase plays an important role in glucose homeostasis and its dysfunction can lead to impaired glycogenesis followed by high levels of blood glucose [17].
The in vivo knockout of Gck yields hyperglycemia in mice aged 6 to 70weeks but does not result in a significant body weight change. However, the induced hyperglycemia selectively impaired the pancreatic islet cells, causing insulin resistance with an increased HOMA-IR and a decreased HOMA-β cell. Based on these results we speculate that the Gck knockout in the hepatocytes leads to higher plasma glucose concentrations and insulin resistance throughout the life of the heterozygous Gck knockout mice.
The current study found strong support for an association between HOMA-IR and diabetic cardiomyopathy [18], and that insulin resistance is associated with diastolic dysfunction and the The 60-week-old Gck knockout mice had hypertrophy and disorder of cardiac myocytes and increased deposition of PAS and Masson positive substances in cardiac tissues. PAS positive material deposition was observed in the 30-week-old Gck knockout mice, but at a lower level than seen in the 60-week-old Gck knockout mice. Our results demonstrate that there is fibrosis in the 60-week-old Gck knockout mice, which further supports the existence of DCM that develops with age in this Gck knockout mice model.

4.2Gck knockout induces changes in the myocardial protein expression profile
Genes associated with myofibrils are involved in myocardial contraction and affect the severity Myosin light chain2 (MYL2) is the regulatory light chain of myosin, which induce structural changes and affects myocardial contraction through phosphorylation. It has been shown that Myl2expression in the myocardium of diabetic rats is upregulated [25], which is consistent with our observations in mice. Although the mechanism is not clear, we speculate that the increase inMYL2 levels in 60-week-old Gck knockout mice affects myocardial fiber assembly leading to myocardial hypertrophy. Increased production of advanced glycation end products (AGEs) secondary to hyperglycemia in diabetic cardiomyopathy can lead to increased myocardial stiffness [4]. AGEs can lead to the cross-linking of collagen molecules, which inhibits the degradation of collagen, thus increasing fibrosis and an increase in myocardial stiffness and impairment of cardiac diastolic function. In addition, the oxidative stress of diabetes leads to increased expression of the receptor for AGEs in the heart, which in turn increases the stiffness of the myocardium [4]. in the myocardium, and Popdc1 is also expressed in skeletal muscle [30].Popdc1 and Popdc2 deletion variants were found to be associated with arrhythmias such as long QT syndrome, sinus bradycardia and atrio ventricular block in animal models such as mice and zebrafish, as well as in human patients [31].Null mutations inPopdc1 increase the sensitivity to oxidative stress and ischemia-reperfusion injury in myocytes[32], suggesting that POPDC is an essential protein for the maintenance of cardiac function and has a protective effect on the myocardium. The Oxygen free radicals can damage proteins, phospholipids, and DNA, either directly or indirectly, by oxidation of lipids to lipid peroxidation or by converting nitric oxide to reactive nitrogen [45].
In this process, the calcium induced opening of mitochondrial membrane permeability transporters is more sensitive, which results in damage to the mitochondrial electron transport chain and affects energy supply.

Conclusions
Liver specific Gck knock-out can induce myocardial fibrosis at an early stage and diabetic myocardial injury at a late stage. Decreased glucokinase activity leads to disorders of glucose metabolism and induces hyperglycemia. Long-term hyperglycemia damages the insulin signaling pathway, AMPK and ACC function decline, and causes cardiac myocyte energy metabolism disorders. The genes associated with mitochondrial oxidative phosphorylation are upregulated, and oxidative stress is increased, which induces oxidative damage to cardiomyocytes. In this process, the proportion of myosin heavy chain and light chain falls out of balance. Myosin imbalance leads to hypertrophy and contraction of the heart cavity, which is related to cardiac function damage in diabetic cardiomyopathy (Fig 8).