Peripheral and cognitive benefits of physical exercise in a mouse model of midlife metabolic syndrome

Despite national and international efforts for the prevention of metabolic syndrome and its underlying diseases/disorders, its prevalence is still rising, especially in the middle-aged population. In this study, we explore the effect of high fat diet on the development of metabolic syndrome in middle-aged mice and to evaluate the potential benefits of voluntary physical exercise on the periphery as well as brain cognitive function, and to explore the potential mechanisms. We found that metabolic syndrome developed at middle age significantly impairs cognitive function and the impairment is associated with gene dysregulation in metabolic pathways that are largely affecting astrocytes in the brain. Eight-week voluntary wheel running at a frequency of three times a week, not only improves peripheral glucose control but also significantly improves learning and memory. The improvement of cognitive function is associated with restoration of gene expression involved in energy metabolism in the brain. Our study suggests that voluntary physical exercise is beneficial for metabolic syndrome-induced peripheral as well as cognitive dysfunction and can be recommended as therapeutic intervention for metabolic syndrome and associated diseases.

Metabolic syndrome (MetS) is defined as having at least three out of five cardio-metabolic conditions that include abdominal obesity, hyperglycemia and impaired insulin sensitivity, hypertriglyceridemia, hypercholesterolemia and hypertension 1,2 . It is by far the most prevalent disease in the world. Based on the National Health and Nutrition Examination Survey 3 , over 30% of adults aged 18 years or older in the United States are estimated to have MetS. In Europe and Latin America, approximately 25% of the adult population meet the criteria for metabolic syndrome 4 . It is projected that the prevalence of MetS will continue to increase worldwide because of the rising rate of obesity and diabetes in developing countries 5,6 .
MetS is associated with increased risk of multiple chronic illnesses including cardiovascular disease [7][8][9] , chronic kidney disease 10,11 and several types of cancers such as liver, colorectal and bladder cancer 12 . Systemic review and Meta-analysis of clinical and population-based studies showed that metabolic syndrome is also associated with increased risk of vascular dementia and increased risk of progression from mild cognitive impairment to dementia 13 . Several longitudinal studies also demonstrated that metabolic syndrome, as a whole, is related to a higher risk of cognitive decline [14][15][16] . Based on these epidemiological data coupled together with an aging population in many Western countries, the potential impact of metabolic syndrome has serious implications on general health and on the health care system. Therefore, safe and efficacious interventions for metabolic syndrome and associated peripheral and central disorders are of fundamental importance.
While the etiology of MetS is poorly understood it is generally acknowledged that both genetic and environmental factors play roles in the development of MetS. Genetic susceptibility to MetS has been under active investigation 17 and select mutations have been linked to the key components of MetS 18 . For example, Polymorphisms of APM1, a gene coding for adipocyte-derived hormone Adiponectin, are significantly associated with

Results
High fat diet induces features of metabolic syndrome. In this study, we initiated the high fat diet treatment when the C57BL/6 mice reached 12 months of age ( Fig. 1A for schematic design of the study). This age in mice recapitulates the physilogical conditions equivalent to middle-age in humans. Following 5 months of high fat diet treatment, the DIM mice became obese and had significantly higher body weight compared to the CTRL mice on 10% fat diet (Fig. 1B, two-tailed t-test, t 24 = 9.090, P < 0.0001). The DIM mice developed impaired IGTT compared to the CTRL mice (Fig. 1C, two-way ANOVA repeated measure, F 1,24 = 8.616, P = 0.0072 for diet effect; F 5,120 = 1.695, P = 0.1410 for interaction). The DIM mice had significantly increased fasting glucose content (Fig. 1D, two-tailed t-test, t 24 = 4.478, P = 0.0002). There were no significant differences in two-hour postprandial glucose content (Fig. 1E, two-tailed t-test, t 24 = 1.508, P = 0.145). We then randomly group the DIM mice to DIM-CTRL and DIM-PE (Body weight 58.2 ± 10.0 g vs. 59.3 ± 8.0 g; fasting glucose 124.7 ± 34.8 mg/dl  Benefits of PE on glucose utilization and body composition. The DIM-PE mice were voluntarily exercising on the running wheel three days a week for 8 weeks. We calculated that the mean exercise intensity is about 4.1 km over 24 h period with large variations between animals (ranging from 0.47 km to 13.0 km). Following 2 months running wheel exercise, the DIM-PE mice had stabilized their body weight ( Fig. 2A, twotailed t-test, t 8 = 0.000, P > 0.999) while the DIM-CTRL mice continued to gain body weight ( Fig. 2A, two-tailed t-test, t 8 = 4.452, P = 0.0021). IGTT test also showed that the DIM-CTRL mice continue to have impaired glucose tolerance (Fig. 2B, left panel, two-way ANOVA repeated measure, F 2,23 = 14.06, P = 0.0001 for diet/PE effect; F ,10,115 = 3.112, P = 0.0015 for interaction) with higher fasting glucose (Fig. 2B, middle panel, One-way ANOVA, F 2,23 = 9.903, P = 0.0008) and higher two-hour post-prandial glucose content (Fig. 2B, right panel, One-way ANOVA, F 2,23 = 6.595, P = 0.0054) while DIM-PE mice had similar fasting glucose level as well as two-hour post prandial glucose level as the LF-CTRL mice (Fig. 2B), indicating the benefit of physical activity on better controls of peripheral glucose level. We then subjected all mice to DEXA scan to evaluate body composition. We found that there were differences in bone mineral content (BMC) between the three groups ( Fig. 2C, One-way ANOVA, F 2,23 = 3.792, P = 0.0377): Compared to CTRL mice, the DIM mice had reduced BMC (Fig. 2C, P = 0.0336) and 8-weeks exercise had a trend of increasing BMC in the DIM-PE group (P = 0.0911). We also analyzed the bone mineral density between the groups (Fig. 2D, One-way ANOVA, F 2,23 = 3.947, P = 0.0336). We did not find significant difference between the CTRL mice and the DIM mice (Fig. 2D), however, 8-weeks exercise significantly increased BMD (Fig. 2D, P = 0.0258). Though exercise stabilizes the body weight in DIM-PE group, there was no difference in lean mass or fat mass between the DIM-CTRL and the DIM-PE group (Fig. 2E,F).

Central benefits of PE on cognitive function.
To test whether MetS developed during middle age affects cognitive function and whether PE has cognitive benefits, we subjected all mice to a contextual fear conditioning test. We found that the DIM mice had short-term memory deficits compared to the CTRL and the DIM-PE mice reversed the MetS-induced learning and memory deficits (Fig. 3A, One-way ANOVA, F 2,21 = 3.976, P = 0.0343 for auditory memory and F 2,21 = 10.52, P = 0.0007 for context memory). Long-term memory test also showed that www.nature.com/scientificreports/ by exercise (Fig. 3F). Rescue assessment showed that DIM-induced downregulated pathways that were related to immune responses (INF-γ signaling, INF-α signaling, cytokine signaling and prostaglandin synthesis) and growth hormone signaling were reversed by exercise (Fig. 3F). DIM-upregulated pathways that were related to mitochondrial function (TCA cycle, oxidative phosphorylation) and NGF transcription were reversed by exercise (Fig. 3F).

Effect of PE on peripheral biochemistry.
To assess the effect of DIM and PE on metabolic hormones, we collected plasma and conducted ELISA using the mouse hormone multiplex kit. We found that compared to CTRL mice, DIM mice had significantly increased levels of Insulin, C-peptide, a proinsulin that is important for insulin synthesis and pancreatic polypeptide (PP), an endocrine regulates pancreatic secretion activities and impacts liver glycogen storage and gastrointestinal secretion (Table 1). There was also a significant increase of leptin, an important hormone secreted mainly by adipocytes and directly reflects the total amount of fat in the body and glucagon-like peptide -1 (GLP-1), a peptide hormone secreted by intestinal L-cells that regulates insulin secretion (Table 1). We found that 8-weeks PE had no effect on the levels of any of these hormones in circulation albeit improvement in the glucose tolerance test (Table 1). Inflammation is closely associated with metabolic syndrome. There was an increase in the levels of IL-6, TNF-α (Table 1) but there was no difference in the level of chemokine monocyte chemoattractant protein-1 (MCP-1) which was shown to play an important role in renal injury and fibrosis in diabetic nephropathy 36 . 8-Weeks PE significantly reduced the plasma level of IL-6, however, had no effect on TNF-α (Table 1). We then measured the level of cholesterol and triglyceride, we found that DIM mice had significantly higher plasma level of total cholesterol and triglyceride and PE significantly lowered the level of triglyceride but not cholesterol (Table 1).

Discussion
Metabolic syndrome is a multifactorial disease that manifests its pathological features both in the periphery and in the central nervous system. Previous preclinical studies using a high fat diet to induce hyperinsulinemia, hyperglycemia and hyperlipidemia often start the diet at a very young age in order to study the effect of lifelong metabolic disturbance. Since the prevalence of MetS increases with age and more middle-aged people are affected by this disease we decided to study the effect of MetS developed during that period of life on periphery and cognitive health.
Physical activity has been recommended to lower the prevalence or risk of MetS [28][29][30][37][38][39][40] . However, epidemiological studies examining the association between physical activity and the risk of MetS reported conflicting findings. Some suggested that PE may benefit metabolic and cardiovascular function and reduce risk factors related to MetS [39][40][41][42][43][44] . Some showed no benefits [45][46][47][48] . The discrepancy is very likely related to the types and intensity of exercises or physical activities involved in the studies. We show that a 5-month western diet led to increased body weight, hyperglycemia, hyperlipidemia and insulin intolerance in middle-aged female mice while 8-week voluntary running wheel exercises stabilize body weight and significantly improve glucose tolerance albeit continuing on the high fat diet. Assessment of plasma levels of molecules that are related to glucose control did not find improvement in the levels of insulin, PP, GLP-1 or glucagon. Previous studies demonstrated that running wheel exercise-mediated improvements in blood glucose is primarily attributed to the reduction in the endogenous glucose production which is largely controlled by insulin and glucogon 49 . It is possible that our assessment of glucose tolerance was conducted immediately after the running wheel exercises while the measurement of these Table 1. Blood biochemistry assessment in control, DIM and DIM-PE mice Plasma levels of metabolic syndrome-associated hormones and cytokines measured at the end of the study. Data represents mean ± STDEV, n = 7-9 per group. www.nature.com/scientificreports/ hormones were conducted weeks after the mice completed their exercise. Moreover, most of the studies showing molecular changes that are associated with improved glucose uptake and glucose metabolism were conducted using forced instead of voluntary exercise [50][51][52][53][54][55] . Forced exercises usually subject mice or rats to motor-driven treadmill and provide a much higher intensity of the physical activity. Generally speaking, the treadmill runs at a rate of 10-25 m per minute and in our voluntary wheel running the mice ran about an average of 4100 m over the 24 h recording period. It is possible that the differences in the intensity of exercise may have different impact on plasma levels of molecules related to glucose control. The prevalence of age-associated bone loss in women is higher than in men and severe osteoporosis in postmenopausal women is also associated with obesity. Previous report showed that female C57BL/6 mice fed with a 10% corn oil high fat diet led to significant bone loss as reflected by lower bone mass density (BMD) 56 . In our study, we found that 6 months high fat diet led to a significantly lower bone mineral content compared to control lean mice while there is no significant difference in BMD. The lack of BMD changes could be due to the different sources of fats used in the diet. Our high fat diet mainly uses animal fats (lard) while corn oil was used in the previous study. Corn oil has high content of omega-6 fatty acids and very low omega-3/omega-6 ratio (1:46), which is postulated to have a big impact on bone losses during aging while the ratio of omega-3/omega-6 in lard is approximately 1:10. Nevertheless, we found two months of wheel-running was able to increase BMD in mice albeit they were continuously fed on the high fat diet. Our data supports the clinical observation that physical exercise is effective in stimulating osteogenesis and reduces the rate of bone loss. The exercise protocol used in the study is voluntary and mice were only subjected to the running wheel three times a week. Therefore, it can be easily recommended for long-term application in humans.
It is well acknowledged that MetS not only affects the periphery but also has a negative impact on the CNS 57-59 . Numerous epidemiological studies have shown that people with MetS have increased risk of developing agerelated dementia [13][14][15][16] . In animal models, it has also been demonstrated that risk factors associated with metabolic syndrome such as diabetes and obesity can induce neurological deficits, including reduced hippocampal dendritic spine density, reduced LTP and impaired learning ability [60][61][62][63] . Previously we found that mice chronically fed on high fat diet starting at a very young age (2 months of age) had abnormal functional connectivity in the brain 64 and synaptic maladaptation 32,33 . In this study, we showed that development of MetS in middle age can also lead to impairment in both short-term and long-term memory. RNA-seq in the hippocampal formation identified several pathways that are significantly altered in the brain of female mice developed MetS at middle age. Most notably, the lipid metabolism pathway and glycogenolysis pathway are significantly downregulated in the MetS brain. The brain is one of the organs that has very high content and rich composition of lipids and it is considered to be autonomous in lipid synthesis. Astrocytes are the main cell type that produce lipids which are taken up by neurons for synapse formation. The homeostasis of lipids in the brain is vital to proper brain function. Disturbance of lipid metabolism has been linked to several neuropathological conditions such as Alzheimer's disease and Parkinson's disease. Glycogen is important energy reserve in the brain and is found predominantly in astrocytes with small amounts in neurons. Specific neurotransmitters and neuronal signaling can trigger glycogenolysis which in turn breaks glycogen down to glucose-1-phosphate and provides fuel for ATPases to pump K + and Ca ++ for proper brain function and memory. Previous studies showed that MetS associated hyperglycemia may disrupt normal TCA cycle in the mitochondrial matrix 65 and disturb oxidative phosphorylation, increase production of reactive oxygen species (ROS), leading to impaired mitochondrial biogenesis 66,67 . Our data suggests that astrocytes are very likely the main cell type in the brain that are affected by these metabolic alterations that may lead to impaired glutamine synthesis, loss of synaptic proteins and eventually synaptic maladaptation and brain dysfunction. Previous studies conducted in young animals with metabolic disturbances showed that physical activity can improve brain function by reducing oxidative stress and lipid peroxidation [68][69][70][71] , modulating immune responses and reducing neuroinflammation 68,69,[72][73][74][75] , and improving insulin signaling and glucose utilization 68,[75][76][77][78][79] . Our study also showed that two months of voluntary wheel running significantly improves learning and memory. MetS-mediated DEGs rescued by physical exercise are enriched primarily in pathways associated with mitochondrial function and immune responses. Physical exercise significantly downregulates MetS-induced upregulation of TCA cycle and oxidative phosphorylation that may help restore proper energy metabolism in astrocytes, maintain redox potential and glutamate homeostasis, hence improved brain function. Our RNA-seq data conducted in middle aged mice is largely consistent with the mechanisms elucidated in previous studies 68-79 that physical exercise may benefit metabolic syndrome-mediated cognitive impairment though modulation of redox regulation, neuroinflammation and energy metabolism.
The prevalence of metabolic syndrome has steadily increased over the years and poses a serious burden on the US healthcare system. Our studies provided preclinical evidence that physical activities, exercised voluntarily and three times a week, provides protection against MetS-induced peripheral and central nervous system dysfunctions. Future study will assess clinical benefits of this kind of exercise regimen in humans. www.nature.com/scientificreports/ of mice continued to be fed on the high fat diet throughout the study. All animals were maintained on a 12:12-h light/dark cycle with lights on at 07:00 h in a temperature-controlled (20 ± 2 °C) vivarium and all procedures were approved by the VAMC IACUC (Approve number WAN-16-061). All proceudres were conducted in accordance with the ARRIVE guidelines and the Guidelines for Animal Care and Use at the James J. Peters VAMC.

Intra-peritoneal glucose tolerance test (IGTT).
IGTT was performed as previously reported 33,80 following 5 months treatment and at the end of the physical exercise. Specifically, mice were given a single dose of intraperitoneal glucose administration (2 g/kg BW) following overnight fasting and postprandial blood was collected from the tail vein periodically over a 2 h period. Blood glucose content was assessed using the Contour blood glucose System (Bayer, IN).
Physical exercise. At  DEXA scan. A small animal dual energy X-ray absorptiometer (DXA, Lunar Pixmus, WI) was used to measure areal bone mineral density (BMD) 81 . Briefly, mice were anesthetized with low ketamine/xylazine and placed on the scanner bed in the prone position, with the limbs and tail stretched away from the body. The scan data was analyzed with PIXImus software (2.10; GE/Lunar) according to manufacturer's instructions.
RNA isolation and RNA-seq. Total RNA from the hippocampus was isolated using the RNeasy Mini Kit (Qiagen, Valencia, CA). All RNA preparations were confirmed with integrity numbers > 8.0 by a Bioanalyzer before proceeding to library construction. Library construction (LncRNA library, Ribo-Zero, 4-5 libraries per test condition), quality assessment and Illumina Novaseq 6000 with 150-pb paired-end reads were conducted by Novogene.

RNA-seq data analysis and bioinformatics.
Quality control was carried out on the raw sequence data coming from the sequencer using the FastQC software 82 . This assesses total sequence, reads flagged as low quality, read length, GC content, per base and per tile quality, per sequence quality score, per base content distribution, per sequence GC content, per base N content, sequence length distribution, sequence duplication levels, overrepresented sequence, adaptor content, and Kmer content. Reads of universal low base quality were discarded, and reads with certain low quantity bases were trimmed. Following QC, RNA reads were mapped to mouse reference genome (mm10) with STAR aligner 83 to obtain gene-level read counts. We retained genes with counts per million reads (cpm) > 1 in at least 4 samples for downstream analysis. Read counts within samples and between samples were then normalized and the differential expression level between groups were examined using the edgeR R-package 84 . Genes with 20% fold change and nominal p < 0.05 were considered statistically significant.
Gene ontology (GO) analysis. GO