Endothelial FoxO proteins impair insulin sensitivity and restrain muscle angiogenesis in response to high fat diet
Emmanuel Nwadozi*1, Emilie Roudier*1, Eric Rullman2, Sujeenthar Tharmalingam1, Hsin-yi Liu1, Thomas Gustafsson$2, Tara L. Haas$1 *,$ these authors contributed equally 1
Angiogenesis Research Group, Faculty of Health, York University, Toronto, Canada
2
Department of Laboratory Medicine, Clinical Physiology, Karolinska Institutet, and Department of Clinical
Physiology, Karolinska University Hospital, Stockholm, Sweden.
Running Title: FoxOs block muscle angiogenesis in high fat diet
Number of Words: 8164
Contact information: Tara Haas, PhD Angiogenesis Research Group Faculty of Health York University Rm. 339-341 Farquharson Building 4700 Keele St. Toronto, ON M3J 1P3 Canada Phone: 416 736-2100 x77313 Email:
[email protected] Page 1 of 23
Nonconventional Abbreviations AOC; Area over the curve; CD, capillary density; C:F, capillary to muscle fibre ratio;; FDR, false discovery rate; FoxO, forkhead boxO; HF, high fat; HPRT-1, hypoxanthine-guanine phosphoribosyltransferase; IL-6, interleukin-6; Itgam, Integrin αM; ITT, insulin tolerance test; LDL, low density lipoprotein; LDLR-/-, low density lipoprotein receptor deficient; Mx1, myxovirus resistance 1/interferon inducible protein p78; NC, normal chow; Rorc, Retinoid-Related Orphan Receptor C; TNF-α, tumor necrosis factor-α; VEGF-A, vascular endothelial growth factor-A
Page 2 of 23
Abstract Skeletal muscle microvascular dysfunction contributes to disease severity in type 2 diabetes. Recent studies indicate a role for Forkhead box O (FoxO) transcription factors in modulating endothelial cell phenotype. Here we hypothesize that high fat (HF) diet generates a dysfunctional vascular niche through an increased expression of endothelial FoxO. FoxO1 protein increased (+130%) in the skeletal muscle capillaries from HF compared to normal chow fed mice. FoxO1 protein was significantly elevated in cultured endothelial cells exposed to the saturated fatty acid palmitate or the pro-inflammatory cytokine tumor necrosis factor alpha. In HF-fed mice, endothelial-directed depletion of FoxO1/3/4 (FoxOΔ) improved insulin sensitivity (+110%) compared to controls (FoxOL/L). Skeletal muscle capillary number increased significantly in HF-FoxOΔ mice. Transcript profiling of skeletal muscle identified significant increases in genes associated with angiogenesis and lipid metabolism in HF-FoxOΔ versus HF-FoxOL/L mice. HF-FoxOΔ muscle also was characterized by a decrease in inflammation-related genes and an enriched M2 macrophage signature. We conclude that endothelial FoxO proteins promote insulin resistance in HF diet, which may in part result from FoxO proteins establishing an antiangiogenic and pro-inflammatory micro-environment within skeletal muscle. These findings provide mechanistic insight into the development of microvascular dysfunction in the progression of type 2 diabetes.
Keywords: capillary; obesity; diabetes; microcirculation; gene expression
Page 3 of 23
INTRODUCTION Skeletal muscle, which represents >40% of whole body mass, is a key insulin target and plays an essential role in establishing systemic metabolic homeostasis. Obesity is a major risk factor for the development of insulin resistance and type 2 diabetes mellitus (1). A spectrum of dysfunctional cellular processes is recognized to underpin the progression of diabetes; however, the role played by the local vasculature in particular has largely been overlooked. The reduced capacity for glucose uptake in skeletal muscle that is characteristic of insulinresistant individuals can be attributed in part to vascular reduced basal and insulin-stimulated muscle blood flow (2, 3), which limits the delivery of both insulin and glucose to the muscle. A growing body of research also indicates that the capillary network itself plays a significant role in establishing skeletal muscle insulin sensitivity, and controls the efficacy of insulin transport to the surrounding myocytes (4–6). Several studies have reported that obese or insulin resistant individuals have a lower skeletal muscle capillary density compared to healthy subjects (7, 8). These findings imply that obesity-associated metabolic disturbances exert negative influences on the skeletal muscle microvascular niche, which in turn can exacerbate the insulin resistant state. However, causal relationships and the underlying cellular mechanisms remain to be established. Forkhead box O (FoxO) transcription factors modulate cell cycle progression, oxidative stress, apoptosis and metabolism (9). Within the endothelium, FoxO proteins regulate the expression of genes involved in vascular homeostasis during development and in adulthood (10–12). FoxO transcription factors act as angiostatic regulators within skeletal muscle; the inducible deletion of endothelial FoxO gene products enhanced capillary growth in response to exercise or muscle ischemia (13, 14). Several lines of evidence indicate that vascular FoxO proteins play a pathological role in the context of metabolic disorders. Endothelial FoxO1 transcriptional activity was augmented in endothelial cells exposed to elevated glucose, leading to increased LDL oxidation (15). Increased endothelial FoxO1 was associated with reduced insulin sensitivity in the adipose tissue of obese individuals, and was demonstrated to impede the activation of endothelial nitric oxide synthase (16). Endothelial deletion of FoxO factors reduced the infiltration of immune cells and development of atherosclerosis in the aorta of LDLR-/- mice (17). All together these observations suggest that endothelial FoxO proteins are induced in obesity, however it remains unknown whether this may promote dysfunction within the skeletal muscle vascular niche. Thus, in the current study, we hypothesized that the metabolic disturbances associated with prolonged high fat (HF) feeding increase endothelial FoxO protein expression, which in turn contributes to development of insulin resistance within skeletal muscle. Here we demonstrate that endothelial FoxO1 protein level is upregulated within the skeletal muscle microcirculation as a consequence of long term HF diet. Similar increases in FoxO1 protein were evoked in cultured endothelial cells exposed to the saturated fatty acid palmitate, or the obesity-associated cytokine TNFα. Page 4 of 23
Endothelial-directed depletion of FoxO1/3/4 (FoxOΔ) improved insulin sensitivity in HF-fed mice. HF-FoxOΔ mice were characterized by increased skeletal muscle capillary number in comparison to the HFFoxOL/L controls. Transcript profiling indicated that skeletal muscle of HF-fed FoxOΔ mice had significant increases in genes associated with angiogenesis and lipid metabolism, and an enriched signature of genes associated with M2 macrophages and inhibition of inflammation, when compared to HF-fed FoxOL/L animals. Thus, the increase in endothelial FoxO proteins that is evoked by chronic metabolic imbalance plays an integral role in disrupting the vascular niche within skeletal muscle, establishing an anti-angiogenic, pro-inflammatory micro-environment. These findings provide mechanistic insight into the contribution of FoxO proteins to the insulin resistance and microvascular dysfunction that develop during the progression of type 2 diabetes.
MATERIALS & METHODS Mouse Models Animal studies were approved by the York University Committee on Animal Care and performed in accordance with the American Physiological Society’s guiding principles in the Care and Use of Animals. Male wild-type FVB/N mice were used in the first experiment while male FVB/N mice harboring an endothelial directed deletion of FoxO1/3/4 (Mx1Cre+;FoxO1,3,4L/L) and corresponding age-matched Mx1Cre;FoxO1,3,4L/L littermates (referred to as FoxoΔ and FoxOL/L, respectively) were used in subsequent experiments. The generation of this transgenic mouse model and genotyping for floxed and deleted alleles has been described previously (11). When 4 weeks old, mice received 3 injections of poly-inosine:cytosine (500μg i.p.; Invivogen, San Diego, CA) to transiently induce Cre expression. FoxOΔ have a 70-80% reduction of FoxO1/3 and undetectable levels of FoxO4 within the endothelium (11, 13). Diet-induced obesity At 7-8 weeks of age, mice initiated normal chow (NC) or high-fat (HF) diets for 8 or 16 weeks. Diets consisted of 10.5% fat, 73.1% carbohydrate and 16.4% protein for NC and 58% fat, 25.5% carbohydrate and 16.4% protein for HF (#D12329 and D12331, respectively; Research Diets Inc., New Brunswick, NJ, USA). Tissue collection was conducted under isoflurane anaesthesia and mice were euthanized by exsanguination. Intraperitoneal insulin tolerance tests (ITT) were conducted during diet weeks 7 or 14 after a 4 hour fast. Blood glucose was measured by glucometer (Aviva Accuchek) at times 0, 20, 40 and 60 minutes post-injection of insulin (0.75U/kg i.p.; Humalog; Lilly, Scarborough, ON, Canada). Isolation of skeletal muscle capillary segments Page 5 of 23
NC and HF -wildtype FVB/N mice underwent cardiac perfusion of polystyrene magnetic beads (8.0x107 beads/mouse; PM-50-10, Spherotech, Lake Forest, IL, USA). These 5-6 μm diameter beads lodge within the capillaries, enabling magnetic collection of capillaries (18). Hindlimb muscles were minced and digested with 0.2% collagenase type II (17101-015; Life Technologies, Burlington, ON, Canada). Magnetic separation was used to collect bead-filled capillary segments. Protein was extracted and analyzed by Western blot. In vivo and ex vivo insulin stimulation Muscles from 8 week HF fed FoxOL/L and FoxoΔ mice were analyzed for insulin sensitivity. For in vivo analysis, mice were maintained under isoflurane anesthesia; one extensor digitorum longus (EDL) muscle was removed and snap frozen, then mice were injected with 0.12U insulin i.p. and after 15 minutes, the contralateral EDL was removed and snap frozen. For ex vivo analysis, extensor hallucis proprius muscles from mice were removed from both legs (prior to insulin injection). After 30 minute (37°C) pre-incubation in HEPES-saline buffer, insulin (25 mU/mL Humalog) was added to one of each muscle pair and muscles were incubated at 37°C for 30 minutes, then snap-frozen in liquid nitrogen for protein analysis. Cultured microvascular endothelial cells Mouse skeletal muscle microvascular endothelial cells were isolated and cultured as previously described (13, 14). Cells were plated to confluence in a 6-well tissue culture dish and treated overnight with recombinant TNFα or IL-6 (PHC3015 and PHC0064 respectively; Life Technologies) or with albumin-conjugated palmitate (19) for 48 hours. Microvascular endothelial cells also were isolated from skeletal muscle of FoxOL/L and FoxOΔ mice to measure the extent of endothelial FoxO1 depletion. Protein analysis Protein extraction and Western blotting were performed as described previously (20). Primary antibodies: pSer473 Akt, pThr308 Akt, FoxO1, FoxO3a, Akt, α/β-Tubulin (#4058, 9275, 2880, 2497, 9272, 2148 respectively; Cell Signaling, Pickering, ON, Canada), β-actin (Sc-47778, Santa-Cruz, Ca, USA). Secondary antibodies: goat anti-rabbit or anti-mouse IgG-HRP (111-035-003, 115-035-003; Jackson ImmunoResearch Laboratories, West Grove, PA, USA). Densitometry analysis was performed using Carestream software (Molecular Bioimaging, Bend, OR, USA). VEGF-A protein was assessed in gastrocnemius muscle homogenates (50 µg protein in 120 mM Tris (pH 8.7), 0.1% Triton-X-100) using a mouse VEGF-A Elisa (DY493-05; R&D systems, Minneapolis, MN, USA).
Page 6 of 23
Histological assessments Transverse cryosections (10 μm thick) from plantaris muscle were stained to detect capillaries and arterioles using isolectin (FITC–Griffonia simplicifolia isolectin B4 (1:100; Vector Laboratories, Burlington, ON, Canada) and Cy3-anti-α-smooth muscle actin (SMA) (1:300; C6198, Sigma-Aldrich Canada, Oakville, ON). Capillary to fiber ratios (C:F), SMA+ vessel density and size were calculated from 3-4 non-overlapping fields of view per muscle (20). FoxO1 positive capillaries were detected by co-staining with isolectin (Rhodamine Griffonia simplicifolia Isolectin B4 (1:100; Vector Laboratories) and rabbit anti-FoxO1 (1:50; PLA0084-100; Sigma-Aldrich) or normal rabbit IgG (#2729, Cell Signaling) and Alexa Fluor488 Goat anti-rabbit IgG (1:300; A11008, Invitrogen). Macrophages were detected using anti-F4/80 (1:50; MF48000, Life Technologies) and FITC-goat anti-rat IgG (112-585-003, Jackson ImmunoResearch Laboratories), co-staining with Rhodamineisolectin. Staining was viewed using a Zeiss Axiovert 200M microscope (20X or 40X objectives) and imaged with a cooled digital CCD camera and Metamorph software, using identical exposure settings across specimens for each respective stain. RNA analysis Total RNA was isolated from gastrocnemius muscle using RNeasy Fibrous (Qiagen Inc. Canada, Toronto, ON, Canada) and analyzed by Taqman quantitative real time PCR (qPCR) on the 7500 Fast Real Time PCR platform (Applied Biosystems) using Mastermix (4444963, Life Technologies) and TaqMan FAM-labelled probe sets for Itgam (Mm00434455_m1), neutrophil elastase (Mm00469310_m1), Rorc (Mm01261022_m1) and Hprt1 (Mm00446968_m1). mRNA was quantified relative to Hprt1 and expressed as 2-ΔCt. Transcriptome analysis Gene expression in the muscle of HF-FoxOL/L and HF-FoxOΔ mice was analysed on the Mouse Gene 2.0 platform at the Centre for Applied Genomics (The Hospital for Sick Children, Toronto, Canada). Total RNA was isolated from plantaris muscles (n=8/group) using RNeasy Fibrous (Qiagen Inc. Canada) and was quantified and quality checked by Agilent BioAnalyzer. Quality control and normalization was carried out using the AffyPLM and Oligo packages from Bioconductor on the R platform. All samples passed QC as assessed by mean-expression, NUSE, RLE and PCA of raw expression values. Robust Multi-array Average (RMA) was used for normalization. Raw data and RMA normalized expression values are publicly available (Gene Expression Omnibus #GSE72576). Annotation and filtering was done using mogene20sttranscriptcluster.db and genefilter packages from Bioconductor. Probe-sets were filtered by quantile, excluding those with an IQR below 0.5 from further analysis. Differential expression of probe-sets was analysed using the LIMMA package (Bioconductor), using the Bayesian adjusted p-value of
