Introduction

Low dietary folate is strongly associated with an increased risk of human vascular diseases including atherosclerosis and stroke (Voutilainen et al. 2000, 2004; Durga et al. 2005). This has been attributed primarily to suboptimal intracellular folate, together with low levels of vitamins B12 and B6, inducing hyperhomocysteinemia (Splaver et al. 2004). However, there is evidence for folate alone influencing human vascular disease. High circulating folate is associated with reduced risk of primary coronary events and carotid intima-media thickness independently of total plasma homocysteine in several prospective human studies (Voutilainen et al. 2000, 2004; Durga et al. 2005). Moreover, supplemental folic acid positively influences vascular function in patients with, or at risk of, vascular disease (Mangoni et al. 2005; Tawakol et al. 2005; Till et al. 2005). These effects appear largely independent of homocysteine.

Determining whether low folate or hyperhomocysteinemia is causal for vascular disease in vivo is difficult due to the intimate metabolic relationship linking folate and homocysteine. Transgenic mouse strains provide a representative model of human atherosclerosis and allow complete investigation of the effect of diet on vascular disease progression. The ApoE knockout mouse spontaneously develops atherosclerotic plaques that are morphologically analogous to lesions found in human tissue (Zhang et al. 2002). We have developed distinct nutritional deficiencies in transgenic ApoE null mice so that they exhibit either mild or moderate hyperhomocysteinemia (McNeil et al. 2011). We have shown that B vitamin deficiency increases atherosclerotic plaque formation in transgenic ApoE knockout mice and that this is strongly linked with accumulation of cholesterol and pro-atherogenic lipoproteins in the adventitial lipid surrounding the aorta (McNeil et al. 2011, 2012).

Proteomic technologies have been widely used to identify quantitative changes in protein expression as a consequence of several human vascular pathologies including atherosclerosis, aorta medial degeneration and myocardial revascularisation (Wu et al. 2007; Delbosc et al. 2008; Farina et al. 2010). Moreover, they have proven to be powerful tools in rodents studies for providing mechanistic insight into how vascular disease progresses and is influenced by environmental factors such as diet (Mayr et al. 2005; Almofti et al. 2006).

Here, we employed a combined biochemical and quantitative global proteomics approach to determine how prolonged nutritional folate deficiency and/or moderate hyperhomocysteinemia differentially modulates global protein expression directly in the aorta of ApoE null mice and to further identify whether these changes are associated with proteins and processes involved in the progression of human atherosclerosis.

Materials and methods

Animal study and diets

The design of this animal study and the composition of the experimental diets employed have been described in detail elsewhere (McNeil et al. 2011, 2012). All procedures were carried out in accordance with the requirements of the UK Animals (Scientific Procedures) Act 1986. ApoE null mice (males, 5 weeks old) were allocated (by weight) to one of six treatment groups (n = 10 animals per group). Diets were based on either a control [C; 4 % w/w lard] or high-fat [HF; 21 % w/w lard and cholesterol (0.15 % w/w)] background with different B vitamin compositions applied to each (1) folic acid and B vitamin replete, (2) folic acid depleted (F–), or (3) folic acid, B6 and B12 depleted (F–B–). Mice were fed these diets for 16 weeks.

Blood and aorta collection

Mice were fasted up to 6 h and killed by exsanguination under terminal anaesthesia. Blood was collected from the vena cava for analysis of B vitamins, homocysteine and lipoproteins. Residual blood was flushed from the heart by injection of Dulbecco’s modified Eagles medium (DMEM) into the left ventricle. The complete aorta was microdissected of tunica adventitia. The thoracic and abdominal aorta and the dissected adventitia were snap-frozen and stored at −80 °C for proteomic and lipid analysis. The aortic arch was fixed in 10 % formalin and atherosclerotic plaque volume quantified by staining plaques in situ with the neutral lipid-targeting lysochrome Oil Red O (ORO, Sigma, Poole, UK), which targets fatty acid deposits in the arterial plaques and solubilising and measuring spectrophotometrically the dye retained in the aorta. Solubilised ORO as a biomarker of atherosclerotic plaque volume was corrected for aorta tissue weight (Beattie et al. 2009; McNeil et al. 2011, 2012).

Preparation of aorta tissue for proteomic analyses

Weighed frozen aorta (approx. 10 mg) was added to extraction buffer containing 7 M urea, 2 M thiourea, 4 % CHAPS and 2 % BioRad Biolite Ampholyte pH 3–10 and homogenised using a Precellys 24 lysis and homogenisation cycler (Stretton Scientific, UK). The homogenate was centrifuged at 14,000 rpm for 5 min at 4 °C and the supernatant frozen at −80 °C. Protein was measured using the BioRad RC DC protein assay.

2D gel electrophoresis of aorta tissues

One 2D gel was run per aorta (100 μg) sample (n = 10 mice per treatment group; 60 gels in total). Proteins were separated by isoelectric focusing in the first dimension [BioRad immobilised pH gradient (IPG) strips (pi range 3–10)] and SDS-PAGE in the second dimension on 8–16 acrylamide gels (18 × 18 cm) as described previously (Duthie et al. 2008). Gels were stained with Coomassie blue and imaged on a BioRad GS710 flat bed imager followed by image analysis using PD Quest version 8.0.1. Spots were excised from the gels using a robotic BioRad spot cutter, trypsinised in a MassPrep Station (Waters, MicroMass, Manchester, UK) and analysed by LC MS/MS using an Ultimate nano LC capillary chromatography system (LC Packings, Camberly, Surrey, UK), combined with an Applied Biosystems 2000 Q-Trap (Warrington, UK). Peak list were generated using Analyst 1.4.2 software set at default. Peptide fragment mass spectra yielded a sequence of separated peptides that were pasted into the fingerprinting web resource programme Mascot dll version 1.6b9 (Matrix Science Ltd, Boston, MA) for protein identification. Protein identification was made using SWISS-PROT (version 2012.09) specifically for Mus musculus (mouse), with expectation values, cut-off scores and threshold values for individual proteins considered. Protein confidence parameters including Mowse scores, expected and actual masses, PI, number of unique peptides matched, percentage sequence coverage and gene and protein accession numbers, together with protein function, are shown for each identified peptide in the experimental results (Supplementary Table 1).

B vitamin and lipoprotein analyses

Blood and tissue folate and B12 were measured by radioassay and plasma total homocysteine by gas chromatography (McNeil et al. 2011, 2012). Lipid was extracted from adventitial tissue (approx. 200 mg) (Folch et al. 1957) and total cholesterol, LDL cholesterol, HDL cholesterol, TG and NEFA fractions measured in plasma and aorta adventitial lipid using a Konelab 20 Clinical Chemistry Analyser (Thermo Scientific, Passau, Germany). Fatty acids were trans-esterified to methyl esters (FAMEs) in methanolic HCL, separated by liquid–liquid solvent extraction and measured by gas–liquid chromatography (GLC; Burdge et al. 2000).

Statistical analysis

For proteomic analysis, 2D gel images were studied using PDQuest software (BioRad). After normalisation and matching, spot densities were exported for data analysis (multivariate) using Excel and “R” (R Foundation for Statistical Computing, Vienna, Austria; version 2.10). Log transformed spot densities from aorta tissue were analysed by two-way ANOVA for the effect of fat intake, vitamin composition and interaction between these factors (P < 0.05). These analyses provide quantitative protein expression data, as densitometry of the Coomassie-stained spots is carried out for each animal in each treatment group individually prior to statistical analysis.

Biochemical data were analysed by two-way ANOVA for the effect of fat intake, vitamin composition and interaction between these factors. Significant differences between all groups were detected using the least significant difference (LSD) post hoc method. Analysis of associations among aorta tissue protein expression and blood B vitamin status markers, blood and vascular lipids, and atherosclerotic plaque volume was carried out for all mice in all treatment groups using Pearson’s correlation coefficients. Spots with densities that differed significantly either due to treatment (P < 0.05) or were highly correlated with B vitamin and lipid status at the end of the study (P < 0.001) were excised from the gel and identified by LC MS/MS. Association networks between biochemical biomarkers and protein expression were displayed using Cytoscape software (version 2.8.3; http://www.cytoscape.org). Principal component analysis (PCA) of the physiological biomarkers measured from each mouse in the study was carried out and the component scores calculated.

Results

B vitamin, homocysteine, lipid status and aorta plaque volume

Principal component analysis (PCA) of biochemical data for all animals revealed that the treatment phenotypes separated differentially according to fat intake and folate and B vitamin treatment (Fig. 1). Aortic plaque volume, blood and lipid biomarkers are shown in Table 1. Folate was depleted more than 65 % in mice fed a folate-deficient diet. B12 decreased more than 70 % in animals fed the combined folate and B vitamin-depleted diet. Folate deficiency increased homocysteine twofold. Homocysteine was elevated further in mice fed a combined B vitamin-depleted diet (approx. fourfold). B12 depletion and hyperhomocysteinemia were exacerbated by high fat. Total cholesterol was increased in aorta adventitial lipid in response to a high-fat diet. Plasma cholesterol, HDL and LDL were elevated (approx. twofold). Folate depletion alone did not impact on circulating or adventitial lipids. However, feeding a combined folate and B vitamin-deficient diet exacerbated cholesterol accumulation and significantly reduced MUFA and increased pro-atherogenic SATFAs in the aorta adventitial lipid. Plaque volume was increased almost twofold (P < 0.0001) in mice fed a HF diet for 16 weeks when compared with animals on a control diet (Table 1). Plaque volume was significantly increased (17 %; P < 0.05) in mice depleted of folic acid. Imposing a combined B vitamin deficiency did not increase plaque development further (Table 1).

Fig. 1
figure 1

Primary component analysis (PCA) score plot of blood and tissue biomarkers significantly altered by folic acid, B vitamins and fat intake in ApoE null mice. Mice were fed a control (C) or high-fat (HF) diet depleted of folic acid (F–) or folic acid and vitamins B6 and B12 (F–B–) for 16 weeks. Symbols represent individual replicate animals within every treatment group

Full size image
Table 1 The effect of folic acid, B vitamins and dietary fat intake on B vitamin, homocysteine and plasma and aorta adventitial lipid status in ApoE null mice
Full size table

Atherosclerotic plaque abundance was correlated strongly with circulating and vascular lipid levels, and weakly with plasma vitamin B12 (Table 2). Plaque formation did not correlate directly either with blood folate or homocysteine. However, whole blood and plasma B12 did strongly influence both circulating and adventitial lipid concentrations, with low B12 associated with high levels of plasma TG, adventitial total cholesterol and SATFAs and, conversely, a lower proportion of MUFAs. Low folate was similarly associated with accumulation of SATFAs in the aorta adventitia. Hyperhomocysteinemia was positively associated with elevated SATFA and reduced MUFA concentrations in the adventitia (Table 2).

Table 2 Correlation analyses of B vitamin, lipid status and atherosclerotic plaque formation in ApoE null mice
Full size table

Aorta tissue proteomics

Comparative quantitative proteomics revealed approximately 800 spots in the aorta samples that matched across all 60 treatment gels. The primary proteins (approx. 35 % of total identified proteins) influenced by treatment in the aorta were involved in regulating cytoskeletal organisation, cell adhesion and cell motility (Fig. 2). Mean (and group SEM) spot densities for aorta proteins that differed between treatment groups (P < 0.05) are shown in Table 3. The independent effects of fat or vitamin treatment on protein expression are described (P values) together with the synergistic effect of a HF and combined vitamin-depleted diet. The effect of B6 and B12 depletion, above folate deficiency alone is also shown. Vitamin depletion alone altered expression of only six aorta proteins [fibrinogen, gelsolin, glutamate dehydrogenase 1, lactate dehydrogenase and talin-1 (all upregulated) and peptidyl-prolyl cis–trans isomerise A (downregulated)]. Feeding a combined vitamin-deficient and high-fat diet had a strong synergistic effect on 25 aorta proteins (including dermatopontin, fibrinogen, glutathione S-transferase Mu 1, peroxiredoxin-2, serotransferrin, transgelin and vimentin; Table 3).

Fig. 2
figure 2

Proteins and major cellular pathways altered by folic acid, B vitamins and fat intake in aorta from ApoE null mice. All proteins (influenced by treatment and associated with B vitamin and lipid status at the end of intervention), together with primary cellular function are presented as percentage of total identified proteins

Full size image
Table 3 Aorta proteins significantly altered in response to folic acid, B vitamins and fat intake
Full size table

Correlation analysis revealed strong (P < 0.001; mean q = 0.03) associations between lipid and vitamin status, and aorta protein expression (Table 4). Protein expression in the aorta was exclusively correlated with aorta adventitial lipid levels (total fat, cholesterol, HDL, NEFA) and with the proportion of saturated and monounsaturated fat. Figure 3 constructed from biochemical and aorta protein correlations demonstrates the impact that B vitamins, homocysteine and methionine cycle intermediates have on lipoprotein accumulation directly in aorta adventitial lipid (e.g., increased SATFAs, decreased MUFAs, increased total cholesterol) and the influence that aorta lipid levels have on the expression of more than 80 aorta proteins (identified by spot number in Supplementary Table 1).

Table 4 Associations (correlation analyses) between aorta lipid and aorta protein expression
Full size table
Fig. 3
figure 3

Association network 1 (ball and stick) of aorta proteins influenced by B vitamin and fat intake in ApoE null mice. Association networks between biochemical biomarkers and protein expression were displayed using Cytoscape software (version 2.8.3; http://www.cytoscape.org). Lines connecting proteins and biomarkers indicate a correlation with R > 0.3 or <−0.3. ACH adventitial total cholesterol, AHDL adventitial high-density lipoprotein, AMF adventitial monounsaturated fatty acid, ANF adventitial non-esterified fatty acid, APF adventitial polyunsaturated fatty acid, ASF adventitial saturated fatty acid, ATF adventitial total fat, ATG adventitial triglyceride, B 12 vitamin B12, HDL high-density lipoprotein, Hcy homocysteine, LDL low-density lipoprotein, SAH S-adenosylhomocysteine, SAM S-adenosylmethionine, SAM: SAH S-adenosylmethionine: S-adenosylhomocysteine ratio, TC total cholesterol, TG triglyceride. Spot number and protein name: 409, serotransferrin; 410, serotransferrin; 411, glyceraldehyde-3-phosphate dehydrogenase; 438, moesin; 468, 78 kDa glucose-regulated protein; 473, heat shock protein 1β, 479, WD repeat-containing protein 1; 503, stress-induced-phosphoprotein 1; 507, transforming growth factor β-induced; 524, pyruvate kinase isozymes M1/M2; 557, fibrinogen Β β polypeptide isoforms; 582, protein-glutamine gamma-glutamyltransferase 2; 604, glutamate dehydrogenase 1; 682, serotransferrin; 711, isovaleryl-CoA dehydrogenase; 752, vimentin; 772, aldose reductase; 812, electron transfer flavoprotein subunit α; 813, S-formylglutathione hydrolase; 829, annexin A1; 875, Rho GDP dissociation inhibitor 1; 887, glutathione S-transferase Mu 1; 890, collagen α-1(XVIII) chain; 909, GTP-binding nuclear protein Ran; 912, isocitrate dehydrogenase; 924, peroxiredoxin-2; 938, vimentin; 1,060, transgelin; 1,085, myosin regulatory light polypeptide 9; 1,125, transgelin; 1,126, myosin light polypeptide 6; 1,169, dermatopontin

Full size image

Discussion

The primary aim of this study was to examine and quantify the differential effects of folate and/or B vitamin depletion on global protein expression in the aorta of ApoE mice and to consider the role of these proteins on cellular processes implicated in human vascular disease. Feeding a folate and/or B vitamin depleted diet against a control fat background altered expression of only very few aorta protein [fibrinogen, gelsolin, lactate dehydrogenase, talin (upregulated) and peptidyl-prolyl cis–trans isomerise (decreased)]. However, when combined with a major dietary challenge (a high-fat diet), B vitamin depletion strongly perturbed many more processes [including oxidative stress, inflammation and vascular smooth muscle cell (VSMC) activation and migration] causally associated both with rodent and human atherosclerosis (see Fig. 2). Suboptimum folate status and hyperhomocysteinemia are linked with hyperlipidemia and oxidative stress, major risk factors for vascular disease (Kris-Etherton et al. 2002; Lohm et al. 2002; Mikael et al. 2006; Devlin et al. 2007). Here, B vitamin depletion was associated with significant changes in vascular adventitial lipid metabolism and protein markers of cellular stress known to be altered in atherosclerosis (Brownlee 2001; Vikramadithyan et al. 2005) only when combined with this additional oxidative insult. Glutathione S-transferase Mu was elevated, as was aldose reductase, a protein that induces atherosclerosis in mice (Vikramadithyan et al. 2005) and is upregulated in human foam cells and plaque macrophages (Gleissner et al. 2008). Peroxiredoxin expression was decreased in this study. Low peroxiredoxin expression is associated with atherosclerosis in mice and humans (Gleissner et al. 2008). Elevated serotransferrin is linked with oxidative stress and chronic disease risk. Multiple aorta spots were identified as serotransferrin in this study and feeding a high-fat/B vitamin-depleted diet increased expression of each of these (P < 0.05). Moreover, serotransferrin expression was positively associated with total cholesterol and HDL accumulation in the adventitia and negatively linked with monounsaturated fatty acid concentration. Collective changes in these proteins indicate an upregulated metabolic response to hyperlipidemia in the aorta and an increased requirement to resist vascular ROS generation.

Inflammation and accelerated growth and migration of VSMC are critical in the progression of atherosclerosis. Here, combined B vitamin depletion altered expression of multiple protein molecular markers indicative both of a pro-inflammatory environment and a pro-proliferative VSMC phenotype. Annexin A1 (lipocortin), a potent anti-inflammatory agent (Perretti and Dalli 2009) is present in human atherosclerotic plaques, with low protein levels indicative of advanced disease (Cheuk and Cheng 2011). Lipocortin was negatively associated with adventitial cholesterol accumulation. Atherosclerosis-driven inflammation increases expression of acute phase proteins including fibrinogen, C-reactive protein and serotransferrin. Hyperfibrinogenemia is a marker for advanced human atherosclerosis and a predictor for cardiovascular events (Kofoed et al. 2003; Sabeti et al. 2005). Protein-glutamine gamma-glutamyltransferase 2 (fibrinogen-stabilising factor) promotes fibrin deposition within the arterial lumen and vascular endothelial cell dysfunction, VSMC proliferation and atherosclerosis (Smith 1995). Here, fibrinogen and fibrinogen-stabilising factor protein expression increased and was positively linked with total and HDL cholesterol and with reduced MUFA concentrations. Specific heat shock proteins (including Hsp60 and Hsp70) are elevated in early human cardiovascular disease and are predictive for atherosclerotic progression (Xu 2002; Pockley et al. 2003). Hsp70 (identified here as 78 kDa glucose-related proteins and Hsp 1B) and stress-induced phosphoprotein 1 (a modulator of Hsp70 activity) were positively associated with total and HDL adventitial cholesterol and negatively linked with MUFA concentration. Rho GDP dissociation inhibitor 1 negatively regulates VSMC adhesion and enhances migration (Qi et al. 2008). Here, feeding a high-fat combined B vitamin-depleted diet significantly downregulated Rho GDP dissociation inhibitor 1 expression in mouse aorta. Conversely, GTP-binding nuclear protein Ran expression is positively associated with VSMC migration (Fuastino et al. 2010). GTP-binding nuclear protein Ran expression was upregulated in this experiment. Actin assembly and disassembly drives vascular cytoskeletal organisation and regulates VSMC phenotype. Many aorta proteins (including transgelin, vimentin, dermatopontin, collagen, moesin, myosin regulatory light polypeptide 6 and 9, and talin) that influence actin filament regulation, cell differentiation, cell adhesion and cell migration were altered by treatment in this study. Reduced expression of transgelin-1 (sm22α) is related to VSMC proliferation and atherosclerotic plaque formation (Feil et al. 2004; Shen et al. 2010). Transgelin expression was strongly downregulated in the aorta of mice fed a high-fat and combined B vitamin-depleted diet. Moreover, a strong negative relationship between transgelin and saturated fatty acid accumulation in adventitial lipid was observed. Overexpression of gelsolin increases vascular endothelial cell motility (Silacci et al. 2003). Gelsolin was elevated in the aorta of ApoE null mice in this study. While collagen deposition is associated both with plaque formation and stability, polymerised collagen type 1 maintains VSMC quiescence (Adiguzel et al. 2009). Here, expression of this anti-proliferative and anti-inflammatory protein was significantly decreased. Dermatopontin, an extracellular matrix protein, regulates collagen deposition and promotes cell adhesion (Okamoto et al. 2010), while moesin, a member of the ezrin/radixin/moesin (ERM) protein family (together with Rho GDP dissociation inhibitor 1), inhibits cell migration (Takahashi et al. 1997). Feeding a high-fat B vitamin-depleted diet decreased aorta moesin and dermatopontin levels and (as described previously) Rho GDP dissociation inhibitor 1 expression. Conversely, expression of talin, a cytoskeletal protein which promotes cell dispersal (Albiges-Rizo et al. 1995) was doubled.

These parallel quantitative changes in numerous proteins implicated in VSMC proliferation and motility provide strong but indirect evidence that feeding a combined B vitamin-depleted high-fat diet may promote an invasive VSMC phenotype in mice vivo. Whether B vitamin depletion impacts on human aorta protein expression and on SMC phenotype in the same way remains to be established. ApoE null mice are recognised as a good model for human vascular disease, spontaneously developing atherosclerotic plaques morphologically similar to lesions found in humans (Zhang et al. 2002). Additionally, here, we have employed a nutritionally relevant nutritional rodent model that better reflects a moderate B vitamin deficiency when compared with previous studies inducing supraphysiologically high levels of homocysteine (McNeil et al. 2011). Furthermore, our findings are consistent with other studies using proteomic technologies to profile aorta or plasma protein expression patterns in several human vascular pathologies. Many proteins, acknowledged as characteristic of vascular disease progression, are identified in this animal study as strongly linked to nutritional B vitamin depletion (noted in parentheses). Medial degeneration in the human ascending aorta alters the expression of proteins critical in vascular remodelling, cytoskeleton organisation and muscle contraction (actin, collagen, profilin, myosin, vimentin), VSMC phenotype (transgelins), protein folding and ROS detoxification (Hsps, peroxiredoxin, protein disulphide isomerase), and iron transport and inflammation (serotransferrin) (Mayr et al. 2005). The HUPO Plasma Proteome Project identified more than 300 human plasma proteins as strong biomarkers of vascular function and/or disease. These include markers of inflammation (transferrin), coagulation (fibrinogen A α-polypetide, fibronectin) and cytoskeletal organisation (actin, myosin, talin and villin) (Sanzen et al. 2010). Protein profiling has also been employed to investigate mechanism of vascular dysfunction in animal models. Serotransferrin, 78 kDa glucose-regulated protein and pyruvate kinase isozyme M1/M2 are elevated in experimentally induced rat heart failure (Ingwall and Weiss 2004), while fibrinogen, collagen, peroxidase and transgelin are overexpressed in atherosclerotic mice (Mayr et al. 2005; Wu et al. 2007). Transgelin-1 is increased in the artery wall of hypertensive rats (Delbosc et al. 2008).

In summary, nutritional B vitamin depletion when combined with a high-fat diet in a mouse model of human vascular disease quantifiably altered the expression of many aorta proteins implicated in human atherosclerosis, primarily proteins associated with regulating the VSMC phenotype. It remains to be established whether these significant changes in vascular protein expression are driven by elevated homocysteine or are a consequence of other B vitamin-sensitive mechanisms.