2126
DOI 10.1002/pmic.200700784
Proteomics 2008, 8, 2126–2135
RESEARCH ARTICLE
Aorta protein networks in marginal and acute zinc deficiency John H. Beattie1, Margaret-Jane Gordon1, Garry J. Rucklidge2, Martin D. Reid2, Gary J. Duncan2, Graham W. Horgan3, Young-Eun Cho4 and In-Sook Kwun4 1
Division of Vascular Health, Rowett Research Institute, Greenburn Road, Bucksburn, Aberdeen, UK Proteomics Unit, Rowett Research Institute, Bucksburn, Aberdeen, UK 3 BioSS, Rowett Research Institute, Bucksburn, Aberdeen, UK 4 Department of Food Science and Nutrition, Andong National University, Andong, Kyungpook, South Korea 2
Human zinc deficiency is a global problem and may influence the development of cardiovascular disease. Our objective was to determine Zn deficiency affected pathways and protein interactions in rat aorta and their likely influence on stress-induced atherogenesis. In two separate studies, rats were given diets acutely (,1 mg Zn/kg) or marginally (6 mg Zn/kg) deficient in Zn. Both studies included Zn adequate controls (35 mg Zn/kg) and the acute deficiency study included a pair-fed group. After 6 wk, proteins from thoracic aorta were separated by 2-DE. Proteins affected by zinc deficiency were identified by principal component analysis. Multiple correlations of identified proteins indicated protein networks of related function. Proteins clusters decreased in zinc deficiency were related to fatty acid and carbohydrate metabolism. Structurally related proteins, including zyxin and over nine transgelin 1 proteins, were either increased or decreased by acute and marginal deficiencies. PKCa was significantly decreased in Zn deficiency suggesting that Zn may regulate the phosphorylation of target proteins. Zn deficiency-related changes in structural, carbohydrate and fatty acid-related proteins may be disadvantageous for maintaining vascular health and are consistent with a protective role for zinc in the development of atherosclerosis.
Received: August 9, 2007 Revised: December 19, 2007 Accepted: January 14, 2008
Keywords: Atherosclerosis / Transgelin / Vascular health / Zinc deficiency
1
Introduction
One-third of the World’s population is thought to be Zn deficient [1] and the pathology of acute Zn deficiency in humans and animals is well documented [2]. However, the Correspondence: Dr. John H. Beattie, Division of Vascular Health, Rowett Research Institute, Greenburn Road, Bucksburn, Aberdeen AB21 9SB, Scotland, UK E-mail:
[email protected] Fax: 144-1224-716629 Abbreviations: ALP, Alkaline phosphatase; APF, pair-fed control group for acute zinc deficiency; AZA, zinc adequate control group for acute zinc deficiency; AZD, acute zinc deficiency; ENH, Enigma homologue protein; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; MZA, zinc adequate control group for marginal zinc deficiency; MZD, marginal zinc deficiency
© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
influence of Zn status on chronic diseases, such as cardiovascular disease, caused by other lifestyle or dietary factors has rarely been considered. A major reason for this is the lack of epidemiological studies investigating associations of Zn deficiency with disease incidence in humans due to the absence of a reliable diagnostic marker of Zn status [3]. Zinc has many biological roles, being a structural and/or functional component in over 7% of the human proteome. It is therefore not surprising that Zn deficiency can potentially have an impact not only on specific Zn-dependent proteins, but also on entire metabolic pathways. Zn has antioxidant and anti-inflammatory properties [4, 5] and Zn status may therefore influence the pathogenesis of chronic diseases that are initiated and driven by such stresses. There is clear evidence that administration of Zn can protect against stress induced by pro-oxidants [6, 7], which may be a direct protection by Zn rather than an indirect effect through, for examwww.proteomics-journal.com
Animal Proteomics
Proteomics 2008, 8, 2126–2135
ple, the induction of the Zn-binding protein metallothionein. Likewise, Zn deficiency increases the susceptibility of cells to the effects of oxidative stress [4]. It is therefore plausible that Zn deficiency may influence atherogenesis and the development of heart disease [8], the severity of which is partly a direct consequence of chronic exposure to oxidative stress. Indeed, the effects of stress from exposure of vascular endothelial cells to pro-oxidants, such as TNFa and linoleic acid are more harmful in Zn deficiency [9]. In LDL receptor knock-out mice acute Zn deficiency elicits changes in key transcription factors, such as PPARg and NFkB, and adhesion molecules such as VCAM1, which are proatherogenic [10]. In human population studies, a strong negative association was found between dietary Zn intake and the incidence of factors related to diabetes and heart disease, particularly in urban localities [11]. While the effects of acute Zn deficiency may be more readily observed than with marginal deficiency, the occurrence of acute deficiency in human populations, even in developing countries with under- or mal-nutrition, is rare. In contrast, marginal Zn deficiency is thought to be widespread and a significant global problem. Major and micronutrients have a substantial impact on the development of atherosclerosis and we hypothesised that inadequate Zn nutrition would compromise vascular health. Our objective was to reveal the complex changes and interactions of vascular proteins in acute and marginal Zn deficiency whose pattern might indicate mechanisms of action and potential sensitivity to proatherogenic stress factors. We used 2-D gel proteomics to construct aorta protein networks that would point to targets of zinc deficiency. This technique utilises statistical approaches whose power is based on the inclusion of adequate replication in the study design. It has yielded a clear insight into zinc-influenced protein interactions and pathways in vascular tissue.
2
Materials and methods
2.1 Animals and diets Sprague–Dawley rats (100 g, age 4 wk) were obtained from SLC (Shizuoka, Japan), and were individually housed in stainless steel wire-bottom plastic cages in an environmentally temperature controlled room at 22 6 0.57C with an alternate 12 h light and dark cycle. All of the animals were given a modified egg albumen and corn oil-based AIN-76 diet that contained 35, 6 or ,1 mg Zn/kg. Zn-adequate and Zn deficient rats were fed ad libitum. The pair-fed rats were given an amount of diet equivalent to that consumed by a paired Zn deficient rat. All rats had free access to purified water from plastic bottles with silicon stoppers. The diet was stored at 47C in plastic containers and handled with plastic zinc-free gloves and appropriate utensils to avoid Zn contamination. Food intake was recorded daily and body weight was recorded weekly. The care and use of the © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
2127
rats strictly followed “The Guiding Principles for the Care and Use of Animals” [12] and the ANU institutional ethical guidelines. In Study 1 (marginal zinc deficiency; MZD), 18 rats were divided randomly into two groups of nine animals, and one group received a marginally zinc deficient diet, containing 6 mg Zn/kg (MZD group), for 43 days. In Study 2 (acute zinc deficiency; AZD), 30 rats were divided randomly into three groups of ten animals and one group received an acutely Zn deficient diet, containing ,1 mg Zn/kg (AZD group), for 39 days. Animals in a second group (pair-fed control group for acute zinc deficiency; APF group) were pair-fed with rats on the AZD diet. Each pair-fed rat was given the same weight of a Zn adequate diet (35 mg Zn/kg) as the amount of AZD diet that had been consumed by its pair on the previous day. The control rats for Study 1 (zinc adequate control group for marginal zinc deficiency; MZA group) and Study 2 (zinc adequate control group for acute zinc deficiency; AZA group) received the Zn-adequate diet ad libitum. All rats were killed using ether anaesthesia and cervical dislocation. 2.2 Blood and tissues Heparinised blood samples were centrifuged at 10 0006g for 15 min to obtain plasma. The thoracic aortas were quickly removed, trimmed of adventitial fat and perfused with icecold 10 mM Tris-HCl buffer Ph 7.4 containing 0.25 M sucrose. The aorta and plasma samples were then snap frozen in liquid nitrogen and stored at 2807C. Proteomic analyses were completed within 5 months of the animal study termination date. 2.3 Diet, plasma and liver zinc analysis Samples of the diets were wet-ashed and Zn concentrations were analysed by inductively coupled plasma emission spectroscopy (Flame Modula S, Spectro, Kleve, Germany). The accuracy of the method was evaluated using bovine liver standard reference material (NIST SRM 1577b, USA). Plasma samples were diluted 1:10 with 0.1 M hydrochloric acid, centrifuged at 25006g, and then analysed for Zn using a Unicam Solaar 969 atomic absorption spectrophotometer (AAS). 2.4 Alkaline phosphatase (ALP) activity assay Plasma ALP was measured enzymatically using p-nitrophenyl phosphate as the substrate and recording OD at 420 nm, as previous described [13]. Protein concentration was estimated by the method of Lowry et al. [14]. 2.5 2-D gel proteomics Each aorta was defrosted and divided on an ice-cold surface into nine equally sized rings. Three rings, one from the anterior end, one from the posterior end and one from the www.proteomics-journal.com
2128
J. H. Beattie et al.
middle section were pooled together and homogenised in 0.5 mL of a solution containing 7 M urea, 2 M thiourea, 4% CHAPS, 2.5% DTT, 10% isopropanol, 5% glycerol, 2% Biolyte ampholite pH 3–10. The proteins were separated by isoelectric focussing in the first dimension using IPG strips (pH 3–10, BioRad, Hemel Hempstead, UK) and SDS-PAGE in the second dimension on 8–16% gradient gels (18618 cm2), as described in detail elsewhere [15]. The IPG strips were rehydrated with 300 mg of aorta protein and gels were stained with CBB. Images were obtained from a high resolution scanner and analysed using PDQuest software (BioRad). Aorta tissue from each rat was individually processed and analysed giving rise to nine replicate gel images per group in Study 1 and ten replicate gel images per group in Study 2. Mean protein spot density analysis could therefore be regarded as quantitative. 2.6 Statistics and bioinformatics After normalisation and spot matching on PDQuest, the densities of all matching spots were statistically analysed using Genstat (VSN International, Hemel Hempstead, UK), SIMCA-P (Umetrics UK, Winkfield, UK) and Excel software. Unpaired t-tests were used to compare protein spots from Zn adequate and Zn deficient rats whereas paired t-tests were used for comparing spots from Zn deficient and pair fed rats. All matched spot data were then subjected to principal component analysis (PCA) of the correlation matrix and presented as 2-D plots of the first and second component scores for observations (rats) and loadings for variables (spot densities). Those spots with high positive or negative scores which were most responsible for the score separation of rats with different Zn intakes were added to the list of significant spots determined by the t-tests. As would be expected, many of the significant spots were also identified by PCA. The densities of identified spots were then subjected to correlation analysis and spot associations were displayed as a network using Cytoscape software (http://www.cytoscape.org/). Proteins were ranked according to an index of the number and strength of correlations in order to indicate possible proximity to target sites for zinc deficiency. 2.7 MS Spots affected by Zn deficiency were cut robotically, proteolysed in a Mass Prep Station and analysed by MALDITOFMS (Voyager-DE PRO in reflectron mode, Applied Biosystems) as previously described in detail [15]. Trypsin was used for proteolysis of all proteins, but selected proteins were also proteolysed with endoproteinases Lys-C, Glu-C, Arg-C and Asp-N from a Protease Profiler Kit (Sigma), using the method supplied by the manufacturer. Most proteins were further analysed by LC-MS/MS using an Ultimate Nano-LC capillary chromatography system (Dionex, Camberly, UK) hyphenated with an Applied Biosystems 4000 Q Trap. Samples processed by the Mass Prep Station (trypsinised and © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Proteomics 2008, 8, 2126–2135
dissolved in 1% TFA) were dried down and dissolved in 0.1% formic acid prior to injection of 5 mL onto a 15 cm675 mm PepMap C-18 nano-column. Laddered peptide fragment mass spectra yielded the sequence of separated peptides and the peptide fingerprinting web resource program “MASCOT” (Matrix Science, Boston, USA) was used to interrogate the MSDB database (Release 20063108) for matching proteins. Search parameters were as follows: fixed modifications, carbamidomethyl (C); variable modifications, oxidation (M); missed cleavages, up to 1; peptide tolerance, 1.2 Da. 2.8 Western blotting Protein extracts were separated on 10% BioRad Criterion XT gels (for PKCa), or gradient 2-D gels as described above (for transgelin), blotted onto PVDF membranes and targets were detected using an established chemiluminescence method (Pierce Biotechnology, Rockford, IL, USA). The primary antibodies were anti-PKCa (C-20, sc-208), diluted 1:400, and antitransgelin (P15, sc-18513), diluted 1:200 (Santa Cruz Biotechnology, Heidelberg, Germany). Images were captured using a Fujifilm LAS-1000plus Luminescent Image Analyser (Raytek Scientific, Sheffield, UK).
3
Results
3.1 Zinc intake and status Zinc analysis of diets with nominal levels of 35, 6 and ,1 mg Zn/kg gave empirical values of 36.4, 5.64 and 0.89 mg Zn/ kg, respectively. Standard reference material analyses yielded Zn concentrations that were 94.5% of the certified value and were within the acceptable reference range. In Study 1 (marginal Zn deficiency), MZA and MZD rats grew at the same rate throughout the study, food intake was the same and there was no evidence of food intake cycling, a characteristic of acute Zn deficiency. Body weight (mean 6 SEM) at termination of the study was 348.3 6 4.9 and 356.0 6 5.0 g for MZA and MZD rats, respectively. However, plasma ALP levels were significantly lower in MZD rats compared with MZA animals (mean 6 SEM: MZA 427.1 6 24.3; MZD 304.2 6 19.2 IU/L; t-test, p,0.05). No effect on food intake or growth combined with lower plasma ALP confirmed their marginally Zn-deficient status. In Study 2 (acute Zn deficiency), weight gain of rats was considerably reduced within days of consuming the AZD diet, as compared with the animals consuming the AZA diet. APF rats showed a very similar pattern of weight gain to the AZD animals, and body weights (mean 6 SEM) at termination of the study were 347.8 6 5.2, 224.9 6 3.9 and 219.7 6 2.5 g for AZA, APF and AZD rats, respectively (ANOVA and Tukey’s comparisons: AZA–AZD and APF– AZD, p,0.05). From day 2 of the study, all rats eating the AZD diet showed a 3–4 days cycling pattern in the amount of food consumed. Plasma Zn levels were significantly lower www.proteomics-journal.com
Proteomics 2008, 8, 2126–2135
(p,0.05) in AZD rats as compared with AZA and APF animals (data not shown). Thymic atrophy is characteristic of acute Zn deficiency and the weight of the thymus from AZD rats was significantly lower (p,0.05) than that from AZA and APF rats (data not shown). Plasma ALP levels were significantly lower in AZD rats compared with both AZA and APF animals (mean 6 SEM: AZA 666 6 29.1, APF 586.8 6 70.2, AZD 333 6 35 IU/L, AZA–AZD and APF– AZD p,0.05). The combination of all these observations indicates that the rats were acutely Zn deficient. 3.2 Aorta proteins in marginal (Study 1) and acute (Study 2) zinc deficiency The certainty of protein identification (.95% in all but four cases) is provided as a MOWSE score, along with peptide sequence information or coverage data, in the Supporting Information files associated with this paper. Of the stained protein spots matched on many or all gels in Study 1 (460 spots) and Study 2 (333 spots), around 11% were significantly (p,0.05) decreased by both AZD and MZD. In contrast, 5.4% of matched spots were significantly (p,0.05) increased by AZD compared to 0.9% for MZD. Spot densities significantly changed by APF as well as AZD were excluded. PCA analysis showed that both AZD and MZD influenced first and second component scores for observations (rats), confirming changes in at least some protein levels due to zinc deficiency (Figs. 1A and C). The proteins most influencing changes in observations scores were identified in loadings plots of the same data (Figs. 1B and D). In Study 2, APF also influenced the observations scores for rats but independently of AZD. Therefore, from loadings plots (Fig. 1D), it was possible to identify spots least affected by APF and most affected by AZD. A loadings plot for Study 1 spot densities also indicated proteins most affected by MZD (Fig. 1B). In Study 1, construction of protein networks from
Animal Proteomics
2129
multiple protein–protein correlations using “Cytoscape” revealed two main clusters of highly correlated proteins (proteins 1–5, 20–47, Fig. 2). The major cluster involved an interaction between about 40 proteins with a core of about 20, many of which were significantly decreased by MZD (proteins 20–47, Fig. 2). The majority of proteins in this core had functions relating to fatty acid and carbohydrate metabolism (Table 1, Fig. 2). In Study 2, construction of protein networks from multiple correlations revealed two main protein clusters (Fig. 3). The most prominent cluster contained nine proteins (proteins 36, 37, 40, 42–47 in Fig. 3), all but one of which were significantly increased in AZD compared with AZA. Most but not all of these proteins were related to the cell structure and included zyxin and several transgelins (Table 1, Fig. 3). Enigma homologue protein (ENH), a PKC binding protein which was increased by AZD, was highly correlated with the transgelins (Fig. 4A) and zyxin. The second most prominent cluster was composed of six proteins, which were all but one significantly decreased by AZD, and were related to fatty acid and carbohydrate metabolism. These included L-3-hydroxyacyl-CoA dehydrogenase (protein 18, Fig. 3), acetyl-CoA Cacetyltransferase (protein 13) and superoxide dismutase 2 (protein 17). Spot density data for proteins identified by statistical analysis as changing significantly in the zinc deficient group compared to the zinc adequate group were also found to be significantly correlated with plasma zinc levels. The correlation coefficients and their statistical significance for highly clustered proteins in the acute deficiency network (Fig. 3) are shown in Table 1. 3.3 Transgelin expression In the acute Zn deficiency study, a total of 13 spots gave peptide sequences that identified them as transgelin with a high degree of probability (p.0.999). Using Western blotting
Figure 1. PCA of matched spot data of aorta proteins from marginal (A and B) and acute (C and D) Zn deficiency studies. In the observations plots (A and C), the symbols represent Zn adequate (d), Zn deficient (s) and pair-fed (m) rats. In the loadings plots (B and D), the proteins selected for identification and network analysis in Figs. 2 and 3 are indicated by open circles (s) whereas those not selected are indicated by closed circles (d). The ellipse shown in the observations plots encloses 95% of the density of a bivariate normal distribution fitted to the data.
© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.com
2130
J. H. Beattie et al.
Proteomics 2008, 8, 2126–2135
Figure 2. Correlation matrix of proteins influenced by marginal Zn deficiency. Open (s) and closed circles (d) are proteins increased and decreased by Zn deficiency, respectively. Double circle lines indicate statistical significance (p,0.05) for an effect of Zn deficiency. Lines connecting proteins indicates a correlation with R.0.6 or, –0.6, and the heavier and more connected the line, the stronger the correlation.
Table 1. Top ten aortic proteins ranked according to the strongest and most numerous correlations with other proteins in acute (AZD) and marginal (MZD) Zn deficiency
Fig. 2 spot no.
MZD
1
32
2
26
3
27
4
28
Short-chain acyl-CoA DHDase Dihydrolipoamide DHDase Guanosine mono-P RDTase Long-chain-acyl-CoA DHDase
5 6
21 33
7
29
8
37
9
38
10
20
Rank
2-Enoyl-CoA hydratase Ubiquinone oxidoreductase Short-chain acyl-CoA DHDase Short-chain acyl-CoA DHDase Propionyl-CoA carboxylase Enoyl CoA hydratase domain 1
Fold change
p
Fig. 3 spot no.
AZD
Fold change
p
Correlation with plasma Zn R
p 0.001
21.6
0.035
45
Transgelin 1–7*
11.6
0.001
20.560
21.5
0.031
40
Zyxin
11.9
0.001
20.639 ,0.001
21.9
0.064
47
12.7
0.004
20.499
21.7
0.008
37
12.6
,0.001
20.675 ,0.001
22.2 21.8
,0.001 0.019
43 44
Chloride intracellular channel protein 4 Protein kinase C-binding protein Enigma Transgelin 1–5* Transgelin 2
11.9 11.7
,0.001 0.011
20.602 ,0.001 20.474 0.008
22.3
,0.001
36
Transgelin 1–6*
11.7
,0.001
20.540
0.002
21.7
0.004
17
0.003
0.521
0.003
21.5
0.104
18
0.008
0.586
0.001
21.7
0.038
13
Superoxide 22.2 dismutase 2 L-3-hydroxyacyl-CoA 22.4 DHDase Acetyl-CoA 23.4 C-acetyltransferase
0.007
0.527
0.003
0.005
Up (1) or down (2) regulation and fold-change from the Zn adequate group are presented along with the statistical significance level (p). Asterisks indicate the transgelin 1 spot number in Fig. 5. Protein spot density data for acutely Zn deficient (AZD) and Zn adequate (AZA) rats were correlated with plasma Zn from the same animals, and the correlation coefficients (R) and significance levels (p) are shown. DHDase, dehydrogenase; RDTase, reductase.
© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.com
Animal Proteomics
Proteomics 2008, 8, 2126–2135
2131
Figure 3. Correlation matrix of proteins influenced by acute Zn deficiency. See Fig. 2 legend for symbol explanation. Asterisk indicates transgelin 1 spot number in Fig. 5. All proteins with the exception of 11, 25, 31, 34, 50 and 54 were significantly (p,0.05) correlated with plasma zinc when pooling data from Zn adequate (AZA) and Zn deficient (AZD) groups. Those proteins increased or decreased by Zn deficiency were negatively or positively correlated with plasma Zn, respectively.
Figure 5. Identification of transgelin 1 protein isoforms in part of a CBB stained gel (A) and a 2-D gel Western blot probed with an antitransgelin 1 antibody (B), showing the same region as the gel. Corresponding spots on each image are indicated by numbers.
3.4 PKCÆ expression
Figure 4. Correlations between proteins significantly affected by acute Zn deficiency (AZD). Zn adequate (AZA) and pair-fed (APF) data are also shown and all correlations were highly significant (p,0.001). Asterisks indicate transgelin 1 spot number in Fig. 5.
of the 2-D gels, a cluster of nine proteins corresponding to the transgelins identified by LC-MS/MS analysis on CBB stained gels were found (Fig. 5). The partial (52–61%) sequences of these proteins showed 100% homology with that of transgelin 1 (Fig. 6). © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The levels of PKCa in aortas from acutely Zn deficient rats were measured by Western blotting. Aorta PKCa levels (relative density) were significantly decreased in the AZD group compared with the AZA and APF groups (mean 6 SEM: AZA 1.35 6 0.27; APF 1.07 6 0.25; AZD 0.47 6 0.05; ANOVA and Tukey’s comparisons: AZA–AZD and APF–AZD, p,0.05).
4
Discussion
Many individual proteins were significantly affected by zinc deficiency, but utilisation of the protein correlation and networking technique has strongly focussed attention on three key cellular functions represented by limited numbers of highly correlated and clustered proteins. Zinc deficiency consistently targeted protein networks with functions relatwww.proteomics-journal.com
2132
J. H. Beattie et al.
Proteomics 2008, 8, 2126–2135
Figure 6. Transgelin isoform partial sequences compared to the published sequence for transgelin 1 (Tagln). Spot numbering (S1–9) is shown in Fig. 5, and arrows indicate a statistically significant (p,0.05) increase (:) or decrease (;) in the isoform, caused by zinc deficiency. NS indicates no significance.
ing to cell structure, fatty acid metabolism and carbohydrate metabolism. Some proteins well known to be regulated by zinc were not detected by our 2-D gel proteomic technique using CBB staining, either because of their low level of expression or their lack of abundance in the proteins extracted from aorta. The following discussion will focus on the proteins at the core of the identified clusters, which are likely to be close to, or are themselves, targets of zinc deficiency. Since these studies use a zinc-deficient rat model, the relevance of the results to human vascular health and atherosclerosis has yet to be verified. Perhaps of most significance for vascular health, clusters of proteins with structural functions were identified in both MZD and AZD. Direct effects of Zn on elements of the cytoskeleton have been previously reported [16, 17]. Since several transgelins were significantly affected by AZD and MZD, it seems likely that they are key targets for changes in Zn status. Transgelin, otherwise known as sm22, was first isolated from chicken gizzard in 1987 [18] and was thought to be specific to smooth muscle. It has a calponin-like domain and may be involved in crosslinking actin filaments [19] and control of vascular smooth muscle contractility [20]. Isoforms of this family of 22 kDa proteins were separated on the basis of pI using 2-D gels, and four proteins were revealed by Western blotting [18]. These were named a, b, d, and g, and sm22a (transgelin 1) was subsequently identified in different species. In hypercholesterolemic ApoE-deficient mice, genetic ablation of sm22a resulted in increased atherosclerotic lesion area and a higher proportion of proliferating smooth muscle cellderived plaque cells [21]. © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
In our acute and marginal Zn deficiency studies we found a total of nine and three spots, respectively, giving an LC-MS/MS sequence identity of transgelin 1. All transgelins increased by AZD were highly positively correlated with each other. In MZD, all putative transgelins were decreased compared with MZA (Fig. 2). We used 2-D gel Western blotting as an independent technique to confirm the identity of putative transgelins affected by AZD. The antitransgelin 1 antibody used was raised against an internal peptide close to the N-terminus. Having established that all nine of the putative transgelin spots identified by LC-MS/MS analysis showed antibody binding, the spots from replicate gels were re-cut, digested with one of five different proteases (trypsin and endoproteinases Arg-C, Lys-C, Glu-C and Asp-N) and reanalysed by LC-MS/MS for final confirmation and to obtain more complete sequence information. All nine proteins digested individually with different proteases once again gave unambiguous identities of transgelin 1, despite the sequences being incomplete (Fig. 6). Transgelin 2 (sm22b) was also detected and significantly increased by AZD, but the sequence was clearly distinct from those of the nine transgelin one forms. Transgelin 3 (SN22), a protein expressed in superior frontal cortex neurones, was not detected in any aorta sample. We could find no evidence, using antibodies to phosphoserine and phosphothreonine, that any transgelin isoforms were phosphorylated (data not shown), which is consistent with observations elsewhere [22]. However, the phosphorylation of ser-181 of transgelin by PKC occurs in vitro [23]. We cannot discount other PTMs but some sm22a isoforms were of lower Mr than 22 kDa and so C- or N-terminally truncated forms of transgelin are a possibility. In that www.proteomics-journal.com
Proteomics 2008, 8, 2126–2135
regard, no peptide was detected within 47 amino acids of the transgelin 1 C-terminus for the isoforms with the lowest apparent molecular weight (Spots 7–9, Fig, 5), and given the considerable interisoform molecular weight discrepancy (up to about 4000 Da from the gel migration positions), a truncated C-terminus would seem the only explanation. Since actin-binding occurs within this region, these isoforms may be nonfunctional. AZD tended to decrease the lower Mr isoforms and increase the most highly expressed isoform with an Mr of 22 kDa. Changes in sm22a expression may impact on atherosclerotic lesion formation [21] and so this complex interrelationship of isoforms in zinc deficiency requires more detailed investigation. Classical PKC isoforms have two Zn fingers in their regulatory domain and the binding of phosphorylation activators such as diacylglycerol cause the release of Zn and a change in the domain conformation [24]. Due to the pivotal role of Zn in this protein, it has been described as a potential target for Zn deficiency. Indeed, PKCa levels decreased in Zn deficient 3T3 cells in culture, although phosphorylation of PKC targets was unaffected [25]. We found that PKCa levels in aorta were significantly decreased by AZD compared with AZA or APF. Also, ENH, a three-LIM domain PKC binding protein [26], was a principal member of the AZD protein network cluster increased by zinc deficiency. It may be significant that several structural proteins are targets of PKC phosphorylation. Thus Zn may be required for the attachment of PKC to the cytoskeleton [27]. In addition to some transgelins (Fig. 4B), ENH was also highly correlated with zyxin (Fig. 3), another key member of the protein network cluster, which was increased in AZD. Like paxillin, zyxin is a multi-LIM domain, Zn finger adaptor protein of focal adhesion complexes [28] and interacts with other LIM domain proteins. It has nuclear export signals and has an as yet unidentified role in the nucleus that does not involve DNA binding. Such LIM domain focal adhesion proteins may act as mechanosensors which transmit signals to the nucleus about changing extracellular environment [29, 30]. The significant increase in zyxin, a Zn-requiring protein, in response to Zn deficiency suggests that interference with the requirement of this protein for Zn is not a factor influencing its levels in aorta. It is possible that zyxin may be acting as a signal transducer for nuclear transcription factors in response to Zn deficiency-induced effects on the cell or cell structure. An association between Zn deficiency and lipid metabolism is well established [31, 32]. Indeed, some biological effects of Zn deficiency can be ameliorated by manipulating the dietary fat composition [33]. As a consequence of the complexity of variables influencing the interaction between Zn and lipid metabolism, apparently contradictory observations have been reported and the precise targets of Zn deficiency have proved elusive. When dietary lipid is in the form of saturated fat, Zn-deficient rats develop fatty livers and accumulate hepatic triglycerides [34]. In the present study, where rats were given corn oil rich in linoleic acid in the diet, © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Animal Proteomics
2133
AZD and MZD did not affect liver appearance, hepatic lipid content or, as shown with mice [35], fatty acid composition. Protein network targets identified by proteomics suggest that Zn deficiency inhibits the b-oxidation of fatty acids in aortas from male rats. This is consistent with the proposed effects of Zn deficiency on hepatic fatty acid metabolism interpreted from microarray and proteomic data [31, 36]. Levels of three of the four key enzymes involved in the b-oxidation of fatty acids, namely acyl CoA dehydrogenase, enoyl CoA hydratase and L-3-hydroxyacyl-CoA dehydrogenase, were significantly reduced by MZD or AZD and were highly clustered and correlated with themselves and with proteins of related function. They are all mitochondrial proteins and L-3-hydroxyacyl-CoA dehydrogenase was highly positively correlated with mitochondrial superoxide dismutase, suggesting a diminution of mitochondrial fatty acid oxidation in Zn deficiency. It is hypothesised that all of the apparently pleiotropic effects of Zn deficiency on hepatic fat metabolism could be explained by an unbalancing of gene transcription control via the peroxisome proliferator-activated receptor, PPARa, and sterol regulatory element binding protein-dependent pathways [36]. PPARs are Zn finger transcription factors and the activity of PPARg is reported to be sensitive to Zn deficiency in vascular endothelial cells [37]. Hepatic PPARg DNA binding was significantly reduced in acutely Zn deficient LDL receptor-null mice compared with that in Zn adequate animals [10]. PPARa promotes hepatic fatty acid oxidation and so its potential inhibition by Zn deficiency may also suppress levels of fatty acid catabolic enzymes. PPARg activation is anti-inflammatory in vascular tissue and so its inhibition by Zn deficiency may be considered proatherogenic. AZD and MZD suppressed the levels of enzymes regulating carbohydrate (CHO) metabolism. Both glycolytic enzymes of the cytoplasm, such as glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and transaldolase, and mitochondrial enzymes of the citric acid cycle, such as fumarate hydratase and malate dehydrogenase, were affected indicating that the impact of Zn deficiency transcended the subcellular enzyme compartmentalisation. Mitochondrial Mn superoxide dismutase (SOD2) was highly positively correlated with fumarate hydratase (Fig. 4D), perhaps indicating less generation of superoxide radicals by oxidative phosphorylation due to suppressed CHO metabolism. A generalised suppression of enzymes involved in CHO metabolism in response to Zn deficiency also occurs in rat epidermis [38]. Zn has insulinomimetic effects on glucose metabolism, and a mechanism involving Zn inhibition of protein tyrosine phosphatases, which dephosphorylate the intracellular signalling domain of the insulin receptor, has been proposed [39]. Insulin signalling is crucial for glucose transporter activity and glucose uptake, and a reduction of intracellular glucose due to Zn deficiency-induced insulin resistance, may suppress levels of CHO metabolism enzymes. In endothelial cells, insulin signalling also influences nitric oxide synthase 3 activity and the production of nitric oxide, which induces www.proteomics-journal.com
2134
J. H. Beattie et al.
vasodilation. Reports on the effects of Zn deficiency on blood pressure are, however, conflicting [40, 41]. It seems likely that there is some interdependency between elements of all target pathways. For example, a link between effects on CHO metabolism enzymes and cell structural proteins was observed through a strong correlation between cofilin, a protein involved in actin-binding and disassembly of actin filaments [42], and GAPDH. Both were significantly decreased by AZD and the high degree of correlation between them (Fig. 4C) was indicative of a high level of interdependency. It may be relevant that glycolytic enzymes are physically located on the cytoskeleton [43]. Triose phosphate isomerase, which like GAPDH utilises glyceraldehyde-3-phosphate as a substrate, can associate with cofilin, and cofilin may feed glycolytic fuel for Na, K-ATPase activity [44]. Zinc deficiency-induced insulin resistance in vascular tissue, which suppresses CHO metabolism and which may lead to reduced nitric oxide production and hypertension [40], may promote the expression of mechanotransducing proteins such as zyxin [30]. In conclusion, Zn deficiency has profound effects on aorta cell structure, fatty acid metabolism and carbohydrate metabolism. Key structural proteins included trangelins and zyxin. Interdependency between some proteins from different functional categories is likely considering the strong correlations observed in their levels within aorta. Structural effects suggest changes in smooth muscle contractility and deficiency-induced suppression of carbohydrate and fatty acid metabolism are indicative of insulin resistance and perturbation of key Zn-dependent transcription factors such as the PPAR proteins and classical PKC isoforms. The Zn deficiency-related changes revealed in this work can best be interpreted as being disadvantageous for maintaining vascular health. They are consistent with a protective role for zinc in atherogenesis and may form the basis for future studies investigating the relevance of zinc status in human atherosclerosis.
All RRI-based authors acknowledge funding from the Scottish Executive Environment and Rural Affairs Department. I. S. Kwun and Y. E. Cho were supported by a grant from the Ministry of Health and Welfare, Republic of Korea (03-PJ1-PG3-220000044). The authors have declared no conflict of interest.
5
References
Proteomics 2008, 8, 2126–2135 [4] Oteiza, P. I., Mackenzie, G. G., Zinc, oxidant-triggered cell signaling, and human health. Mol. Aspects Med. 2005, 26, 245–255. [5] Mocchegiani, E., Costarelli, L., Giacconi, R., Cipriano, C. et al., Nutrient-gene interaction in ageing and successful ageing. A single nutrient (zinc) and some target genes related to inflammatory/immune response. Mech. Ageing Dev. 2006, 127, 517–525. [6] Chung, M. J., Walker, P. A., Brown, R. W., Hogstrand, C., Zincmediated gene expression offers protection against H2O2induced cytotoxicity. Toxicol. Appl. Pharmacol. 2005, 205, 225–236. [7] Zhou, Z., Wang, L., Song, Z., Saari, J. T. et al., Zinc supplementation prevents alcoholic liver injury in mice through attenuation of oxidative stress. Am. J. Pathol. 2005, 166, 1681–1690. [8] Beattie, J. H., Kwun, I. S., Is zinc deficiency a risk factor for atherosclerosis? Br. J. Nutr. 2004, 91, 177–181. [9] Meerarani, P., Ramadass, P., Toborek, M., Bauer, H. C. et al., Zinc protects against apoptosis of endothelial cells induced by linoleic acid and tumor necrosis factor alpha. Am. J. Clin. Nutr. 2000, 71, 81–87. [10] Reiterer, G., MacDonald, R., Browning, J. D., Morrow, J. et al., Zinc deficiency increases plasma lipids and atherosclerotic markers in LDL-receptor-deficient mice. J. Nutr. 2005, 135, 2114–2118. [11] Singh, R. B., Niaz, M. A., Rastogi, S. S., Bajaj, S. et al., Current zinc intake and risk of diabetes and coronary artery disease and factors associated with insulin resistance in rural and urban populations of North India. J. Am. Coll. Nutr. 1998, 17, 564–570. [12] National Research Council, Guide for the Care and Use of Laboratory Animals, National Academy Press, Washington D.C. 1996. [13] Bessey, O. A., Lowry, O. H., Brock, M. J., A method for the rapid determination of alkaline phosphatase with 5 cubic millimeters of serum. J. Biol. Chem. 1946, 164, 321–329. [14] Lowry, O. H., Rosebrough, N. J., Farr, A. L., Randall, R. J., Protein measurement with the folin phenol reagent. J. Biol. Chem. 1951, 193, 265–275. [15] de Roos, B., Duivenvoorden, I., Rucklidge, G., Reid, M. et al., Response of apolipoprotein E*3-Leiden transgenic mice to dietary fatty acids: Combining liver proteomics with physiological data. FASEB J. 2005, 19, 813–815. [16] Gaskin, F., Kress, Y., Zinc ion-induced assembly of tubulin. J. Biol. Chem. 1977, 252, 6918–6924. [17] Mills, J. W., Zhou, J. H., Cardoza, L., Ferm, V. H., Zinc alters actin filaments in Madin-Darby canine kidney cells.Toxicol. Appl. Pharmacol. 1992, 116, 92–100. [18] Lees-Miller, J. P., Heeley, D. H., Smillie, L. B., Kay, C. M., Isolation and characterization of an abundant and novel 22 kDa protein (SM22) from chicken gizzard smooth muscle. J. Biol. Chem. 1987, 262, 2988–2993.
[2] Prasad, A. S., Zinc deficiency. Br. Med. J., 2003, 326, 409–410.
[19] Shapland, C., Hsuan, J. J., Totty, N. F., Lawson, D., Purification and properties of transgelin: A transformation and shape change sensitive actin-gelling protein. J. Cell. Biol. 1993, 121, 1065–1073.
[3] Hambidge, M., Biomarkers of trace mineral intake and status. J. Nutr. 2003, 133, 948S–955S.
[20] Zeidan, A., Sward, K., Nordstrom, I., Ekblad, E. et al., Ablation of SM22alpha decreases contractility and actin contents
[1] WHO, Quantifying selected major ri sks to health. The World Health Report 2002, WHO, Geneva 2002.
© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.proteomics-journal.com
Proteomics 2008, 8, 2126–2135 of mouse vascular smooth muscle. FEBS Lett. 2004, 562, 141–146. [21] Feil, S., Hofmann, F., Feil, R., SM22a modulates vascular smooth muscle cell phenotype during atherogenesis. Circ. Res. 2004, 94, 863–865. [22] Gimona, M., Sparrow, M. P., Strasser, P., Herzog, M., Small, J. V., Calponin and SM 22 isoforms in avian and mammalian smooth muscle. Absence of phosphorylation in vivo. Eur. J. Biochem. 1992, 205, 1067–1075. [23] Fu, Y., Liu, H. W., Forsythe, S. M., Kogut, P. et al., Mutagenesis analysis of human SM22: Characterization of actin binding. J. Appl. Physiol. 2000, 89, 1985–1990. [24] Korichneva, I., Hoyos, B., Chua, R., Levi, E., Hammerling, U., Zinc release from protein kinase C as the common event during activation by lipid second messenger or reactive oxygen. J. Biol. Chem. 2002, 277, 44327–44331. [25] Chou, S. S., Clegg, M. S., Momma, T. Y., Niles, B. J. et al., Alterations in protein kinase C activity and processing during zinc-deficiency-induced cell death. Biochem. J. 2004, 383, 63–71. [26] Nakagawa, N., Hoshijima, M., Oyasu, M., Saito, N. et al., ENH, containing PDZ and LIM domains, heart/skeletal muscle-specific protein, associates with cytoskeletal proteins through the PDZ domain. Biochem. Biophys. Res. Commun. 2000, 272, 505–512. [27] Forbes, I. J., Zalewski, P. D., Giannakis, C., Betts, W. H., Zinc induces specific association of PKC with membrane cytoskeleton. Biochem. Int. 1990, 22, 741–748. [28] Wang, Y., Gilmore, T. D., Zyxin and paxillin proteins: Focal adhesion plaque LIM domain proteins go nuclear. Biochim. Biophys. Acta. 2003, 1593, 115–120. [29] Kadrmas, J. L., Beckerle, M. C., The LIM domain: From the cytoskeleton to the nucleus. Nat. Rev. Mol. Cell. Biol. 2004, 5, 920–931. [30] Cattaruzza, M., Lattrich, C., Hecker, M., Focal adhesion protein zyxin is a mechanosensitive modulator of gene expression in vascular smooth muscle cells. Hypertension 2004, 43, 726–730. [31] Daniel, H., tom Dieck, H., Nutrient-gene interactions: A single nutrient and hundreds of target genes. Biol. Chem. 2004, 385, 571–583. [32] Cunnane, S. C., Role of zinc in lipid and fatty acid metabolism and in membranes. Prog. Food Nutr. Sci. 1988, 12, 151– 188.
© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Animal Proteomics
2135
[33] Kettler, S. I., Eder, K., Kettler, A., Kirchgessner, M., Zinc deficiency and the activities of lipoprotein lipase in plasma and tissues of rats force-fed diets with coconut oil or fish oil. J. Nutr. Biochem. 2000, 11, 132–138. [34] Eder, K., Kirchgessner, M., Zinc deficiency and activities of lipogenic and glycolytic enzymes in liver of rats fed coconut oil or linseed oil. Lipids 1995, 30, 63–69. [35] Beattie, J. H., Gordon, M.-J., Reid, M. D., Rucklidge, G. J. et al., Hepatic responses to dietary stress in zinc and metallothionein-deficient mice. Exp. Biol. Med. 2006, 231, 1542– 1547. [36] tom Dieck, H., Doring, F., Fuchs, D., Roth, H. P., Daniel, H., Transcriptome and proteome analysis identifies the pathways that increase hepatic lipid accumulation in zinc-deficient rats. J. Nutr. 2005, 135, 199–205. [37] Meerarani, P., Reiterer, G., Toborek, M., Hennig, B., Zinc modulates PPARgamma signaling and activation of porcine endothelial cells. J. Nutr. 2003, 133, 3058–3064. [38] Im, M. J., Hsu, J. M., Hoopes, J. E., Enzyme activities in the epidermis of zinc-deficient rats. J. Nutr. 1975, 105, 1391– 1394. [39] Haase, H., Maret, W., Intracellular zinc fluctuations modulate protein tyrosine phosphatase activity in insulin/insulin-like growth factor-1 signaling. Exp. Cell Res. 2003, 291, 289–298. [40] Tomat, A. L., Weisstaub, A. R., Jauregui, A., Pineiro, A. et al., Moderate zinc deficiency influences arterial blood pressure and vascular nitric oxide pathway in growing rats. Pediatr. Res. 2005, 58, 672–676. [41] Sato, M., Kurihara, N., Moridaira, K., Sakamoto, H. et al., Dietary Zn deficiency does not influence systemic blood pressure and vascular nitric oxide signaling in normotensive rats. Biol. Trace Elem. Res. 2003, 91, 157–172. [42] Huang, T. Y., Der Mardirossian, C., Bokoch, G. M., Cofilin phosphatases and regulation of actin dynamics. Curr. Opin. Cell. Biol. 2006, 18, 26–31. [43] Beitner, R., Control of glycolytic enzymes through binding to cell structures and by glucose-1,6-bisphosphate under different conditions. The role of Ca21 and calmodulin. Int. J. Biochem. 1993, 25, 297–305. [44] Jung, J., Yoon, T., Choi, E. C., Lee, K., Interaction of cofilin with triose-phosphate isomerase contributes glycolytic fuel for Na,K-ATPase via Rho-mediated signaling pathway. J. Biol. Chem. 2002, 277, 48931–48937.
www.proteomics-journal.com
