Unless indicated, all reagents used for biochemical methods were purchased from Sigma-Aldrich (St. Louis, MO, USA). Materials and reagents for SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis) were from Bio-Rad (Bio-Rad Laboratories, Hercules, CA, USA). The iTRAQ reagent multi-plex kit, containing the iTRAQ reagents, was bought commercially (Applied Biosystems, Foster City, CA, USA).

Experimental procedures, including the killing of animals, were in accordance with the International Guiding Principles for Animal Research (WHO) and were approved by the local Institutional Animal Care & Use Committee (NTU-IACUC). One-year-old male Wistar Kyoto rats (~250 g) were obtained from the laboratory animal centre (National University of Singapore) and randomly assigned to control and tianma-treated groups (10 each). According to previous reports and our own recent pilot studies, the average daily dose of tianma per rat was 2.5g/kg body weight [12, 18, 20-22]. They were fed a normal chow (Funabashi SP, Japan) and tap water was given freely. Room temperature (RT) was kept at 21 ± 2 ºC, with 60 % humidity, and a 12 h light/dark cycle. Tianma-feeding was done orally (intragastric administration) with a blunt needle syringe by dispensing the tianma solution for the period of three months. Control rats were treated with the same volume of the solvent only. Animals were sacrificed by CO₂ asphyxiation. All efforts were made to minimize animal suffering and to reduce the number of animals used.

The rhizome of Gastrodia elata (tianma), grown under standardized conditions [23], was collected from Zhaotong City, China and provided by Dr. Jun Zhou (Kunming Institute of Botany, Chinese Academy of Science, Yunnan, P.R. China). The species was identified and chemically analyzed as reported previously [12, 24]. A voucher specimen (0249742) was deposited in the herbarium of the Kunming Institute of Botany (Chinese Academy of Science, Yunnan, P.R. China). In this study, tianma was prepared according to previous reports [12, 16, 19, 20, 22, 25-27]. Whole dried tubers of the tianma were hammered into smaller pieces and subsequently ground to fine powder. 7.5 g of tianma powder was mixed with 100 ml sterilized Milli-Q water and boiled for 1 h at 100 ºC. The solution was centrifuged at 5000xg for 10 min at RT. The supernatant was filtered with a Whatman filter paper-1 (GE Healthcare, Chalfont St Giles, UK), yielding approximately 85 ml. The tianma solution was concentrated at 60 ºC under vacuum and the final volume was reduced to 10 ml for further applications.

Briefly, the thoracic aorta was excised from the rat gently after dissection, immediately immersed into cold physiologic saline solution (PSS, 119 mM NaCl, 25 mM NaHCO₃, 4.7 mM KCl, 1.18 mM KH₂PO₄, 1.17 mM MgSO₄, 2.5 mM CaCl₂, 0.026 mM EDTA and 5.5 mM glucose; the solution was gassed with 5 % CO₂ in O₂ to maintain pH at 7.4) and kept on ice for further careful cleaning with scissors and forceps under the microscope to remove fat and connective tissues as well as blood clots. The aorta was then cut into eight segments of 2 mm rings and kept in 4 °C-cold PSS. The integrity of the endothelium was tested by constricting with 10 µM PE and after steady contraction obtained, relaxed with 10 µM ACh [28, 29]. The myograph (Danish Myo Technology, ADInstruments S.E.Asia, Subang Jaya, Selangor, Malaysia) chambers were filled with ‘Krebs’ solution and the aorta rings mounted onto the hooks inside the chambers for isometric tension recording [30]. The tension was adjusted to 1 g as the baseline. The tissues were then washed twice with 124 mM K⁺ PSS to ensure the viability. For experimental set I, the segments were initially loaded to an optimum stretch, which was previously determined by using high-K⁺ physiological solution (80 mM) as the contracting agent after applying different passive tensions. The initial stretch and the length of the segments (2 mm) were consistently maintained across all arterial rings of either group. The aorta rings were pre-contracted with 80 mM K⁺ PSS until they reached a plateau. Thereafter, an ACh-mediated relaxation was performed as a concentration response curve (CRC) by adding increasing concentrations of ACh in half-log concentration increments, i.e. 10^-8 M, 3x10^-8 M, 10^-7 M, 3x10^-7 M, 10^-6 M, 3x10^-6 M, 10^-5 M, 3x10^-5 M, and 10^-4 M. After the ACh CRC, the tissues were end-relaxed by applying 5x10^-5 M sodium nitroprusside (SNP). For experimental set II, the aorta rings were pre-contracted with 10^-6 M PE and the ACh CRC applied. Upon reaching the PE plateau, the ACh CRC was performed by half-log (10^-8 M to 10^-4 M) applications of ACh. Again, after the ACh CRC, the tissues were further relaxed by applying 5x10^-5 M SNP. For data analysis the weight of the dried aorta rings were measured for appropriate normalization. Contraction responses to PE were calculated as percent of its maximal contraction. Relaxant responses to ACh were calculated as percent inhibition of the K⁺-/PE-induced contraction. Data analysis was done using GraphPad Prism for Windows (Graph-Pad Software, San Diego, CA, USA). Data from several (8) vascular rings of the same rat were averaged and presented as the datum for 1 rat, with the n value (= 5) representing the number of rats for each group (control or tianma-treated). Differences were considered statistically significant at P < 0.05.

For this purpose, PE was initially added at a final concentration of 10^-7 M to the bath to contract the ring, and force was allowed to stabilize for 5 min. ACh at a final concentration of 10^-5 M was then added to the pre-contracted rings for 5 min to test for endothelial integrity and aortal ring viability. If a ring failed to contract in response to PE or failed to relax in response to ACh, it was replaced with another aortic ring from the same rat. After the initial test for vessel viability and endothelial integrity, the ring was washed 3 times with PSS, allowed to equilibrate, and then re-washed with fresh PSS at 10 min intervals until the measured active force stabilized at 0 g. The maximum contraction achievable by the ring was then determined by filling the bath with 80 mM K⁺ and adding increasing concentrations of PE up to a final concentration of 10^-4 M (a PE CRC with half-log concentrations was performed (i.e. 10^-8 M, 3x10^-8 M, 10^-7 M, 3x10^-7 M, 10^-6 M, 3x10^-6 M, 10^-5 M, 3x10^-5 M, and 10^-4 M)). Maximal contractile force generated in response to the combination of 80 mM K⁺ and PE was normalized to the wet weight of the aortic ring (determined at the end of the experiment). After determining the maximum contraction of the aortic rings, the vessels were allowed to stabilize and washed with PSS every 10 min until the measured active force returned to 0 g.

For the aorta-tissue-specific proteome analyses, aorta tissues were isolated from tianma-treated and control rats. Briefly, the thoracic aorta was excised from the rats gently after dissection, immediately immersed into liquid nitrogen, and then powdered using a mortar and pestle. Upon the addition of lysis buffer (2 % SDS, 0.5 M Triethyl ammonium bicarbonate buffer (TEAB), 1 Complete™ protease inhibitor cocktail tablet (Roche, Mannheim, Germany) and 1 PhosSTOP phosphatase inhibitor cocktail tablet (Roche)), the samples were vortexed for 1 min and incubated on ice for an additional 45 min prior to homogenisation (sonication parameters: amplitude, 23 %; pulse: 5 s/ 5 s for 5 min) using a Vibra Cell high intensity ultrasonic processor (Jencon Scientific Ltd, Leighton Buzzard, Bedfordshire, UK). After centrifugation (20,000 x g / 4 °C / 30 min), supernatant was collected and stored at -80 °C until further use. The protein concentration was quantified by a ‘2-D Quant’ kit (Amersham, Piscataway, NJ, USA) according to the manufacturer’s protocol.

A detailed description of the 2D-LC-MS/MS-iTRAQ procedures [19, 31-33], including post-proteomic data verification by SDS-PAGE - western blot analysis [34-36], can be found in the supplementary content document as described previously.

Anti-Anxa2 (Annexin-A2, 1:4000, polyclonal; Abcam, Cambridge, UK), anti-Des (Desmin, 1:1000, rabbit polyclonal; Cell Signaling Technology Inc., Danvers, MA, USA), anti-Gapdh (1:1000, mouse monoclonal; Abcam), anti-Gstm2 (Glutathione S-transferase mu 2, 1:1000, goat polyclonal; Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA), anti-Serpinh1 (Serpin H1, 1:1000, rabbit polyclonal; Abnova, Taipei, Taiwan), anti-Vcl (Vinculin, 1:4000, monoclonal; Abcam).

The mechanical responses of the vessels were measured as force and expressed as active wall tension, which is the increase in measured force divided by twice the segment length [30]. By using a computer program (GraphPad, Institute for Scientific Information, San Diego, CA, USA), the CRCs were fitted to the classical Hill equation, as described earlier [29]. The results are expressed as mean ± SD. Differences between means were analyzed using either one-way analysis of variance (ANOVA) followed by a Bonferroni t-test, Student’s t-test or paired t-test (SPSS (Statistical Products and Service Solutions) for Windows Version 19 was used to perform ANOVA, optionally followed by Fisher’s Protected Least Significant Difference (PLSD) post hoc tests, when warranted). For the western blot analyses the Student’s t-test was applied accordingly. For the iTRAQ analysis ProteinPilot Software 3.0 was used as described above. To be considered statistically significant, we required a probability value to be at least < 0.05 (95 % confidence limit, *P < 0.05).

In order to investigate the vasodilatory effects of tianma on blood vessels, we used myography to perform contractility/relaxation reactivities by using isolated thoracic aortic rings from rats treated with/without tianma (~2.5g/kg/day for three months). As shown in Figs. (2 and 3), in the endothelium-intact thoracic aortic rings, ACh-induced smooth muscle relaxation was obviously enhanced by tianma treatment in both K⁺ (80 mM)- (Fig. 2) and PE- (10^-6 M) (Fig. 3) induced pre-contraction. It was noticed that in PE-induced pre-contraction,tianma treatment enhanced ACh-induced endothelium-dependent smooth muscle relaxation up to about 20 %, whereas in K⁺-induced pre-contraction, tianma treatment only enhanced ACh-induced endothelium-dependent smooth muscle relaxation to about 10 %. The different mechanisms of K⁺- and PE-induced smooth muscle contraction could account for the different enhancing relaxatory effects by tianma.

Fig. (2). Tianma enhanced ACh-induced vasorelaxation in endothelium-intact rat thoracic arterial rings (I). Analysis of ACh induced relaxations in KCl (80 mM) pre-contracted isolated endothelium-intact arterial rings from control and tianma-treated rats. One-year-old rats were treated with or without tianma for three months at ~2.5g/kg/day and the thoracic aortas were isolated. The endothelium-intact arterial rings were first pre-contracted by 80 mM of K+. Increasing concentrations of ACh were added and then the percentage of ACh-induced relaxation to K+ contraction was measured as described in the materials and methods. Data points represent means ± SD of measurements in 8 arterial rings from 5 rats of each group (*P < 0.05 compared with the control group).

Fig. (3). Tianma enhanced ACh-induced vasorelaxation in endothelium-intact rat thoracic arterial rings (II). Analysis of ACh induced relaxations in PE (10-6 M) pre-contracted isolated endothelium-intact arterial rings from control and tianma-treated rats. One-year-old rats were treated with or without tianma for three months at ~2.5g/kg/day and the thoracic aortas were isolated. The endothelium-intact arterial rings were first pre-contracted by PE (10-6 M). Increasing concentrations of ACh were added and then the percentage of ACh-induced relaxation to PE (10-6 M) contraction was measured as described in the materials and methods. Data points represent means ± SD of measurements in 8 arterial rings from 5 rats of each group (*P < 0.05 compared with the control group).

Tianma almost doubled the contractile forces of the thoracic aortic rings in response to PE and promoted the arrival of the plateau (maximal contractile force) (Fig. 4). The concentration of PE required for reaching the plateau in the tianma treated group is about 10^-6 M, which is about 10 times less than in the control group (10^-5 M). Thus, the potential vasodilatory effect of tianma on smooth muscles could eventually enhance the maximal contractile response to PE, which could be interpreted as an effect that may enhance blood vessel elasticity.

Fig. (4). Tianma increased the contractile force and elasticity of the thoracic aortic rings. Comparison of maximum contractile force developed in response to 80 mM K+ and increasing concentrations (CRCs) of PE in aortic rings from tianma-treated rats and controls. The isolated endothelium-intact arterial rings were studied with the endothelium left intact and data are presented as tension (g) per milligram tissue wet weight (g/mg wet weight of aorta). Values represent an average of 2x8 individual experiments. An asterisk indicates a significant difference from the control value (*P < 0.05 compared to the control group).

iTRAQ analysis was performed on the purified protein extracts from rat thoracic aortas with or without tianma treatment to understand the mechanism of tianma-mediated relaxation in vascular smooth muscle strips on the cellular and molecular level. Four samples set into two batches (each batch (B-I, B-II) contained five tianma-treated samples and five (untreated) control samples) were subjected to iTRAQ analysis in order to ensure the results were statistically meaningful. Identified proteins were also visualized in a virtual two-dimensional protein gel, JVirGel, and categorized by their calculated isoelectric points and molecular weights [37]. The protein spots were well separated without aggregation, indicating a well-qualified whole cell proteomic pattern (Fig. 5).

Fig. (5). Simulated 2D gel presentation of rat thoracic aorta-derived quantified proteins. MW and pI values generated from JVirGel (http://www.jvirgel.de/). The proteins identified by LC-MS/MS were uploaded onto JVirGel, online software used to create a 2D gel image. This image confirmed that the tissue lysis performed was adequate and the entire proteome within cells was extracted.

We identified a total of 1298 proteins through iTRAQ analysis out of which 1166 proteins were quantified (with a strict cutoff of unused ProtScore ≥ 2 as the qualification criteria, which corresponds to a peptide confidence level of 99 % and an applied FDR (false discovery rate) of 0.33% (<1.0%)). Comparing the expression levels between tianma-treated and untreated samples, 54 proteins showed an altered expression level (30 proteins were down-regulated and 24 proteins were up-regulated, (Supplementary data: Table 1)). We used Panther, UniProt, and NCBI online databases for classifying the altered proteins based on their role and localization. Approximately 33.30 % of the altered proteins were cytoskeletal, 26 % were catalytic and 10 % were extracellular matrix (ECM) proteins (Fig. 6).

Fig. (6). Pie chart depicting the iTRAQ identified proteins characterized by the molecular function GO category. Proteins identified and quantified by iTRAQ, were classified in terms of their role in biological processes (A) and molecular functions (B).

We focused our interest on the structural protein subgroup: In tianma treated aortic rings, actin (Acta2) decreased to about half the level of the control (with p<0.02) while other structural proteins, like desmin (Des), vinculin (Vcl), PDZ and LIM domain protein 1 (Pdlim1), tubulin beta 6 (Tubb6), alpha-parvin (Parva) and microtubule-associated protein-4 (Map4) were also down-regulated by tianma treatment upto 30 % - 60 %, whereas Myo1c and lamin-B1 (Lmnb1) were up-regulated. Most ECM/cell-surface proteins were also up-regulated. Elastin (Eln) and proline arginine-rich end leucine-rich repeat protein (Prelp) expressions were increased to more than twice of the control while fibulin-5 (Fbln5), biglycan (Bgn) and fibromodulin (Fmod) expressions were also increased to more than 1.3 times of the control. In contrast, fibronectin (Fn1) and periostin (Postn) were down-regulated to about half of the control level (Fig. 7).

Fig. (7). Effect of tianma on the expression levels of cytoskeletal and ECM proteins. Protein expression levels were quantitatively analyzed ex-vivo by using iTRAQ after tianma treatment for three months as described in the experimental procedures. The Y-axis shows the iTRAQ ratios of proteins between tianma-treated and untreated controls. Values above 1.2 indicate up-regulation and below 0.83 indicate down-regulation of the proteins. Proteins shown (from left to right): black bars: down-regulated: Acta2, Actb, Tubb6, Pdlim1, Des, Vcl, Parva, Cryab, Map4, Coro1c, Eml2, up-regulated: Myo1c, Pxn, Lmnb1, Flnc. White bars: up-regulated: Eln, Fbln5, Prelp, Bgn, Fmod; down-regulated: Fn1, Postn.

Following the database search and classification of proteins, western blots were performed on randomly selected proteins to further verify the iTRAQ results (Fig. 8). While Gstm2, Des and Vcl were down-regulated by tianma treatment, the smooth muscle cell-specific serine (or cysteine) peptidase inhibitor Serpinh1 (also known as 47 kDa heat shock protein; beta enolase repressor factor 1; collagen binding protein 1) and Anxa2 were up-regulated. Notably, the western blot images correlated very well and thus confirmed the iTRAQ data obtained.

Fig. (8). Western blot validation of iTRAQ results. Randomly selected proteins significantly regulated in tianma-stimulated rat aortas compared with controls. Rats were treated, proteins extracted and western blot performed as described in materials and methods. A: Serpinh1, Anxa2 and Gstm2 expressions were up-regulated, while Vcl and Des expressions were down-regulated. Gapdh, that did not show any change in the iTRAQ data set, was used as reference protein. B: Quantitative analyses of the western blots shown in A. Western blots were performed at least four times for statistical quantification and analyses (n=4). Values (= relative protein expression) represent the ratio of densitometric scores for the respective western blot products and statistical error was indicated as mean ± SD (*P < 0.05, compared with controls) using the Gapdh bands as reference. C: The histogram indicates a similar close relationship between iTRAQ and western blot expression ratios. Tianma-stimulated and control aorta iTRAQ expression ratios from selected proteins were consistent with the western blot results and thus validated a strong agreement in the expression data.

The STRING (Search Tool for the Retrieval of Interacting Genes) analyses provided us an essential systems-level understanding of cellular events, including both physical and functional interactions, in a tianma-treated aorta [38]. Our current study revealed the functional link among ECM and cytoskeletal proteins such as Bgn, Fn1, Vcl, Acta2, Actb, Tubb6, Des, Myo1c or Pxn and their potential link to other metabolically modulated proteins such as Suclg1, Sdha and Slc25a4/5 (Fig. 9).

Fig. (9). STRING-9.0 analysis (rattus norvegicus at http://string-db.org/; default mode) of tianma-modulated proteins in rat aorta. Different line colors represent the types of evidence for the association. Network display: Nodes are either colored (if they are directly linked to the input as in supplemental table 1) or white (nodes of a higher iteration). Edges, i.e. predicted functional links, consist of multiple lines: one color for each type of evidence.

Further bio-computational network analysis of the proteins identified in tianma-stimulated aortic tissue using the Ingenuity Pathways Analysis (IPA) offered us additional valuable clues about the complex interactive link of various identified proteins within their interactive protein networks obtained from other cellular metabolic information (Fig. 10).

Fig. (10). IPA network analysis of proteins identified by iTRAQ in the aorta and their potential functional link to other proteins in human cells (in various tissues, under various conditions). The six major IPA-provided networks were analyzed based on the data of proteins expressed in rat aorta. The network (i) considered 14 proteins (Actin, BCAR1, CANX, CASP3, CAV1, CRYAB, GAPDH, GLMC, GCLM, GSTM1, GSTM5, GSTP1, ILK, ITGB1, KCNJ11, MYO1C, NFE2L2, PARVA, PIK3R1, PPP1CA, PTK2, PXN, SERPINH1, SLC25A4, SUCLG1, TNF, TNS1, UQCRC1, VCL) involved in necrosis/cell death, cell morphology. Network (ii) considered proteins (26S Proteasome, TUBB6) involved in the cardiovascular system development and function. Network (iii) considered proteins (RHOA, ROCK1) involved in cell morphology, cellular development, nervous system development and function. Network (iv) considered proteins (FABP4, LIPE) in-volved genetic disorders and lipid metabolic diseases. Network (v) considered proteins (NRF1, SDHA) involved in lipid metabolism, small molecule biochemistry, connective tissue development and function. Network (vi) considered proteins (ANXA2, IGF1R, INSR) involved in organ morphology, cell-to-cell signaling and interaction and embryonic development. The solid lines show direct proteinprotein interaction, while dotted lines show indirect interactions between the proteins. Proteins identified by iTRAQ, as shown in Supplemental Table 1, are pre-sented in grey-color circles, while additional interaction proteins were added by the IPA network analysis in white color.