Mineralocorticoid receptor expression in human venous smooth muscle cells: a potential role for aldosterone signaling in vein graft arterialization

Richard Bafford, Xin Xin Sui, Min Park, Takuya Miyahara, Brenna G. Newfell, Iris Z. Jaffe, Jose R. Romero, Gail K. Adler, Gordon H. Williams, Raouf A. Khalil, Michael S. Conte


Experimental studies have suggested a role for the local renin-angiotensin-aldosterone system in the response to vascular injury. Clinical data support that aldosterone, via activation of the mineralocorticoid receptor (MR), is an important mediator of vascular damage in humans with cardiovascular disease. In mineralocorticoid-sensitive target tissue, aldosterone specificity for MR is conferred enzymatically by the cortisol-inactivating enzyme 11β-hydroxysteroid-dehydrogenase-2 (11βHSD2). However, the role of MR/aldosterone signaling in the venous system has not been explored. We hypothesized that MR expression and signaling in venous smooth muscle cells contributes to the arterialization of venous conduits and the injury response in vein bypass grafts. MR immunostaining was observed in all samples of excised human peripheral vein graft lesions and in explanted experimental rabbit carotid interposition vein grafts, with minimal staining in control greater saphenous vein. We also found upregulated transcriptional expression of both MR and 11βHSD2 in human vein graft and rabbit vein graft, whereas control greater saphenous vein expressed minimal MR and no detectable 11βHSD2. The expression of MR and 11βHSD2 was confirmed in cultured human saphenous venous smooth muscle cells (hSVSMCs). Using an adenovirus containing a MR response element-driven reporter gene, we demonstrate that MR in hSVSMCs is capable of mediating aldosterone-induced gene activation. The functional significance for MR signaling in hSVSMCs is supported by the aldosterone-induced increase of angiotensin II type-1 receptor gene expression that was inhibited by the MR antagonist spironolactone. The upregulation of MR and 11βHSD2 suggests that aldosterone-mediated tissue injury plays a role in vein graft arterialization.

  • renin-angiotensin system
  • peripheral vascular disease
  • 11β-hydroxysteroid-dehydrogenase-2
  • vascular injury response

peripheral arterial disease and coronary heart disease are associated with significant morbidity and mortality with an estimated 8 to 16 million affected individuals in the United States (25a, 37). Recent data suggest that there are ∼80,000 lower extremity bypass grafts and 450,000 coronary artery bypass grafts performed annually in the U.S. (25a, 34, 42). Despite the increasing use of percutaneous methods to treat peripheral arterial disease and coronary heart disease, surgical bypass with autogenous vein remains the most durable and effective therapy. However, up to 30–50% of peripheral vein grafts will fail within five years, leading to significant morbidity, reinterventions, limb loss, and diminished quality of life (3). Neointimal hyperplasia within the graft is the mechanism presumed responsible for most graft failures that occur during this period. After a vein graft is implanted, an injury/healing response occurs that involves a complex and dynamic interplay of processes including hemodynamic adaptation, cellular proliferation, inflammation, extracellular matrix deposition, oxidative stress, and fibrosis (31, 38). This adaptive remodeling is regulated by a host of cytokines and growth factors that act in an autocrine and paracrine manner. The regulation of this vein arterialization response is incompletely understood and of central importance to improve clinical outcomes. To date there are no therapeutic options that can substantially reduce neointimal hyperplasia and improve vein bypass graft patency.

The mineralocorticoid aldosterone is emerging as a significant regulator of pathophysiological changes in the cardiovascular system (7, 16). Aldosterone is a steroid hormone that acts via the mineralocorticoid receptor (MR), a member of the nuclear hormone receptor family. MR binds aldosterone and cortisol with equal affinity, but the presence of the cortisol-inactivating enzyme 11β-hydroxysteroid-dehydrogenase-2 (11βHSD2) confers aldosterone specificity for MR in tissues where they are coexpressed (12, 13). Clinical studies have shown that MR antagonism significantly decreases the incidence of cardiovascular ischemia and mortality in cardiovascular patients out of proportion to modest changes in blood pressure (7, 39, 40), supporting a direct role for aldosterone and vascular MR in human vascular disease. There is growing evidence that aldosterone can act directly on MR in the vascular cells to induce inflammatory and oxidative damage and to promote vascular hyperplasia and fibrosis (8, 20, 21, 41, 46, 47, 49). All the key elements necessary for aldosterone/MR signaling are expressed within the arterial wall (9, 18, 27); however, the role of aldosterone and MR signaling in venous tissue and in vein graft disease has not been explored. We hypothesize that aldosterone-specific MR activation occurs in the setting of vein graft arterialization and contributes to the injury response in venous bypass grafts.


Cell culture.

Primary cultures of human greater saphenous vein (GSV) smooth muscle cells (hSVSMCs) were obtained by medial explant from discarded operative specimens (48) according to a protocol approved by the Institutional Review Board of Brigham and Women's Hospital. Cells were maintained in Dulbecco's modified Eagle's medium (DMEM; BioWhittaker, Walkersville, MD) plus 10% fetal bovine serum (BioWhittaker) and used between passages 2–5.

Human vein graft lesions.

Specimens of normal GSV (N = 6) and excised lesions of diseased lower extremity vein grafts (human vein graft; N = 17, 84–1,330 days post-bypass) were obtained under an Institutional Review Board-approved protocol of Brigham and Women's Hospital. Tissue for immunohistochemistry was fixed in 10% formalin. Total RNA was isolated from freshly harvested vein grafts and control tissue using TRIzol (Invitrogen Life Technologies, Carlsbad, CA) according to manufacturer's protocol.

Rabbit vein graft model.

All animal experiments were performed in accordance with the Guide for the Care and Use of Laboratory Animals (1), under a protocol approved by the Harvard Medical Area Standing Committee on animals. New Zealand White rabbits (n = 15), weighing 3 to 3.5 kg and maintained on a normal chow diet, were used in all experiments. An anastomotic cuff technique was used to create a carotid artery interposition vein graft from the ipsilateral jugular vein (22). Tissue destined for immunohistochemistry was perfusion fixed in 10% formalin and processed as previously described (48). Total RNA from freshly harvested vein grafts and control tissue was isolated with TRIzol (Invitrogen Life Technologies) according to the manufacturer's protocol.

Analysis of gene expression by RT-PCR.

Cells were plated onto six-well plates at ∼70–80% confluency in DMEM plus 10% fetal bovine serum and allowed to attached overnight. Cells were then made quiescent by placing in serum-free media for 48–72 h before the start of the experiments. Total RNA from cells stimulated with aldosterone (0.1 μM; Sigma-Aldrich, St. Louis, MO) alone or in the presence of spironolactone (10 μM; Sigma-Aldrich) for up to 24 h was isolated with TRIzol per the manufacturer's instructions. One microgram of total RNA [from either vascular smooth muscle cells (VSMCs) or human/rabbit vein graft tissue] was used to generate cDNA for subsequent RT-PCR reactions. The following primer sequences were employed—angiotensin II (ANG II) type-1 receptor (AT1R): forward, 5′-GTC ACC TGC ATC ATC ATT TGG-3′, and reverse, 5′-TCA TAA GCC TTC TTT AGG GCC TTC 3′; 11βHSD2: forward, 5′-GAC CAA ACC AGG AGA CAT TAG C-3′, and reverse, 5′-ATG TAG TCC TTG CCG TAG GC-3′; MR: forward, 5′-CAG AGC TGG CAG AGG TTC TA-3′, and reverse, 5′-TGG ACG CTA ACG AGT GTG TA-3′; GAPDH: forward, 5′-TTA GCA CCC CTG GCC AAG G-3′, and reverse, 5′-CTT ACT CCT TGG AGG CCA TG-3′; and β-actin: forward, 5′-GAG ACC TTC AAC ACC CCA GCC ATG-3′, and reverse, 5′-AGC CAG GTC CAG ACG CAG GAT-3′. PCR parameters included an initial 3-min denaturation step at 95°C, followed by a cycling program of 95°C for 1 min, 55°C for 1 min, and 72°C for 1 min for 35–40 cycles. PCR products were resolved on 2% agarose gels.


Paraffin-embedded human and rabbit vein grafts were deparaffinized and rehydrated in xylene before antigen retrieval with either proteinase K (5 μg/ml; 20 min at room temperature; Fischer Bioreagents) or sodium citrate (15 mM; 40 min at 95–100°C). Sections were quenched with 1.5% H2O2 for 10 min, blocked in 10% horse serum, and incubated with primary antibodies: mouse anti-human monoclonal antibody to MR (H10E4C9F, 1:25, Affinity Bioreagents), anti-human AT1R (sc-597, 1:50, Santa Cruz), or an IgG isotype control for 1 h at room temperature. Bound primary antibody was detected using a biotinylated secondary antibody kit (ABC standard kit, Vector). Sections were counterstained with Gill's hematoxylin, followed by dehydration in graded ethanol and xylene.

MR reporter assay.

Primary human saphenous vein smooth muscle cells (P5–8) were grown in DMEM (GIBCO) with 10% fetal bovine serum. Five days before adenoviral infection, the cells were split into 12-well plates in DMEM with 10% fetal bovine serum. Cells were transduced with an adenoviral MR response element (MRE)-luciferase gene reporter containing the mouse mammary tumor virus promoter at a multiplicity of infection of 200 as previously described (20). Twenty-four hours after infection, the indicated concentration of vehicle, ligand, and/or inhibitor was added for 18 h. Cells were lysed, and luciferase activity was determined in duplicate as previously described (20).

Statistical analysis.

Data are reported as means ± SE. Within-group differences were assessed with one-factor ANOVA with Student-Newman-Keuls posttest (Fig. 5). A value of P < 0.05 was considered significant.


MR is expressed in human vein tissue after bypass grafting.

To determine whether MR is expressed in human vein grafts, excised specimens of diseased grafts ranging from 84–1,330 days postoperative (N = 16) were immunostained with a monoclonal antibody against MR or an isotype IgG control. MR staining occurred in a transmural pattern in all lesions examined (Fig. 1, B, C, E, and F). MR staining was not detectable in human saphenous vein tissue that had not been grafted (Fig. 1D).

Fig. 1.

Detection of immunoreactive mineralocorticoid (MR) in human peripheral vein graft lesions: IgG control (A) and MR staining of 112-day vein graft at 4× (B) and 10× (C) magnification; normal greater saphenous vein (GSV) stained for MR (D); MR staining in 390-day vein graft (10×) (E); and 84-day vein graft (20×) demonstrating marked neointimal concentration of MR expression (F). L, lumen.

MR and 11βHSD2 are coexpressed in human venous smooth muscle cells (SMCs) and in human vein grafts.

We used RT-PCR to determine whether the components necessary for aldosterone-specific signaling were expressed in human venous tissue. Cultured hSVSMCs showed strong expression of both MR and 11βHSD2, also seen in control cultured human embryonic kidney cells (HEK-293) (Fig. 2). RNA isolated from human vein graft lesions demonstrated a significant expression of both MR and 11βHSD2, whereas ungrafted human GSV had minimal MR expression and no detectable 11βHSD2.

Fig. 2.

MR and 11β-hydroxysteroid-dehydrogenase-2 (11βHSD2) are expressed in vascular smooth muscle cells (VSMCs) and diseased human vein grafts (HVGs). RNA was analyzed by RT-PCR as described in text. GSV showed minimal MR expression and no 11βHSD2 expression. Human embryo kidney 293 (HEK-293) cells serve as a positive control. SMC, smooth muscle cell. (Figure is a composite created from 3 experimental gels.)

MR and 11βHSD2 expression are coordinately upregulated in an in vivo vein graft model.

To assess the early temporal pattern of MR expression in arterialized vein, a rabbit model was employed. Rabbit vein grafts (n = 15) were explanted at 1, 7, 14, 30, and 51 days postimplantation and processed for immunohistochemistry along with appropriate controls. Immunostaining for MR demonstrated a transmural expression pattern detectable at 7 days post-bypass and beyond (Fig. 3A). In contrast, MR staining was not seen in normal rabbit jugular vein (Fig. 3C). RT-PCR of RNA isolated from explanted rabbit vein grafts demonstrated that the coexpression of MR and 11βHSD2 occurred by day 7 after vein grafting and continued to be strongly expressed through 51 days (Fig. 4). The single-band MR product from rabbit tissue corresponds to the expected 806-bp product (nucleotides 2962–3767, GenBank Acc. No. M16801), which spans a segment of the COOH-terminal aldosterone-binding domain and a portion of the 3′-UTR. Sequencing of this rabbit PCR product demonstrated 85% overall nucleotide homology to the human MR cDNA sequence (92% nucleotide, and 96% amino acid homology within the translated region of the cDNA).

Fig. 3.

Rabbit vein grafts express MR early after implantation. A: immunostaining for MR in 7-day rabbit vein graft. B: IgG control. C: ungrafted rabbit jugular vein stained with MR.

Fig. 4.

MR and 11βHSD2 are expressed early in the arterialization response. Representative RT-PCR of rabbit vein grafts shows temporal pattern of increased expression of both MR and 11βHSD2 beginning at 7-days post-bypass and persisting through 51 days. C, normal vein; d, day. N = 3.

Human venous smooth muscle cells express functional MR.

To explore the function of MR in venous SMCs, an adenovirus containing a MRE driving a luciferase gene reporter was transduced into hSVSMCs (Fig. 5). There was a dose-dependent increase in gene activation with aldosterone stimulation that was significant at 10 nM and peaked at 100 nM concentrations. Aldosterone-stimulated gene transcription was inhibited in the presence of the competitive MR antagonist spironolactone (10 μM), supporting a role for MR. Similar experiments using a non-MR-binding response element (estrogen response element-luciferase reporter) did not show aldosterone-induced activation in hSVSMCs (data not shown). These data support that the MR in hSVSMCs is capable of aldosterone-induced MR-mediated gene activation.

Fig. 5.

Human venous smooth muscle cells exhibit aldosterone (Aldo)-induced MR-mediated gene activation. VSMCs infected with adenovirus MR response element-luciferase and stimulated with Aldo in the presence or absence of 10 μM spironolactone (Spirono). Bars indicate fold activation of luciferase activity over vehicle alone (means ± SE). *P < 0.05 vs vehicle. N = 4.

Aldosterone increases AT1R expression in human venous smooth muscle cells via an MR-dependent pathway.

To further investigate the functional significance of MR-related gene activation in hSVSMCs, a relevant target gene (AT1R) in the renin-angiotensin-aldosterone system (RAAS) signaling pathway was explored. hSVSMCs were stimulated with aldosterone (100 nM) in the presence or absence of the MR antagonist spironolactone (10 μM) for 24 h, and RNA was analyzed by RT-PCR with primers specific for human AT1R. Aldosterone stimulation resulted in an increase in AT1R gene expression that was abrogated by the addition of spironolactone (Fig. 6). This suggests that aldosterone upregulation of AT1R is mediated by an MR-sensitive pathway in hSVSMCs and supports that MR signaling may augment RAAS activation in venous SMCs. Further evidence for the relevance of this observation was found by immunostaining for both MR and AT1R in human venous tissue (Fig. 7). The expression of both genes within the neointima and media of grafted veins, but not ungrafted saphenous vein, is demonstrated. While the degree and extent of AT1R expression appear greater than that of MR in the graft lesion, there is considerable spatial overlap. Taken together with the cell culture data, these findings suggest a potential mechanism for the amplification of RAAS signaling within the arterialized vein.

Fig. 6.

Aldo-MR signaling induces angiotensin II type-1 receptor (AT1R) gene expression in VSMCs. Representative RT-PCR of human venous smooth muscle cells stimulated for 24 h in the presence of 0.1 μM Aldo (A) or 0.1 μM Aldo + 10 μM Spirono (A/S). SF, serum-free control. (Intervening lanes between A and A/S were removed in construction of final figure.) Data are representative of greater than 3 independent experiments.

Fig. 7.

Distribution of MR and AT1R expression in human veins and vein grafts. Top: sections of ungrafted human saphenous vein. Bottom: sections from a HVG explanted at 84 days from implantation. A and D: MR staining. B and E: AT1R staining. C and F: IgG isotype control.


Signaling through the RAAS is important in mediating pathophysiological processes in cardiovascular tissue. Here we show for the first time that the components of this system are present in venous bypass grafts and may be important in the healing response of arterialized veins. Our immunohistochemical data show that MR is expressed in a transmural distribution across various time points in both human and rabbit vein grafts. We also demonstrate the coexpression of MR and 11βHSD2, the enzyme responsible for conferring aldosterone selectivity on MR, in human vein graft lesions, rabbit vein bypass grafts, and human venous smooth muscle cells. Functional studies using the adenovirus MRE-luciferase constructs show that venous smooth muscle cells are capable of MR-mediated gene transcription. Finally, we demonstrated that AT1R, a critical receptor in RAAS-mediated cardiovascular injury, is upregulated by aldosterone via an MR-dependent pathway and is expressed in a similar distribution in vein graft lesions. Importantly, in ungrafted veins, MR (and AT1R) is barely detectable and 11βHSD2 is absent. Taken together, these data suggest that MR and 11βHSD2 are upregulated in venous tissue after grafting, allowing for aldosterone-selective MR activation in grafted venous tissues. Thus activated MR may be involved in the local regulation of vascular gene expression, resulting in changes in cell proliferation and ultimately graft remodeling.

Aldosterone, via activation of MR, has been shown to directly induce inflammation, endothelial dysfunction, and fibrosis (7, 8). Clinical studies have demonstrated a reduction in mortality in patients with left ventricular dysfunction due to CHF (Randomized Aldactone Evaluation Study) and post-myocardial infarction (Eplerenone Post-Acute Myocardial Infarction Heart Failure Efficacy and Survival Study) with MR blockade (39, 40). More recently, multiple large clinical trials of the cholesteryl ester transfer protein inhibitor Torcetrapib were terminated early because of increased cardiovascular mortality, despite very favorable lipid profiles (2, 6, 33), likely due to the off-target aldosterone-raising effects of the drug (32). The results of these studies suggest that MR activation by aldosterone is an important mediator of vascular remodeling, inflammation, oxidative stress, and fibrosis.

The pathophysiological actions of aldosterone appear to be independent from its classical actions on water and sodium homeostasis and can act via genomic and nongenomic pathways. The components necessary for aldosterone signaling have been demonstrated in cardiovascular tissues. MR expression has been identified in isolated endothelial and smooth muscle cells, as well as in intact tissues such as the heart, aorta, and pulmonary artery (8, 26). The cortisol-inactivating enzyme 11βHSD2 has been demonstrated in human arterial endothelial and smooth muscle cells and in mouse vessels (8, 9, 20). MR binds aldosterone and cortisol with equal affinity. However, cortisol has 100–200-fold higher plasma free level and presumably a similar intracellular ratio. Aldosterone specificity for MR is conferred by the presence of the cortisol-inactivating enzyme 11βHSD2 in tissues where they are coexpressed. The precise mechanism for 11βHSD2-induced aldosterone selectivity is not completely understood, but it has been thought to involve the conversion of cortisol to cortisone, thus allowing the preferential binding and activation of MR by aldosterone. However, stoichiometrically, there does not appear to be adequate levels of 11βHSD2 to convert all the cellular cortisol to cortisone: there would still be a 10-fold excess relative to aldosterone. Funder (15) has recently suggested that 11βHSD2 may be important in aldosterone selectivity by inhibiting the ability of cortisol to act as an MR agonist rather than excluding cortisol from binding MR. In our study, the parallel upregulation of 11βHSD2 and MR suggests the potential for aldosterone-specific signaling within the arterialized vein.

A role for MR signaling in vascular remodeling after vessel injury has been demonstrated in porcine coronary artery angioplasty and stenting models. Ward et al. (49) showed that after porcine coronary angioplasty, treatment with the selective MR antagonist eplerenone resulted in significantly increased vessel area and luminal area and decreased neointimal to vessel area ratio over controls. Additionally, intimal and medial collagen content was decreased in the eplerenone treatment vessels. These data suggest an attenuation of constrictive remodeling and collagen accumulation with MR antagonism. Wakabayashi et al. (47) demonstrated the suppression of neointimal hyperplasia by MR antagonism in a porcine coronary artery stent model. Eplerenone-treated animals had significantly increased luminal area and decreased neointimal area with less accumulated collagen type I and III at 4 wk poststenting. The eplerenone-treated group had downregulated TGF-β mRNA expression, implying that the effects seen were achieved by the inhibition of TGF-β with subsequent inhibition of collagen type I and III expression.

ANG II has a well-characterized role in mediating vascular injury (23, 29, 45). Several studies have suggested that the RAAS is locally active in the healing vein graft wall and has an important role in vein graft remodeling (4, 19, 24, 35). In addition, the inhibition of components of the RAAS has been shown to limit the progression of vein graft disease in experimental animal models (10, 14, 36). However, these experimental findings have not been translated into clinically significant improvement in vein graft patency (5).

ANG II induces VSMC proliferation and hypertrophy and promotes superoxide generation and reactive oxygen species-mediated injury (11, 25, 45). ANG II also stimulates the local production of growth factors (e.g., PDGF, TGF-β1) and matrix-related proteins (e.g., collagen, fibronectin, plasminogen activator inhibitor-1) that are associated with the development of hyperplastic lesions and fibrosis (45). In cultured VSMCs and in intact vessels, the majority of effects appear to be mediated by AT1R activation. We show that aldosterone-stimulated hSVSMCs have increased transcription of AT1R that was abrogated by the MR-antagonist spironolactone. Furthermore, we show that AT1R expression in the grafted vein follows a similar distribution pattern as MR. This implies a functional relevance to aldosterone/MR signaling in these tissues that may promote or enhance the activation of ANG II signaling, thus potentiating vascular injury.

There are several studies that suggest that ANG II and aldosterone may act in a synergistic manner to activate signaling pathways that regulate vascular cell proliferation and inflammation (20, 30, 41, 44, 51, 52). Aldosterone stimulation has been shown to upregulate AT1R, enhance ANG II signaling, and increase angiotensin-converting enzyme expression in rat vascular tissue (17, 28, 43). In rat aortic smooth muscle cells, a combined stimulation with ANG II and aldosterone resulted in VSMC proliferation via an ERK1/2 pathway that was inhibited by the addition of AT1R and MR antagonists. This effect on proliferation was not seen with ANG II or aldosterone stimulation alone, suggesting a synergistic interaction (30). Jaffe and Mendelsohn (20) have shown in human arterial SMCs that ANG II can induce MR-dependent gene expression. This gene expression can be abolished by spironolactone and the AT1R inhibitor losartan, suggesting an interplay between the ANG II and aldosterone signaling pathways.

This study demonstrates the expression of functional MR in hSVSMCs and the coexpression of MR and 11βHSD2 in vein bypass grafts. The major limitations of this study are its descriptive nature and the limited sample of diseased human vein grafts for analysis. A critical question is whether there is differential expression of these genes between normally functioning and diseased (e.g., stenotic) vein grafts; however, we are unable to obtain specimens of nondiseased human grafts, and the animal models employed typically do not develop stenosis. Finally, these studies do not provide functional correlates of the molecular and histological changes observed, although the reporter gene studies provide a functional validation in cultured venous cells. Future studies using MR antagonists or genetic manipulation of this pathway in vivo are needed to elucidate the role of MR activation in vein graft remodeling.


The components necessary for aldosterone-selective MR signaling are upregulated in arterialized vein grafts, suggesting that these tissues may constitute a novel target for aldosterone-mediated injury. Local aldosterone/MR signaling may potentiate ANG II activity via the upregulation of AT1R in vein grafts. Drugs that inhibit this pathway are clinically available and safe and have been shown to improve cardiovascular outcomes in other patient populations. Future studies should examine the therapeutic potential of targeting these pathways to prevent vascular disease in arterialized veins after venous bypass surgery.


Primary support for this work was from the Carl and Ruth Shapiro Family Foundation (to M. S. Conte); the Harvard-Longwood Research Training Program in Vascular Surgery Grant 5T32-HL07734 (to R. Bafford); and National Institutes of Health Grants HL-086907 (to G. K. Adler and G. H. Williams), HL-74892 (to I. Z. Jaffe), HL-065998 (to R. A. Khalil), and R01-HL-096518 and R21-ES-014462 (to J. R. Romero).


No conflicts of interest, financial or otherwise, are declared by the author(s).


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 25a.
  27. 26.
  28. 27.
  29. 28.
  30. 29.
  31. 30.
  32. 31.
  33. 32.
  34. 33.
  35. 34.
  36. 35.
  37. 36.
  38. 37.
  39. 38.
  40. 39.
  41. 40.
  42. 41.
  43. 42.
  44. 43.
  45. 44.
  46. 45.
  47. 46.
  48. 47.
  49. 48.
  50. 49.
  51. 51.
  52. 52.
View Abstract