OUP user menu

Cyclooxygenase-dependent vasoconstricting factor(s) in remodelled rat femoral arteries

Akiko Hirao , Kazunao Kondo , Kazuhiko Takeuchi , Naoki Inui , Kazuo Umemura , Kyoichi Ohashi , Hiroshi Watanabe
DOI: http://dx.doi.org/10.1093/cvr/cvn111 161-168 First published online: 2 May 2008


Aims Denudation and regeneration of the vascular endothelium are important in the pathogenesis of atherosclerosis. The aim of this study is to clarify the mechanisms of functional alterations in remodelled arteries following endothelial injury.

Methods and results Non-mechanical endothelial injury was induced by 540-nm light irradiation of rose Bengal in femoral arteries of Wistar rats. Endothelium-dependent vasodilation was assessed by the response to acetylcholine (ACh) 1, 2, and 4 weeks after the injury. In control arteries, ACh-induced relaxation was mainly nitric oxide-dependent at all study time points. In injured arteries, this response was completely restored at 1 week, but was more dependent on KCl-sensitive endothelium-derived hyperpolarizing factor production during the first 2 weeks. Cyclooxygenase (COX) isoforms 1 and 2 were detected in the endothelium of injured arteries, and inhibition of prostanoids production with the non-specific COX inhibitor indomethacin substantially enhanced the ACh-induced vasorelaxation response in injured arteries, but did not affect control arteries. Similar effects were observed with the COX-1 inhibitor SC-560, the COX-2 inhibitor NS-398, the thromboxane (TX) A2/prostaglandin (PG) H2 receptor antagonist SQ29548 and the PGF receptor antagonist AL-8810. However, the TX synthetase inhibitor OKY-046 had no effect on ACh-induced relaxation in injured arteries.

Conclusion In remodelled arteries following photochemical endothelial injury, the vasoconstrictive prostanoids PGH2 and PGF, but not TXA2, contribute to changes in endothelium-dependent vascular response via COX-1- and 2-dependent pathways.

  • Endothelium
  • Cyclooxygenase
  • Vasoconstricting factor(s)
  • Remodeling
  • Prostanoids

1. Introduction

The vascular endothelium plays a pivotal role in the control of vascular homeostasis. Endothelial cell injury leads to various cardiovascular diseases such as atherosclerosis, hypertension, and stroke. The turnover of endothelial cells is accelerated under most of these conditions,1 and the primary endothelial cell damage is in turn aggravated, constituting a vicious cycle of pathological events. Inasmuch as it produces alterations in vascular tone and inflammatory responses, regeneration or repair of the injured endothelium may be critical in the pathogenesis of these disorders.2 Nevertheless, the precise functional disturbances of regenerated endothelium responsible for these alterations are not entirely clear.

A number of experimental models of endothelial denudation have been developed to study regenerated endothelial cell functions, including wire-mediated injury,3 air-drying,4 and balloon catheterization.5 Most of these models, however, involve mechanical stress to the vessel under study, and the observed functional alterations in the regenerated endothelium might have been interfered by those in adjacent structures such as smooth muscle. In this regard, non-mechanical approaches to cause endothelial injury, such as oxidative stress, would appear to give a more genuine picture of pathological endothelial denudation. We have previously developed a non-mechanical method of endothelial denudation that makes use of the photochemical reaction stimulated by 540-nm light of systemically administered rose Bengal.6 Excitation at 540 nm stimulates rose Bengal to produce singlet molecular oxygen by energy transfer to molecular oxygen. The reactive oxygen species (ROS) reacts with structural proteins and lipids in the endothelial cell membrane to initiate peroxidative reactions that lead to direct endothelium injury. Thus, in this model, endothelial denudation is induced intravascularly and there is no mechanical stress on the vessel under study.

We have now used this model to examine whether photochemical injury impairs endothelium-dependent and independent vasodilatation responses and to evaluate the relative contribution of the incurring changes in the production of endothelium-dependent relaxing factors, nitric oxide (NO), endothelium-derived hyperpolarizing factor (EDHF), and prostanoids to this process.

2. Methods

2.1 Animal model

Male Wistar rats (6 weeks old, weighing 120–160 g) were anaesthetized with intraperitoneal injection of sodium pentobarbital (50 mg/kg), and a cannula was inserted into the jugular vein for injection of the photosensitizing dye rose Bengal. The right femoral artery was carefully exposed, rose Bengal (10 mg/kg) was infused for 1 min, and 540-nm transillumination of the artery was performed at 0.9 W/cm2 for 20 min using a xenon lamp equipped with a heat-absorbing filter and a green filter (Hamamatsu Photonics, Japan). The sham-operated left femoral artery was used as a control in following tension measurement. The surgical wound was then closed and the rats were allowed to recover from anaesthesia. These protocols were approved by our Institutional Animal Care and Use Committee and conformed with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996).

2.2 Morphological studies and immunohistochemical staining

Rats were studied at different time points after the induction of endothelial injury, including before, 1 day, and 1, 2, and 4 weeks after photochemical irradiation. The chest and abdominal cavities were opened, and a catheter was inserted into the left ventricle. The whole body blood vessels were first washed with physiological saline and then perfusion fixed with 1% formaldehyde in 0.1 mol/L PBS, pH 7.4. Then the right, injured femoral artery was removed and fixed further by immersion overnight in the same fixative. Fixed femoral artery segments were embedded in paraffin and sliced consecutively into 5-μm thick sections. For morphometric analysis, the deparaffinized sections were stained with haematoxylin and eosin. The cross-sectional intimal and medial areas in a given photomicrograph were determined with Win Roof image analysis software (Mitani Corporation, Japan). For von Willebrand factor (vWF) staining, the deparaffinized sections were treated with proteinase K (DakoCytomation, Denmark) for 5 min at room temperature, and incubated overnight at 4°C with polyclonal rabbit anti-human vWF antibody (DakoCytomation), an endothelial marker, at a dilution factor of 1:700. For cyclooxygenase (COX)-1 staining, the deparaffinized sections were incubated overnight at 4°C with polyclonal rabbit anti-murine COX-1 antibody (Cayman Chemical, USA) at a dilution factor of 1:200. For COX-2 staining, the deparaffinized sections were treated with 1 mM EDTA in 10 mM Tris buffer (pH 9.0) for 20 min at 95°C, and incubated overnight at 4°C with a polyclonal rabbit anti-murine COX-2 antibody (Cayman Chemical) at a dilution factor of 1:200. After washing with PBS, these samples were exposed for reaction with a goat anti-rabbit secondary antibody (Nichirei, Japan) for 30 min at room temperature. Finally, immunoreactions were visualized with 3,3′-diaminobenzidine (Nichirei), and the sections were counterstained with haematoxylin. Negative control samples were incubated with a rabbit non-immune IgG.

2.3 Vessel preparation and measurement of isometric tension

Rats were asphyxiated with CO2 and exsanguinated 1, 2, or 4 weeks after endothelial injury. The right injured and left control femoral arteries were removed and cut into rings at a length of 1.5 mm. Arterial ring preparations were mounted on an isometric myograph (Primetech, Japan) filled with oxygenated (5% CO2/95% O2) Kreb's solution of the following composition (mM): NaCl 120.0, KCl 4.7, MgSO4 7H2O 1.2, KH2PO4 1.2, CaCl2 2.5, NaHCO3 25.0, and glucose 10.0, at 37°C. Isometric tension was continuously recorded on a polygraph (Nihon Kohden, Japan). The rings were stretched until an optimal resting tension of 0.4 g was loaded. After an equilibration period of 60 min, the rings were contracted by phenylephrine (PE, 10 µM) two times at 45-min intervals. After the second contraction by 10 µM PE reached a stable state, endothelium-dependent and independent relaxation was induced in a cumulative manner with acetylcholine (ACh, 100 pM to 100 µM) and sodium nitroprusside (SNP, 100 pM to 10 µM), respectively. In our preliminary experiments, this dose of PE produced near-maximum contraction in each group. Prior to PE contraction, the rings were incubated for 30 min with one of the following agents: NO synthase inhibitor Nω-nitro-L-arginine methyl ester (L-NAME), non-selective COX inhibitor indomethacin, COX-1 inhibitor SC-560, COX-2 inhibitor NS-398, thromboxane (TX) synthetase inhibitor OKY-046, a TX A2 (TXA2)/prostaglandin H2 (PGH2) receptor antagonist SQ29548, and a PG F2 alpha (PGF) receptor antagonist AL-8810. KCl solution (40 mM) was prepared by substitution of KCl for NaCl on an equimolar basis. SQ29548 and KCl were purchased from Cayman Chemical and Wako (Japan), respectively. Other drugs used in the tension experiments were purchased from Sigma-Aldrich (USA). Stock solution for indomethacin was prepared in 96% ethanol. PE, L-NAME, and OKY-046 were dissolved in deionized water. The solvent of other agent stocks was dimethyl sulphoxide. The final concentration of respective vehicles did not exceed 0.1% in organ bath medium.

2.4 Data analysis

Levels of vascular relaxation are expressed as the percentage of the preceding contraction induced by 10 µM PE. The pD2 value, which is the negative logarithm of the molar concentration that caused half-maximal response, and the area over each individual curve (AOC) were calculated with Prism 4 software (GraphPad, USA). Differences in AOC values for ACh-relaxation curves in the absence and presence of L-NAME were used to estimate the contribution of NO. Differences in AOC values for ACh-induced vasodilatation in the presence of L-NAME and the combination of L-NAME and indomethacin were used as an estimate for the contribution of prostanoids. AOC values for ACh-induced vasodilatation in the presence of L-NAME and indomethacin were used as an estimate for the contribution of EDHF. Then, the relative contribution of NO, EDHF, and prostanoids to the vasodilatation response was estimated as the percentage of the AOC in ACh concentration-relaxation curve in each artery. Significant differences were analysed by using Student's t-test or ANOVA followed by Dunnett's post-hoc test. P-values <0.05 were considered statistically significant.

3. Results

3.1 Time course of re-endothelialization and arterial thickening

Re-endothelialization was evaluated immunohistochemically with anti-vWF antibodies, a marker of endothelial cells.7 vWF immunoreactivity was present along the luminal arterial surface before injury (Figure 1A), was no longer detected the day after the injury (Figure 1B and G), but was observed again 1 week later (Figure 1C and H), and continued to be present at 2- and 4-week time points after injury (Figure 1D and E). These results indicate that photochemical injury could cause complete endothelial denudation, and regeneration of the denuded endothelium was almost complete 1 week after injury.

Figure 1

Immunohistochemical analysis showing the time course of endothelial regeneration using the anti-von Willebrand factor antibody in rat femoral arteries before and after photochemical injury. (A) Right femoral artery before photochemical denudation. (BF) Right femoral artery at 1 day (B and G), 1 week (C and H), 2 weeks (D), and 4 weeks (E and F) after photochemical injury. (F) Negative control staining. (G and H) Cross sections of the whole arteries corresponding to (B) and (C). Squares in (G and H) indicate the location of (B and C), respectively. The internal elastic lamina is indicated by the arrows. The luminal arterial surface of control (A) is covered with a thin monolayer of endothelium, which may make it difficult to distinguish the endothelial monolayer from the subendothelial tissue stained by deep brown and nuclei stained by strong blue. In (B) soluble von Willebrand factor in blood may attach to collagen on the exposed subendothelial surface even after complete denudation of endothelium, which makes luminal surface slight brown.

Neointima formation in injured arteries before and 1, 2, and 4 weeks after rose Bengal stimulation was assessed by morphometric analysis. No intimal thickening was observed before injury (Table 1). However, the intimal area showed the tendency to grow from 1 to 4 weeks after injury (P < 0.1); the intima–media ratio at 4 weeks was significantly larger than at 1 week (P < 0.05). In contrast, the respective medial area was virtually unaltered throughout the experimental period.

View this table:
Table 1

Time course of changes in intimal and medial area following photochemically induced endothelial injury

Time of sacrificeMedial area (×0.01 mm2)Intimal area (×0.01 mm2)Intima–Media ratio
Before5.565 ± 0.3060 ± 00 ± 0
1 week5.872 ± 0.185n.s.0.808 ± 0.096*0.136 ± 0.013*
2 weeks5.911 ± 0.209n.s.0.952 ± 0.086*0.161 ± 0.013*
4 weeks5.799 ± 0.126n.s.1.022 ± 0.063*0.176 ± 0.010*#
  • Data are presented as mean ± SEM from 6–7 rats. n.s., not significant. *P < 0.01 vs. data in arteries before endothelial injury by Dunnett's multiple comparison and #P < 0.05 vs. 1-week group by t-test.

3.2 Contractile and vasodilator responses of smooth muscle cells

For evaluation of vasodilator responses, an arterial segment was precontracted by 10 µM PE, an alpha-1 adrenergic receptor agonist, in every study. The vasoconstriction responses to PE of the injured arteries were similar to those of control arteries throughout the experimental periods (Figure 2A). These results indicate that the smooth muscle cells in the injured arteries maintained their contractility, at least in response to the high dose of PE (10 µM).

Figure 2

Contractile responses to 10 µM PE (A) in control femoral arteries (open column) and injured arteries (closed column) at 1, 2, and 4 weeks after photochemical injury. Concentration-response curve to the vasodilator SNP (B) in control arteries (a) and injured arteries (b) at 1 week (open triangle), 2 weeks (closed square), and 4 weeks (open circle) after photochemical injury. Results are presented as mean ± SEM from 6 rats. There is no significant difference between the two groups at any study time points with respect to both PE contraction and SNP vasodilation. gf, gram-force.

Endothelium-independent vasodilation capability was assessed by observing the vasodilation responses to cumulative additions of the NO donor SNP to PE-precontracted arteries. The maximal vasodilator response (Emax) and the pD2 value of SNP-induced concentration response curves in injured arteries were not significantly different to those of control arteries at any time point of the study period; the pD2 value and Emax at 1, 2, and 4 weeks after injury were 8.16 ± 0.05 and 105 ± 1, 8.12 ± 0.07 and 101 ± 1, 8.18 ± 0.05, and 101 ± 1% in control arteries, and 8.18 ± 0.13 and 104 ± 2, 8.21 ± 0.18 and 97 ± 4, and 8.13 ± 0.10 and 103 ± 2% in injured arteries (Figure 2B). These results demonstrate that the smooth muscle cells in the injured arteries maintained their relaxation capability in response to vasodilating substances such as NO.

3.3 Endothelium-dependent relaxation to ACh

Endothelium-dependent relaxation was evaluated by observing the vasodilation responses to cumulative additions of ACh to PE-precontracted arteries. Compared with control arteries, ACh-induced maximal relaxation (Emax) of the injured arteries was abolished after 1 day (data not shown), significantly enhanced after 1 week (P < 0.01), identical at 2 weeks, and significantly reduced at 4 weeks (P < 0.05) after injury (Figure 3Ab open circles); Emax was 99 ± 1, 99 ± 2, and 99 ± 1% in control arteries, and 109 ± 2, 94 ± 2, and 91 ± 3% in injured arteries at 1, 2, and 4 weeks after injury, respectively. On the other hand, the pD2 values in injured arteries were significantly decreased at any time point of the study period compared with control arteries, and lowered in a time-dependent manner; pD2 values in control vs. injured artery were 7.54 ± 0.02 vs. 7.05 ± 0.08 (P < 0.01) at 1 week, 7.41 ± 0.09 vs. 6.82 ± 0.17 (P < 0.05) at 2 weeks, and 7.43 ± 0.05 vs. 6.75 ± 0.16 (P < 0.01) at 4 weeks after injury, respectively. The change of pD2 values indicates that the sensitivity to ACh of the injured arteries was gradually attenuated.

Figure 3

Concentration-response curve to ACh in control arteries (a) and injured arteries (b) at 1 week (A), 2 weeks (B), and 4 weeks (C). The relaxation was carried out in absence (open circle) or presence of either 100 µM L-NAME (open triangle), L-NAME + 10 µM indomethacin (closed square) or 100 µM L-NAME + 10 µM indomethacin + 40 mM KCl (closed diamond). Relative contribution of NO (c, hatched column), prostanoids (c, closed column, PNs), and EDHF (c, open column) to the ACh-induced relaxation was calculated by AOC of each concentration response curves (a and b). Results are expressed as percentage of PE-induced precontraction (a and b) and percentage of total response to ACh (c). Data are mean ± SEM from 6 rats; *P < 0.05l; **P < 0.01 vs. corresponding data in control arteries.

3.4 The relative contribution of different mediators of endothelial-dependent relaxation

In order to evaluate the relative contribution of NO, prostanoids, and EDHF to the endothelium-dependent relaxation induced by ACh, we generated concentration responses to ACh in the presence or absence of L-NAME, indomethacin, and high KCl in injured and control arteries precontracted with PE and calculated AOC values for each concentration response curves. Inhibition of NO synthase with L-NAME (100 µM) attenuated the relaxation to ACh in both injured and control arteries. Inhibition of both COX and NO synthase with indomethacin (10 µM) and L-NAME (100 µM) did not change the maximal response to ACh in control arteries, but enhanced it in injured arteries. In the presence of similar doses of L-NAME and indomethacin, addition of the depolarizing agent KCl (40 mM) totally abrogated the ACh-induced relaxation in both arteries (Figure 3Ab). L-NAME at 100 µM and indomethacin at 10 µM have been shown to cause complete inhibition of endothelial NO synthase, and both the constitutive and inducible forms of COX, respectively.8,9 A defining property of the EDHFs is the ability to induce endothelium-dependent hyperpolarization of smooth muscle cells independently of COXs and NO synthase. Despite controversy over the identity of the EDHFs, ample evidence exists that this effect results from the opening of K+ channels in the smooth muscle cell membrane10 and is abolished by K+ concentrations higher than 25 mM.11 We therefore chose to use 40 mM KCl to prevent the hyperpolarizing effect of EDHFs.

Analysis of AOC values (Figure 3C) indicates that ACh-induced vasorelaxation response was mainly dependent on NO at all study time points in control arteries. In injured arteries, however, EDHF contributed more to this response at 1- and 2-week time points; the respective ACh responses from injured and control arteries were 52 ± 2 vs. 31 ± 6% at 1 week; 54 ± 8 vs. 24 ± 5% at 2 weeks. This effect disappeared at 4 weeks after injury, when no difference from control values was observed (43 ± 7 vs. 32 ± 3%). Thus, EDHF becomes an important factor during the first two weeks after injury but reclines after 4 weeks as NO resumed its dominant role. As for prostanoids, vasodilator prostanoids such as prostacyclin appear to play little role in ACh-induced relaxation in control arteries. On the other hand, in injured arteries, participation of vasoconstrictive prostanoids (represented by the minus portion in Figure 3, bottom bar graphs) became statistically significant −17 ± 6% at 2 weeks and −14 ± 6% at 4 weeks after injury. This effect was not significant 1 week after the injury. Furthermore, the analysis of Emax in response to ACh (Figure 4) showed similar statistical differences to the AOC analysis. In addition, there was no significant difference between controls and injured arteries as regards PE and KCl-induced contractions in the presence of L-NAME and/or indomethacin. These results suggest that the production or activity of vasoconstricting metabolites from the COX pathway is enhanced in injured arteries. Interestingly, the maximal relaxation response to ACh was well preserved in injured arteries, where the production of vasoconstrictive prostanoid(s) was enhanced, suggesting a counter-balancing role of the vasorelaxing factors EDHF and NO to maintain vascular homeostasis.

Figure 4

Maximal relaxation in response to ACh in the presence of 100 µM L-NAME (L), L-NAME + 10 µM indomethacin (I) or 100 µML-NAME + 10 µM indomethacin+40 mM KCl (K) in control arteries (A) and injured arteries (B) at 1 day, 1 week, 2 weeks, and 4 weeks after photochemical injury. The responses are presented as percentage of the ACh-induced maximal relaxation in the absence of inhibitors in each artery. Data are mean ± SEM from 6 rats; n.s., not significant. *P < 0.05; **P < 0.01 vs. corresponding data in control arteries.

3.5 The pathways of the vasoconstrictive prostanoids

To determine which isoform of COX is responsible for the effect of indomethacin, the effects of the COX-1 inhibitor SC-560 and the COX-2 inhibitor NS-398 on the maximal relaxation induced by cumulatively applied ACh in injured arteries were examined in injured arteries 2 weeks after injury. It has been shown that NS-398 at 10 µM selectively inhibits COX-2 activity,9 and SC-560 has IC50 values of 9 nM for COX-1 and 6.3 µM for COX-2.12 Thus, in this study we used 0.3 µM SC-560 and 10 µM NS-398 to selectively inhibit COX-1 and COX-2, respectively. At these doses these drugs individually enhanced the maximum relaxation to ACh in the presence of 100 µM L-NAME. However, when both inhibitors were used, ACh-stimulated relaxation was not further enhanced (Figure 5A). This suggests that the involved vasoconstrictive prostanoids are produced by the coordinated activities of both COX-1 and COX-2.

Figure 5

Maximum relaxation responses to ACh (100 pM to 100 µM) in the presence of 100 µM L-NAME only (open column) or with (closed column) either 10 µM indomethacin (INDO), 0.3 µM SC-560, 10 µM NS-394, the combination of 0.3 µM SC-560 and 10 µM NS-398 (SC + NS) (A), 10 µM OKY-046, 0.1–1 µM SQ29548 or 1–10 µM AL8810 (B). Results are expressed as percentage of the maximal relaxation to ACh in the absence of any inhibitors and are presented as mean ± SEM from 6–9 rats. *P < 0.05 and **P < 0.01 vs. data obtained in the absence of prostanoid inhibitors by unpaired t-test and Dunnett's multiple comparison, respectively.

To identify the prostanoids responsible for the observed changes in ACh-induced relaxation in injured arteries, the effects of OKY-046, SQ29548, or AL-8810 on ACh-induced relaxation were compared in injured arteries 2 weeks after injury. Previous studies have shown that OKY-046 is a selective TX synthase inhibitor with an IC50 of 11 nM,13 SQ29548 is a potent TXA2/PGH2 receptor antagonist with a Kd of 10 nM,14 and AL-8810 has a selective antagonistic activity at FP receptor (PGF receptor) even at 10 µM.15 Both SQ29548 (0.3, 1 µM) and AL-8810 (1, 10 µM) enhanced ACh-induced relaxation as did indomethacin, whereas OKY-046 (10 µM) did not (Figure 5B). PE-induced contractions in the presence of SC-560, NS-398, OKY-046, SQ29518, and AL-8810 in combination with L-NAME were comparable to the prior PE-induced contraction in the presence of L-NAME. In addition, all reagents used in tension studies did not affect the resting tone of either control or injured arteries at any study time point. These results suggest that PGH2 and PGF, but not TXA2, are possible candidates for the vasoconstrictive prostanoids pertaining to ACh-induced relaxation.

3.6 COX-1 and COX-2 expressions

Analysis of COX-1 and COX-2 protein expression was performed immunohistochmically on sections of femoral arteries including before (control), 1, 2, and 4 weeks after surgery. COX-1 expression represented by brown colour staining was observed mainly in thin endothelial monolayer but also slightly stained in sub-endothelial tissue including smooth muscle cells. Positive staining for COX-1 was also observed within both the intimal and medial areas of injured arteries at all study time points. On the other hand, COX-2 represented by brown colour staining in panel E-H was not observed in control arteries, but was detected strongly in the surface neointimal area including regenerated endothelial cells in injured arteries (Figure 6).

Figure 6

Immunohistochemical staining with the anti-COX-1 antibody (AD) and the anti-COX-2 antibody (EH) in injured arteries. Right femoral artery before photochemical denudation (A and E), and at 1 week (B and F), 2 weeks (C and G), and 4 weeks (D and H) after photochemical injury. The internal elastic lamina is indicated by the arrows. Bars = 20 µm.

4. Discussion

Important findings of the present study are the possible involvement of COX-dependent vasoconstricting factors and EDHF up-regulation in the ACh-stimulated vascular response in remodelled arteries injured by a non-mechanical method. This study is the first to demonstrate that COX-1 and COX-2-dependent vasoconstricting factors such as PGHB2B and PGFB2αB are related to vascular tone in the arterial segment of the regenerated endothelium.

In the present study, a non-specific COX inhibitor indomethacin significantly augmented the ACh vasodilator response at 2- and 4-week time points after injury under a condition where NO production is inhibited with L-NAME. In addition, this effect is mimicked by inhibition of COX-1 and COX-2, but not by inhibition of TX synthase even when the inhibitor OKY-046 was used at the maximally effective concentration of 10 µM (IC50 of 11 nM). However, the PGHB2B/TXAB2B receptor antagonist SQ29548 enhanced relaxation dose-dependently. These findings suggest that the observed effect of SQ29548 was due to the inhibition of PGHB2B but not TXAB2. PGFB2αB receptor antagonism also induced vascular relaxation dose-dependently. PGHB2 acts upstream to PGFB2αB in the PG production cascade, and our results indicate that inhibition of either product exerts similar effect on the endothelium-dependent vasodilation response in injured arteries. PGHB2B is not only the precursor for all other active prostanoids, but also exhibits strong vasoconstricting activity by itself as well as PGFB2αB and TXAB2.16 Since endothelial cells are able to release a large quantity of untransformed PGHB2, 17 we cannot exclude the possibility that both untransformed PGHB2B and PGFB2αB rather than TXAB2B work as endothelium-derived constricting factors (EDCF)18 in photochemically injured arteries. COX-1 and COX-2 are the isoenzymes that metabolize arachidonic acids to PGGB2 and PGHB2, and finally lead to the production of PGFB. Our tension measurements indicate that these isoforms work coordinately to produce EDCFs in injured arteries, because there was no additive effect when COX-1 and COX-2 inhibitors were applied in combination. Thus it is reasonable that COX is responsible for the production of the vasoconstrictive prostanoid PGFB through the intermediate metabolite PGHB2. Identification of the involved vasoconstrictive prostanoids as well as their roles in vivo is should be the work of the near future.

Our immunohistochemical studies detected both COX-1 and COX-2 proteins in the neointima including the endothelium in injured arteries. Prostanoids have been thought to participate in the pathogenesis of many cardiovascular disorders.19 Although several laboratories have reported that PG formation is enhanced in remodelled arteries injured by balloon angioplasty,20,21 the functional role of vasoconstrictive prostanoids in these remodelled artery models has been focused on the acceleration of platelet aggregation, proliferation of smooth muscle cells and inflammation. Our study now demonstrates that vasoconstrictive prostanoids are responsible for the regulation of vascular tone in remodelled arteries. This finding is in agreement with the previous reports using animal models such as hypertension and heart failure, in which vasoconstrictive prostanoids interfered with ACh-induced relaxation.22,23 It has been reported that inducible NO synthase (iNOS) may enhance COX-2 production and the hyperactivity of COX-2 catalysis may be a source of ROS.24 In the remodelled artery, increased iNOS and ROS is thus an intriguing probability.

We have observed that, at 1 and 2 weeks after photochemical injury, when arachidonic acid derivative(s) with vasoconstrictor properties appeared to oppose the vasodilator action of ACh, the maximal ACh response was well preserved. Thus, compensatory mechanism(s) counteracting vasoconstrictive prostanoid(s) must have been available in the remodelled arteries. In fact, the relative contribution of EDHF to the ACh response was augmented between 1 and 2 weeks after injury. It is generally believed that endothelium-derived relaxing factors work in concert to control vascular tone. Thus, up-regulation of EDHF has been observed in various condition(s) with reduced NO production, such as in eNOS (endothelial NO synthase) knockout mice,25,26 congestive heart failure,27 and ischemia-reperfusion;28 NOS inhibition is observed in states of increased release of prostacyclin, a vasodilative prostanoid;29 and prostacyclin levels appear to be reduced in states of increased NO production.30,31 In our study, the enhancement in EDHF production apparently compensated for the increased production of vasoconstrictive prostanoid(s), thereby preserving the maximal vasodilator response to ACh.

In this study, components of the vascular response was dissected using cumulative addition of the inhibitors of the NO-, PG-, and EDHF pathways. With this approach it is difficult to assess the precise proportional contribution of NO, PG, and EDHF, since the effects might differ when the inhibitors are given individually. Nevertheless, qualitatively it is clear from these studies that regenerated endothelial cells produce not vasodilative but vasoconstrictive prostanoids.

In some mechanical endothelium-denudation models, the maximal vasodilator response to ACh is impaired acutely and lasts at least for 3 weeks after injury.32 This is clearly different from our results, and the simplest explanation would be the apparent impact of mechanical denudation on the underlying smooth muscle cells in these earlier studies. Mechanical injury such as balloon angioplasty may damage smooth muscle cells physically by hyperdistending the vessels, and may also impair functions of these cells. For example, it is noted that in the balloon angioplasty model, reduction in ACh-induced relaxation was accompanied by the attenuation of PE-induced contraction.32 On the other hand, the maximal vasodilator response to ACh was preserved in the silastic collar model with neointimal formation, in which a collar was positioned around the studied artery to prevent mechanical stress to the smooth muscle layer.33 In our model, there was apparently no mechanical stress on the smooth muscle cells, and the PE-induced contraction was well preserved in injured arteries.

We observed significant attenuation in ACh sensitivity as early as 1 week after the injury, when intimal thickening had already been provoked. Also, the sensitivity to ACh decreased as neointima thickening increased over time. This indicates that the development of intimal hyperplasia has profound impact on the reactivity to ACh. Intimal thickening was initiated in our model as early as in mechanical denudation models.34,35 It is possible that in isolated artery segments, the thickened neointima may disturb the diffusion of NO and other endothelium-derived relaxing factors towards the smooth muscle cells or impair the gap-junction communication between endothelial cells and smooth muscle cells, the putative pathway for EDHF. In fact, not only the sensitivity but the maximal relaxation response to ACh was also significantly reduced at 4 weeks after injury. Morphologically, intimal thickening in our model seemed to be less than that in the balloon angioplasty model.21,34 The development of neointimal formation may also affect maximal responses to ACh as well as the sensitivity.

In humans, ACh is clinically used as a probe for testing endothelial function. Many studies have shown that ACh-induced vasodilatation is diminished in conductance and resistance vessels of patients with atherosclerosis,36 hypertension,37 diabetes,38 and congestive heart failure.39 These abnormalities in endothelium-dependent relaxing responses in humans have usually been attributed to decreased activity of endothelium-derived relaxing factors, especially NO.40 However, recent studies have suggested that increased production of COX-dependent vasoconstricting factors may be responsible for abnormal responses to ACh in patients with congestive heart failure39 or hypertension.37 In line with these, aspirin has been shown to improve ACh-mediated relaxation in patients with atherosclerosis, possible via inhibition of COX-dependent vasoconstriction.41 Accumulating evidence suggests that vasoconstrictive prostanoid(s) are involved in the impairment of vascular relaxation in cardiovascular diseases.

In summary, we have demonstrated that in rat femoral arteries remodelled after non-mechanical endothelial denudation, the sensitivity to ACh decreases as the neointima thickens, whereas the maximal vasodilator response to ACh is preserved, probably by up-regulation of the EDHF component. Mechanistically, vasoconstrictive prostanoid(s) rather than vasodilating prostanoid(s) are augmented in response to ACh in remodelled arteries after endothelial denudation, and the candidates of vasoconstrictive prostanoids are PGHB2B and PGFB2αB, but not TXAB2B, produced by the COX-1 and COX-2 pathway.


This study was supported by Ministry of Education, Culture and Science of Japan Grants-in Aid 18390074 to H.W.


The authors are grateful to Dr Quang-Kim Tran (University of Missouri-Kansas City) for his helpful insights.

Conflict of interest: none declared.


View Abstract