Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The serine proteases, thrombin and trypsin, are potent vasoactive agents (Muramatsu et al., 1992; Godin et al., 1995; Hwa et al., 1996; Komuro et al., 1997; Bhattacharya and Cohen, 2000). Protease-induced vascular relaxant and contractile effects are mediated by hydrolysis of protease-activated receptors (PARs), although the role of each of the four PARs cloned to date has not been precisely defined. Protease-induced relaxation was endothelium-dependent (Muramatsu et al., 1992) and mediated predominantly via activation of PAR-2 (Hollenberg et al., 1996; Hwa et al., 1996), whereas protease-induced vasoconstriction was endothelium-independent (Sakiyama et al., 1991) and mediated predominantly via PAR-1 activation (Muramatsu et al., 1992).

Furthermore, thrombin and trypsin exhibited unique pharmacology as vasoactive proteins. Trypsin was 2000-fold more potent as a vasorelaxant than as a vasoconstrictor, whereas thrombin was only 7.8-fold more potent as a relaxant agonist (Bhattacharya and Cohen, 2000). In addition, the time course of thrombin-induced relaxation (t_1/2 = 22 s) differed from trypsin-induced relaxation (t_1/2 = 62 s), although thrombin- and trypsin-induced vasoconstriction was longer (12-14 min), but identical (Bhattacharya and Cohen, 2000). Therefore, thrombin- and trypsin-induced vasorelaxation physiologically and pharmacologically differed from vasoconstriction.

In this study, we further explored the pharmacology of relaxation and contraction to these proteases by using an active site protease inhibitor. LY287045 (Fig. 1), a tripeptide arginal (D-MePhg-Pro-Arg-H H₂SO₄), is a reversible inhibitor of thrombin (Shuman et al., 1992; Smith et al., 1996). The aldehyde carbonyl group of LY287045 forms a reversible hemiacetal bond with the oxygen of serine¹⁹⁵ in the catalytic center of thrombin (Smith et al., 1996).

View larger version (11K): [in this window] [in a new window]   Fig. 1.   Chemical structure of the sulfate salt of the tripeptide-arginal LY287045.

The effect of LY287045 on thrombin-induced relaxation and contraction was compared because the two proteolytic processes are physiologically and pharmacologically different (Bhattacharya and Cohen, 2000). Furthermore, Winn et al. (1993) suggested that protease-induced vasoconstriction may involve both proteolytic and nonproteolytic mechanisms using thrombin as the agonist. In addition, thrombomodulin, a nonactive site thrombin inhibitor, attenuated thrombin-induced vasoconstriction more potently than relaxation (Bhattacharya and Cohen, 2000), whereas argatroban, a thrombin inhibitor binding close to the catalytic site, inhibited thrombin-induced relaxation more potently than contraction (Winn et al., 1993). These studies indicate that thrombin inhibition may not affect thrombin-induced contraction and relaxation similarly. Since the catalytic triad of serine¹⁹⁵, histidine⁵⁷, and aspartate¹⁰² was similar in thrombin and trypsin (Pavone et al., 1998), the effect of LY287045 to alter trypsin's proteolytic activity and trypsin-induced vasomotility in rat and rabbit aorta was also examined. In addition, increases in both thrombin and trypsin have been implicated in angiogenesis (Koshikawa et al., 1997; Tsopanoglou and Maragoudakis, 1999) and serine proteases are thought to play a significant role in the inflammatory responses (Cirino et al., 1996, 2000). Because of the potential importance of serine proteases to several pathologies, we studied the attenuating effect of LY287045 on both thrombin- and trypsin-induced effects.

Thus, the major objectives of this study were to 1) compare the effects of LY287045 on thrombin- and trypsin-induced vascular relaxation and contraction and 2) explore the mechanism of action of LY287045-induced effects. Relaxation was studied using the endothelial intact rat aorta, whereas contraction was examined using endothelium denuded rabbit aorta, two established in vitro models for the study of protease-induced responses (Muramatsu et al., 1992; Godin et al., 1995; Komuro et al., 1997). The most interesting aspect of the present study was an inhibitory effect of LY287045 on thrombin- and trypsin-induced vasorelaxation, but not vasoconstriction. Our mechanistic studies ruled out a role for nitric oxide and cyclooxygenase products like prostacyclin in the marked effect of LY287045 to inhibit selectively thrombin- and trypsin-induced aortic relaxation. Similarly, the selective effect of LY287045 on vasorelaxation was not due to an effect of the compound on the relaxant PAR-2 or due to alteration of norepinephrine-induced contraction prior to examining protease induced relaxation. The novel finding that inhibition of both trypsin and thrombin will not affect all responses to these proteases similarly raises the hope that protease-dependent pathologies could be therapeutically targeted by development of pharmacophores with selectivity and minimal side effects.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Aortic Tissue Preparation. Thoracic aortae were dissected from male New Zealand White rabbits (2-3 kg) (Harlan Sprague-Dawley, Indianapolis, IN) and male Sprague-Dawley rats (0.25-0.35 kg) (Harlan Sprague-Dawley, Indianapolis, IN). Rats were sacrificed by cervical dislocation, while rabbits were euthanized by intravenous injection of a lethal dose of sodium pentobarbital (65-100 mg/kg) into the ear vein according to animal use protocols approved by the Lilly Animal Care and Use Committee. The thoracic aorta was dissected free of surrounding tissue in modified Krebs' buffer (mM): KCl (4.6), KH₂PO₄ (1.2), MgSO₄ (1.2), NaCl (118.2), glucose (10.0), CaCl₂ · 2H₂O (1.6), NaHCO₃ (24.8), and cut into 3- to 5-mm rings. For endothelium denudation, rat aortic rings were rotated 10 times on fine-point serrated forceps. Tissues were then placed between two stainless steel hooks and mounted in 10-ml organ baths filled with buffer solution. Baths were maintained at 37°C and bubbled with a 95:5% O₂:CO₂ mixture (pH 7.4). Tissues were equilibrated for 1 h, and optimum passive force was produced by successively increasing the initial force to 6 g in each tissue with intermittent tissue washes.

Aortic Contractile Responses. Contractile responses were examined in endothelial-denuded rabbit aorta, a vascular model previously used to explore protease-induced contractile activity (Sakiyama et al., 1991; Godin et al., 1995; Komuro et al., 1997; Bhattacharya and Cohen, 2000). Presence or absence of endothelium was determined by adding carbamylcholine (10⁶ M) to tissues precontracted to steady state with norepinephrine (10⁷ M). For each concentration of protease, noncumulative contraction was measured at steady state and was expressed as percentage of the maximal force to KCl (67 mM) generated initially in each tissue. Contraction to only one concentration of thrombin or trypsin was generated in each tissue. In some specified experiments, two protease contractile responses were generated in the same tissue separated by 90 min to establish the extent to which tachyphylaxis occurred in the aorta.

Aortic Relaxant Responses. Relaxant responses to the proteases were examined in endothelial-intact rat aorta, a vascular model used to explore protease-induced relaxant activity (Muramatsu et al., 1992). The presence or absence of endothelium was determined by adding carbamylcholine (10⁶ M) to tissues precontracted to steady state with norepinephrine (10⁷ M). Tissues were contracted with norepinephrine (10⁷ M) to a steady state, followed by a single concentration of thrombin, trypsin, or PAR-2 activating peptide (PAR2-AP; H-Ser-Leu-Ile-Gly-Arg-Leu-OH). Carbamylcholine and prostacyclin-induced relaxant responses were measured cumulatively and noncumulatively, respectively, after the tissues were contracted to steady state with phenylephrine (3 × 10⁷ M). Maximal relaxation for each concentration of agonist was expressed as the percentage decrease in norepinephrine- or phenylephrine-induced force. Relaxation to only one concentration of thrombin or trypsin was generated in each tissue.

Effect of LY287045, L-NAME, or Indomethacin on Protease-Induced Aortic Responses. LY287045 was incubated with the tissues for 60 min at 37°C before addition of thrombin, trypsin, carbamylcholine, or PAR2-AP. In some experiments, thrombin was incubated with LY287045 or vehicle (control responses) for 30 min at 35°C before tissue exposure. Both L-NAME (10⁴ M) and indomethacin (10⁵ M) were incubated with the tissues for 30 min at 37°C before addition of the agonist.

Inhibition of Thrombin and Trypsin. The ability of various concentrations of LY287045 to inhibit thrombin (5.9 nM)- and trypsin (1.4 nM)-induced hydrolysis of the chromogenic substrate Bz-Phe-Val-Arg-pNA was determined at room temperature from kinetic studies using 0.19 mM substrate and a ThermoMax plate reader (Molecular Devices, San Francisco, CA). The hydrolysis kinetics were monitored for rates of p-nitroanilide formation, and the affinity of LY287045 for each protease was calculated from the inhibition kinetics as apparent association constants, K_assoc values, as previously described (Smith et al., 1996; Sall et al., 1997).

Aortic Data Acquisition and Analysis. For all experiments, isometric force was measured with Sensotec transducers coupled to MP100 data acquisition software (BIOPAC Systems, Inc., Santa Barbara, CA). Data were analyzed off-line and expressed as the mean ± S.E.M. Data represent aortic responses from the number of animals indicated with the number of tissues shown in parentheses. Statistical comparisons were performed with Student's t test using SigmaStat software. Differences between mean values were considered statistically significant when p < 0.05.

Proteins and Chemicals. Norepinephrine, phenylephrine, indomethacin, L-NAME, carbamylcholine, and porcine trypsin were obtained from Sigma (St. Louis, MO). Human and bovine -thrombin were purchased from Enzyme Research (South Bend, IN), and bovine trypsin was purchased from Worthington Biochemicals (Lakewood, NJ). The PAR2-AP (H-Ser-Leu-Ile-Gly-Arg-Leu-OH) was purchased from BACHEM (King of Prussia, PA). LY287045 was synthesized in the Lilly Research Laboratories. Chromogenic substrate Bz-Phe-Val-Arg-pNA was purchased from Midwest Biotech (Fishers, IN).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Comparison of Aortic Contraction to Repeated Administration of Thrombin and Trypsin. Thrombin- and trypsin-induced vasomotility is mediated by proteolysis of contractile PARs, and based on this, is subject to the development of tachyphylaxis (Sakiyama et al., 1991). To understand the extent of tachyphylaxis to both thrombin- and trypsin-induced contraction in the aorta, we studied the effect of repeated administration of equieffective concentrations of thrombin (10⁷ M) and trypsin (10⁶ M) in endothelium-denuded rabbit aorta. Although thrombin (10⁷ M) and trypsin (10⁶ M) produced marked vascular contractions initially, a second challenge of thrombin and trypsin, respectively, did not induce a significant increase in aortic tone (Fig. 2). Moreover, thrombin failed to contract tissues previously exposed to trypsin. Surprisingly, in contrast to these results, trypsin (10⁶ M) elicited significant vasoconstriction (54.7 ± 4.9%) after previous contraction to thrombin (10⁷ M) (46.4 ± 10.9%). Thus, trypsin is capable of producing a contraction after thrombin-induced proteolysis, raising the possibility that trypsin-induced aortic contraction may involve additional proteolytic or nonproteolytic mechanisms. Because of these data documenting tachyphylaxis to both thrombin and trypsin, subsequent protocols used only a single thrombin or trypsin response in each tissue.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Effect of repeated administration of thrombin (107 M) and trypsin (106 M) in endothelial denuded rabbit aorta. Contraction produced by an initial challenge of thrombin or trypsin (filled bars) and that followed by a second challenge of the serine proteases (open bars) was represented as a percentage of KCl (67 mM)-induced maximal force (3.8 ± 0.4 g). Endothelial denudation was confirmed by the loss of carbamylcholine (106 M)-induced relaxation (3.8 ± 2.1%) in norepinephrine (107 M)-treated tissues contracted to a steady state (3.7 ± 0.2 g). Bars are mean values, and vertical lines represent the standard error of the mean of 3 tissues from 3 rabbits.

Effect of LY287045 on Thrombin-Induced Relaxation and Contraction. LY287045 (10⁷ M) significantly attenuated thrombin-induced relaxation in endothelium-intact rat aorta (Fig. 3; top). The EC₃₀ of thrombin-induced relaxation was 3.0 × 10⁹ M, whereas the EC₃₀ of thrombin in the presence of LY287045 (10⁷ M) was 7.2 × 10⁸ M. Therefore, the negative logarithm of the apparent antagonist dissociation constant of LY287045 (log K_B) was 8.4. The presence of vascular endothelium in these tissues was established by carbamylcholine (10⁶ M)-induced relaxation of norepinephrine (10⁷ M)-contracted tissues (Fig. 3, top; inset). In contrast to its marked effect on thrombin-induced relaxation, LY287045 (10⁷ M) did not inhibit thrombin-induced contraction (data not shown). Even higher concentrations (10⁶ M) of LY287045 did not inhibit thrombin-induced contraction in endothelium-denuded rabbit aorta (Fig. 3; bottom). Endothelium denudation was established by the inability of carbamylcholine (10⁶ M) to relax norepinephrine (10⁷ M)-contracted tissues (Fig. 3, bottom; inset).

View larger version (29K): [in this window] [in a new window]   Fig. 3.   Effect of LY287045 on thrombin-induced relaxation in endothelium-intact rat aorta (top) and contraction in endothelium-denuded rabbit aorta (bottom). Relaxant effects were represented as the percentage of norepinephrine (107 M)-induced force (1.8 ± 0.06 g) (top), whereas contractile effects were represented as the percentage of KCl (67 mM)-induced maximal contraction (4.8 ± 0.2 g) (bottom), in vehicle or LY287045-treated tissues. The status of the endothelium was determined by carbamylcholine (106 M)-induced relaxation of norepinephrine (107 M)-contracted tissues (insets). Points are mean values, and vertical lines represent the standard error of the mean. The number of animals and number of rings (in parentheses) are indicated for each point. *p < 0.05.

7

7

6

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Effect of preincubation of LY287045 (107 M) (white bar) or vehicle (black bar) with thrombin (107 M) on relaxation (left) to thrombin in endothelium-intact rat aortic rings graphed as a percentage of norepinephrine (107 M)-induced force (2.1 ± 0.1 g). Endothelial function was assessed by carbamylcholine (106 M)-induced relaxation (84.2 ± 3.1%) of norepinephrine-contracted (107 M) tissues. Effect of preincubation of LY287045 [107 M (white bar) and 106 M (hatched bar)] or vehicle (black bar) with thrombin (107 M) on contraction (right) to thrombin in endothelium denuded rabbit aorta. Contraction was represented as a percentage of KCl (67 mM)-induced maximal contraction (5.9 ± 0.3 g). Endothelial function was assessed by carbamylcholine (106 M)-induced relaxation (2.6% ± 1.2%) of norepinephrine (107 M)-contracted tissues. Bars are mean values, and vertical lines represent the standard error of the mean for the number of animals and number of rings indicated in parentheses.

Effect of LY287045 on Trypsin-Induced Relaxation and Contraction. The effect of LY287045 was next studied on trypsin-induced vasomotility because the active site catalytic triad is similar between thrombin and trypsin. As with thrombin, trypsin-induced vascular relaxation was significantly inhibited by LY287045 (10⁷ M) (Fig. 5; top). The negative logarithm of the apparent dissociation constant of LY287045 in attenuating trypsin-induced relaxation was 8.6 (log K_B), similar to that for inhibition of thrombin-induced relaxation. Furthermore, LY287045 (10⁶ M) did not inhibit trypsin-induced vasoconstriction consistent with its inability to inhibit thrombin-induced contraction (Fig. 5, bottom). Most surprisingly, LY287045 (10⁶ M) potentiated trypsin-induced contraction. For both relaxation and contraction, the status of the vascular endothelium was determined by carbamylcholine (10⁶ M)-induced relaxation (Fig. 5; insets).

View larger version (31K): [in this window] [in a new window]   Fig. 5.   Effect of LY287045 on trypsin-induced relaxation in endothelium-intact rat aorta (top) and contraction in endothelium-denuded rabbit aorta (bottom). Relaxant effects were represented as a percentage of norepinephrine (107 M)-induced force (1.5 ± 0.05 g) (top), whereas contractile effects were represented as a percentage of KCl (67 mM)-induced maximal contraction (4.3 ± 0.3 g) (bottom), in vehicle or LY287045-treated tissues. The status of the endothelium was determined by carbamylcholine (106 M)-induced relaxation of norepinephrine (107 M)-contracted tissues (insets). Points are mean values, and vertical lines represent the standard error of the mean. The number of animals and number of rings (in parentheses) are indicated for each point. * p < 0.05.

Effect of LY287045 on Thrombin- and Trypsin-Induced Proteolysis. Because LY287405 inhibited relaxation to both thrombin and trypsin, we wanted to establish the relative activity of LY287405 as an inhibitor of proteolysis induced by thrombin and trypsin. The inhibition of LY287045 on thrombin and trypsin's proteolytic activity is depicted in Table 1 as association constants (K_assoc) of LY287045 with the serine proteases. The higher the K_assoc values, the better the association of LY287045 with thrombin or trypsin. As seen in Table 1, LY287045 inhibited the proteolytic activity of thrombin as well as trypsin. Thus, LY287045 was indeed a thrombin and a trypsin inhibitor, consistent with the effects of LY287045 on thrombin- and trypsin-induced relaxation.

Effect of LY287045 on the Nitric Oxide Pathway. Since LY287045 (10⁷ M) inhibited protease-induced vascular relaxation without attenuating protease-induced vascular contraction, we considered the possibility that LY287045 might be inhibiting nitric oxide formation, action, or release since thrombin-induced relaxation is thought to be mediated by an endothelium-dependent mechanism. Carbamylcholine-induced relaxation in rat aorta also results from nitric oxide formation and release (Khan et al., 1992) and L-NAME, a nitric oxide synthase inhibitor, attenuated cholinergic-dependent vasorelaxation (Cox et al., 1995; Hamilton et al., 1999). Thus, the effect of LY287045 on carbamylcholine-induced vascular relaxation was examined to obtain an independent assessment of the effect of LY287045 on nitric oxide-mediated relaxation. LY287045 (10⁷ M and 10⁶ M) did not inhibit carbamylcholine-induced relaxation of the rat aorta (Fig. 6), suggesting that LY287045 did not affect mechanisms associated with nitric oxide-induced relaxation in rat aorta.

View larger version (22K): [in this window] [in a new window]   Fig. 6.   Effect of LY287045 or vehicle on carbamylcholine-induced relaxation of phenylephrine (3 × 107 M)-contracted rat aortic rings (1.07 ± 0.08 g) with intact endothelium. The presence of functional endothelium was confirmed by the ability of carbamylcholine (106 M) to relax (81.3 ± 5.6%) norepinephrine (107 M)-contracted tissues (0.93 ± 0.09 g). Points are mean values, and vertical lines represent the standard error of the mean. The number of animals and number of rings (in parentheses) are indicated for each point.

4

6

View larger version (16K): [in this window] [in a new window]   Fig. 7.   Effect of L-NAME (104 M) and LY287045 (106 M) on trypsin (106 M)-induced contraction in endothelium denuded rabbit aorta. Vasoconstriction was expressed as the percentage of KCl (67 mM)-induced maximal contraction (3.7 ± 0.3 g). Bars are mean values and vertical lines represent the standard error of the mean of the number of animals and number of rings indicated in parentheses. *p < 0.05.

Effect of LY287045 on the Prostacyclin Pathway. We next considered the possibility that LY287045 inhibited relaxation to thrombin and trypsin in the rat aorta by a direct effect to alter prostacyclin formation, release, or activity. This was examined in both the rat and rabbit aorta. First, we examined the ability of prostacyclin itself to relax the endothelium intact rat aorta. Prostacyclin did not significantly relax the rat aorta, in the presence or absence of LY287045 (10⁷ M), although marked relaxation could be demonstrated to carbamylcholine (10⁶ M) in the same tissues (Fig. 8).

View larger version (24K): [in this window] [in a new window]   Fig. 8.   Effect of prostacyclin with and without LY287045 (107 M) to relax rat aorta with intact endothelium (top). Relaxation was expressed as the percentage of phenylephrine (3 × 107 M)-induced force. The status of the endothelium was determined by carbamylcholine (106 M)-induced relaxation of norepinephrine (107 M)-contracted tissues (insets). Effect of indomethacin (105 M) on trypsin-induced contraction in endothelium-denuded rabbit aorta (bottom). Contractile responses to trypsin (106 M), either with or without indomethacin, were expressed as the percentage of KCl (67 mM)-induced maximal contraction (3.8 ± 0.1 g). Points (top) or bars (bottom) are mean values, and vertical lines represent the standard error of the mean for the number of animals and number of rings (in parentheses) indicated.

5

5

6

Effect of LY287045 on PAR-2-Activating Peptide-Induced Aortic Relaxation. Since PAR-2 is the protease receptor thought to play a predominant role in vascular relaxation (Hwa et al., 1996), we tested the possibility that LY287045 had selective inhibitory effect on relaxation due to antagonism of PAR-2. A peptide sequence corresponding to the tethered PAR-2 sequence (PAR2-AP) was used to induce vascular relaxation in endothelium-intact rat aorta (Fig. 9). PAR2-AP (10⁴ M and 5 × 10⁴ M) relaxed the rat aorta, although it was less potent than trypsin or thrombin as previously reported (Hollenberg et al., 1996). However, LY287045 (10⁷ M) did not alter PAR2-AP-induced vascular relaxation, demonstrating that LY287045 had no effect on PAR-2. Therefore, the ability of LY287045 to selectively inhibit protease-induced relaxation was not a result of the direct inhibition of the protease receptor thought to be responsible for relaxation.

View larger version (25K): [in this window] [in a new window]   Fig. 9.   Effect of LY287045 (107 M) on relaxation to PAR2-AP in endothelial-intact rat aorta. Relaxation was expressed as a percentage of norepinephrine (107 M)-induced force (1.33 ± 0.05 g). Endothelial function was confirmed by the ability of carbamylcholine (106 M) to relax (75.5 ± 2.9%) norepinephrine (107 M)-contracted tissues (1.2 ± 0.05 g). Bars are mean values, and vertical lines represent the standard error of the mean of the number of animals and number of rings indicated in parentheses.

Effect of LY287045 on Norepinephrine-Induced Contraction in Rat Aorta. Since LY287045 possessed a selective effect to inhibit protease-induced relaxation, we wanted to rule out the possibility that this was an artifact related to differences in the contractile tension developed to norepinephrine in the presence and absence of LY287045. Tissues contracted to greater force will relax less than tissues contracted with less force (Cohen and Berkowitz, 1974). For this reason, it becomes important to ensure that LY287045 did not enhance norepinephrine-induced force in the rat aorta before the addition of the relaxant proteases. As seen in Table 2, tissues exposed to LY287045 (10⁷ M) or vehicle, contracted to a similar force in response to norepinephrine (10⁷ M), thus ruling out the possible effect of LY287045 on tissue contractility (Table 2).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

LY287045 selectively attenuated the proteolytic activity of both thrombin and trypsin (Table 1), but not other serine protease-like factors Xa, XIa, XIIa, and kallikrein (Smith et al., 1996). LY287045 inhibited trypsin and thrombin by virtue of its ability to bind reversibly to the active site of the proteases (Smith et al., 1996), in contrast to D-phenylalanyl-L-prolyl-L-arginine chloromethyl ketone (PPACK), a tripeptide that covalently and irreversibly inhibits trypsin and thrombin proteolysis (Bode et al., 1992) and argatroban, an arginine-based inhibitor that binds competitively and selectively to thrombin, but not trypsin (Winn et al., 1993). In fact, argatroban has been proposed to interact with a site distinct from the active site, but close to the catalytic area that can mask the enzymatic activity of thrombin (Kikumoto et al., 1984). Although the effects of PPACK and argatroban on protease-induced vascular activity have been studied (Ku, 1986; Winn et al., 1993), the fact that thrombin inhibitors can vary by structure, protease selectivity, and binding characteristics has led to our studies with LY287045 to further understand the vascular consequence of inhibition of both proteases.

LY287045 (10⁷ M) attenuated both thrombin- and trypsin-induced relaxation whereas an even higher concentration of LY287045 (10⁶ M) did not inhibit either thrombin- or trypsin-induced contraction (Figs. 3 and 5). The effect of LY287045 to inhibit relaxation did not change with preincubation of LY287045 and thrombin, consistent with previous information documenting a time-independent and rapid association of LY287045 with thrombin (Smith et al., 1996). The fact that LY287045 selectively inhibited thrombin-induced relaxation, but not contraction, was consistent with the effects of argatroban (Winn et al., 1993), which was more potent in inhibiting thrombin-induced relaxation than contraction. Unlike LY287045, argatroban did not inhibit trypsin-induced relaxation (Winn et al., 1993), consistent with its selectivity as a proteolytic inhibitor of thrombin (Kikumoto et al., 1984; Hauptmann and Sturzebecher, 1999).

The ability of LY287045 to inhibit thrombin-induced relaxation, but not contraction, contrasts with thrombomodulin (Bhattacharya and Cohen, 2000), which potently inhibited vasoconstriction. The ability of thrombomodulin to selectively alter thrombin-induced contraction was attributed to a dominant role of the exosite-1 domain in contraction, the proposed site for thrombomodulin/thrombin and PAR-1/thrombin interactions (Bhattacharya and Cohen, 2000). The fact that LY287045, an active site inhibitor dramatically blocked vasorelaxation, but not contraction is consistent with the possibility that the active site of thrombin is critical for relaxation, whereas other sites on these proteases may be more critical for contraction.

Additional studies were directed toward understanding how LY287045 could inhibit protease-induced relaxation, but not contraction by ruling out the possibility that LY287045 possessed additional pharmacological attributes that explain its selective ability to inhibit protease-induced vascular relaxation. Two approaches were taken to study the effect of LY287045 on the nitric oxide pathway, since thrombin- and trypsin-induced vascular relaxation is endothelium-dependent (Muramatsu et al., 1992; Hwa et al., 1996). First, LY287045 (10⁷ M and 10⁶ M) did not inhibit carbamylcholine-induced relaxation, a response mediated by nitric oxide release (Khan et al., 1992), suggesting that LY287045 was not inhibiting nitric oxide synthase or nitric oxide-induced relaxation. We also considered the possibility that inhibition of nonendothelial nitric oxide by LY287045 might enhance contraction to thrombin and trypsin in de-endothelialized rabbit aorta. Nonendothelial sources of nitric oxide have been proposed in blood vessels (Ignarro et al., 1999). If LY287045 were inhibiting the nitric oxide pathway, vascular contractility to trypsin and thrombin would be enhanced, opposing the inhibition of LY287045 on protease-induced contraction. If this hypothesis were valid, then L-NAME (10⁴ M), a nitric oxide synthase inhibitor, should potentiate trypsin-induced vasoconstriction. However, this did not occur, suggesting that inhibition of nitric oxide did not alter rabbit aortic contractility to trypsin. Thus, in both the rat and rabbit aorta, it is unlikely that LY287045 exerted a direct effect on the nitric oxide pathway.

Next, we considered the possibility that LY287045 could be inhibiting prostacyclin formation, or release. Prostacyclin relaxed many vascular beds (Trachte, 1986; Malomvolgyi et al., 1988) and was proposed to be involved in trypsin-induced vascular effects (Savion and Naveh-Floman, 1985). However, prostacyclin did not relax the rat aorta, and indomethacin (10⁵ M), a cyclooxygenase inhibitor that blocked prostacyclin synthesis, did not potentiate trypsin-induced vascular contractility. The inability to demonstrate aortic relaxation to prostacyclin was consistent with other reports using rat aorta (Borda et al., 1983; Dam et al., 1986). Thus, prostacyclin did not play a major role in aortic relaxation or contraction to these proteases, ruling out the possibility that the selective effects of LY287045 on vascular relaxation could be attributed to any direct effect of LY287045 on the prostacyclin pathway.

Next, we considered the possibility that LY287045 might exert a direct effect to block PAR-2, the receptor proposed to mediate vascular relaxation (Hollenberg et al., 1996; Hwa et al., 1996). However, LY287045 (10⁷ M), in a concentration that inhibited relaxation to both thrombin and trypsin, did not inhibit relaxation to the PAR2-AP. These data rule out the possibility that the selective effect of LY287045 to inhibit protease-induced vascular relaxation resulted from its ability to inhibit PAR-2 activation.

Lastly, we ruled out the possibility that the selective effect of LY287045 to inhibit protease-induced relaxation was related to differences in the contractile force of norepinephrine. Relaxation of vascular tissue is inversely related to initial contractile force (Cohen and Berkowitz, 1974). However, LY287045 (10⁷ M) did not alter norepinephrine (10⁷ M)-induced force, since thrombin and trypsin-induced vascular relaxation was endothelium-dependent (Muramatsu et al., 1992; Hwa et al., 1996).

To explain the selective effect of LY287045 to inhibit thrombin-induced relaxation, it is possible that LY287045 may be altering a component of the endothelium, namely endothelium-derived hyperpolarizing factor. Recently, endothelium-derived hyperpolarizing factor has been suggested to play a role in thrombin- and trypsin-induced vascular relaxation (Hamilton and Cocks, 2000). In addition, it is possible that the inability of LY287045 to inhibit contraction to trypsin and thrombin may be related to the difference in time course of contraction relative to relaxation. The possibility exists that LY287045 might dissociate from thrombin and trypsin over the time course required for vascular contraction, resulting in reduced or no inhibition, especially in light of the fact that PPACK, an irreversible thrombin inhibitor, attenuated vasoconstriction (Winn et al., 1993). It would be judicious to study a slowly reversible and a tightly bound active site inhibitor of thrombin and/or trypsin to test the hypothesis of time-dependent dissociation of LY287045 from the protease catalytic pocket. Alternatively, protease activity may be substrate sensitive such that PAR-2, but not PAR-1, is restricted from active site proteolysis by LY287045. A careful observation of the amino-terminal exo domain of PAR-1 and PAR-2 reveals that the extracellular amino acids are not conserved between the two receptors (Nystedt et al., 1994) and may be folded differently, resulting in a varying degree of accessibility of the cleavage site to the active site of thrombin or trypsin. Lastly, the possibility that thrombin and trypsin may induce vascular contraction via a mechanism that, in part, involves a nonproteolytic effect exists and remains to be carefully studied. Repeated challenge with thrombin and trypsin indicated that trypsin produced vasoconstriction after tachyphylaxis had been demonstrated to thrombin (Fig. 2), raising the possibility that trypsin can, in part, contract vascular tissue by a nonproteolytic mechanism. Although the mechanism by which LY287045 selectively inhibits vascular relaxation, but not contraction, remains to be understood, LY287045 provides a useful tool to probe the vasodilatory effects of thrombin and trypsin without altering vasoconstrictor effects.

These studies suggest that LY287045 may be useful in diseases associated with high protease levels and a predominant hypotensive effect. In this regard, proinflammatory cytokine treatment has been documented to up-regulate relaxant PARs (Cirino et al., 2000; Cocks and Moffatt, 2000), and hypotension is a well documented effect of the inflammatory response associated with endotoxemia (Cicala et al., 1999). To the extent that thrombin- and trypsin-induced vascular relaxation may be associated with hypotensive inflammatory responses, LY287045 may serve as a useful adjunct to anti-inflammatory therapy. As a matter of fact, the augmentation of trypsin-induced vasoconstriction in the presence of LY287045 (Fig. 5) suggests that LY287045 or similar molecules may be useful in inflammatory diseases associated with hypotension due to the dual ability to suppress relaxation and potentiate contraction of blood vessels. In addition, a role for thrombin and trypsin in angiogenesis is well documented (Koshikawa et al., 1997; Tsopanoglou and Maragoudakis, 1999). Angiogenesis is known to be associated with activation of vascular endothelial growth factor (Tsopanoglou and Maragoudakis, 1999), and endothelial cell-derived trypsin has also been implicated in tumor angiogenesis (Koshikawa et al., 1997). To the extent that endothelial growth and formation in developing capillaries are promoted by thrombin and trypsin, both from the perspective of enhancing growth factor activity and possibly via relaxation of formed vasculature, LY287045 which inhibits both trypsin- and thrombin-induced endothelial effects might be an effective antiangiogenic agent.

In summary, these studies have defined a unique property of the arginal tripeptide LY287045 as a selective inhibitor of both thrombin- and trypsin-induced vasorelaxation. The fact that this agent, while acting at the active site of thrombin and trypsin (Smith et al., 1996), did not inhibit all trypsin and thrombin responses coupled to previous studies with thrombomodulin, argatroban and PPACK, support the contention that protease inhibition can result in inhibition of only certain selective responses to trypsin and thrombin and not others. This observation raises the possibility that protease inhibitors can be developed to antagonize some, but not all, actions of trypsin and thrombin. These studies with LY287045 also suggest that this agent may be a useful tool to probe the mechanisms for the interaction of trypsin and thrombin with the PARs associated with contraction and relaxation.