INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

HEME OXYGENASE (HO) is an enzyme responsible for the degradation of protoheme IX to form biliverdin, free iron, and carbon monoxide (CO). The enzyme exists in three isoforms: HO-1 (28), HO-2 (15), and HO-3 (16). Among them, HO-1, which is identical to heat shock protein 32, is induced by a variety of stimuli such as cytokines, heavy metals, and oxidants (14). This inducible isoform can also be upregulated by protoheme IX (iron protoporphyrin IX) (15), the substrate for HO. Approximately 70% of the heme degraded in the body is derived from hemoglobin from senescent erythrocytes (14).

Liver is one of the major organs in which the heme molecules are oxidatively degraded by HO, and its capacity to decompose the heme is increased by the HO-1 induction. Recent studies also revealed that induction of HO-1 occurred in the liver in response to endotoxemia (6) and hemorrhagic shock (2). The hepatic HO-1 expression could also be markedly elevated when amounts of free heme molecules in circulation are acutely increased. Such circumstances involve massive hemolysis by sepsis (27), excessive blood transfusion, and splenectomy (35).

Recently, several lines of evidence indicated that CO produced by the HO reaction contributes to regulation of cell functions by activating soluble guanylate cyclase (4, 36) or by hyperpolarizing membrane potentials through the mechanisms involving stimulation of potassium channels (21). Although these previous studies suggest a possible role of the endogenous CO-generating system in regulation of cell and organ functions, the relationship between actual CO amounts released through the endogenous origins and resultant influences on the cell function have not yet been fully investigated. We have shown in the liver that CO endogenously generated by the HO reaction is necessary to maintain sinusoidal vessels in a relaxing state (32, 33). Administration of zinc protoporphyrin IX (ZnPP), a potent inhibitor of HO, or of free oxyhemoglobin (HbO₂), a trapping agent of CO, markedly increased the vascular resistance concurrent with sinusoidal constriction. HbO₂ can diffuse across the fenestrated endothelium of the liver into the space of Disse and thereby entrap CO in the extrasinusoidal space. Considering that hepatocytes that constitutively express HO-2 are the major sites for CO generation, these results suggested that CO generated by HO-2 serves as a regulator of sinusoidal tone under ordinary conditions (10). Such a physiological role of CO constitutively generated by HO-2 raised a question as to whether the liver undergoing stress conditions alters its vascular function through overproduction of CO as a consequence of the HO-1 induction. The present study was thus undertaken to demonstrate time history and topographic distribution of the HO-1 expression, their relationship with quantitative evaluation of local CO generation, and its functional link to vascular tone.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Reagents. Anti-rat HO-1 and HO-2 antibodies were prepared and purified as described elsewhere (10). ZnPP was purchased from Aldrich Chemical (Milwaukee, WI). Other reagents used were purchased from Sigma Chemical (St. Louis, MO) unless otherwise mentioned.

Preparation of hemoglobin and liposome-encapsulated hemoglobin vesicles. To examine the role of endogenously generated CO in the regulation of vascular tone in the liver, free and liposome-encapsulated hemoglobins, hereafter designated as free Hb and HbV, respectively, were used as a tool to trap CO in and around the sinusoidal space. Free Hb, HbV, and methemoglobin (metHb) were prepared using outdated human erythrocytes as described previously (23, 24). metHb was used as a tool to scavenge nitric oxide (NO) but not CO. The mean value of the diameter of HbV used in the present study was ~0.25 µm. Previous studies (22) revealed that, when administered into the systemic circulation, HbV with this size was partly captured in tissue macrophages including Kupffer cells but did not enter the hepatic parenchyma. On the other hand, free Hb was readily trapped by the parenchymal cells because of the presence of fenestration in sinusoidal endothelium, the diameter of which ranges from 0.10 to 0.15 µm (18, 37). The oxygen-binding affinity exhibited no substantial differences between HbV and free Hb, as described previously (23). Both free Hb and HbV were used for experiments within 7 days after the preparation procedures.

Animal preparation and hemin treatment. Male Wistar rats (260-280 g) were obtained from Saitama Animal Laboratory (Saitama, Japan). All animals were allowed free access to laboratory chow and tap water. Rats were pretreated with hemin via an intraperitoneal injection (40 µmol/kg). They were anesthetized intraperitoneally with pentobarbital sodium (50 mg/kg), and the common bile duct was cannulated by a polyethylene (PE)-10 catheter. The livers were excised and perfused with Hb- and albumin-free Krebs-Henseleit bicarbonate-buffered solution (pH 7.4, 37°C) gassed 95% O₂-5% CO₂ containing 30 µM taurocholate as described previously (25, 40). The perfusate was pumped through the liver with a peristaltic pump at a constant rate of 4.0 ml · min¹ · g liver weight¹ in a single pass mode while the inlet perfusion pressure was monitored (40).

Experimental protocols. Five main protocols were employed to compare the differences in the vascular resistance between the control and the hemin-treated livers. Livers in the first, second, and third groups were treated with HbO₂, HbV-O₂, and metHb, respectively. Concentrations of the Hb derivatives were all 0.5 g/dl and were usually perfused for 10 min. In the fourth group, 1 µM ZnPP, an HO inhibitor, was applied to the perfusate. In the fifth group, the same concentration of ZnPP was perfused with CO supplemented at 4 µM, the concentration that can sufficiently compensate the CO suppression by ZnPP. Effects of these reagents on the vascular resistance were also examined in livers pretreated with endothelin-1 (ET-1); in these experiments, the reagents were added to the perfusate for 5 min followed by the infusion of 1 nM ET-1 for 1 min. The dose of ET-1 was estimated to be ~30 pmol/min for 1 min. This dose of ET-1 was reported to evoke site-specific constriction of microvessels including portal venules and sinusoids but not of larger portal veins (11, 34). We confirmed that no significant reduction of bile flow was observed after injecting this dose of ET-1 (data not shown). This observation convinced us that there was little involvement if any of vasoconstriction at the level of the portal veins. In some experiments, desired concentrations of N-nitro-L-arginine methyl ester (L-NAME) were added to the perfusate to suppress endogenously generated NO. ET-1 and L-NAME were purchased from Sigma. In all experiments, bile samples were collected every 5 min to monitor the bile output.

Digital microfluorography of sinusoids in perfused liver. To examine alterations in the diameter of sinusoids, digital microangiography was carried out in perfused rat livers by using real-time laser confocal microscopy as described elsewhere (10, 29). Briefly, the surface of the perfused liver was observed through an inverted-type intravital microscope assisted by a line-scan laser confocal imager (Insight/TMD300, Meridian Far East, Tokyo, Japan), and the hepatic microcirculation was visualized by a 10-bit color chilled, intensified charge-coupled device camera (C5810-01, Hamamatsu Photonics, Hamamatsu). This system allowed us to acquire the confocal fluorescence images at a video-rate speed (30 frames/s); images were optically sliced to the desired width, which ranged from 1 to 20 µm. To visualize the hepatic sinusoidal vessels, 100 µl of 1% of FITC-dextran solution were injected transportally every 2 or 5 min after the start of experiments, and the microvasculature was visualized by epi-illumination at 488 nm using an argon laser light source. The width of the confocal plane along the z-axis was set up to 2 µm to collect the fluorescent images.

Determinations. Rat liver homogenates were subjected to SDS-PAGE, transferred to a Millipore Immobilon polyvinylidene difluoride transfer membrane, and then served as samples for Western blot analysis, as described previously (10). HO-1 and HO-2 proteins were visualized using specific antibodies against rat HO-1 and HO-2, namely, GTS-1 and GTS-2, respectively (10).

Immunohistochemistry of liver HO-1 and HO-2. Frozen sections (7 µm thick) were prepared from rat livers. After fixation with acetone, the sections were treated with avidin, biotin, and normal horse serum to minimize the nonspecific staining. These tissues were incubated with monoclonal antibodies to GTS-1 or GTS-2, dissolved in 1% BSA-PBS at a final concentration of 1 µg/ml, for at least 2 h at 25°C. After several washes with PBS, the sections were stained with anti-mouse IgG for 1 h (Vectastain Elite ABC kit, Vector, CA). To prevent endogenous peroxidase reactions, the samples were pretreated with 0.3% H₂O₂ in cold methanol for 30 min and subsequently incubated with avidin and horseradish peroxidase-conjugated biotin for 30 min. Finally, 0.1 mg/ml of 3,3'-diaminobenzidine tetrahydrochloride were applied to sections for 5 min. The sections were counterstained with methyl green after fixation with 20% formaldehyde for 20 min, and slides were coverslipped with aqueous mounting medium. The immunohistochemical staining patterns were examined using the samples collected from at least three individual rats. To confirm the specificity of immunohistochemical localization, antibodies preabsorbed with an excess of adequate antigens in advance were used.

Statistical analyses. All data in the present study are expressed as means ± SE of measurements. Differences in the mean values among the groups were analyzed by one-way ANOVA combined with Scheffé's-type multiple comparison test. P values <0.05 were considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

HO-1 expression and CO generation in the hemin-treated liver. Figure 1 illustrates time history of the HO-1 expression in the liver of rats treated with an intraperitoneal injection of 40 µmol/kg hemin. As shown in Fig. 1A, the HO activity in the tissue increased time dependently and reached a maximum at 12 h, plateaued until 18 h, and then was followed by a reduction at 24 h. Results for the immunoreactive HO-1 protein also displayed a time course similar to the results for HO activity, as assessed by Western blot analysis (Fig. 1B). Because the expression of HO-2, as assessed by Western blot analysis (data not shown), showed no significant changes during the 18-h period, these results indicate that an increase in the HO activity is attributable to the increase in HO-1 protein expression. On the other hand, the time course of the CO flux in the venous effluents exhibited quite different pictures, as indicated in Fig. 1C. At 6 h, the flux displayed no significant elevation, in contrast to the HO-1 activity, which was 2.7-fold greater than that in the control, as shown in Fig. 1A. However, at 12 h, when the enzyme activity reached a maximum level, the CO flux did not show any significant changes. However, at 18 h, the CO flux exhibited a marked elevation, being as much as 4.5-fold greater than the control flux. The elevation was markedly repressed at 24 h, at which time HO-1 protein expression exhibited a fall nearly to control levels (Fig. 1B).

View larger version (23K): [in this window] [in a new window]   Fig. 1.   Time history of the heme oxygenase (HO) activity, the protein expression of HO-1, and carbon monoxide (CO) generation in livers from the hemin-treated rats. A-C: temporal alterations in the microsomal enzyme activity, Western blotting analysis, and flux of CO measured in the venous samples collected from the perfused liver, respectively. D: time course of contents of cytochrome P-450 per mg microsomal protein. Note that a marked increase in the CO flux occurred at 18 h, whereas the enzyme expression and activity reached a maximum level at 12 h. Data in A, C, and D represent means ± SE of measurements from 4-5 separate livers. * P < 0.05 compared with control values measured at time 0.  P < 0.05 compared with values at 12 h. # P < 0.05 compared with values at 18 h.

Reduction of the basal vascular resistance in the CO-overproducing liver. We then investigated whether the liver undergoing the HO-1 induction alters its organ functions, such as vascular resistance and bile formation. As seen in Table 1, these functional parameters exhibited no significant changes in the 12-h hemin-treated liver. However, the liver exposed to the 18-h hemin treatment displayed a 13% reduction of the basal vascular resistance. Bile output was elevated concurrently with reduction of the resistance, showing a 10% increase vs. control values. Although small, these changes were statistically significant. We also examined differences in the biliary flux of bilirubin IX as an index of the HO-mediated heme degradation, which became 4.8-fold greater at 18 h than that shown in the controls, the value being stoichiometric to the increase in the venous CO flux. These results indicate that alterations in the vascular resistance and bile output are the events occurring in concert with the overproduction of CO.

View this table: [in this window] [in a new window]   Table 1.   Changes in the baseline vascular resistance, bile output, and biliary flux of bilirubin IX in the hemin-challenged rat liver

Intralobular distribution of HO-1 expression in the hemin-pretreated liver. Figure 2 illustrates immunohistochemical analysis using a monoclonal antibody against HO-1. In the control liver, HO-1 was densely stained in cells located in nonparenchymal regions, whereas only little staining was observed in parenchymal cells (Fig. 2A). The sites of the HO-1 expression corresponded to the distribution of Kupffer cells as described previously (10). In contrast, HO-2 was mainly observed in the parenchymal cells (data not shown) (10). After the 18-h hemin treatment, HO-1 expression was demonstrated not only in Kupffer cells but was also prominent in hepatocytes throughout the entire lobule (Fig. 2B). On the other hand, HO-2 expression in the hemin-treated liver displayed no detectable changes between the control and the 18-h hemin-treated livers (data not shown). Together with the data from Fig. 1, A-C, these results suggest that the overproduction of CO in the livers of hemin-treated rats is ascribable to a marked induction of HO-1 in parenchymal cells and to the increase in availability of cytochrome P-450-derived heme.

View larger version (143K): [in this window] [in a new window]   Fig. 2.   Immunohistochemical determination of HO-1 expression in the hemin-pretreated rat livers. The 7-µm-thick frozen sections of rat livers were immunostained with GTS-1. A and B: samples collected from the control and 18-h hemin-treated livers, respectively. Bar = 100 µm. P and C denote portal and central venules, respectively.

Effects of HbO₂ and ZnPP on vascular resistance. We investigated whether reduction of vascular resistance observed at 18 h of treatment is indeed controlled by mechanisms involving CO. Figure 3 illustrates alterations in the baseline vascular resistance and the sinusoidal caliber in response to administration of 0.5 g/dl HbO₂, a CO-trapping reagent. Figure 3A shows that, immediately after the start of administration, the vascular resistance exhibited a marked elevation, being 33% greater than the baseline level. When HbO₂ was eliminated from the perfusion circuit at 10 min after the administration, the increased resistance decreased back to the baseline level within 20 min, showing that the HbO₂-induced elevation of resistance is a reversible event. On the other hand, metHb, a reagent that could scavenge NO but not CO, did not alter the vascular resistance. Furthermore, administration of NO synthase inhibitors such as L-NAME and aminoguanidine did not increase the resistance (data not shown). These results suggest that CO rather than NO plays a major role in lowering the baseline vascular resistance under the current experimental conditions. To specify the role of CO generated in intra- and extravascular compartments, effects of liposomal encapsulation of HbO₂ on the resistance were also examined. Administration of HbV-O₂ induced no significant changes in vascular resistance, indicating that CO generated in the extravascular space is important in lowering the baseline vascular resistance.

View larger version (48K): [in this window] [in a new window]   Fig. 3.   Effects of oxyhemoglobin (HbO2), methemoglobin (metHb), and liposome-encapsulated hemoglobin (HbV) on the vascular resistance in perfused rat liver exposed to the 18-h hemin treatment. In A, open triangles denote data collected from the HO-1-induced liver receiving no hemoglobin derivatives.  and , Data from livers receiving HbO2 and HbV, respectively. , Data from metHb-treated liver. Results are means ± SE of 6 separate experiments, expressed as percentages vs. basal vascular resistance in the 18-h hemin-treated liver (see Table 1). * P < 0.01 compared with control (). B: time course of alterations in the sinusoidal caliber during HbO2 administration. Twenty individual sinusoids in a representative liver preparation were chosen for measurements, and their changes in caliber were determined before and 2, 5, 10, 15, and 20 min after the start of the HbO2 administration. HbO2 was eliminated from the perfusion circuit at 10 min. Bold lines indicate means ± SE of the 20 measurements in the representative experiment.

View larger version (27K): [in this window] [in a new window]   Fig. 4.   Effects of zinc protoporphyrin IX (ZnPP), an inhibitor of HO, on baseline vascular resistance in the 18-h hemin-pretreated rat liver (H18). Baseline vascular resistance was plotted as relative values vs. those in control. Namely, the value 100 indicates 2.42 ± 0.13 cmH2O · g liver · min · ml1. ZnPP was administered at final concentrations of 1 mM. When added, 4 µM CO was applied to the perfusate 5 min before ZnPP administration. Data are means ± SE of measurements from 6-9 separate livers. * P < 0.05 compared with control value.  P < 0.05 compared with values measured in 18-h hemin-treated liver. # P < 0.05 compared with values in 18-h hemin-treated liver perfused in the presence of 1 µM ZnPP.

Overproduction of CO diminished ET-1-elicited sinusoidal constriction. The above studies with different Hb derivatives showed that the reduction of basal vascular resistance in the HO-1-induced liver is attributable to relaxation of microvessels by overproduced CO. This finding tempted us to examine whether the CO overproduction can predispose sinusoids to actively relax even under stimulated conditions. To this end, we applied ET-1 to constrict microvessels. This vasoconstrictive peptide can increase sinusoidal tone at a dose that does not evoke constriction of larger portal veins (34) but elicits that of microvessels involving portal venules (11) and/or sinusoids (3). As shown in Fig. 5, 1.0 nM but not 0.5 nM ET-1 evoked an increase in the vascular resistance. In contrast, the liver undergoing the 18-h hemin treatment displayed a 45% decrease in the ET-1-elicited elevation of the resistance, whereas that exposed to the 12-h treatment exhibited no changes. We did not observe interlobular heterogeneity of the perfusion in these livers as assessed by the transportal injection of trypan blue at the end of experiments (data not shown). In addition, during the ET-1 administration, the bile output did not show any detectable decrease. On the other hand, administration of 5 nM ET-1 evolved a twice greater increase in the resistance, which coincided with a marked reduction of the bile output and an interlobular heterogeneity in the perfusion. Considering that greater doses of ET-1 are known to evoke cholestatic changes in parallel with a marked constriction of portal veins (3, 34), these results suggest that the increase in vascular resistance elicited by ET-1 at <1 nM is ascribable to vasoconstriction of microvasculature, including portal venules and/or sinusoids, rather than to vasoconstriction of larger portal veins. On the basis of these data, we compared the differences between the control and the 18-h hemin-treated livers in the ET-1-induced vascular response at 1 nM.

View larger version (31K): [in this window] [in a new window]   Fig. 5.   Differences in responses of endothelin-1 (ET-1)-induced elevation of the vascular resistance in HO-1-induced livers. ET-1 was administered transportally at concentrations ranging from 0.1 to 1 nM for 1 min. Values of change in () vascular resistance were determined by calculating the difference between the resistance measured at 4 min after ET-1 injection and that measured before injection.  and , Data collected from control and 12-h hemin-treated livers, respectively. , Data from 18-h hemin-treated livers. Note that the liver undergoing 18-h hemin treatment specifically decreased its vasoconstrictive sensitivity to ET-1. * P < 0.05 compared with control value.  P < 0.05 compared with values measured in 18-h hemin-treated liver. # P < 0.05 compared with values in 18-h hemin-treated liver perfused in the presence of 1 µM ZnPP. Inset: effects of N-nitro-L-arginine methyl ester (L-NAME), a nitric oxide synthase inhibitor, on ET-1-induced elevation of resistance.  and , data collected from 12-h hemin-treated livers in the absence or presence of 1 mM L-NAME, respectively. Closed and open arrows indicate time points when L-NAME and ET-1 were administered to the perfusate, respectively. Mean value of the vascular resistance induced by 1 nM ET-1 in 18-h hemin-treated liver (see  in main panel) was expressed as 100%. Data are means ± SE of 5 separate experiments.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Previous studies suggested that, under various disease conditions, HO-1 is induced and thereby causes alterations in cell and organ functions (1, 2, 5, 20). However, outcome of the functional alterations on the enzyme induction was not sufficiently elucidated in these previous studies. The present study provides substantial evidence that CO overproduced by the inducible HO serves as an endogenous vasodilating factor that alters the hepatic vascular resistance. Quantitative determination of CO as a terminal product of the HO pathway disclosed that, under our experimental conditions, actual generation of the product rather than amounts of the enzyme expressed in the tissue dictated changes in the vascular function. Immunohistochemical analysis illustrates that both parenchymal and nonparenchymal cells constitute major cellular components responsible for the regional CO production. Furthermore, entrapment of CO generated in situ by HbO₂ as well as enzyme inhibition by ZnPP elicited marked elevations of resistance, suggesting that CO is a determinant for the decreased vascular resistance in the HO-1-induced liver, at least under the current experimental conditions using isolated perfused preparations.

We reported that CO generated by the constitutive isozyme of HO, i.e., HO-2, plays a major role in maintaining sinusoids in a relaxing state under ordinary conditions (10, 32). The biological significance of the HO-1-mediated vasorelaxing mechanism depicted in this study appears to be distinct from that of the HO-2-mediated one, although these two isozymes share the same reaction product as the vasorelaxing reagent. Because of the greater Michaelis-Menten constant value for protoheme IX (16), an induction of HO-1 under disease conditions that induce overloading of heme iron observed in sepsis and hemolysis has been thought to play an important role in normalizing the amounts of intracellular free heme molecules. At the same time, the enzyme induction could help ameliorate tissue susceptability to oxidative stress through a reduction of excessive heme loading as well as generation of anti-oxidant bile pigments such as bilirubin (31, 39). With overloading of free protoheme IX, the liver can utilize several possible mechanisms to reduce the amount of free heme molecules overloaded in hepatocytes and to degrade them. First, once uptaken into the parenchyma, protoheme IX could be degraded by HO-2 constitutively expressed in the cells. Second, free protoheme IX can be captured by empty heme-binding pockets of apo-heme proteins (e.g., cytochrome P-450 and tryptophan 2,3-dioxygenase) in the cells (7, 38). A portion of excessive heme molecules that escaped from the aforementioned mechanisms could be excreted directly into the bile (30). Finally, protoheme IX evokes transcriptional upregulation of HO-1 in the cells and helps its oxidative degradation through the newly induced enzyme (15). When the greater catalytic ability of HO-1 over HO-2 (15) is considered, this mechanism could serve as an important fail-safe mechanism that is necessary not only to ameliorate the toxicity of protoheme IX as a prooxidant reagent but also to help guarantee an ample perfusion in the hepatic vascular system through the mechanism involving CO under the aforementioned disease conditions. Considering our previous data (10, 32), these results suggest that HO-1 and HO-2 have distinct roles in regulation of hepatic microvascular tone under pathological and physiological conditions, respectively.

It should be noted that CO overproduction occurred with a certain lag time after the enzyme induction reached a maximum level. Although whole pictures of the mechanisms are presently unknown, there are several possible mechanisms to be considered. First, the HO reaction requires NADPH-cytochrome P-450 reductase for an electron donor system. At 12 h, when the amounts of cytochrome P-450 become maximum, HO could share this electron-donating system with cytochrome P-450 monooxygenases and would not exert its maximum catalytic activity. Second, differences in availability of free heme molecules as a substrate of HO between naive heme proteins and those undergoing oxidative insults should be considered. Previous data showed that cytochrome P-420, a denatured heme protein derived from oxidative modification of cytochrome P-450, serves as a better substrate for the HO reaction than the naive protein (12). It has also been suggested that, once exposed acutely to xenobiotics such as acetaminophen, hepatic cytochrome P-450 can undergo oxidative destruction through its active intermediates and can induce CO overproduction mediated by HO-2, even when the inducible enzyme is not overexpressed (17). These previous observations led us to speculate that availability of NADPH-mediated electron donation and oxidative susceptability of cytochrome P-450 proteins could be factors that determine temporal discrepancy between the HO-1 expression and CO generation. Although further investigation is obviously necessary to solve these questions, the current results shed light on the importance of measurements of products such as CO and/or bilirubin IX to study the effects of HO-1 induction on organ functions.

It has been shown in the liver that at least two distinct compartments in the vascular system are responsible for controlling the resistance: portal veins and microvessels such as portal venules, sinusoids, and central venules. The current study provided several lines of data showing that, in the HO-1-induced liver, microvessels constitute a major vascular compartment sensing the regional CO generation. First, when administration of HbO₂ or ZnPP abolished reduction of the basal resistance and further increased the resistance, presumably by canceling the HO-2-derived CO, there was neither a reduction of the basal bile output nor an interlobular heterogeneity of the perfusion, whereas a decrease in the perfusion caused by vasoconstriction at the portal levels is known to result in these changes (34). Second, elimination of CO by HbO₂, which can diffuse into the space of Disse, but not by HbV, which is restricted to access the space, preferentially constricted sinusoids in the HO-1-induced liver. Finally, when undergoing perturbation by nanomolar levels of ET-1, a substantial difference in the vascular resistance became evident between the control and the HO-1-induced liver. This event can only be explained by differences in HO-1 expression and CO generation but not by differences of HO-2 expression between the groups, inasmuch as no fundamental differences in the HO-2 expression in the parenchyma were observed immunohistochemically. We were unable to observe a reduction of the ET-1-elicited constriction of the preterminal portal venules, a putative resistant vessel responding to low doses of this peptide constrictor (11), in the hemin-treated liver because intravital microscopy of the hepatic microcirculation under epi-illumination in rats did not allow us to observe these vessels in portal regions directly (32). However, it appears possible that vascular tone of such portal venules can be regulated by overproduced CO, since these vessels stand in a periportal domain adjacent to hepatocytes that abundantly express HO-1 in the hemin-treated liver, as shown by immunohistochemistry in Fig. 2. Further evaluation is obviously necessary to specify which portion(s) of microvessels is mainly responsible for the CO-mediated attenuation of the ET-1-induced constriction. In summary, the current results shed light on a possible role of CO overproduction in the mechanisms for hyporeactivity of the hepatoportal vascular system observed in a variety of experimental disease models accompanied by HO-1 induction, such as carbon tetrachloride-induced cirrhosis (8), portal hypertension (9), and hemorrhagic shock (2). Further investigation is necessary to confirm whether the CO-mediated vasorelaxing mechanisms are applicable to blood-perfused innervated livers in vivo under these disease models.