Hydrogen sulfide bypasses the rate-limiting oxygen activation of heme oxygenase

Discovery of unidentified protein functions is of biological importance because it often provides new paradigms for many research areas. Mammalian heme oxygenase (HO) enzyme catalyzes the O2-dependent degradation of heme into carbon monoxide (CO), iron, and biliverdin through numerous reaction intermediates. Here, we report that H2S, a gaseous signaling molecule, is part of a novel reaction pathway that drastically alters HO's products, reaction mechanism, and catalytic properties. Our prediction of this interplay is based on the unique reactivity of H2S with one of the HO intermediates. We found that in the presence of H2S, HO produces new linear tetrapyrroles, which we identified as isomers of sulfur-containing biliverdin (SBV), and that only H2S, but not GSH, cysteine, and polysulfides, induces SBV formation. As BV is converted to bilirubin (BR), SBV is enzymatically reduced to sulfur-containing bilirubin (SBR), which shares similar properties such as antioxidative effects with normal BR. SBR was detected in culture media of mouse macrophages, confirming the existence of this H2S-induced reaction in mammalian cells. H2S reacted specifically with a ferric verdoheme intermediate of HO, and verdoheme cleavage proceeded through an O2-independent hydrolysis-like mechanism. This change in activation mode diminished O2 dependence of the overall HO activity, circumventing the rate-limiting O2 activation of HO. We propose that H2S could largely affect O2 sensing by mammalian HO, which is supposed to relay hypoxic signals by decreasing CO output to regulate cellular functions. Moreover, the novel H2S-induced reaction identified here helps sustain HO's heme-degrading and antioxidant-generating capacity under highly hypoxic conditions.

Hydrogen sulfide (H 2 S) has emerged as the third gaseous mediator molecule alongside nitric oxide (NO) and carbon monoxide (CO) in mammalian signaling to modulate a variety of biological processes (1)(2)(3)(4)(5)(6). H 2 S is endogenously synthesized by cystathionine ␤-synthase (CBS), 3 cystathionine ␥-lyase, and 3-mercaptopyruvate sulfurtransferase. Among the gaseous mediators, only H 2 S can be ionized in aqueous solution to exist mainly as its monoanion form (SH Ϫ ) with negligible amounts of S 2Ϫ present under physiological pH (pK a values for H 2 S and SH Ϫ are 7.0 and Ͼ17, respectively) (6). The SH Ϫ anion is known to possess high reducing ability and nucleophilicity. One of the most established targets of SH Ϫ is a cysteine residue of proteins to form persulfide (-SSH). Functions of target proteins such as ATP-dependent potassium (K ATP ) channels (1) are modulated by the covalent modification, which may proceed through nucleophilic attack of SH Ϫ to oxidized forms of the cysteine residue, although the detailed mechanism has yet to be identified (7). In addition, SH Ϫ is reported to react with a variety of small biomolecules to afford reactive sulfur species, including polysulfides (S n 2Ϫ ) (6,8,9). Considering these unique reactivities, H 2 S could have further signaling targets, whose identification is crucial for the H 2 S biology. We have envisaged a new interplay of H 2 S within the gaseous mediator systems because of their tight interactions, especially in vasorelaxation and oxygen (O 2 ) sensing. For instance, CO binding to a prosthetic heme of CBS inhibits the H 2 S synthesis to cause vasoconstriction in neurovascular units (10). The biological source of CO is heme degradation catalyzed by heme oxygenase (HO), which exists as two isoforms, inducible HO-1 and constitutive HO-2 (11). HO activity could have O 2 dependence that is a prerequisite to function as a primary O 2 sensor in the cerebral blood vessels as well as in the carotid body (10,12). These intricate interplays may be an inherent feature of the gaseous mediators and suggest the existence of further unidentified cross-talk. Here, we report that H 2 S opens a novel pathway of HO to produce a new heme catabolite.
The new HO reaction, termed an S-HO reaction hereafter, is predicted on the basis of mechanistic knowledge of HO cataly-sis as well as H 2 S reactivity. HO catalyzes degradation of heme into biliverdin (BV), ferrous iron, and CO by three successive monooxygenation steps via hydroxyheme and verdoheme intermediates (see Fig. 1A) (13,14). These O 2 activation steps are performed by the substrate heme itself, so that HO catalysis actually involves numerous reaction intermediates (Fig. S1) (13,14). Among them, we have initially postulated the verdoheme intermediate, which may exist in either a ferric or ferrous state (Fig. S1), as a potential target of H 2 S. In the enzymatic reaction, the macrocycle of verdoheme is cleaved by reductive O 2 activation on its central iron (15,16). Anomalously slow O 2 binding on ferrous verdoheme is regarded as a major rate-determining step of HO to achieve the O 2 dependence of its overall activity (15,17). The other two O 2 reactions with heme and hydroxyheme are not rate-limiting because of their high reactivity and/or affinity (18,19). Chemically, the verdoheme ring could be cleaved even by O 2 -independent hydrolysis (20). Although the verdoheme hydrolysis proceeds only under physiologically irrelevant, strong alkaline conditions, we hypothesized that H 2 S efficiently promotes "thiolysis," the hydrolysis-like ring cleavage (see Fig. 1A), because of its strong nucleophilicity even under physiological pH. As expected from the reaction design, the S-HO reaction is found to produce an unprecedented sulfur-containing biliverdin termed S-biliverdin (SBV). H 2 S reacts with ferric, but not ferrous, verdoheme with drastic modulation of the HO catalytic properties due to circumvention of the ratelimiting O 2 reaction (Fig. 1).

Novel S-HO reaction under single-turnover condition
Effects of H 2 S on HO catalysis were first examined under single-turnover conditions where ferric heme complexed with rat HO-1 was converted once into products. In the absence of H 2 S, reduction of ferric heme-HO-1 by its physiological redox partner, NADPH-cytochrome P450 reductase (CPR), resulted in an intricate absorption spectral change to eventually afford BV having intense visible absorption around 688 nm ( Fig. 2A) (21). Temporal emergence of sharp absorption at 641 nm is due to transient accumulation of CO-bound verdoheme (Fig. S1). In the presence of H 2 S, the starting heme complex showed a spectral shift typical for partial binding of H 2 S to the ferric heme iron (Fig. 2B) (22). The heme degradation with H 2 S proceeded through distinct intermediate(s) absorbing around 754 nm to yield product(s) having broader absorption at 697 nm. Although H 2 S hardly affected amounts of CO produced through the heme degradation (Fig. S4D), HPLC analysis detected two new products other than BV in the presence of H 2 S (Fig. 2D), whose absorption spectra are similar to each other but different from that of BV (Fig. S4A). Further product analysis identified these two new products as SBV regioisomers, 1-SBV and 19-SBV (Fig. 1B). Positive mass signals of the new products were both found at 599.24 (Fig. S4C) (Table S1). 1-SBV and 19-SBV standards synthesized according to a previous report (23) (Figs. S2 and S3) are indistinguishable from the two new products in terms of the HPLC retention time (Fig. 2D) and absorption spectra (Fig. S4, A and  B). These results strongly suggest H 2 S-dependent ring cleavage as hypothesized.

Formation and properties of SBR
In mammals, BV is rapidly reduced by biliverdin reductase (BVR) to a yellow pigment, bilirubin (BR; Fig. 1C and Fig. S4E). The single turnover reaction of the purified HO enzyme with additional BVR yielded BR having typical absorption around

Alternative heme degradation induced by H 2 S
462 nm (Fig. 2C). The SBV standards were also reduced by BVR to corresponding 1-and 19-SBR isomers (Fig. 1C), whose absorption peaks were observed around 530 nm to have distinct orange colors (Fig. S4, E-G). Structures of the SBR isomers were further confirmed by various product analyses (Fig. S5, A-C and Table S2). Due to the large red shifts of the absorption bands, the SBR formation under the single-turnover condition is easily visualized by an unusual shoulder around 530 nm (Fig.  2C). The single turnover rate slightly decreased in the presence of H 2 S as judged by absorbance increase either at 460 or 530 nm (approximately half; Fig. 2C). SBR is also produced by a macrophage cell line, RAW264.7. This cell line is reported to secrete BR into cell culture media when incubated with hemin (24). Our HPLC analysis detected only BR as a product of the normal heme catabolism by RAW264.7 (Fig. 2E). Addition of H 2 S, however, resulted in the formation of the two SBR isomers as well as BR (Fig. 2E), assuring the progression of the S-HO reaction in mammalian cells.
The SBR isomers share similar properties with BR as expected from their structural similarity (Fig. 1C). Highly hydrophobic SBR and BR require bovine serum albumin (BSA) for their solubilization in neutral buffers. BR is known to be excreted into bile upon glucuronidation in liver ( Fig.  S5D) (25). Both isomers of SBR were similarly converted by UGT1A1 to their mono-and diglucuronides (Fig. S5, E-G). BR is also proposed to be a potent antioxidant through a redox cycling with BV ( Fig. S6A) (26,27). Catalytic removal of radical species was suggested for BR and the SBR iso-mers in the 2,2Ј-azobis(2-amidinopropane) dihydrochloride (AAPH) assay (Fig. S6), implying their comparable potencies as antioxidants.

Catalytic properties of the S-HO reaction
The H 2 S effects on HO catalytic properties were examined in the presence of BVR (17). As shown in Fig. 2F, the product ratio of total SBR over all three products (%SBR) rose as the H 2 S concentration increased (EC 50 ϭ 34 M). To our surprise, the catalytic activity of HO-1 was severely suppressed by H 2 S ( Fig.  2F; IC 50 ϭ 48 M) despite its small effect on the single turnover rate (Fig. 2C). Although the EC 50 and IC 50 values are similar at the basal condition ([CPR] ϭ 100 nM), these parameters showed distinct behavior when the CPR concentration was lowered to 20 nM (EC 50 and IC 50 ϭ 9.6 and 200 M, respectively). This observation reveals different mechanisms for the SBR induction and HO inhibition by H 2 S. It should also be noted that %SBR was saturated at ϳ45% under normal air. Incomplete formation of SBR is due, at least in part, to an unexpected action of O 2 because %SBR increased with lowering O 2 concentrations (Fig. 2G). O 2 dependence of the catalytic activity was also largely affected by H 2 S (Fig. 2G). The normal HO reaction without H 2 S was slowed down at O 2 concentrations less than 130 M due to the rate-limiting reaction of the ferrous verdoheme intermediate with O 2 (Fig. 1A) (15,17). In contrast, very little O 2 dependence was observed in the presence of H 2 S, consistent with circumvention of the slow O 2 -dependent ring cleavage (Fig. 1A).

Alternative heme degradation induced by H 2 S Ferric verdoheme as a branch point intermediate
We finally examined reaction mechanism of the S-HO reaction. In our initial design, H 2 S was postulated to attack ferrous verdoheme (Fig. 1A); however, this intermediate showed negligible spectral change upon addition of H 2 S and CPR under anaerobic condition (Fig. S7A). Further O 2 addition initiated the reaction but afforded only BR (Fig. S7A). Instead, more electrophilic ferric verdoheme, which was generated prior to the ferrous form (Fig. S1), was found to rapidly react with H 2 S. The H 2 S addition to ferric verdoheme diminished its characteristic absorption at 683 nm with appearance of featureless, "flat" absorption in a 600 -800-nm range (Fig. 3A). Resulting intermediate X was reduced to yield a mixture of BR and SBR under normal air, whereas the SBR formation became dominant at lower O 2 concentration (Fig. 3D), consistent with the O 2 dependence of %SBR observed in the catalytic assay (Fig. 2G). Other HO intermediates generated prior to ferric verdoheme, including hydroxyheme (Fig. S1), did not react with H 2 S or only reversibly interacted to eventually afford the same intermediate X (Fig. S7, B-D). These results unambiguously identify ferric verdoheme as a branch point intermediate for the BV and SBV pathways of HO as shown in Fig. 4.
Conversion of ferric verdoheme to the intermediate X required approximately 1 molar eq of H 2 S (Fig. 3B). Mass analysis also revealed incorporation of one sulfur atom into ferric verdoheme ( Fig. 3C and Fig. S9A). Apparent rate constants for the formation of X show pH-dependent transition with a pK a value of 7.0 (Fig. 3E). This change is attributed to ionization of H 2 S (pK a ϭ 7.0) because no significant spectral change was observed for ferric verdoheme in this pH range. The alkaline form (SH Ϫ ; 29 mM Ϫ1 s Ϫ1 ) was ϳ300-fold more reactive than the acid form (H 2 S; 0.094 mM Ϫ1 s Ϫ1 ), indicating SH Ϫ as a major reactant. This conclusion rationally explains the low reactivity of other biological thiols. Glutathione (GSH), L-cysteine, and homocysteine, which have higher pK a values (9.7, 8.4, and 8.9, respectively), showed only limited reactivity with ferric verdoheme (Fig. S8B). Moreover, these alkyl thiols did not convert ferric verdoheme to X but reduced it to the ferrous state, leading to the normal BV pathway (Fig. S8A and Fig. 4). Similar simple reduction was also observed for sodium polysulfides

Alternative heme degradation induced by H 2 S
(Na 2 S 2 , Na 2 S 3 , and Na 2 S 4 ) as shown in Fig. S8, C and D. Therefore, H 2 S is the specific thiol that induces SBV formation.

Ring opening mechanism in the H 2 S-mediated reaction
Under anoxic conditions, reduction of the intermediate X dominantly produced SBR (Fig. 3D), strongly suggesting an O 2 -independent hydrolysis-like mechanism for the SBV production (Fig. 4). In this mechanism, nucleophilic addition of SH Ϫ to a ␤-pyrrole carbon atom of ferric verdoheme should be followed by deprotonation to cleave the macrocycle. A resulting Fe 3ϩ -SBV complex could be reduced by one electron to afford SBV with release of the ferrous iron. As expected from this scheme, the anoxic conversion of X to SBR consumed three electrons, two of which were utilized for reduction of SBV to SBR (Fig. 3F). In the presence of O 2 , the mechanism became more intricate to afford BV in addition to SBV. The BV formation from X required O 2 -induced extrusion of the sulfur atom. X itself has no reactivity with O 2 as judged from the negligible O 2 dependence of its absorption spectra (Fig. 3D). Instead, a reduced form of X may bind O 2 on its central iron to liberate the SH Ϫ moiety. This putative reaction yielded oxyferrous verdoheme, which is an intermediate proposed in the normal HO reaction (Fig. 4) (15). The verdoheme regeneration from X is possible because pyridine extraction of X from the protein yielded a bispyridine complex of verdoheme (Fig. S9B). It should be also mentioned that one oxygen atom of O 2 was inserted in the conversion of X to BV (Fig. S9C), consistent with the O 2 activation on the verdoheme iron (15).

Discussion
In this study, we report the new reaction pathway of mammalian HO induced by H 2 S. The gaseous signaling molecule drastically alters the reaction mechanism of HO not only to modulate the catalytic activity and its O 2 dependence but also to produce the new sulfur-containing heme catabolites, 1-and 19-SBV isomers (Fig. 1). It is quite surprising that the classical HO enzyme, whose function and mechanism have been studied in detail over several decades, has an unidentified reaction pathway that is just now being revealed. Recently, noncanonical heme degradation has also been identified for a few bacterial enzymes; however, these enzymes are specifically designed to promote unusual reactions (28 -30). We have succeeded in inducing the alternative reaction of the canonical HO enzyme by the rational selection of the small biological molecule to attack its reaction intermediate. H 2 S appears to have broad reactivities with many biological compounds. For example, heme is known to interact with H 2 S to cause catalytic oxidation of the sulfide (22) and covalent modification of the heme periphery to generate sulfheme (31). These reactions appear to be negligible upon induction of the S-HO reaction (Figs. 2B and 3B), probably due to the facile and specific reactivity of the HO intermediate with H 2 S. Our approach to target the enzyme intermediate thus allows identification of highly specific interactions and could be applied widely to search for hidden functions, especially of well-studied enzymes. Besides the SBV pathway, the HO reaction may branch into unidentified pathways at its numerous reactive intermediates (Fig. S1) by choosing other appropriate small molecules. Moreover, H 2 S could react with enzymes other than HO to bring out their unidentified reactions. To discover further unknown targets, possible effects of H 2 S should be extensively studied for reactions involving hydrolysis or a step replaceable with a hydrolysis-type reaction in addition to searching for electrophilic intermediates.
We have almost fully elucidated the formation mechanism of SBV. The SH Ϫ ion binds to ferric verdoheme and cleaves the macrocycle in an O 2 -independent manner. The only major ambiguity remaining is the nature of the observed intermediate X. The intermediate X generated in the reaction between ferric verdoheme and H 2 S can be either an SH Ϫ -adduct of verdoheme or an Fe 3ϩ -SBV complex (Fig. 4), both of which match the mass data ( Fig. 3C and Fig. S9A). The SH Ϫ -adduct form well explains the O 2 -dependent SH Ϫ extrusion, whereas ring closure of Fe 3ϩ -SBV seems not to occur easily. Nevertheless, the absorption spectrum of X, especially its flat absorption in the visible region, is very similar to that of the Fe 3ϩ -BV complex ( Fig. 3A and Fig. S9D) (32), strongly suggesting the Fe 3ϩ -SBV structure for X. The ring closure may proceed more easily than we expected to achieve equilibrium of the open and closed ring structures for X and/or its reduced form.
Successful detection of SBR assures that the S-HO reaction can take place in mammalian cells (Fig. 2E). Nevertheless, we cannot specify the H 2 S concentration required for the SBV induction (EC 50 ) in vivo. Our mechanistic analysis identifies ferric verdoheme as a branch point intermediate for the normal and SBV pathways (Fig. 4). Because the SH Ϫ addition and reduction of ferric verdoheme are essentially irreversible, a major determinant of the EC 50 value should be relative rates of the two competing reactions. Slower reduction by CPR hence results in the lowered EC 50 value as shown for the in vitro assay. Some naturally occurring mutations of CPR are known to significantly suppress electron transfer to HO (33), probably leading to hypersensitivity toward H 2 S. Hypoxia is also expected to decelerate the electron transfer by inducing liberation of HO from the endoplasmic reticulum (34). The cytosolic HO enzyme could interact only weakly with CPR remaining on the endoplasmic reticulum membrane. The SBV formation could be enhanced under hypoxia because of the slow reduction as well as the low O 2 concentration. Nonmammalian HOs with greater diversity in structure and reactivity may have larger deviations in the H 2 S requirement to stimulate the S-HO reaction. Even non-HO-type heme-degrading enzymes involving ferric verdoheme in their catalysis are possible targets of H 2 S (35). Our finding clearly indicates the necessity to take the possible interplay between H 2 S and heme degradation into account in all related research areas.
The new S-HO reaction could have large impacts on physiological functions of H 2 S and HO through drastic changes in both the product and catalytic activity of HO. Properties of the new heme catabolites have been examined mainly for their reduced form, SBR, which as well as BR should be the major form in mammalian cells. Although BR is crucial in mammals as a causative agent of jaundice and a potent antioxidant (26), SBR shows solubility, glucuronidation potency, and redox properties similar to those of BR (Figs. S5, D-G, and S6). This observation implies that H 2 S does not greatly alter antioxidative

Alternative heme degradation induced by H 2 S
functions of the HO reaction in mammals. Biological roles of SBV are yet to be examined, especially in biosynthesis of phytochromes and photoreceptor proteins in plants, bacteria, and fungi, which utilize BV or its derivatives as a pigment for detecting red and far-red light (36).
Our catalytic assay reveals the O 2 -dependent change of the SBR production ratio (%SBR) due to the O 2 -induced return to the BV pathway (Figs. 2G and 4). Considering the similar properties of BR and SBR, this product change itself may have a biological significance only when a receptor system specific to the new catabolite exists. More importantly, the O 2 dependence of the HO activity was lost in the presence of H 2 S due to circumvention of the rate-limiting O 2 activation by ferrous verdoheme (Figs. 2G and 4) (10,15,37). This activity change is expected to have large impacts on gas signaling, namely the O 2 sensing in the brain microvessels as well in as the carotid body. In these systems, HO-2, a constitutive isoform of HO, serves as a primary oxygen sensor, and its CO product is considered to transduce the O 2 signal in concert with H 2 S (10, 12, 38). A major H 2 S generator in brain is CBS (39), whose prosthetic heme group binds CO to directly inhibit the enzyme activity under normoxia (40). CO is also reported to be a physiological inhibitor of cystathionine ␥-lyase in the carotid body despite the absence of an apparent CO-binding site (41). Under hypoxic condition, the HO activity decreases to relieve the CO inhibition of the H 2 S-generating enzymes. Consequently, hypoxia stimulates the H 2 S synthesis to enhance vasodilation in brain and neural activity of the carotid body through regulation of potassium channels (10,38). The faster formation of H 2 S together with its slower decay through O 2 -dependent processes is expected to promote the O 2 -independent S-HO reaction under hypoxia (Figs. 2G and 4). This maintains the HO activity at a certain level even under highly hypoxic conditions to avoid a complete stop of CO synthesis and thus overproduction of H 2 S. An H 2 S concentration ceiling may be important to prevent respiratory disorder (42) and to reduce its toxicity. Likewise, under inflammatory conditions in which cytokines activate macrophages, the cells induce HO-1 (43), which might protect against oxidative stress through the action of SBV and SBR under hypoxic conditions. Another striking effect of H 2 S is severe suppression of the HO catalytic activity (Fig. 2F), which is not explained by the mechanism shown in Fig. 4. Considering the weak inhibition of the single turnover reaction (Fig. 2C), interference by H 2 S appears to occur at a step prior to formation of the heme-HO complex. Under the catalytic conditions, excess hemin weakly binds to BSA supplemented to solubilize the BR and SBR products. The hemin-BSA complex exhibits a significant spectral change when incubated with H 2 S under hypoxic conditions but not under air, suggesting temporal reduction of hemin, O 2 binding, and rapid auto-oxidation. This implies catalytic production of reactive oxygen species, which may inactivate the HO enzyme. The inhibitory effect of H 2 S in vivo might be distinct from our in vitro observation because it would largely depend on the status of "free heme" that interacts with various cellular factor(s).

Materials
18 O 2 and 34 S 8 were purchased from Spectra Stable Isotope ( 18 O content, Ͼ95%) and ISOFLEX ( 34 S content, 99.9%), respectively. Liquid NH 3 (20 ml) was added dropwise to 34 S 8 (100 mg, 2.94 mmol) and sodium (135 mg, 5.89 mmol) in a round-bottom flask at Ϫ78°C. The reaction mixture was stirred at room temperature under an argon atmosphere until liquid NH 3 entirely vaporized. The residue was dried under reduced pressure to afford Na 2 34 S (anhydrous) as pale yellow powder (233 mg, quantitative). Microsomes containing human UGT1A1 and reagents required for glucuronidation were obtained from BD Gentest. Preparation of rat HO-1, its hemin complex, human CPR, and BVR was carried out as described previously (44 -46). Other chemicals and proteins were obtained from Wako, Oriental Yeast, Sigma-Aldrich, or ICN and used without further purification. RAW264.7 was purchased from the American Type Culture Collection (ATCC). The basal buffer used throughout this study was 0.1 M HEPES, pH 7.5, at 20°C unless otherwise stated.

Heme degradation by purified enzyme systems
Single turnover reactions to yield BV/SBV were typically

Product analysis of enzymatic reactions
Heme catabolites were concentrated by solid-phase extraction with Supelclean LC-18 columns (Supelco) as described elsewhere (15). The BR/SBR extraction was operated under minimum light to suppress their photoisomerization. Benzophenone was added as an internal standard at the end of the reactions for quantitative experiments. HPLC analyses of the extracted products were performed on a Shimadzu LC-20 HPLC system equipped with an SPD-M20A photodiode array detector at a flow rate of 1.0 ml/min. The BV/SBV products were separated on a Tosoh ODS-80Ts reversed-phase column (4.6 ϫ 150 mm) with an isocratic elution of 70% methanol and 30% 0.1 M ammonium acetate (v/v). The BR/SBR analysis was performed typically on the ODS-80Ts column with a linear gradient from 75% methanol, 25% 20 mM triethylamine, and acetic acid (v/v), pH 5.0, to 85% methanol (v/v) over 15 min.
ESI-MS spectra were measured on a Bruker micrOTOF-Q-II mass spectrometer equipped with an Agilent 1100 HPLC system at a flow rate of 0.2 ml/min. LC separation of BV/SBV was conducted on the ODS-80Ts column (2.0 ϫ 150 mm) with an

Alternative heme degradation induced by H 2 S
isocratic elution of 20% 0.1 M ammonium acetate and 80% methanol (v/v). BR/SBR and their glucuronides were analyzed with an Agilent ZORBAX Extend-C18 reversed-phase column (2.1 ϫ 150 mm) using a linear gradient from 35% methanol and 65% 20 mM aqueous ammonia (v/v) to 50% methanol (v/v) over 20 min. Mass spectra of the intermediate X were measured by direct infusion at 3 l/min of a reaction solution containing 3 M heme-HO-1, 10 M H 2 O 2 , and 20 M H 2 S. Typical parameters for the MS measurements were as follows: end plate offset, Ϫ500 V; capillary, 4500 V; in-source collision-induced dissociation, 150 V; nebulizer gas, 0.4 bar; dry gas, 4.0 liters/min; dry gas temperature, 180°C.
CO was quantitated using an H64L variant of myoglobin having high CO affinity as reported earlier (28). After completion of single turnover reactions of 4 M heme-HO-1, solid sodium dithionite and then H64L myoglobin (final concentration, 6 M) were added to record absorption spectra of H64L partially bound with CO. Subtraction of a spectrum of ferrous H64L without exogenous ligands generated difference spectra in which a positive peak indicates generation of the CO-bound form and a negative peak is due to disappearance of ferrous deoxy H64L.

Synthesis of 1-and 19-SBV
1-and 19-SBV isomers were synthesized as their dimethyl esters (total yields from BV, 22 and 15%, respectively) according to a previous report (23). 1 H and 13 C NMR of the dimethyl esters was measured in CDCl 3 with JEOL JMN-LA400 and JMN-LA600 Fourier transform NMR spectrometers (Figs. S2 and S3). Chemical shifts were referenced internally to CDCl 3 . The SBV esters were hydrolyzed, purified, and weighed by a Sartorius ultra-microbalance SE2 to determine extinction coefficients (Fig. S4, F and G).

Glucuronidation
Approximately 20 M SBV isomers were reduced by BVR to corresponding SBR isomers in the presence of 20 M NADPH and 30 M BSA. The freshly prepared SBR was glucuronidated at 37°C by BD Supersomes UGT1A1 according to a protocol provided by the manufacturer. Aliquots of samples were analyzed by HPLC and ESI-MS.

AAPH assay
Each of 10 M freshly prepared BR and SBR isomers was incubated at 37°C with 50 mM AAPH in 0.1 M potassium phosphate, pH 7.5, containing 200 mM NaCl and 15 M BSA with or without 100 M NADPH and 50 nM BVR. Absorption spectral changes during the reaction were monitored at 468, 541, and 521 nm for BR, 1-SBR, and 19-SBR, respectively.

Heme degradation by mammalian cells
RAW264.7 cells were maintained in a 37°C humidified incubator containing 5% CO 2 and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, penicillin-streptomycin, and glutamine. 9 ϫ 10 6 cells were grown in a 75-cm 2 cell culture flask for 24 h and then incubated with 20 M hemin for 12 h to induce HO-1. For efficient removal of BR generated during the induction period, the cells and the flask were washed three times with the culture medium containing fetal bovine serum rather than PBS. Heme degradation was performed by incubating the cells in fresh medium (18 ml) containing 5 M hemin and 2 mM H 2 S at 37°C. The screw cap of the flask was closed during the reaction to suppress volatilization of H 2 S. The medium was recovered after 2-h incubation, centrifuged (3,000 rpm for 5 min), and supplemented with propiophenone as an internal standard. Further addition of a half-volume of saturated guanidine hydrochloride was followed by solid-phase extraction and HPLC analysis as noted above.

Reaction of ferric verdoheme-HO-1 with H 2 S
The ferric verdoheme complex was prepared by H 2 O 2 immediately prior to use and used within 5 min after adding catalase to remove the residual peroxide. Reactions with alkyl thiols were performed in the presence of CO to suppress auto-oxidation of ferrous verdoheme. Kinetic analysis of ferric verdoheme-HO-1 with H 2 S was performed at 20°C using a UNISOKU RSP-601 stopped-flow apparatus equipped with a built-in rapid scan spectrophotometer. The ferric verdoheme complex (3 M) was mixed with equal volumes of H 2 S solution (final concentration, 50 -200 M). Observed rate constants determined from absorbance change at 683 nm exhibited good linear relationships with the H 2 S concentrations to afford apparent reaction rate constants (k app ) at varied pH. pH dependence of the k app values (Fig. 3E) is well-fitted by assuming an acid-base transition of H 2 S (pK a 7.0). Buffers used for the kinetic analysis were: 0.1 M citrate, pH 6.0 -7.0; phosphate, pH 6.5-7.5; HEPES, pH 7.0 -8.0; and Tris-HCl, pH 7.5-8.5.

Product formation from X
The intermediate X was prepared anaerobically from ferrous verdoheme as noted above. The reductive titration to SBR was performed under anoxic condition while monitoring absorbance increase at 541 nm. Formal two-electron reduction of SBV to SBR was confirmed for the stable 1-SBV isomer. To examine O 2 dependence of the product distribution, the anoxic X was mixed with air-saturated buffer to fill the screw-top cuvette. Products were quantitated by HPLC analysis. For isotope labeling of BV generated through O 2 -dependent reaction of X, the intermediate was prepared with a slight excess of non-

Alternative heme degradation induced by H 2 S
labeled H 2 O 2 and air, which was followed by addition of catalase and bubbling with excess 18