Structure of the Novel Heme Adduct Formed during the Reaction of Human Hemoglobin with BrCCI, in Red Cell Lysates*

It was previously shown that the reductive debromi- nation of BrCCl, to trichloromethyl radical by human hemoglobin leads to formation of dissociable altered heme products, two of which are identical to those formed from myoglobin and one which is novel. In this study, we have elucidated the structure of this novel adduct with the use of mass spectrometry, as well as ‘H and lSC NMR as a substitution product of a -C(Cl)=CCl, moiety for a P-hydrogen atom on the prosthetic heme’s ring I vinyl group. From studies with the use of W- enriched BrCCl,, it was determined that the added carbon atoms were derived from 2 eq of BrCCl,. A mecha- nism that involves multiple reductive events and a radical cation heme intermediate is proposed. Consist- ent with this mechanism, cellular reductants were found to selectively enhance the amount of this novel dissociable heme adduct. These studies reveal fine dif-ferences between myoglobin and hemoglobin in the ac- cessibility of reactive intermediates to the ring I vinyl group, as well as the potential importance laser (op- erating at 308 nm, 180 mJ/8-ns pulse) and a Millipore Extrel (Madison, WI) FTMS-2000 dual cell spectrometer equipped with a 7-tesla super- conducting magnet. Source cell spectra were obtained with the use of a 9-V gated trapping sequence (23, 24), maintaining the decelerating potential for 30 ps, followed by trapping at 1 V for 100 ms, prior to spectral data acquisition. For broadband spectral measurements, a 200-V peak-to-peak chirp excitation from 0 to 500 kHz at a 300 Hdps sweep rate was applied, followed by detection. The spectrum of the heme adduct resulted from averaging 2 time domain scans, acquiring 131,072 data points during a 435-ms observation period per scan and the spectrum of the 13C-enriched heme adduct resulted from averaging 4 time domain scans, acquiring 65,536 data points during a 164-ms observation period per scan. The averaged time domain data were aug-mented by an equal number of zeros, and base line was corrected prior to magnitude mode Fourier transformation. No apodization was used. Resolution was estimated as the ratio of peak position to peak width at half-height. High resolution analyzer cell spectra of the heme adduct were obtained with the use of the heterodyne method, employing a 149.6-kHz reference frequency and an excitation sweep from 0 to 200 kHz at 180 Hdps. A 100-ps deceleration period was employed, prior to

Hemoglobin is a major target protein for reactive metabolites of drugs (l), environmental toxicants (2-41, and carcinogens (5)(6)(7)(8)(9). The high abundance, availability, and long half-life of human hemoglobin have led to the use of altered hemoglobins as indicators of exposure to these toxic agents (10,11). In addition, some hemoglobin-based red cell substitutes, which are being developed for clinical use (12, 131, are also covalently altered by reactive intermediates with potentially deleterious consequences (14). Furthermore, it has been speculated that covalent alteration of hemoglobin by reactive intermediates is involved in the formation of hemichromes ( E ) , which are thought to play a role in Heinz body formation and hemolysis (16). Thus, it is important to study the mechanisms involved in the covalent alteration of hemoglobin and the potential biological consequences of such alterations. One model system for such studies involves the covalent alterations of hemoglobin caused by trichloromethyl radicals.
The reductive debromination of BrCCl, to trichloromethyl radical by human hemoglobin and myoglobin has been shown previously to lead to the formation of dissociable and protein-* This work was supported by National Institutes of Health Grant GM-44606 (to the University of California, Riverside; C. L. W.) and National Science Foundation Grant DIR 90-16567 (to Johns Hopkins School of Medicine; R. J. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
bound prosthetic heme-derived products resulting from the regiospecific attack of a trichloromethyl radical on the ring I vinyl group of heme (15,(17)(18)(19)(20). For the most part, the structures of the products formed from these two hemoproteins are highly similar; in fact, two of the dissociable adducts, an a-hydroxy-P-trichloromethylethyl heme derivative (compound 3) and a P-carboxyvinyl heme derivative (compound 1) are formed from both proteins (Scheme I). In contrast, an a,P-bistrichloromethylethyl substitution product of the ring I vinyl of heme was unique to the myoglobin reaction, whereas a novel heme adduct of unknown structure was observed as a major product from the hemoglobin reaction (21). To help explain how identical, as well as unique, products could form from the same reaction with two proteins, we set out to determine the structure of this novel heme product with the use of mass spectrometry, as well as oneand two-dimensional 'H and I3C NMR. The product resulted from the substitution of a -C(Cl)=CCl, moiety for a P-hydrogen atom on the ring I vinyl group of heme. Thus, it appears that the major difference between hemoglobin and myoglobin with respect to modification of the vinyl group is the inaccessibility of trichloromethyl radical at the a-but not the p-carbon region of the ring I vinyl moiety of hemoglobin, presumably due to steric constraints. The mechanism of formation of the novel adduct apparently can involve multiple reductive events, which was consistent with the finding that cellular reductants could selectively enhance the amount of this adduct. However, this interpretation is complicated by the finding that reductants also lead to degradation of some of the other altered heme products. This observation, however, may be of importance in studies on alteration of the prosthetic heme of hemoglobin or possibly other hemoproteins by a variety of agents, especially under biological conditions, where a variety of reductants and reductases could be present. EXPERIMENTAL PROCEDURES Materials-Pyridined, and Br'3CCl, (99 atom %) were purchased from Merck. Methanold, and deuterium oxide were purchased from Cambridge Isotope Laboratories (Woburn, MA). Stannous chloride, human hemoglobin, horse heart myoglobin, and 2,5-dihydroxybenzoic acid were from Aldrich, Sigma, U. S. Biochemical Corp., and Fluka, respectively. Red cell lysates were prepared as described previously (15).
Preparation of Altered Heme Adduct-The reaction of human hemoglobin with BrCCl, in red cell lysates was performed as previously reported (15). In short, red cell lysates were made anaerobic, and the resulting ferrous deoxyhemoglobin (250 PM) in a total volume of 500 ml of 50 m~ potassium phosphate, pH 7.4, was reacted with BrCCl, or Brl3CC1, (1 mmol was added) at room temperature for 1 h. Extraction of the reaction mixture with 2-butanone under acid conditions gave a mixture of heme adducts in the organic phase (17). This organic phase was diluted with water (1:lO) and placed on a Buchner funnel containing C18 resin (50 g, Whatman). The heme products were bound to the C18 material, washed with a liter of 0.1% trifluoroacetic acid (solution A), and then eluted by 20 ml of 0.1% trifluoroacetic acid in acetonitrile (solution B). The eluant was dried under vacuum in a Speed Vac (Savant, Farmingdale, MA). The sample was redissolved in methanol and injected onto a C18 Partisil M20 ODs-3 (Whatman, 2.5 x 50 cm) HPLC' column equilibrated with solution A. The flow rate was 18.0 mI/min. A linear gradient to 35% solution B (l.OOO%/min), then to 48% solution B (0.186%/min), then to 50% solution B (0.029%/min), and then to 100% solution B (1.429%/min) was run. One-minute fractions were collected and the product of interest eluted between 172 and 188 min. To check for purity, an aliquot of each fraction of interest was injected onto an analytical HPLC column (Vydac C,, 0.46 x 25 cm) equilibrated with 36% solution A and 64% solution B. The flow rate was 1.0 ml/min. A gradient (Waters curve 8) to 51% solution B was run over 20 min, and then a linear gradient to 100% solution B was run over the next 10 min. The fractions that were found by use of the analytical HPLC method to contain only the heme adduct of interest were pooled and dried under vacuum in a Speed Vac apparatus. Approximately 5 mg of purified material was obtained per batch. NMR and mass spectra were obtained on these samples.
Preparation ofAltered Heme Adduct in the Presence of Glutathione or Ascorbate--Red cell lysates were diluted with 50 m~ potassium phosphate, pH 7.4, so that the hemoglobin concentration was 200 p~ as determined by the chromophore at 576 nm (22). Ascorbate or glutathione was added to the reaction mixture and subsequently made anaerobic and treated with BrCC1, as described above. The altered heme adduct isolated from these reactions had a molecular ion identical to the heme adduct isolated from reactions without exogenous reductant.
Plasma Desorption Mass Spectrometry-Positive ion mass spectra were obtained on a Bio-Ion Nordic AB (Uppsala, Sweden) model BIN-10K plasma desorption mass spectrometer. Samples were dissolved in solutions of 0.1% trifluoroacetic acid in water/methanol(3:1, dv). Solutions (2-5 pl) were placed on a nitrocellulose-coated, aluminized Mylar sample foil, spin-dried after 5 min, and then washed with 10 pl of 0.1% trifluoroacetic acid in water to remove salts. Sample foils were then placed in the mass spectrometer and the spectra recorded at an accelerating voltage of 16 kV.
Fast Atom Bombardment Mass Spectrometry-Positive ion fast atom bombardment mass spectra were obtained with a Kratos MSBORF (Kratos Analytical, Ltd., Manchester, United Kingdom) double focusing instrument with a mass range of 10000 atomic mass units at full accelerating voltage (8 kV). The instrument was fitted with a model BllNF saddle-field fast atom gun (Ion Tech, Ltd., Teddington, United Kingdom) and a post accelerator detector, which was operated at 14 kV. Xenon was used to bombard the samples at 8 kV. The samples were dissolved in methanol and applied to a gold fast atom bombardment probe in a matrix of 3-nitrobenzyl alcohol. The mass spectra were acquired at a scan rate of 30 s/decade with a resolution of 1 in 2500. All data were acquired and processed with the Kratos DS-90 data system. Fourier Transform Mass Spectral Measurements-Samples were prepared by dissolving approximately 10-100 pg of analyte in 500 pl of a 50 m~ 2,5-dihydroxybenzoic acid matrix solution containing 0.1% trifluoroacetic acid in methanol. The resulting solutions were sprayed as aerosols onto a rotating stainless steel probe tip for homogeneous deposition. Matrix-assisted laser desorptionhonization Fourier transform mass spectra (23) were obtained with 355 nm radiation (maximum -5 mJ/pulse) from a Lambda Physik (Giittingen, Germany) FL-2001 dye laser, pumped by a Lambda Physik EMG 201-MSC excimer laser (operating at 308 nm, 180 mJ/8-ns pulse) and a Millipore Extrel (Madison, WI) FTMS-2000 dual cell spectrometer equipped with a 7-tesla superconducting magnet. Source cell spectra were obtained with the use of a 9-V gated trapping sequence (23,24), maintaining the decelerating potential for 30 ps, followed by trapping at 1 V for 100 ms, prior to spectral data acquisition. For broadband spectral measurements, a 200-V peak-to-peak chirp excitation from 0 to 500 k H z at a 300 Hdps sweep rate was applied, followed by detection. The spectrum of the heme adduct resulted from averaging 2 time domain scans, acquiring 131,072 data points during a 435-ms observation period per scan and the spectrum of the 13C-enriched heme adduct resulted from averaging 4 time domain scans, acquiring 65,536 data points during a 164-ms observation period per scan. The averaged time domain data were augmented by an equal number of zeros, and base line was corrected prior to magnitude mode Fourier transformation. No apodization was used. Resolution was estimated as the ratio of peak position to peak width at half-height. High resolution analyzer cell spectra of the heme adduct were obtained with the use of the heterodyne method, employing a 149.6-kHz reference frequency and an excitation sweep from 0 to 200 k H z a t 180 Hdps. A 100-ps deceleration period was employed, prior to The abbreviations used are: HPLC, high performance liquid chromatography; NOESY, nuclear Overhauser effect spectroscopy. collecting 131,072 data points over a 1.465s period. Data thus obtained were processed as for broadband spectra. Poly(ethy1ene glycol) 1000 was used as an external calibrant.
NMR Spectrometry-NMR spectra were obtained on a Varian XL200 spectrometer in pyridine-d, solution following reduction by stannous chloride (17); 3-5 mg of sample was added to 6 9 mg of SnCl, in a volume of 0.5 ml. Typically, 256 free induction decays were collected with an accumulation time of 4 s for one-dimensional spectra. COSY spectra were taken with the use of a data matrix 1024 by 1024 with 256 t , increments of 16 free induction decays each. Prior to NOESY studies the solutions were degassed by three freeze-exhaust-thaw cycles. Phase-sensitive NOESY spectra were obtained by the method of States et al. (25), in overnight runs, with the use of 1024 x 1024 data matrices and a mixing time of 0.8 s. This allowed collecting 256 t, increments each with 64 free induction decays.
Other Procedures-Visible absorption spectra were obtained with a Hewlett Packard (Rockville, MD) model 8450A diode array spectrophotometer. HPLC was performed with the use of a Waters instrument (Millipore Corp., Milford, MA) consisting of a 600E gradient system controller and a 490E variable wavelength detector. The data were collected with the use of Nelson 760 Series system (PE Nelson, Cupertino, CAI.

RESULTS
The HPLC profiles at 220 and 405 n m of untreated and BrCC1,-treated human ferrous deoxyhemoglobin from red cell lysates are shown in Fig. 1 (panels A and B , respectively). The untreated sample exhibited two major 220 nm-absorbing peaks corresponding to the a and P subunits of hemoglobin (26). Under the acidic conditions of the chromatography, the heme readily dissociated from the untreated protein as seen by the major peak at 405 nm (peak 21, as previously found (17). After treatment with BrCCl,, the P-chain was altered and several 405 nm-absorbing peaks were observed. The structures of compounds corresponding to peaks 1, 3,4, and 5 h a v e been determined previously (Scheme I) (15,17). Peaks 1 and 3 corre- sponded to the p-carboxyvinyl (compound 1) and a-hydroxy-ptrichloromethylethyl (compound 3) derivatives of the I vinyl moiety of heme, respectively (171, and peaks 4 and 5 corresponded to two protein-bound heme adducts due to cross-linking of cysteine 93 of the p-chain to the heme (15). The co-elution of the p-chain with the a-chain after treatment with BrCC1, (Fig. 1, panel B) reflects the nearly complete alteration of the p-chain by heme, as previously described in a report on the characterization of the protein-bound heme adduct (15). In this report, with the use of a different chromatographic condition optimized for the protein-bound heme adducts, the a-chain and the altered p-chains were clearly resolved and the association of heme chromophore with the altered @-chains was evident (15). The peaks eluting between peaks 5 and 6 were unstable and were not isolated. The compound (compound 6) corresponding to peak 6 is the focus of this study.
The structure of compound 6 was determined (Fig. 2) as described below. The visible absorption spectrum of compound 6 in methanol had maxima a t 406, 500, and 620 nm. This spectrum was consistent with an intact porphyrin n system. The bathochromic shift of the Soret maximum for the heme adduct was consistent with an electrophilic substituent on the heme ring (27).
Analysis of the reaction mixtures by plasma desorption mass spectrometry showed a molecular ion a t rnlz 744 that increased in intensity when glutathione was included in the reaction mixture (data not shown). A Fourier transform mass spectrum of compound 6 showed a MH' ion a t mlz 744.1 (Fig. 3, panel B).
The cluster of ions found in the molecular ion region appeared to be due predominantly to chlorine-35 and chlorine-37 isotopes in the molecule, consistent with the distribution of ions predicted from the molecular formula of compound 6 (Fig. 3, panel  A). Furthermore, the mass spectrum of compound 6 isolated from a reaction mixture treated with I3C-enriched BrCC1, gave molecular ions with an additional 2 mass units, showing that the added carbon atoms in compound 6 were derived from 2 eq of BrCCl, (Fig. 3, panel C ) .
The 'H NMR spectrum of the altered heme (Fig. 4, panel A ) was for the most part similar to native heme (17). The four singlets near 10 ppm corresponded to the meso protons and the four singlets near 3.5 pprn corresponded to the four methyl groups. The peaks of the p-methylene groups of the propionic residues were visible between the sharp methyl peaks and the a-methylene protons gave a broad triplet at 4.5 ppm with integrated intensity of 4. However, multiplets with chemical shifts near 8.5 ppm for the a-proton and 6.0 and 6.3 ppm for the P-protons at vinyl substituents corresponded to only a single unaltered vinyl group. Thus one vinyl group was altered, giving rise to protons that exhibited a pair of doublets at 8.84 and 7.96 ppm in the NMR spectrum. The large coupling between these protons (15.5 Hz) indicated a trans stereochemistry.
To determine which vinyl group was modified and the nature of this modification, two-dimensional NOESY and COSY experiments were conducted. In the NOESY experiments, nuclei close enough to each other to allow polarization transfer through space give rise to off-diagonal peaks at their corresponding chemical shifts (28). Thus for the altered heme, as seen in Table I  methyl groups, which were thus identified as those on rings I and IV. The y-meso proton was identified by interaction with the two methylene groups of the propionic acid residues, and Further information on the nature of this modification was obtained by analysis of the heme adduct from the reaction with Brl3CCl3. In the 'H NMR spectrum (Fig. 4,panel B), the olefinic signals at 8.84 and 7.95 ppm showed evidence of further coupling. In the 13C NMR spectrum of the product (Fig. 5), doublets appeared with chemical shifts (131.5 and 117.8 ppm) characteristic of olefinic carbons and coupled by 105 Hz, a value suitable for one-bond coupling of olefinic carbons with several negative substituents (29). To provide detailed information on this coupling, we performed a FLOCK experiment, which provides a heteronuclear correlation spectrum between protons and carbons coupled to them by smaller couplings (30). When optimized for couplings near 10 Hz, the spectrum showed that only the carbon of 131.5 ppm was coupled to both of these protons by couplings of this size. Smaller couplings are observed of the carbon resonating at 117.8, 3.3 Hz (3J(H,C)) and 1.8 Hz (4J(H,C)) ( Table I). The small vicinal coupling implies that the dihedral angle in the center of the diene must be held near 90" to accommodate the bulk of the substituents. The evidence clearly shows that the vinyl group of ring I has been altered by the subst~tution of a second vinyl group bearing three chlorine atoms (Fig. 2).
It was discovered that the amount of compound 6 increased approximately %fold when reduced glutathione was added in excess of 2.5 m~ to the reaction mixture (Fig. 6, panel A), whereas addition of oxidized glutathione at a concentration of 10 m M had little effect (Fig. 6, punel B). Glutat~one did not appear to be obligatory for formation of compound 6, since red cell lysate passed over a Sephadex G-25 column to remove glutathione gave the altered heme adduct albeit in low yield ( Fig. 6, panel B, nested-smw). Similar results were obtained with purified hemoglobin from Sigma (Fig. 6, panel B 1. Studies with the use of ascorbate gave a more dramatic rise (approximately 10-fold) in the amount of compound 6 in red cell lysates (Fig. 7). The amount of compound 3 remained constant (Fig. 71, whereas heme and the other altered heme products (compounds 1, 4, and 5 ) decreased (Fig. 7). Similar results were obtained with glutathione (data not shown), except that the amount of the altered heme products corresponding to compounds 1,3,4, and 5 remained fairly constant at low glutathione concentrations (0.5-2.5 m~) , but at higher concentrations ~~ of glutathione (5-25 mM), degradation of compounds 1,4, and 5 the remaining meso protons showed interaction with the was observed while compounds 3 and 6 remained relatively constant. A marked loss of heme was found over the entire concentration range of glutathione (0.5-25 m~) with approximately 40% loss a t a concentration of 0.5 m~.

DIS~USSION
In this study the structure of the novel altered heme adduct formed in the reaction of human hemoglobin with BrCCl, in red cell lysates has been shown to be derived from the substitution of a -C(Cl)=CCl, moiety for a P-hydrogen atom on the ring I vinyl group of heme. The regiospecificity of this reaction was the same as that found for all of the other altered heme prod-ucts that have been characterized from the reaction of BrCCl, with myoglobin (17,201 and hemoglobin (151 and suggests that the initial steps involved in the formation of all the heme adducts may be similar. A mechanism that accounts for the formation of the previously characterized heme adducts of hemoglobin (Scheme 11,paneZ A ) (15) involves the initial addition of the trichloromethyl radical on the ring I vinyl moiety with subsequent delocalization of the electron to form a trichloroheme cationic species. Water or cysteine could add to this cation to form the previously characterized tric~oromethyl alcohol (compound 3) or protein-bound heme adduct (compound 5), respectively (15). The trichlorovinyl-heme species could undergo reductive dechlorination and subsequent delocalization of the electron to form a dichloro-heme cationic species, which could then form the acrylic acid (compound 1) or protein-bound heme adduct (compound 4) (15, 19, 20).
A mechanism for the formation of compound 6 that is consistent with the above mechanism is depicted in Scheme I1  (panel B ) . The dichloro-heme cationic species described above catalyzes a reductive debromination of BrCC1, to form a second equivalent of CC1, radical, which subsequently adds to the dichlorovinyl moiety. The cation radical species could abstract a €Imost likely from the protein and not water (31), and subsequently form the pentachloro compound. Another redox cycle, with reducing equivalent derived from ascorbate or glutathione, or possibly other cellular reductants, leads to intramolecular reductive dechlorination (19,201. Another redox cycle would give the ferric tetrachloro-heme anion, in a manner similar to that described for reductive metabolism of halogenated hydrocarbons by P450 cytochromes (32), with subsequent elimination of C1-to give the observed trichloro product, compound 6. Alternatively, the tetrachloro-heme radical could abstract another H' with subsequent dehydrochlorination reaction to give compound 6.
Although the addition of exogenous reductants was found to selectively enhance the amount of compound 6, which tends to P N\ Fo'+3 ,Y SCHEME II. A proposed mechanism for the formation of compound 6.
favor the multiple reductive mechanism shown above, the in-sistent with previous reports on the susceptibility of this adterpretation is complicated by the finding that reductants also duct to redox cycling and self destruction (14, 33,34). In addilead to losses of some of the altered heme products. This loss tion to this complication, it is not clear how this product can was most prominent for the protein-bound heme adduct, con-form from purified preparations of hemoglobin without exog-enous reductants, unless reducing equivalents can somehow transfer between proteins. It is evident from our current findings that multiple catalytic cycles are involved in the covalent alteration of the prosthetic heme moiety to form compound 6. Although the exact mechanism remains to be determined, it is clear that the level of reductants have a dramatic effect on the course of heme alteration, which is reflected in the profile of heme products observed. These findings should be considered in the interpretation of studies on heme alteration of other hemoproteins in biological systems where a variety of cellular reductants and reductases are usually present.