Isolation and Characterization of the Unique Prosthetic Group of a Green Hemoprotein from Human Erythrocytes*

A simplified and streamlined purification scheme has been developed for the large scale isolation of a green hemoprotein from human erythrocytes. The isolation procedure involves hypotonic lysis, freezing of the hemolysate at -6O”, centrifugation, direct chromatography of the supernatant on DEAE-cellulose, and subsequent cation exchange, anion exchange, and gel filtration chromatography. Approximately 2 pmol of this anionic hemoprotein were isolated per liter of packed erythrocytes. The previous inability to separate the prosthetic group from the protein by conventional solvent extraction procedures was shown to be a consequence of the highly polar character of the heme and not due to covalent linkage between heme and protein. The polar nature and marked lability of the heme necessitated development of techniques for the extraction, purification, and derivatization of the prosthetic group. The heme was separated from the protein by membrane filtration in the presence of pyridine and alkali or by disc gel electrophoresis in the presence of cyanide. The heme was methylated with trimethyloxonium tetrafluoroborate and further derivatized. The heme derivatives were purified on columns of Sephadex LH-20 or alumina. of the

A simplified and streamlined purification scheme has been developed for the large scale isolation of a green hemoprotein from human erythrocytes. The isolation procedure involves hypotonic lysis, freezing of the hemolysate at -6O", centrifugation, direct chromatography of the supernatant on DEAE-cellulose, and subsequent cation exchange, anion exchange, and gel filtration chromatography. Approximately 2 pmol of this anionic hemoprotein were isolated per liter of packed erythrocytes. The previous inability to separate the prosthetic group from the protein by conventional solvent extraction procedures was shown to be a consequence of the highly polar character of the heme and not due to covalent linkage between heme and protein. The polar nature and marked lability of the heme necessitated development of techniques for the extraction, purification, and derivatization of the prosthetic group. The heme was separated from the protein by membrane filtration in the presence of pyridine and alkali or by disc gel electrophoresis in the presence of cyanide. The heme was methylated with trimethyloxonium tetrafluoroborate and further derivatized. The heme derivatives were purified on columns of Sephadex LH-20 or alumina.
Chromatography of the heme, heme methyl ester, acetylated heme ester, and the corresponding porphyrin derivatives suggests that the heme contains three carboxyl groups and one or more polar, acetylatable functional groups, probably hydroxyl groups. Spectral characterization of these compounds, as well as the derivatives resulting from reaction of the heme with NH,OH, NaHSO,, and Na,S,O,, show that the prosthetic group is a previously undescribed, formyl-containing heme that can be clearly distinguished from heme a, Spirogruphis heme, and all other naturally occurring prosthetic groups.
A green hemoprotein with spectral properties distinct from any previously known protein was isolated from human erythrocytes by Morrison in 1961 (1). The protein was obtained in soluble form by freezing and thawing red cells that had been suspended in water containing a little toluene. In collaboration with Dr. Morrison, we further purified and studied this protein (2 tion procedures failed to liberate the heme from the protein, and the side chains of the heme appeared to be highly labile. In this paper we report greatly improved yields of hemoprotein by a simplified procedure suitable for very large scale isolations. We also report the extraction of heme from the protein, the purification of the heme, and the properties of this previously unreported, highly atypical prosthetic group. Preliminary reports of some phases of this work have previously appeared in abstract form (3)(4)(5)(6). RESULTS AND The procedure described here is greatly improved over that described previously (2) in that larger preparations can be carried out more rapidly at less cost with a greater and more consistent yield. The greatly improved yield was in part a consequence of the substitution of DEAE-cellulose chromatography for the time-consuming cation exchange chromatography as the first step in the procedure and substitution of ultrafiltration for the ammonium sulfate precipitation and dialysis procedures. The use of a pH gradient in place of an ionic strength gradient to elute the hemoprotein from DEAE-cellulose greatly reduced the amount of contaminating salt which had to be subsequently removed from the sample and thus the time necessary to accomplish this desalting. However, the major factor responsible for allowing us to consistently obtain a good yield of hemoprotein was the use of the -60" treatment.
We have demonstrated that freezing at -20" or omission of the freezing step resulted in no recovery of the green hemoprotein by the described procedure.
The previously observed inconsistency of yield following rapid lysis in a dry ice-acetone bath (2) undoubtedly resulted from variation in the final temperature of the hemolysate. The lability and the highly unusual properties of the heme prosthetic group of this protein necessitated the application of new techniques for its isolation, purification, and derivatization. The classical extraction of heme prosthetic groups into nonpolar solvent in the presence of acid was unsuitable for this protein because of the heme's acid lability and insolubility in nonpolar solvents.
Our isolation of heme was accomplished by dissociating the heme from the protein with alkali and then separating heme from protein by membrane filtration, or by dissociating with cyanide and separating by disc gel electrophoresis. Likewise, the classical methods for heme purification were unsuitable for this water-soluble heme, and it was necessary to use anion exchange chromatography for purification. Derivatization of the acid-labile and polar heme to products which could be highly purified and characterized by mass spectroscopy was not possible by methods in the literature. The only published methods for heme esterification use nonpolar solvent and mineral acids, conditions which were shown to alter the structure of the prosthetic group described in this paper; diazomethane has been reported to be completely unsuitable for heme methylation (16). Whereas methylation of porphyrins with diazomethane proceeds under mild conditions and without problem, a good preparation of porphyrin from the heme of the green hemoprotein was not possible because the acidic conditions for the conversion result in heme degradation.
Thus, the present study of the prosthetic group was made possible by the development of the method for heme methylation in neutral, aqueous solution with trimethyloxonium tetrafluoroborate (5,15). The methylation appeared to proceed to near completion, as evidenced by the chloroform solubility of the heme compounds after the reaction and by the paper chromatographic analysis of the single identifiable product (Table IV)  with the same RF as protoheme. In analogy with heme a and protoheme, reaction of the prosthetic group with trimethyloxonium tetrafluoroborate or with methanol/sulfuric acid yielded a product that migrated at the solvent front in the lutidine/water/ammonia system and migrated away from the origin in the propanol/kerosene system (Table IV). Such chromatographic behavior shows that the ionic groups on the molecule are completely methylated by these procedures.
Since the trimethyloxonium tetrafluoroborate reaction in aqueous solution is specific for carboxyl groups (14), and since carboxyl groups are the only ionic residues which have been found on heme prosthetic groups, the chromatographic data are best explained by the presence of three carboxyl groups. Presence of Formyl Group in Conjugation with Tetrapyrrole Ring-The hemoprotein, the isolated heme, and the heme methyl ester gave similar pyridine hemochrome spectra with absorbance maxima at wavelengths longer than those of monoformyldeuteroheme, but shorter than those of diformyldeuteroheme (Table I) Tables I and II). The spectral findings suggest the presence of one formyl group and one weaker electron-withdrawing group. The isolated prosthetic group is similar to formylvinyldeuteroheme in terms of pyridine hemochrome spectra of the heme and heme oxime (Table III). The neutral spectrum of the porphyrin derived from the prosthetic group has absorbance maxima similar to those of porphyrin isomers with one formyl and one vinyl group and also to porphyrins with one formyl group and one acetyl group (Table II). However, the "etio"-type spectrum (peak intensities of IV > III > II > I) shown by the porphyrin is in contrast to the "oxorhodo" spectra (III > II > IV > I) of the P-formyl,-B-vinyl and the 2-acetyl,6-formyl derivatives of deuteroporphyrin II and the "rhodo" spectra (III > IV > II > I) of the 2formyl,4-vinyl and the 2-vinyl,4-formyl derivatives of deuteroporphyrin IX. Of the model compounds with similar absorbance maxima reported in the literature, only 2(4)-formyl, 4(2)-acetyldeuteroporphyrin IX (with its electron-withdrawing groups on adjacent pyrroles) shows the same etio spectrum. The data suggest that the prosthetic group under study possesses one or more electron-withdrawing groups on the pyrrole (or pyrroles) adjacent to the formyl-substituted pyrrole and that the "rhodofying" effect of this group (or groups) must be greater than that of a single vinyl group.
Presence on Heme of Acetylatable, Polar Group-Methylation converts the very water-soluble and chloroform-insoluble prosthetic group into a methyl ester which is insoluble in water and soluble in chloroform.
As described above, the chromatographic properties show that the methylated product contains no free carboxyl group. However, the products resulting from methylation with oxonium salt, methanol/sulfuric acid, and methanol/hydrochloric acid all migrate on paper chromatography in nonpolar solvents at a slower rate than the methyl esters of protoheme, heme a, and 2-hydroxymethyl,4-vinyldeuteroheme ( hydroxyl group. Chromatography in the propanol/kerosene system shows that the methylation product resulting from reaction with oxonium salt has greater polarity than the products of the two other methylation reactions. It would appear that the acidic conditions of the methanol/sulfuric acid and methanol/hydrochloric acid reactions result in the modification or cleavage of a polar portion of the molecule.
Acetic anhydride converts the methyl ester obtained by the oxonium salt reaction from a polar compound with an RF of 0.05 in the propanol/kerosene system to a derivative with an RF of 0.72, demonstrating that one or more polar groups of the prosthetic group are acetylatable and suggesting the presence of multiple hydroxyl groups or possibly an amino group. However, even after the methylation-acetylation reaction sequence the derivative is still a more polar compound than 2,4-diformyldeuteroheme as evidenced by its failure to migrate in the chloroform/kerosene solvent. It is unclear whether this polar, acetylatable group is the same or a different group than the second electron-withdrawing group.
Evidence that Heme of Green Hemoprotein is Distznct from all Known Prosthetic

Groups and Synthetic
Hemes-The prosthetic group of the green hemoprotein is clearly distinguished by spectral properties and reactivity from protoheme, heme c, the prosthetic group of lactoperoxidase (26), siroheme (27), and all other hemes without carbonyl groups in conjugation with the tetrapyrrole nucleus. Moreover, this prosthetic group is distinguished from heme a, by pyridine hemochrome spectrum, water solubility, chloroform insolubility, and paper chromatographic properties of the heme, by the pyridine hemochrome spectrum of the heme oxime, by the paper chromatographic properties of the heme methyl ester, and by the spectrum of the porphyrin in nonpolar solvents. These same properties show the heme to be distinct from synthetic mono-and diformylhemes, synthetic mono-and diacetylhemes, and the cryptohemesp derived from protoheme (28).
Whereas the prosthetic group is spectrally similar to Spirogruphis heme (2- Gference prior to the isolation of heme a and was initially postulated by Negelein to be the prosthetic group of cytochrome oxidase (29). The isolation of cryptoheme a has been reported from several sources and by a number of laboratories but the significance of these findings has yet to be explained (19,(30)(31)(32). This compound appears to be a 2,4-substituted derivative of deuteroheme IX with a formyl group, an olefinic group, and a large molecular weight side chain (33). As shown in Tables I  and III, cryptoheme a and the new heme from the erythrocyte protein show similar pyridine hemochrome spectra before and after oxime formation.
However, in contrast to the properties of the new heme, cryptoheme a is soluble in diethyl ether, is stable in acid, migrates as a dicarboxylic tetrapyrrole, does not contain a polar side chain, and gives a porphyrin which shows a "rhodo"type spectrum (33). Thus, the heme we have isolated is not cryptoheme a but the hemes may have structural features in common.
Since acetic acid converts the prosthetic group to a derivative which then behaves on paper chromatography as a dicarboxylic heme, it is even possible that cryptoheme a is derived from this heme by the acid conditions of heme extraction. We conclude that the prosthetic group of the erythrocyte hemoprotein contains one formyl group, a polar and acetylatable side chain, and either a ketonic or olefinic side chain. The best candidate for the polar side chain would be a moiety with multiple hydroxyl groups. The formyl group and the second electron-withdrawing side chain must be on adjacent pyrroles. Since all mammalian heme prosthetic groups studied to date are derivatives of deuteroheme IX substituted in the 2, 4, and, in some cases, the 8 position, we postulate that the heme under study has the structure CH3> TR2 C"2 C"2

Loo-L
We further suggest as a working hypothesis that R, is a ketonic group, R, is a formyl group, and multiple hydroxyl groups are present on R, or R1.