Bovine Erythrocyte Superoxide Dismutase

The complete primary structure of reduced and S-carboxymethylated bovine erythrocyte superoxide dismutase has been derived through analysis of peptides from peptic, plasmin, and hydroxylamine digests of the intact polypeptide chain, and from chymotryptic, subtilisin, and dilute acid digests of derived peptides. From these data, the following unique amino acid sequence of 151 residues was deduced, corresponding to a molecular weight of 15,600, in each of the 2 apparently identical subunits of the protein molecule. (Cmc denotes S-carboxymethylcysteine.)

cyanogen bromide digests of the reduced and S-carboxymethylated protein, and from individual tryptic digests of the two cyanogen bromide peptides, after maleylation (1). Approximately 60% of the amino acid residues were placed in sequence within these various peptides, by manual or automated sequencing procedures, and about one-half of the tryptic peptides were ordered through direct analysis of overlapping peptides, through compositional analysis of overlap peptides, or through end group analysis (Ref. 1,Fig. 6). Determination of the total primary structure reported here required characterization of the peptides derived from several other types of digests, including cleavage with pepsin, plasmin, chymotrypsin, subtilisin, hydroxylamine, and dilute acid. The assignment of the single cysteine and cystine functions to the 3 half-cystine residues per subunit, identified here as S-carboxymethylcysteine, is reported in the following article (2).

Enzyme Preparation
Bovine erythrocyte superoxide dismutase was purchased from Truett Laboratories (Dallas, Texas) and purified further by chromatography on DEAE-cellulose (3) or by passage through a sulfanilamide affinity column (4) for removal of the carbonic anhydrase.
All dismutase preparations showed a single protein band when subjected to electrophoresis on acrylamide gels (5) and had specific activities between 3500 and 4000 (AA&ho = 0.0125 per min) units per mg in the xanthine oxidase-cytochrome c assay system (3). The reduced and S-carboxymethylated derivative was prepared as described earlier (1). However, the apoenzyme was not isolated, but was prepared in situ, by including EDTA (7 to 8 mM). in the 6 M auanidine hvdrochloride-Tris (0.5 to 1.0 M) buffer sdiution used is the denaturing solvent for the reduction and subsequent alkylation reactions.
The initial steps in the primary structural analysis of bovine erythrocyte superoxide dismutase were reported in the preceding article (1). Peptides were isolated from tryptic and NH1 to COOH terminus (e.g. Pl, P2). Fragments of these peptides, produced by subsequent cleavage by a different agent, are designated by a hyphen and a similar letter and numeral abbreviation, following the designation of the parent peptide (e.g. Bl-C3). If nonspecific cleavage occurred in these subfragmentations, a lower case letter distinguishes the secondary peptides which overlap the initial fragment (e.g. T4-A2-Sla, T4-A2-Slb). The abbreviations for the cleavage agents are: A, dilute acid; B, cyanogen bromide; C, chymotrypsin; H, hydroxylamine; L, plasmin; MT, trypsin after maleylation; P, pepsin; S, subtilisin; T, trypsin.
Computer Comparison of Primary Structures Sequence relatedness was assessed semiquantitatively with a modification of a computer program described previously (12). In this program, the sequence homology between two proteins is measured by the minimum number of base differences between the RNA triplet codons for the amino acids being compared.
A comparison length (12) of 20 residues was employed.

Purification
of Pepti&s-Combinations of gel filtration on Sephadex G-50 and G-75 and ion exchange chromatography on SE-and QAE-Sephadex were, in general, effective for purification of the peptides encountered.
However, certain fractions from complex digests, most notably the peptic digest, remained impure even following successive gel filtration and cation and anion exchange chromatography.
The purification of peptides of medium-to-large size is often an obstacle to the progress of sequence analysis.
In this study, peptides containing 45 to 60 residues, from plasmin and hydroxylamine digests of RCM superoxide dismutase, were efficiently purified by chromatography on DEAE-cellulose.
To minimize adsorptive losses on the ion exchanger, urea was present in all of the elution buffers used with DEAE-cellulose.
The concentration employed, 0.5 M urea, appeared to be adequate for this function, while not being so high as to generate technical problems in desalting the chromatographic fractions. Ion exchange chromatography of peptides, utilizing urea as a non-ionic solubilizer, has been often used (13-15), but usually with higher concentrations (6 to 8 M). The tables of amino acid compositions, of peptides from the various digests, are contained in the miniprinted supplement at the end of this article.2 Sequence Analysis of Pep&s-Purified peptides were submitted to sequence analysis using the Edman degradation, in its manual or automated form, and exopeptidase digestion.
Examples of the application of these procedures appeared in the preceding article (1) ; the data which substantiated the sequence conclusions drawn here are entirely contained in the miniprinted supplement.2 Completion of the primary structural analysis required (a.) sequence analysis of unknown regions within tryptic peptides, including gaps encountered in sequenator runs, and (b) ordering of the tryptic peptides through the identification of overlapping peptides.
In tryptic peptide T4 (l), the gaps at cycles 24 and 27 (residues 44 and 47) were identified as histidine and glutamine through characterization of peptic peptide, P5. The ability to perform an extended Edman degradation of hydroxylamine peptide, Hl, with the sequenator was of considerable use in elucidating unknown regions of sequence within peptide T7. Sequenator analyses were also critical in identifying some overlap sequences for tryptic peptide alignment, and very convenient in providing the amide assignments, through direct identification of asparagine-and glutamine-PTH derivatives.
The peptic digest provided a number of small peptides which were indispensable in confirming the alignment of the tryptic peptides.
Lastly, it may be noted that the COOH-terminal section of T4 was inaccessible to sequenator degradation (1). The identification of residues 53 to 66 required subtilisin digestion of an acid cleavage fragment of T4.
Reconstruction of the Amino Acid Sequence-The complete amino acid sequence of bovine erythrocyte superoxide dismutase has been deduced from the data reported in this and in the preceding article (1). The sequence and summaries of the peptide characterizations which uniquely define it are presented in Fig. 1. The 11 tryptic peptides (1) are unambiguously aligned by direct analysis of overlapping sequences and by compositional analysis of overlapping peptides. Peptide Tl is placed at the NH2 terminus by virtue of its acetylated a-amino group and is positioned before T2 by Pl sequence data and Bl-Cl composition data. The overlap between T2 and T3 is identified by sequence analysis of Bl-C2 and that between T3 and T4 by sequence analysis of P2 and the composition of Bl-C3. The T4-T5, T5-T6, and T6-T7 alignments are demonstrated by direct sequencing of Bl-C4, L2, and P6, respectively.
Sequence data for Bl-C6 and the compositions of P7 and P8 serve to position T7 and T8, and sequence data for B2 and the composition of P9 to position T8 and T9. The alignment of T9 and TlO is established by end group analysis and the composition of B2-MT2. Sequence analysis of H2 and the composition of L3 identify the alignment of TlO before Tll. Peptide Tll is placed at the COOH terminus because its COOH-terminal residue is identical with that of B2 and also that of the intact polypeptide chain. Of these 10 tryptic peptide alignments, only that between T2 and T3 is not rigorously proven.
The single residue overlap provided by Bl-C2 and the consistent composition of Bl-MT1 do not entirely exclude the possibility that between T2 and T3 is a peptide not isolated from the tryptic digest. However, the close fit of the 3A electron density map of crystalline bovine superoxide dismutase with the proposed sequence for this regiona suggests that this unlikely loss of a tryptic peptide has not occurred.
A molecular weight of 15,600 per subunit polypeptide chain, and 31,200 for the bovine apoenzyme, is calculated from this amino acid sequence.
These values are in good agreement with molecular weight analyses of holosuperoxide dismutase by standard physical chemical procedures, which determine a value of 32,000 to 34,000 (3,16,17).
In Table I the amino acid composition of bovine dismutase, determined from the primary structure, is compared with that calculated from amino acid analysis of acid hydrolysates of the RCM protein (1). Duplicate samples were hydrolyzed for 24, 48, and 96 hours. Linear extrapolation to zero hour hydrolysis was used to calculate values for threonine and serine; valine and isoleucine values were taken from analysis of the 96-hour hydrolysates.
For all other amino acids the average of the values from the three different times were taken, except for tyrosine, which was assayed spectrophotometrically (18). Specilficity Achieved by Selective Cleavage Agents-The ease with which the primary structure of a protein is reconstructed is related in no small measure to its susceptibility to enzymatic and chemical methods for cleaving peptide bonds in a selective and limited fashion.
Scissile peptide bonds in bovine dismutase are summarized in Table II, as inferred from those peptic, plasmin,  chymotryptic, subtilisin, hydroxylamine, and dilute acid peptides which were isolated.
The peptic cleavage sites are in accordance with the known preference of that protease for cleavage on the amino or the carboxyl side of hydrophobic residues (19). 1 designates the site of cleavage cyte superoxide dismutase, and the peptides characterized for its by cyanogen bromide. The span of each peptide, which was derivation.
The half-cystine residues were identified as their isolated, is designated by a horizontal line, with arrows on either S-carboxymethyl derivatives.
Cleavage agents are abbreviated end; vertical cross-hatching appears beneath the residues which by T, trypsin; C, chymotrypsin; B, cyanogen bromide; P, pepsin; were placed in sequence by manual or automated characteriza-MT, trypsin following maleylation; A, dilute acid; S, subtilisin; tions of the specific peptide designated.
Pepsin catalyzed cleavage on both the amino and carboxyl side of Leu-36 (producing P3 and P4) and of Tyr-108 (producing P7 and P8). Of the chymotryptic cleavage sites detected, the relative lability of the Leu-Ser bond may be noted. This bond (as Leu-65 to Ser-66) was nearly quantitatively cleaved in the brief chymotryptic digest of T4, and the efficient cleavage of Leu-Ser bonds at both the NH2 and the COOH terminus of the sequence of Bl-C4 (Leu-65 to Ser-66 and Leu-104 to Ser-105) permitted its isolation in high yield. Plasmin, well known for it selective proteolysis in blood-clotting phenomena (20), may find increased use in future protein sequence studies. Although only three plasmin peptides were purified, it is evident from these alone that plasmin is a protease much more selective than trypsin, because each peptide contained one or more bonds which were susceptible to trypsin. Hydroxylamine was initially reported as a highly selective chemical cleavage agent for asparaginyl peptide bonds, showing a strong preference for 21,22). In addition to Asn-Gly, cleavage of an Asn-Ala bond has been identified in this study. Hydrolysis with dilute acetic acid has been found to be a selective method for excision of aspartyl residues from peptides (23,24). Dilute acid cleavage of T4 resulted in the expected excision of Asp-25, and of Asp-50, but showed two instances of lack of strict specificity. (a) There was some evidence for deamidation of Asn-51 and subsequent acid cleavage of the resultant Asp-51 to Thr-52 bond. Edman degra-dation of peptide T4-A2, obtained by gel filtration, showed Asx NH*-terminal, and, in addition, a smaller amount of threonine. (b) There was no evidence for excision of Asp-40 between Gly-39 and His-41. An incorrect identification of residue 40 as aspartic acid, rather than asparagine, could account for this observation. However, enzymatic hydrolysis of peptic peptide P4 was in accord with sequenator analyses of T4 in assigning aspartic acid. Ionic interaction with the adjacent imidazole side chain of His-41 may have reduced the participation of the Asp-40 side chain in intramolecular catalysis. The specificity of trypsin was in complete accord with expectations (25). Thus, cleavage was efficiently affected after all lysine and arginine residues except (a) Lys-73, which was followed by the sequence, Asp-Glu-Glu, (b) Lys-120, which was followed by a proline, and (c) Lysdl, which was the COOH terminus of the peptide chain,

DISCUSSION
In recent years, bovine erythrocyte superoxide dismutase has been actively studied in a number of laboratories (e.g. [26][27][28][29]. This copper-and zinc-containing protein has been found to be suitable for application of highly refined spectroscopic techniques (30-32) and of methods of protein chemical characterization (33, 34). The enzyme has been useful as a biological probe-of the involvement of superoxide radicals in biochemical and chemical reactions (26,35) and has been crystallized in a habit amenable to 52 60 70 77 Thr-Gln-Gly-~s-Thr-Ser-Ala-Gly-Pro-His-Phe-Asn-Pro-Leu-Ser-Lys-Lys-His-Gly-Gly-Pro-Lys-Asp-Glu-Glu-Arg- x-ray structure analysis (36). Dismutases have been isolated from different biological sources, possessing different metal COfactors (37,38), and the x-ray crystal analysis has progressed to a 3 A resolution electron density map0 (39). Amidst the present state of knowledge of bovine superoxide dismutase, the primary structure reported here will not only provide additional insight into the structure and function of this enzyme, but will also serve as a focal point for comparative structural studies among superoxide dismutases' (49). The following article (2) describes the assignment of the cysteine and cystine functions to the 3 halfcystine residues in the subunit sequence.
Some comments upon (a) the uniqueness and (b) the accuracy of the reported amino acid sequence are appropriate at this point.
Regarding the conflicting reports about the occurrence of tryptophan in bovine dismutase (16,28,41), little doubt remains that none is present. In the preceding article, it was noted that no tryptophan was found in hydrolysates of the protein (1). In all of the amino acid analyses, exopeptidase digestions, and Edman degradations of the peptides reported here, there was also no evidence for the presence of tryptophan.
The accumulated evidence from characterization of all of the peptides is summarized in Fig. 1. All of these data are consistent only with the unique amino acid sequence shown. In this primary structure investigation no peptides were isolated from digests by any of the nine different cleavage agents used which could not be placed in this sequence. Since it was not possible to isolate every tryptic peptide in quantitative yield, a definitive statement of the absolute absence of microheterogeneities cannot be made. The reported presence of 1 to 2 residues of hexose per dismutaee molecule (16) cannot be excluded, either. However, all of the serine, threonine, and asparagine residues directly identified by sequenator analysis, and 2 of the asparagine residues, whose carboxyl site was cleaved by hydroxylamine, are presumably not glycosylated.
As noted in the previous article (l), several tryptic peptidez were isolated in yields exceeding 59%. In addition, there were a number of instances where the same region of sequence was found in peptides isolated from different digests. Furthermore, some of the peptides found to be of homogeneous sequence by sequencer analysis were purified solely by gel filtration procedures which would not be expected to separate microheterogeneous forms containing amino acid substitutions at specific residue loci. Lastly, it may be noted that the 2 subunits of the dismutase molecule appear to have the same over-all conformation in the 5.5 A resolution electron density map (39) and to have the same polypeptide backbone conformation in the 3 A resolution map." All of these observations lend credence to the conclusion drawn here: that the 2 subunits of bovine erythrocyte dismutase possess the same, unique primary structure of 151 residues.
Regarding the accuracy of the sequence reported herein, the data may stand for themselves. Confidence in the accuracy of primary structural information obtained through sequenator analysis has increased as the use of this propitious instrument has increased since it was first reported (42). Although most of the sequenator analyses were performed only once, no errors were ever found when time-proven manual procedures were used to sequence identical regions in overlapping peptides. Furthermore, all of the gaps encountered were subsequently found to correspond to real residues, removed by the Edman degradation but not identified in the course of the sequenator analysis, rather Amide assignments made by direct identification of the amino acid-PTH derivatives were unambiguous.
One amide assignment made by exopeptidase digestion is possibly questionable.
Residue 47 was assigned as glutamine because (a) only a fractional amount of glutamic acid was liberated by aminopeptidase digestion of P5, while other residues were present in integral amounts, and (b) glutaminyl residues adjacent to histidine are known to be relatively susceptible to deamidation (43). The agreement is not exact between the composition of bovine dismutase, as determined from the primary structure and as identified from amino acid analysis of acid hydrolysates of the reduced and S-carboxymethylated protein ( Table I). The sequence is assuredly a more reliable means to obtain an accurate composition, and some discrepancies merit comment.
The low values for valine and isoleucine are the result of incomplete liberation of these amino acid residues, even after 96 hours of hydrolysis. The protein contains: 1 Ile-Ile, 1 Val-Val, 1 Val-Ile, 2 Ile-Val, and 1 Val-Val-Val sequence.
The low values found for lysine and histidine are unexpected, but could be attributed to alkylation during the S-carboxymethylation reaction, As noted previously (l), the lysine and histidine contents of unmodified bovine dismutase (16) are in exact accord with the results of sequence analysis.
It has been suggested that the amino acid sequences of bovine erythrocyte superoxide dismutase and carbonic anhydrase may be homologous because of the striking similarity in the magnetic circular dichroism spectra of derivatives of these proteins in which the naturally occurring zinc (II) is replaced by cobalt (II) (32). In fact, no sequence homology whatsoever was detected on comparison of the dismutase sequence with that of human carbonic anhydrase B (44) by computer technology (12). The human enzyme sequence was employed because the bovine sequence has not been reported.
In view of the striking similarity that is frequently found on comparing sequences of other proteins isolated from bovine and from human sources (45), a similar lack of homology may be anticipated between the sequences of bovine dismutase and bovine carbonic anhydrase.
It is possible that the three-dimensional conformations around the zinc binding sites of bovine dismutase and carbonic anhydrase are indeed similar, as suggested by the magnetic circular dichroism studies, while other portions of their structures differ. Such a case of local conformation identity has been observed in the virtually identical three-dimensional structures of the coenzyme binding sites in dehydrogenases of apparently unrelated amino acid sequences (46). An ultimate disposition to these projections is possible now that the crystal structure of bovine dismutase has been solved at a resolution of 3 A.a The crystal structure of human carbonic anhydrase C is known at 2 A resolution (47). No search was undertaken for sequence homologies between bovine dismutase and other proteins; however, it may be mentioned that computerized comparison procedures showed no internal homology within the dismutase sequence itself. Two repeating tripeptide sequences are evident on inspection: Lys-Gly-Asp, at residues 8 to 11 and 23 to 25, and Ile-Val-Asp, at residues 94 to 96 and 97 to 99.   The peptides from which these cleavages were deduced, are summarized in Fig. 1 Superhyperfine electron paramagnetic resonance spectra of superoxide dismutase have been interpreted in favor of 3 nitrogen ligands to each copper atom (26,31). Chemical modification of bovine superoxide dismutase, through photo-oxidation and with diazotised sulfanilic acid (34), has implicated histidine residues as part of the catalytic site of the enzyme, and in light of the spectroscopic studies, as possibly involved in metal binding.
The amino acid sequence reported here excluded the possibility that the metal ions are bound via the a-amino nitrogen atoms, for the a-amino groups are acetylated.
However, a histidine-rich sequence located within peptide T4 (His-Gly-Phe-His-Val-His, residues 41 to 46) was an obvious candidate for one or more of the metal binding ligands in this enzyme. In fact, His-44 and His-46 (as well as His-61 and His-118) are observed to be ligands of the copper atom in the 3 A resolution electron density map of bovine dismutase.a Certain regions of predominantly polar character are evident, and have been helpful in assigning exterior loops in the molecule as current x-ray crystal studies have progressed from 5.5 A (39) to 3 A resolution.a The sequence between residues 67 and 78 is strikingly polar, with ionic side chains in 9 of the 12 amino acids. A significant, but somewhat less concentrated polar region, is contained between residues 118 and 134, with 9 ionic side chains among 17 amino acids. In the 3 A resolution electron density map, both of these sequences occur in exterior loops, which are exposed to solvent.a The most outstanding feature of the amino acid composition of bovine erythrocyte dismutase, and of other copper-zinc dismutases as well (48~50), is the predominance of glycine.
In the bovine enzyme, nearly one-sixth of the amino acids are glycine, and the 25 residues of glycine per subunit are distributed relatively evenly throughout the amino acid sequence. This feature could permit many bends in the polypeptide backbone, allowing multiple contacts between nonadjacent regions in the primary structure.
Such interactions could contribute to the remarkable stability of this dismutase.
In the 5.5 A resolution electron density map of the crystalline enzyme, sharp turns in the subunit polypeptide chain are in fact evident in at least four places (39). In the 3 A resolution electron density map, these and other turns have been observed, and glycine residues have been found in most of them.a The most notable difference between the amino acid composition of the copper-zinc bovine dismutase and that of the bacterial manganese-and iron-containing dismutases (38, 51, 52) is the preponderance of alanine and the lesser amount of glycine in the bacterial enzymes. Although dismutases of both classes of metal cofactors have comparable subunit molecular weight and an identical catalytic action (26), the bacterial enzymes do not display the unusual stability typified by the bovine erythrocyte enzyme (26). It may be queried whether the three-dimensional structures of the two classes of dismutase enzymes are distinctly different.
To answer these intriguing questions of protein evolution and the chemical basis of dismutase action, comparative studies have been initiated through amino acid sequence analysis (40) and x-ray crystal structure determination.a