Synthesis of either Fe- or Mn-superoxide dismutase with an apparently identical protein moiety by an anaerobic bacterium dependent on the metal supplied.

Superoxide dismutase of Propionibacterium shermanii, an anaerobic that produces an iron superoxide dismutase, was purified from cells grown in iron-free conditions. The enzyme isolated was found to contain manganese and to have spectral and catalytic properties very similar to those of typical Mn-superoxide dismutases. Its electrophoretic mobility, molecular weight, and subunit size were identical with those of the Fe-enzyme. Amino acid compositions were practically indistinguishable in either case. The NH2-terminal sequence was found to be identical. The catalytic activity of an apoprotein sample prepared from the purified holoenzyme was restored by adding either Mn(II) or Fe(II). Only the metal/protein ratio varied from approximately 1 per subunit in the case of the Fe-enzyme to approximately 2 for the Mn-enzyme. It is concluded that this bacterium can accommodate either Fe or Mn on identical, or very slightly dissimilar, proteins forming active sites with the properties found in specific metallodismutases.

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance wit.h 18 U.S.C. Section 1734 solely to indicate this fact.

MATERIALS AND METHODS
Propionibacterium Jreudenreichii sp. shermanii PZ3 was grown in a complex medium as previously described (5) and in an iron-free synthetic medium containing MnSOI, CoC12, ZnCL, and CuSO, as transition metal salts. Superoxide dismutase activity was tested and units were defined by the cytochrome c-xanthine oxidase method (6).
For the isolation of the enzyme, typically 400-450 g of frozen cell paste was suspended in 1 liter of 0.05 M K phosphate buffer, pH 7.8, and then disrupted by repeated pressing with a Manton-Gaulin press. DNA was removed by incubation with 20 mg of DNase (Serva) in the presence of 1 mM MgClp for 6 h at 30 "C. The suspension was then exposed to controlled heat treatment, for 3 min at 70 "C, after addition of 0.1 M KC1 to the buffered suspension. After centrifugation of denatured protein, the supernatant was extensively dialyzed against the same buffer. To the dialyzed solution, ammonium sulfate was added in steps a t 4 "C. Superoxide dismutase activity was optimally recovered in the fractions precipitated in the 70-9576 concentration range. This ammonium sulfate precipitate was exhaustively dialyzed against 0.05 M phosphate buffer, pH 7.8, and the dialyzed solution was passed through a Sephadex G-100 column (5 X 80 cm). Active fractions were pooled, concentrated by filtration through an Amicon PM 30 membrane and then applied to a DE50 column, equilibrated with the usual buffer. Protein was eluted by a linear 0.2-0.9 M KC1 gradient in the same buffer. The active fractions were pooled, concentrated, and rechromatographed on the Sephadex G-100 column. The purification data are shown in Table I. Electrophoresis was performed on polyacrylamide gels as described by Davis (7) and in the presence of sodium dodecyl sulfate, as described by Weber and Osborn (8). Gels were stained for superoxide dismutase activity as previously described (91. Protein concentration was estimated colorimetrically (10). ESR spectra were recorded on a Varian E-9 spectrometer and optical spectra on a Beckman UV 5230 spectrophotometer. Metals were determined by atomic absorption spectrometry with a Hilger and Watts Atomspek, model H 1120. Ultracentrifuge analysis was conducted as previously described (1 1) on a Beckman model E analytical ultracentrifuge. A partial specific volume of 0.738 ml/g was calculated from the amino acid composition and used for the molecular weight calculations. Amino acid analysis was performed with an LKB 4400 amino acid analyzer equipped with a Spectra-Physics System I computing integrator. The NHn-terminus was determined according to Gray (12). NHZ-terminal sequence analysis was performed with an LKB 4020 solid phase sequencer converted to microsequencing and programmed as described by Hughes et al. (13). Protein samples were demetallized by treatment with 70% formic acid for 2 h at 4 "C, followed by dialysis against 5070 acetic acid, lyophilized, and then carboxymethylated (14). Coupling of samples was carried out with aminopropyl-glass activated by reaction with pphenylene diisothiocyanate, according to the procedures recommended in the LKB Solid Phase Sequencing Handbook. Identification of the 4-N,N-dimethylaminoazobenzene-4'-thiohydantoins was carried out by TLC using solvent systems I1 and 111 of Wilson et al. (15).

RESULTS AND DISCUSSION
Superoxide dismutase, either purified from iron-supplemented or iron-deprived cells, gave a single band on polyacrylamide gel electrophoresis with the same mobility (Fig. 1). The The increase of total activity over that of the crude extract after some preliminary purification steps is often observed in bacterial superoxide dismutase (23). It has recently (25) been ascribed to a production of 0 2 by membrane fragments still present in crude extracts and artifactually minimizing the real superoxide dismutase activity. Production of 0, by P. shermanii membranes was actually observed (unpublished data from this laboratory). Therefore, the total activity of the second step is taken as the starting reference point in the yield column. One and 10 pg of protein were applied to each gel for activity and protein staining, respectively.

In Vivo Exchange of Mn for Fe in a Superoxide Dismutase
same elution volume was measured for either protein by gel exclusion chromatography (Fig. 2), indicating a comparable molecular weight of approximately 73,000. This value was confirmed by ultracentrifuge analysis (25000). Sodium dode-cy1 sulfate-polyacrylamide gel electrophoresis of enzyme samples purified from either cell culture and denatured by 10 min of boiling in the presence of P-mercaptoethanol gave a single band corresponding to a subunit molecular weight approximately half that of the native enzyme (232,000). Metal analysis of purified proteins gave a content of 0.13% iron and 0.02% manganese for the enzyme isolated from iron-containing cultures, and of W8%, manganese and <0.01% iron from the enzyme purified from iron-deprived cells. Thus, Mn is present in the superoxide dismutase synthesized in the absence of iron as approximately 1.8 Mn per subunit, that is nearly twice the metal/protein ratio of the Fe-superoxide dismutase.
The incorporation of iron into superoxide dismutase was verified as occurring as well when iron salts were added to the synthetic medium used for the iron-lacking culture. Therefore, synthesis of either Mn-or Fe-superoxide dismutase was not dependent on the medium, but only on metal availability.
The questions arise whether: (i) the Mn-enzyme synthesized by iron-deprived cells has properties typical of Mn-superoxide dismutases already described; (ii) the Fe-and Mn-enzymes have comparable activity; (iii) the two enzymes consist of an identical protein moiety, as suggested by the same electrophoretic mobility, molecular weight, and subunit size. In particular, they indicate for the Mn-enzyme of P. shermanii the presence of Mn(II1) (16), which is unique to superoxide dismutase.
(ii) The catalytic rate constants were measured at pH 7.4 by the comparison procedure of Forman and Fridovich (18) and gave h = 8.5 X 10" M"S" for both the Fe-and Mnenzyme. These values are comparable to those reported for typical Fe-and Mn-superoxide dismutases (16-20).
(iii) The amino acid composition of the two proteins is shown in Table 11, together with those of Mn-and Fe-super-  oxide dismutases from E. coli, which are established to be different proteins, in spite of their high level of homology. In the case of E. coli superoxide dismutases, clearcut differences can be seen in the half-Cys and Met content and significant differences (~2 5 % ) in the Leu and Lys content. No such differences could be observed in the case of P. shermanii proteins. Furthermore, the NHs-terminal sequence was found to be identical for both proteins, namely Ala-Val-Tyr-Thr-Leu-Pro-Asp-Leu-Pro-Tyr-Asp-Tyr. This sequence is homologous to that of other Fe-or Mn-superoxide dismutases (2), nevertheless it is different from any other reported sequence, by at least 3 residues from the closest one, and by 5-7 from most of them. In a large series of NH2-terminal sequences of this class reported recently (2), this is the average difference in the first 12-residue segment between any two members, including Mn-superoxide dismutases from the same bacterium, i.e. E . coli (4 differences).
An obvious way to verify the strong indication that the same protein can accomodate in P. shermanii either Fe or Mn is to prepare the apoenzyme from the purified holoenzyme and then show that either Mn or Fe ions can restore the catalytic activity. However this approach proved very difficult with this protein. Dialysis against EDTA at low pH, in the presence of variable amounts of urea or guanidine, did remove the metals, but the protein precipitated upon raising the pH back to neutrality. Diethyldithiocarbamate produced metalfree derivatives unable to reconstitute the holoprotein. CNwas totally ineffective. o-Phenanthroline, in the presence of ascorbate, gave a substantial metal removal and this apoprotein was able (Table 111) to recover catalytic activity upon addition of either Fe(I1) or Mn(I1).

CONCLUSIONS
From these results it appears that P. shermanii can produce, as a function of the metal supplied, either Fe or Mnsuperoxide dismutases, which reproduce spectral properties and catalytic efficency of typical Fe-and Mn-superoxide dismutases. This seems to exclude a strict species specificity of Mn-and Fe-superoxide dismutase, as well as a dependence of the type of metal on special conditions of growth other than metal availability. As a matter of fact, no certain rule has been provided so far to predict the presence of either enzyme in different bacteria. E. coli contains both enzymes, but only the Mn-enzyme is inducible by oxygen (4). Mn-and Fe enzymes are present in both Gram-positive and Gram-negative bacteria, although Gram-negative bacteria tend to contain Fe-superoxide dismutase and most Gram-positive bacteria contain only Mn-SOD (15). Only one rule seemed to apply: no Mnsuperoxide dismutase has so far been reported in anaerobes, but even this generalization is apparently contradicted by the synthesis of a typical Mn-superoxide dismutase in iron-deprived cultures of the anaerobe P. shermanii.
Another relevant result of the present work is the strong indication of identical protein moieties for either Feor Mnsuperoxide dismutase of P. shermanii. Only the full amino acid sequence can give unequivocal evidence in this regard, but all the data reported here agree on substantial identity. In particular, Mn-and Fe-superoxide dismutases with identical electrophoretic mobility (Fig. 1) have been reported only in case of artificial metal substitution. Moreover, the apoenzyme of P. shermanii could be reconverted into an active holoenzyme by addition of either Mn(I1) or Fe(I1). This result is not the rule with other Feor Mn-superoxide dismutases, which have a rather strict metal specificity and, when they bind other metals, do not produce enzymatic active derivatives (3,24). Of the two naturally produced P. shermanii superoxide dismutases, the Mn-superoxide dismutase had a higher metal/ protein ratio. Nevertheless, addition of Mn to the purified Feprotein did not lead to binding of extra Mn. In turn, addition of excess iron to the Mn-protein did not lead to either iron binding or manganese loss. Conformational effects of Fe-or Mn-binding may play a role in the properties of the protein moiety, as also suggested by the significantly less stability The purified holoenzyme was dialyzed for 24 h at 4 "C against 1 mM o-phenanthroline and 10 mM ascorbate in 0.05 M acetate buffer, pH 5.5. A longer dialysis time led to precipitation.
The apoprotein was dialyzed for 24 h at 4 "C against I mM FeSOd or MnS04 at pH 5.5 and then exhaustively against many changes of 0.05 M phosphate buffer, pH 7.8.
In Vivo Exchange of M n for Fe in a Superoxide Dismutase observed in the case of the Mn protein, for example, to denaturation a t slightly alkaline pH.