Derivatives of Clostridium acidi-urici Ferredoxin Containing Altered Amino Acid Sequences SEMISYNTHETIC SYNTHESIS, BIOLOGICAL ACTIVITY, AND STABILITY

The semisynthetic syntheses and some properties of derivatives of Clostridium acidi-urici ferredoxin that contain amino acid deletions or replacements in the peptide chain are described. All 16 stable derivatives prepared, with the exception of [Trp’lferredoxin, were fully active as electron carriers in the two enzymatic assay systems tested: the phosphoroclastic system and the ferredoxin-dependent reduction of cytochrome c. [Trp2]Ferredoxin had 7O':i of the activity of native ferredoxin in both assay systems. The stability in aerobic solution of [Ala’lferredoxin, which had had its natural alanyl NH,-terminal residue removed and then replaced chemically, is the same as that of the native ferredoxin (half-life of approximately 54 days). The relative stabilities of derivatives with a replacement or deletion of the NH,-terminal residue are as follows: The data indicate that a large bulky residue, but not a negatively charged residue, is tolerated in position 1 the peptide

The single iron-sulfur cluster in Chromatium high potential iron-sulfur protein appears identical at the present x-ray resolution to the two (Fe&S,*)-clusters in P. aerogenes ferredoxin (15) yet the oxidation-reduction potentials of the two proteins are +350 and -427 mv, respectively (16,17). The nature of the polypeptide chain must, therefore, have a profound effect on the oxidation-reduction properties of these (Fe,S,*)-containing proteins, since the geometry of the clusters is apparently the same. Carter et al. (15)

Ferredorin
Actioity Assays-The ability of the various ferredoxin derivatives to function as electron carriers was assayed in the phosphoroclastic enzyme system as described by Rabinowitz (4. 38). The ferredoxin-dependent cytochrome c reduction assay was also used as described previously (30 and purified as described under "Methods." In Fig. 1 the spectrum of native C. acidi-urici ferredoxin is compared with the three general types of absorption spectra that were found for derivatives modified in position 1 or 2 of the peptide chain. Spectrum 2, Fig. 1 It has the same characteristics as native ferredoxin shown in spectrum 3, Fig. 1. At the same concentration, the spectra of des- at the same concentration as holoferredoxin in spectrum 3, Fig. 1. As previously reported (5), most of the absorbance at 280 nm of ferredoxin is from the (Fe,S,*)-clusters.
However, as shown in Fig. 1 some information on the aromatic amino acid content of a purified ferredoxin may be derived from an examination of its absorption spectrum.
The amino acid compositions of the reconstituted derivatives with substitutions in positions 1 or 2 of the peptide chain are given in Tables I and II, respectively. Prior to analysis, the iron and sulfide were removed by precipitation of the protein with 5% trichloroacetic acid (5). The amino acid analyses of the apoferredoxin derivatives prior to reconstitution and purification were similar to those in Tables I and II. However, the yields of derivatives were generally lower than the 70% yield reported (5) for the reconstitution of untreated native C.
This suggests that some degradative side reactions may occur during the synthesis. Material with apparently the correct amino acid composition, but that cannot be reconstituted to form a stable ferredoxin, may also be formed to a small extent. This material is removed by the purification prscedure after reconstitution. These analyses and spectra indicate that the methods used are reliable for the preparation of these ferredoxin derivatives.

Activity of Derivatives in Phosphoroclastic
Assay-The ability of the various derivatives to function as electron carriers in the phosphoroclastic assay system was determined. All of the derivatives containing substitutions or deletions in position 1 have the same activity as native C. acidi-urici ferredoxin (Fig. 2) Assay-In view of the suggestions that aromatic residues may function in electron transfer in clostridial-type ferredoxins (12, 28) and the possibility that electron transfer might not be the rate-limiting step in the phosphoroclastic assay system, the activity of some of the derivatives was tested in a ferredoxindependent cytochrome c reduction assay. This "artificial" system involves the reduction of ferredoxin by spinach ferredoxin-TPN reductase in the presence of TPNH. The reduction of cytochrome c by the reduced ferredoxin is measured by the increase in absorbance at 550 nm. In both the cytochrome c and phosphoroclastic assay systems ferredoxin acts catalytically. The activity of some of the derivatives tested in the cytochrome c assay is shown in Fig. 5  "The numbers in parentheses are the number of residues of each amino acid found in the sequence of C. acidi-urici ferredoxin by Rall et al. All of these derivatives except [Glu']-and des-Ala'-ferredoxin are as active as native ferredoxin. Preincubation experiments indicated that in very dilute solutions, the latter two derivatives were rapidly inactivated and had halflives of less than 1 min under the assay conditions. Therefore, the apparently diminished activity of [Glu'lferredoxin (80%) and des-Ala'-ferredoxin (40%) probably results from their rapid denaturation under these assay conditions. Other experiments shown later also indicate that des-Ala'-and [Glu'lferredoxin are very unstable.
Stability of Deriuatiues-The stability of the various derivatives in solution was determined as described under "Methods." The decrease in activity with time of the derivatives was followed with the phosphoroclastic assay system. The loss of A,,, of the derivatives was also examined, but it was a reliable indication of the integrity of the ferredoxin only until one-third to one-half of the nntial A,,, was lost. Thereafter, as previously reported by Hong and Rabinowitz (37) the biological activity of ferredoxin is lost more rapidly than the A,,, indicating that degraded ferredoxin still has some residual A,,,. All of the derivatives except for [Ala'lferredoxin, which had its natural Ala' chemically removed and then replaced and served as a "control" of secondary reactions during the chemical synthesis, were less stable than native C. acidi-urici ferredoxin. Table III shows the stabilities of the derivatives with amino acid replacements in position 1 of the peptide chain, as well as the stability of the extremely labile des-Ala'-ferredoxin that lacks its NH*-terminal alanyl residue. These results indicate that the diminution of the length of the polypeptide chain by a single residue or the nature of the NH,-terminal amino acid residue has a great affect on the stability of this ferredoxin. The stabilities of derivatives with amino acid replacements in position 2 of the peptide chain are shown in Table IV. The derivatives containing Ala' and aromatic amino acids other than tyrosine in position 2 have approximately one-half the stability of native C. acidi-urici ferredoxin.
As previously reported (30), an aromatic residue is not required in position 2, but the nature or presence of the residue in this position is important since neither [Gly2]-nor des-(Ala'-Tyr*)apoferredoxin formed stable derivatives upon reconstitution. The half-lives of all of the derivatives and of native C. acidi-urici ferredoxin were generally 2 to 3 times longer if the derivatives were stored under an atmosphere of argon (Tables III and IV). The absolute stability of these proteins probably depends on the degree of anaerobiosis. Although care was taken to remove oxygen, these samples were probably not completely oxygenfree since it is doubtful that the procedure maintains strictly anaerobic conditions. DISCUSSION At least three types of iron-sulfur proteins that contain (Fe&S,*)-clusters are known. Ferredoxins from Desulfouibrio gigas (40), Bacillus polymyxa (41, 42), and Desulfovibrio desulfuricans (43) most likely contain a single (Fe,S,*)-cluster and have oxidation-reduction potentials of approximately -400 mv. The high potential iron-sulfur protein from Chromatium also contains a single (Fe,S,*)-cluster (44) but has an oxidation-reduction potential of +350 mv (16). Clostridial-type ferredoxins, as shown for P. aerogenes ferredoxin (12,13), contain two (Fe,S,*)-clusters and have oxidationreduction potentials in the range of -383 to -490 mv (8,14). X-ray crystallographic analysis (15) indicates that the (Fe,S,*)cluster in Chromatium high potential iron-sulfur protein is structurally the same as the two (Fe&S,*)-clusters in P. aerogenes ferredoxin despite the greatly different oxidationreduction potentials of the two proteins. Herskovitz et al. (45) Synthesis of C. acidi-urici

Ferredoxin
Derivatives 1679  cThe cysteine values reported are those found in the normal hydrolysate and are generally approximately 75% of the value found for cysteic after performic acid oxidation of native apoferredoxin. d This value represents 3-NHl-tyrosine, which elutes near histidine and was clearly resolved from other amino acids.
proposed that the oxidation states of the (Fe,S,*)-clusters in the reduced form of high potential iron-sulfur protein and the oxidized form of P. aerogenes ferredoxin are the same. Carter et al. (15) have advanced a "three-state hypothesis" for the (Fe,S,*)-cluster and proposed that his difference in oxidationreduction potentials of the two proteins may be explained by the existence of different sets of oxidation states of the (Fe,S,*)-clusters in each protein. Cammack (46) has provided support for this by showing that Chromatium high potential iron-sulfur protein in 70% aqueous dimethyl sulfoxide can be reduced with dithionite to a form that exhibits a reduced ferredoxin-like electron paramagnetic resonance signal. This oxidation state of high potential iron-sulfur protein is different from the reduced or oxidized forms isolated normally (16).
These compounds contain clusters that are structurally and electronically similar to the (Fe,S,*)-clusters in P. aerogenes ferredoxin and Chromatium high potential iron-sulfur protein at appropriate oxidation states. Further studies on these model compounds should contribute even more to understanding the electronic properties of (Fe&S,*)- clusters. Of course, the synthesis of these compounds is proof that a peptide chain is not necessary for the existence of an (Fe,S,*)-cluster.
However, to date, no water-soluble model compounds have been reported. Furthermore, it is not known how the peptide chain constrains the oxidation states normally present in proteins containing (Fe,S,*)-clusters and if there are important interactions of amino acid side chain groups with the (Fe,S,*)-cluster.
Generally, alanine has been found to be the NH,-terminal residue of clostridial-type ferredoxins (18)(19)(20)(21)(22)(23)(24)(25). Peptostreptococcus elsdenii ferredoxin contains Met' and, to date, is the only known exception (26). In the structure of P. aerogenes The stabilities were determined as described in the text by following uith time the loss in biological activity of ferredoxin and its derivatives. t, is equal to the time in days for one-half of the activity to be lost. The proteins were approximately lo-' M in 0. ferredoxin determined by x-ray methods by Adman et al. (12), the o-amino group of Ala' is hydrogen-bonded to the y-car-boxy1 of Aspa'. The position of Asp3' in P. aerogenes ferredoxin is homologous to position 39 of other clostridial-type ferredoxins and either Asp (18-21, 23, 24, 26) or Glu (22,25) occur in this The stabilities were determined as described in the text by following the loss in biological activity of ferredoxin and its derivatives. tH is equal to the time in days for one-half of the activity to be lost. The proteins were approximately lo-@ M in 0. IPro*]- aerogenes ferredoxin is located between Asp'? and Asp", the COOH terminus of the protein (12). The studies reported in this paper show that the presence or nature of the NH,-terminal amino acid residue of C. dcidi-urici ferredoxin does not affect the electron transfer ability of this protein in the assays tested (Figs. 2 and 4). The aerobic stability in aqueous solution of C. acidi-urici ferredoxin is, however, greatly affected by the nature of the NH, terminus of the protein (Table III). [Ala'lFerredoxin, which had its normal Ala' chemically removed and replaced, is fully active and stable in,dicating that the procedures used are reliable for the synthesis of derivatives. one-half as stable as is native ferredoxin.
Assuming the tertiary structure of C. acidi-urici ferredoxin is like that of P. aerogenes ferredoxin, possibly there may be an unfavorable interaction of the electronegative sulfur atom of Met' with the nearby carboxyl group of Asps0 and the COOH terminus of this protein, Ala". This would also explain the greatly decreased stability of [Glu'lferredoxin (17%) since the y-carboxyl group of Glu' would very likely interact with the nearby carboxyl groups. [Gly'lFerredoxin is less stable (39%) than native ferredoxin suggesting that the methyl group of Ala' in native ferredoxin is important in shielding the nearby (Fe,S,*)-cluster or is necessary for providing a barrier between the carboxyl groups of Asp" and Ala55. The greatly decreased stability of des-Alar-ferredoxin (2%) probably results partly from the same factors responsible for the decreased stability of [Gly'lferredoxin.
More important, des-Alar-ferredoxin does not have an a-amino group in the correct position to form a hydrogen bond, so that the NH, terminus probably is not held to the body of the protein. Hong and Rabinowitz (37) previously found that derivatives of C. acidi-urici ferredoxin containing an additional amino acid residue on the NH, terminus or derivatives with the a-amino group of Ala' blocked are less stable than native ferredoxin. They proposed that the amino group of Ala' was involved in a hydrogen bond that was important for the stability of this ferredoxin. Storage of the Synthesis of C. acidi-urici

Ferredonin
Derivatives 1681 derivatives reported in this paper under anaerobic conditions increased their stability about 2-to 3-fold (Tables III and IV). This indicates that oxygen contributes to the destruction of ferredoxin.
All clostridial-type ferredoxins of known sequence contain either tyrosine (18, 20, al), phenylalanine (19), or histidine (23,24,26) in position 2 of the peptide chain except for ferredoxins from the photosynthetic bacteria Chromatium (22) and Chlorobium limicola (25). These latter two ferredoxins contain LeuZ and Tyr30. However, they both contain more than the "normal" 55 amino acid residues present in most clostridialtype ferredoxins and at least two other aromatic residues occur in the peptide chain of each of these ferredoxins. We previously described the semisynthetic synthesis of a modified C. acidiurici ferredoxin (30) and recently that of a modified Clostridium M-E ferredoxin (31). These two modified ferredoxins contain Leu* that has been substituted for a tyrosyl residue occurring in this position of the peptide chain of each native protein. Both of these [Leuz]ferredoxins are as active as an electron carrier as is native ferredoxin in the enzymatic assays tested. Since Clostridium M-E [Leu'lferredoxin does not contain an aromatic residue, this shows that aromatic residues are not essential for electron transfer in clostridial-type ferredoxins. Therefore, as suggested earlier (30), the possibility should be considered that electron transfer occurs via cysteinyl sulfur atoms, some of which appear to be exposed to the solvent in P. aerogenes ferredoxin (12). It was also found that neither C. acidi-urici des-(Ala'-Tyrl) apoferredoxin, which lacks the two NH,-terminal residues, or [GlyZ]apoferredoxin, which contains Ala' and Gly2, form stable derivatives upon reconstitution (30). Since the x-ray (12) and Y-NMR (27729) studies on clostridial-type ferredoxins show that the aromatic residue in position 2 of the peptide chain is near an (Fe&S,*)-cluster, the inability of these two derivatives to form stable derivatives indicates that the residue in position 2 acts as a hydrophobic shield for the cluster. Therefore, the peptide chain serves not only to solubilize the (Fe,S,*)-cluster in an aqueous solution but also to protect it from harmful species such as oxygen and probably water. This agrees with the observation of Malkin and Rabinowitz (51) that C.
acidi-urici ferredoxin is more labile in aerobic than anaerobic solutions containing urea or guanidine hydrochloride. McDonald et al. (52) have also reported that C. pasteurianum ferredoxin is more oxygen-sensitive when the protein is perturbed with aqueous dimethyl sulfoxide. The proposal that one of the roles of the peptide chain is to shield the (Fe,S,*)-cluster is supported by recent experiments by Maskiewicz et al. (53,54) that demonstrate that the (Fe&S,*)-cluster can hydrolyze via a mechanism involving attack by protons.
The data in this paper show that [Phe2]-and [His2]ferredoxins are as active as electron carriers as native C. acidi-urici ferredoxin in the two enzymatic assays tested (Figs. 3 and 5). In addition [2-NH,-Tyr3"]-, [2-F-Phe2]-, and [3-F-Phe*]ferredoxins are also fully active in the phosphoroclastic assay, which was the only assay in which they were tested. These results are in accord with our earlier evidence that aromatic residues are not essential for electron transfer in clostridialtype ferredoxins (30,31), since some differences in rates might be expected for the derivatives reported in this paper if electron transfer via an aromatic residue were the rate-limiting step. [Phe'lferredoxin is approximately one-half as stable as is native C. acidi-urici ferredoxin (Table IV). Since [Phe2]ferredoxin differs from native ferredoxin only by the lack of a hydroxyl group on the aromatic ring of the residue in position 2, this decreased stability is somewhat surprising. Possibly, Phe* does not shield the neighboring (Fe,S,*)-cluster as well as Tyr*, since the water structures around the residue in position 2 of the two proteins probably differ. Magnetic resonance data (27) suggest that the protein structure of pheZ]ferredoxin is the same as native C. acidi-urici ferredoxin. It was suggested (55,56) that the heat stability of ferredoxins from the thermophilic bacteria Clostridium tartariuorum and Clostridium thermosaccharolyticum might partially result from the presence of histidine in these ferredoxins and that the substitution of His' for Tyr' may stabilize clostridial-type ferredoxins, since His' might be in a position to hydrogen bond to the sulfur atom electrons of Cys" (56). This hypothesis is not supported by the results reported here in which C. acidi-urici [His*]ferredoxin is only approximately 41?c as stable as native ferredoxin.
Tryptophan has not been found yet in any native clostridialtype ferredoxin of known sequence. C. acidi-urici [Trp*]ferredoxin is the only ferredoxin derivative that is less active than native ferredoxin. It has approximately 70% of native ferredoxin activity in both the phosphoroclastic and cytochrome c reduction assays. Although the rate-limiting step is not known in either of these relatively complex assays, this suggests that electron transfer and not binding is the rate-limiting step in these two assays, since different binding constants between the various enzymes in the two assay systems and [Trpz]ferredoxin might be expected. lTrp']Ferredoxin is possibly less active than native ferredoxin because the substitution of the larger tryptophan residue for Tyr' has altered the conformation of this ferredoxin.
The suggestion that the conformation of [Trp*]ferredoxin is altered is in accord with the decreased stability (48%) of [Trp*]ferredoxin and, as reported elsewhere (32), the different oxidation-reduction potential of [Trp*]ferredoxin relative to native C. acidi-urici ferredoxin. The studies described in this paper show that the replacement of amino acid residues in the NH,-terminal portion of C.
acidi-urici ferredoxin has very little effect on the electron transfer -ability of the protein in two enzymatic assays tested except in the case of [Trp'lferredoxin.
However, most of the substitutions decrease the stability of the protein in aqueous solution. It is likely, at least in some cases, that the decreased stability results from unfavorable interactions of amino acid side chain residues or exposure of the neighboring (Fe,S,*)cluster to attack by harmful species such as oxygen and probably water. It is also possible, especially in the case of prp*]ferredoxin, that the gross conformation of some region of the protein may be altered and that this alteration decreases the stability of ferredoxin. As discussed in the following paper (32), alterations in the conformation of C. acidi-urici ferredoxin also may be responsible for the changes in oxidation-reduction potential of this protein depending on the particular amino acid replacement or ionic composition of the solution. Therefore, the peptide chain not only serves to solubilize the (Fe,S,*)-clusters and to protect them, but it also exerts a fine control on the oxidation-reduction potential of C. acidi-urici ferredoxin (32) and, as shown for [Trp'lferredoxin in this paper, can also influence the biological activity of ferredoxin.