Kinetic Evaluation Of Fet3p Structure And Function

A yeast system has been developed capable of producing the soluble recombinant Fet3p needed for detailed structural, spectroscopic and kinetic analysis (4,J,13,21). This system has been successfully used to produce the copper-site-deleted mutants whose spectral properties were described in some detail earlier (4,J,13,21). The amino acid replacements made in these mutants were shown in Fig. 3. This system is based on a truncation of the FET3 gene at nucleotide +1666 (at amino acid residue 555); a nucleotide sequence encoding the FLAG epitope was appended at that point. The result of this manipulation was the production of a Fet3p that lacked its carboxyl-terminal membrane-spanning domain. Instead, it was epitope tagged at its C-terminus. Lacking its membrane domain, this protein was secreted into the growth medium rather than remaining tethered to the plasma membrane. This strategy is illustrated in Fig. 16. The protein produced was easily recovered from the growth medium by ion-exchange chromatography. The yield of pure protein was 5-10 mg/L, depending on the precise protein species being produced. This system has two significant advantages over the Baculovirus one. First, the protein is recovered directly from the growth medium without protease treatment. More importantly, unlike the Baculovirus expression system, the protein is processed normally within the yeast cell and therefore contains all of its copper prosthetic groups.

This protein has been used to determine the precise kinetic constants for the ferroxidase reaction catalyzed by Fet3p. These constants are given in Fig. 17; this figure also includes the kinetic constants for iron uptake by the yeast cell. The strong similarity between the two sets of kinetic constants is consistent with the inference that the latter values are simply a reflection of the former, that is, that the ferroxidase reaction catalyzed by Fet3p [reaction (1)] is the rate-limiting step in iron uptake by the yeast cell. In particular, note that the Km for O2 is essentially the same in both reactions, approx 1 |M. This underscores the tight coupling that must exist between iron uptake through the iron permease, Ftrlp, and the ferroxidase reaction catalyzed by Fet3p, as is indicated by the diagrams in Fig. 17.

The yeast-expression system has allowed for the production of the 100-mg quantities of Fet3p required for detailed structure-function studies. These studies have been of two general types: kinetic and spectroscopic/biophysical. Most recently, diffraction-quality crystals of Fet3p have been produced. A sample of these are shown in Fig. 18. These particular crystals are approx 300 |m in length and diffract to approx 3 A, indicating that crystallographic structure determination of Fet3p will be accomplished in the near future (Taylor et al., unpublished results).

This structure will show whether any of the homology modeling studies alluded to above provided any real insight into Fet3p structure and function. As noted, the model based on the C. cinereus laccase is the more reliable based on the paradigms used to select among possible homologs in choosing a basis set for model building. In fact, this model did locate a pair of aspartic acid residues at positions 278 and 283 in the encoded protein that were on one side of a shallow cleft with His489 on the opposite side. His489 provides one of the two type-1 histidine imidazole ligands and would be the His that is at or near the protein surface. In this model, D278, in particular, is immediately adjacent to the site of entry of the electron from the ferrous ion substrate and thus could be part of the substrate-binding site for Fe2+ (Fig. 15B). In addition, the aspartic acid residues lie within a region of the protein that secondary-structure prediction indicates may be a-helical. This would put both D residues on the same face of this putative helix, indicating that both could be part of the "ferroxidase" site. Furthermore, these D residues are conserved in all of the yeast "Fet3p" homologs (Fig. 14).

The possible role of either E185 or D278 in the oxidase activity of Fet3p was investigated in the recombinant protein produced by the method outlined in Fig. 16. Both proteins had intact and spec-troscopically normal type-1, type-2, and type-3 copper sites as determined by visible aborbance and cwEPR (e.g., as in Figs. 1 and 2). In contrast to the Baculovirus-produced protein, the E185 mutant exhibited only a modest decrease in activity (approx 50% of wild type under V/Km conditions). Under the conditions of this assay, the D278 mutant was fully wild type in regard to enzymic activity. The former result was in contrast to that reported by Buonacorsi di Patti et al. (19); their mutant protein had only approx 6% wild-type activity. However, their expression system did not give them sufficient protein for spectral assay so that they had no independent method by which to determine the state of the copper sites. In any event, none of these attempts to locate "ferroxidase" residues based on homology modeling have yielded much insight and thus the question "what makes Fet3p a ferroxidase" remains unanswered.

That Fet3p and hCp do indeed have specificity for Fe2+ as a substrate can be demonstrated directly by kinetic analysis. The copper-site mutants generated in the yeast-recombinant system have been particularly useful in two different types of experiments designed to explore the substrate specificity and overall reaction mechanism for Fet3p. First, the T2D protein allows for investigation of the electron transfer to the type-1 Cu(II) in the absence of turnover. This is because in the multicopper oxidase reaction, electron transfer from the type-1 copper—as Cu(I)—to the trinuclear cluster where O2 is reduced, requires the type 2-Cu(II) (1,3). Thus, the addition of Fe(II) or other substrate (o-dianisidine, p-phenylenediamine, hydroquinone) to the fully oxidized T2D protein results in the reduction of the type-1 Cu(II) without further turnover (5). This reaction is shown in Fig. 19A with Fe(II) as substrate and in Fig. 19B with hydroquinone as the substrate (22). The results demonstrate clearly the strong selectivity that Fet3p has toward Fe(II) in comparison to hydroquinone. The reaction with Fe(II) is complete within the instrument dead time (approx 2 ms); therefore, the kobs must be >1200 s-1. Under similar conditions of reactant concentrations, the kobs for reduction of the type-1

Fig. 15. Computer modeling of the ferroxidase site in Fet3p. Model proposed by Buonaccorsi di Patti et al. (21) based on the structure of ascorbate suggested that Glu185 was part of the Fet3p ferroxidase site. A model of Fet3p based on the more similar laccase structure (9) or a de novo model generated in InSightII places this residue >20 A away from the type 1 Cu(II) (panel A). Instead, this model places residue D278 within 7 A of this copper site (panel B) (Severance and Kosman, unpublished).

Fig. 15. Computer modeling of the ferroxidase site in Fet3p. Model proposed by Buonaccorsi di Patti et al. (21) based on the structure of ascorbate suggested that Glu185 was part of the Fet3p ferroxidase site. A model of Fet3p based on the more similar laccase structure (9) or a de novo model generated in InSightII places this residue >20 A away from the type 1 Cu(II) (panel A). Instead, this model places residue D278 within 7 A of this copper site (panel B) (Severance and Kosman, unpublished).

Cu(II) in Fet3p by hydroquinone was 0.008 s-1. The large kobs for the reaction with Fe(II) suggests that electron transfer takes place within a Fet3p-Fe(II) substrate complex, although it does not require it.

Using this stopped-flow technique, Machonkin and his co-workers have measured the rate constant for the Fet3p-hydroquinone reaction and compared it to the type-1 reduction rates for hCp and Lac (22). The purpose of this comparison was to establish any possible correlation between solvent accessibility of the type-1 sites in these proteins (e.g., see comparisons shown in Figs. 5 and 9) and their reactivity. The second-order rate constants for the reaction of hCp and Fet3p with hydroquinone were 6.1 x 104 M-1s-1 and 3.5 x 105 M-1s-1, respectively. The value for the C. cinereus Lac reaction was >107 M-1s-1. This pattern of reactivity parallels the solvent accessibility of the type-1 site in the three proteins in that this site in hCp is completely buried while one edge of this site in C. cinereus is fully solvent accessible. As suggested by the ESEEM data, the Fet3p type-1 Cu(II) falls in between these two extremes (14).

There was no similar correlation between reactivity toward Fe(II) and solvent exposure. Fet3p and hCp exhibited similar rates of type-1 Cu(II) reduction by Fe(II) (kobs > 1200 s-1), whereas the rate with C. cinereus Lac was >23 s-1. In other words, laccases can use Fe(II) as the substrate but have no

High Copy Plasmid Expressed in bfet3 Strain under Control of Constitutive TransactivatioD

\ J TRANSMEMBRANE

\JFET3S DOMAIN DELETED

High Copy Plasmid Expressed in bfet3 Strain under Control of Constitutive TransactivatioD

\ J TRANSMEMBRANE

\JFET3S DOMAIN DELETED

S'Acc ggt gac tac aaggac gac gat gac aag taa taaGTG3' sense 3'TGGCCactg atg ttc ctg ctg eta ctg ttc att att CACS' antfeense T554 G555 DYKD DD DK stopstop

S'Acc ggt gac tac aaggac gac gat gac aag taa taaGTG3' sense 3'TGGCCactg atg ttc ctg ctg eta ctg ttc att att CACS' antfeense T554 G555 DYKD DD DK stopstop

Fig. 16. Strategy for production of a recombinant, soluble Fet3 protein in yeast. (From ref. 4.)

better than 1-2% of this activity in comparison to Fet3p and hCp. In addition, they are at least 100-fold better than the ferroxidases in the turnover of bulky organic reductants. Combining the structure and reactivity features of these proteins indicates that the type-1 sites in the ferroxidases are less accessible to these large reductants while possessing specificity elements that support the recognition and binding of Fe(II) as the substrate. As outlined earlier, some of these elements may have been identified in hCp; they remain uncharacterized in Fet3p.

Theoretically, the redox potential of the type-1 Cu(II) could provide the specificity toward a particular reductive substrate in comparison to another. Type-1 copper sites do exhibit a remarkable variability in reduction potential, from 240 mV in nitrite reductase (23,24) to >1 V for a noncatalytic type-1 copper in hCp (2J). This variability is certainly the result, in part, of the presence of the methionine ligand found in the four-coordinate type-1 sites (as in hCp and AO) that is absent in Lac and Fet3p, for example (cf. Figs. 4 and 9). Mutagenesis studies indicate that this ligand "tunes" the E°' down by approx 100 mV (26). On the other hand, the ferroxidase-active type-1 sites Fet3p and hCp have essentially identical reduction potentials (427 and approx 450 mV, respectively) despite the fact that the latter has Met coordination, whereas the former does not (6,22). In summary, avail-

Fig. 17. Kinetic equivalence of ferroxidation and iron uptake in yeast. The ferroxidation kinetics were determined using soluble recombinant Fet3p, whereas 59Fe uptake was determined in whole yeast cells. (Data are taken from ref. 4.)

able data do not suggest that a simple linear free-energy relationship links substrate specificity to type-1 reduction potential in the multicopper oxidases, nor does the presence or absence of a Met ligand predict the E°'. Instead, ferroxidase activity appears to be an acquired trait resulting from protein elements that are in addition to the copper ligands and environment that directly modulate the spectral and redox properties of the type-1 copper atom itself.

9. WHERE DO WE GO FROM HERE?

The experimental tools and reagents are now in hand to develop a complete understanding of how the ferroxidases, Fet3p and hCp, and their congeners in the same and other organisms work, both in molecular and physiologic terms. Most certainly, a high-resolution structure of Fet3p will be available, as will data on structure-function studies on mutant forms of recombinant hCp. This information will go a long way to fill the current gaps in our understanding of ferroxidase structure and function. There will be data on the electrophysiology of Fe(II) oxidation by Fet3p and iron uptake through Ftr1p in a heterologous eukaryotic system (e.g., Xenopus oocytes or transfected Caco-2 cells). These studies would test models for the coupling mechanism that links Fe(II) oxidation to iron uptake. The cellular role of other ferroxidases (e.g., hephaestin) in human iron metabolism will have been established and that of the other ferroxidase/permease pair in yeast, Fet5p/Fth1p. The links between ferroxidases and the Fe2+ transporters—those in both the plasma and intracellular membranes—in the various cell types in mammals will become clearer.

Finally, as this new information becomes available, we will be come to an understanding of the evolution of the multicopper oxidase that had ferroxidase activity. Copper and iron, of course, are intimately associated with aerobic metabolism, although copper and iron enzymes are not confined to obligate aerobes. True, the ferroxidase reaction is explicitly an aerobic process. However, taking Fet3p as an example (Km for O2 = 1 mM, equivalent to a partial pressure under an 0.1% O2 atmosphere), the ferroxidase reaction can proceed under microaerobic conditions. These conditions do not

Fig. 18. Diffraction-quality crystals of Fet3p. These crystals, which are bright blue, diffract to approx 3 A. but do not have suitable temperature behavior for analysis using synchrotron radiation. This is likely the result of the presence of approx 10% residual carbohydrate following endoglycosidase treatment. A recombinant form of Fet3p lacking carbohydrate is under construction.

Fig. 18. Diffraction-quality crystals of Fet3p. These crystals, which are bright blue, diffract to approx 3 A. but do not have suitable temperature behavior for analysis using synchrotron radiation. This is likely the result of the presence of approx 10% residual carbohydrate following endoglycosidase treatment. A recombinant form of Fet3p lacking carbohydrate is under construction.

obtain for yeast under most lab conditions, but they certainly do for hCp because free O2 in the plasma is strongly buffered by hemoglobin and they certainly would have obtained in the early stages of geologic aerobiosis. Certainly, one of the early events in aerobiosis would have been evolution of the copper- and iron-requiring processes of respiration that enormously extended the energy-production capacity of facultative aerobes. Mechanisms for supplying iron to cells to support these respiratory functions would have been part of this early evolution. In this view, the ferroxidases may have been the first of the multicopper oxidases and may have been selected for specifically as a component of iron uptake in these first eukaryotes. There is an irony in this view in that the target for this cellular iron, whose uptake required the action of a copper ferroxidase, was itself a copper- and iron-dependent enzyme—cytochrome-c oxidase—and both enzymes were dioxygen reductases. On the other hand, this may not be irony, but a very reasonable evolutionary trick. After all, both enzymes are also multinuclear metallo-oxidases with surprising homologies despite their disparate overall structures and cofactors. Reasonably, they are part of the same adaptive stream; perhaps they also part of the same evolutionary tree. This question, too, may be brought closer to resolution in these next few exciting years of research on ferroxidases and iron homeostasis.

0 0

Post a comment