THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 264. No. 13, Issue of May 5, pp. 727&7284,1969 Printed in U.S.A.

Structural and Spectroscopic Characterization of Exogenous Ligand Binding to Isolated Factor F430 and Its Configurational Isomers*

(Received for publication, August 25, 1988)

Andrew K. ShiemkeS, Warren A. Kaplant, Cristi L. Hamilton*, John A. Shelnuttq, and Robert A. Scott* From the $Departments of Chemistry and Biochemistry and Center for Metalloenzyme Studies, University of Georgia, Athens, Georgia 30602, the $School of Chemical Sciences, University of Illinois, Urbana, Illinois 61801, and the llFuel Science Division 6211, Sandin Natioml Laboratories, Albuquerque, New Mexico 87185

Binding of axial ligands to the nickel(I1) of isolated factor Faso from the methyl reductase enzyme of Meth- anobacterium thermoautotrophicum is demonstrated. Evidence of bis-ligand coordination is obtained from the x-ray absorption, optical, and resonance Raman spectral characterization of F490 and its 12,13-diepi- meric isomer in the presence of a large excess of cyanide, pyridine, or 1-methylimidazole. Significant broadening and 6-10-nm red shifts of the main 430- nm optical absorption band and shifts of up to 30 cm" for the high-frequency Raman lines are observed upon coordination of these axial ligands. The Raman spectra of native Frso and the diepimer with a particular axial ligand are nearly identical. Nickel x-ray absorption edge spectra of the diepimer in the absence and pres- ence of these exogenous ligands are indicative of con- version from a square-planar to a tetragonally dis- torted octahedral geometry. Analyses of the nickel ex- tended x-ray absorption fine structure data for the ligated diepimer complexes yield detailed structural information for these complexes. Implications of these data with respect to the enzymatic mechanism and the structure of the enzyme-bound factor are discussed.

Methanogens constitute a diverse class of archaebacteria capable of living autotrophically on Hz and CO,. Energy is derived through the reduction of carbon dioxide to methane, with Hz acting as the ultimate source of reducing equivalents (1). The methanogenesis cycle involves several enzymes and cofactors (2, 3), a few of which have been purified to homo- geneity. One of the enzymes that has been purified is CH3SCoM' methyl reductase (4). This terminal enzyme in

* The XAS studies were performed at the Stanford Synchrotron Radiation Laboratory, which is funded by the Department of Energy under Contract DE-AC03-82ER-13000, Office of Basic Energy Sci- ences, Division of Chemical Sciences and the National Institutes of Health, Biotechnology Resource Program, Division of Research Re- sources. XAS studies (under R. A. s.) at Georgia are supported by National Science Foundation Grant DMB 86-45819 (formerly DMB 85-02707). This work was also supported by National Science Foun- dation Presidential Young Investigator Award CHE 87-15889 (to R. A. S.) and by Department of Energy Contract DE-AC04- 76DP00789 and the Gas Research Institute Contract 5082-260-0767 (to J. A. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The abbreviations used are: CHsSCoM, Z-methylthioethanesul- fonic acid; methyl reductase, CHsSCoM methyl reductase enzyme; HSHTP, N"mercaptoheptanoy1-0-phospho-L-threonine; XAS: x-ray absorption spectroscopy; EXAFS, extended x-ray absorption fine structure; FT, Fourier transform; RR, resonance Raman; Pi, phos- phate buffer; HPLC, high performance liquid chromatography; OEpc, octaethylpyrrocorphin.

the methanogenesis cycle catalyzes the reduction of CH3SCoM to methane and coenzyme M (2-mercaptoethane- sulfonic acid). Methyl reductase is a large complex enzyme of molecular weight -300,000, with an azp2-y~ subunit structure. The isolated enzyme is found to contain two tightly bound equivalents of coenzyme M (5), as well as two equivalents of the nickel-tetrapyrrole factor F430, and two equivalents of N" mercaptoheptanoyl-0-phospho-L-threonine (HSHTP) (6).

The structure of the pentamethyl ester of F430 has been determined by noncrystallographic means (7, 8), and the structure of the parent penta-acid factor is shown in Fig. 1. F430 is unique among naturally occurring tetrapyrroles, being much more reduced than the more common porphyrins, chlor- ins, and bacteriochlorins. F130 and tunichlorin (9) are the only tetrapyrroles from living systems having nickel as the central metal. Little is known about the mechanism of CH3SCoM reduction. Recent results from the laboratories of Wolfe (10) and Thauer (11) have shown that methyl reductase activity can be reconstituted in a system containing the purified enzyme, CH~SCOM, and HSHTP (the cofactor previously referred to as component B) (6, 12, 13). HSHTP is now recognized to be the reductant in the in vitro system, methane being generated from CH3SCoM along with the mixed disul- fide CoM-S-S-HTP (11, 14, 15). Thus, it appears that both CH3SCoM and HSHTP (or a derivative of HSHTP (16)) are physiological substrates for methyl reductase.

Although the precise function of F430 is unknown, this unusual nickel-tetrapyrrole factor is thought to be the site of substrate (CH3SCoM) reduction. Whole cell EPR spectra (17) indicate that at least one Ni(1) form of methyl reductase is active in methanogenesis. On the basis of the role of other metal tetrapyrroles in biological systems, one may imagine that F430 could perform substrate binding, methyl group trans- fer, electron transfer, or some combination of these functions. We have undertaken a study of the ligand binding capabilities of isolated F430 in order to evaluate the likelihood of the central metal participating in substrate binding. In addition, it is known that the nickel in F430 is pseudo-octahedrally coordinated in the holoenzyme (18), the two axial ligands presumably being supplied by the protein. The nature of these ligands is unknown, and this study of exogenous ligand bind- ing by isolated F430 is intended to illuminate this question.

When isolated from the holoenzyme, F430 undergoes com- plex chemistry. The factor is sensitive to heat and may thus be converted to configurational isomers, characterized by epimerization of the acid side chains at &carbon positions 12 and 13 (cfi Fig. 1) of pyrrole ring C (19). The 13-monoepimer is the initial product of this thermal isomerization, but it is rapidly converted to the 12,13-diepimer (20). In addition, the native F430 factor is sensitive to oxygen and is slowly oxidized

7276

Exogenous Ligand Binding to Factor F430 7277

FIG. 1. Structure proposed for factor F4so on the basis of NMR spectroscopy (from Ref. 7).

to 12,13-didehydro F430 (also known as Fsm) in which a double bond is formed between @-carbons 12 and 13 of pyrrole ring C (19). Although there does not appear to be any biological significance to these isomeric derivatives of F430, they have been valuable in the spectroscopic characterization of the factor. The spectroscopic, structural, and chemical character- istics of the configurational isomers of F430 are significantly different from those of the native factor. As we have previ- ously reported, F430 is pseudo-octahedrally coordinated in aqueous solution, whereas the diepimer of F430 has the nickel in a square-planar geometry (19, 21). In this communication we show that in the presence of high concentrations of strongly coordinating ligands, the spectroscopic properties of F430 and its epimeric isomers are nearly identical. Thus we are able to conclude that the structural differences between F430 and its configurational isomers have only subtle effects on the central metal manifested predominantly in a difference in affinity for axial ligands.

EXPERIMENTAL PROCEDURES

Chromatographic media (phenyl-Sepharose, Q-Sepharose, QAE- Sephadex A-25) were purchased from Pharmacia LKB Biotechnology Inc. Chemicals and solvents were reagent grade or better, and the solvents (pyridine and 1-methylimidazole) were vacuum distilled prior to use.

Methanobacterium thrmoautotrophkurn (strain AH) cells were the kind gift of Dr. R. S. Wolfe (University of Illinois). The cells were grown on Hz and COz as energy and carbon sources. The cells were grown at 65 “C, harvested, and stored under Nz at -20 “C as described previously (22).

Methyl reductase was purified from the cells, and F130 was extracted from the holoenzyme as previously described (21). The 12,13-diepimer of F430 was purified from the pool of “free” F430 according to published procedures (21). In the final purification step of “free” F4,, the 13- monoepimer and F6W eluted from the QAE-Sephadex column with the ammonium formate buffer at -1.3 and -1.5 times, respectively, the elution volume of native F130. Ammonium formate was removed by multiple steps of lyophilization and dissolution with 18 megaohm. cm water. Fsso was identified on the basis of its purple color and its reported optical spectrum (19). The optical spectrum of the mono- epimer is essentially identical to the spectrum of the diepimer (191, but these two isomers could be distinguished by their different elution volumes on reversed-phase HPLC chromatography. The monoepimer could be quantitatively converted to the diepimer by heating at 80 “C for -4 h, whereas the diepimer is inert to thermal degradation under these conditions.

Homogeneity of F430 and its isomeric derivatives was determined (both before and after x-ray absorption and Raman experiments) by reversed-phase chromatography using either a PepRPC HR 5/5 col- umn with the Pharmacia FPLC System or a DeltaPak Cla column (3.9 X 150 mm) with the Waters 600 Multisolvent Delivery HPLC System (Waters, a Division of Millipore, Inc., Milford, MA). Flow rates of 0.7-1.0 ml/min were used for the analysis, and a two-part linear gradient from 0 to 7% CH3CN in 9 ml and 7 to 16% CH&N

in 33 ml was found to be sufficient to separate the various isomers of F430.

Samples for spectroscopy (XAS, RR, and optical) were prepared by taking an aliquot of purified F4S0 (or one of its isomers) to dryness by lyophilization in a SpeedVac concentrator (Savant Instruments Inc., Farmingdale, NY). The freeze-dried factor was then dissolved in the desired buffer or solvent and stored frozen at 195 or 77 K until the spectra were recorded.

RR spectra were obtained on pairs of samples simultaneously, using a split cell designed for a Raman difference spectrometer described previously (23). Samples were 100-200 PM in Fm and were excited with the 441.6-nm line of a helium-cadmium laser. Laser power was -40 milliwatts, and spectrometer resolution was 4 cm”. Ultraviolet-visible spectra were obtained at ambient temperature with a Cary 219 spectrophotometer.

XAS data were collected as fluorescence excitation spectra using an argon-filled ionization chamber fluorescence detector (EXAFS Co., Seattle, WA) at the Stanford Synchrotron Radiation Laboratory on a wiggler side station (VII-3) or bending magnet end station (11- 2) under dedicated conditions (3.0 GeV, -40 mA) using Si[220], Si[400], or S i [ l l l ] monochromator crystals. Concentrations of F130 samples varied from -4 to 20 mM in the different solvents. Samples were maintained at 10 K during data collection using a custom- designed (now Oxford Instruments CF1208) liquid helium cryostat. Edge spectra were recorded at high energy resolution (vertical beam aperture of 1 mm) on beamline VII-3, whereas EXAFS spectra were recorded at lower resolution (2-mm aperture) on beamline 11-2. Each edge spectrum is the average of 2-4 15-min sweeps, and each EXAFS spectrum is the average of 10-20 25-min sweeps. Energy calibration was performed by the internal calibration method, assigning the first inflection point of a nickel foil standard to 8331.6 eV (24).

Reduction and analysis of the EXAFS data followed our standard protocol (24, 25). For curve-fitting analysis, Ni-(N,C) scattering functions were extracted from the first shelloof Ni(0Epc) (Ni-N = 1.909 A), [Ni(corphin)SCN]~ (Ni-N = 2.085 A), Ni[2,2’-bis(2-imida- zolyl)bip~enyl]~(C104)z (Ni-N = 1.963 A), and from KZNi(CNL (Ni- C = 1.88 A) using complex Fourier back-transformation (24). Second- shell scattering functions-were extracted in the same manner from Ni(0Epc) (Ni-C = 2.94 A) and from KZNi(CN1) (Ni-N = 3.03 A). The Ni(0Epc) model complex was the kind gift of Dr. A. E. Eschen- moser and corresponds to complex 2 of Ref. 26. Published structural parameters (27) were used for extraction of nickel EXAFS phase and amplitude functions from this model. The [Ni(corphin)SCN]Z model corresponds to the dimer of compound 1 of Ref. 26 and was also the kind gift of Dr. A. E. Eschenmoser. Structural parameters for this model were determined from atomic coordinates obtained from the Cambridge Data Base. The Ni[2,2’-bis(2-imidazolyl)biphen- y1]2(C104)~ complex (28) was the generous gift of Dr. H. J. Schugar. Structural parameters were obtained from the crystallographically determined atomic coordinates.* KzNi(CN)4 was synthesized accord- ing to published procedures, and structural parameters were obtained from the literature (29).

RESULTS

XAS Evidence of Exogenous Ligand Binding to the 12,13- Diepimer of F430-We reported recently the spectroscopic differences between F430 and its 12,13-diepimer and ascribed these spectral differences to alterations in the nickel coordi- nation sphere of the two isomers (21). In particular, it was concluded that the factor in its native configuration displays a 6-coordinate pseudo-octahedral geometry, whereas the di- epimer contains four-coordinate square-planar nickel in aqueous solution. These conclusions are based on the appear- ance of a sharp pre-edge feature characteristic of square- planar nickel at 8336 eV in the x-ray absorption spectrum of the diepimer (Fig. 2a) and the absence of this feature in the edge spectrum of the native factor. The x-ray absorption edge spectrum of the 13-monoepimer also exhibits the intense pre- edge feature at 8336 eV, allowing the conclusion that this derivative also contains square-planar nickel.

Strong evidence for axial-ligand coordination to F430 and its isomers is obtained via XAS. This technique is sensitive

H. J. Schugar, personal communication.

7278 Exogenous Ligand Binding to Factor F430

Energy (eV)

FIG. 2. Nickel K-edge x-ray absorption spectra of the 12,13-diepimer in (a) 10 mM Pi, pH 7 (-), and neat pyridine (---) and (b) neat 1-methylimidazole (-) and 5 M aqueous KCN, pH 12 (---). Spectra were obtained at 10 K, and the aqueous samples contained 20% (w/w) glycerol.

to changes at the nickel and is therefore a direct probe of the metal coordination state. In this study our XAS results focus predominantly on the complexes of the diepimer due to its greater stability (toward epimerization as well as oxidation), its greater solubility in coordinating solvents, and its greater prevalence in cell extracts, relative to the native factor. Axial- ligand binding to the diepimer of F430 is dramatically demon- strated in the XAS near-edge and EXAFS spectra of this square-planar derivative in the presence of certain coordinat- ing bases. Addition of a large excess of cyanide to the diepimer results in a loss of the 8336-eV pre-edge feature (Fig. 2b, dashed l i n e ) and yields a spectrum that has the sharp feature- less edge rise of nickel in a 6-coordinate octahedral geometry. Similarly, when the diepimer is dissolved in neat pyridine or 1-methylimidazole the edge spectra (Fig. 2a, dashed line, and 26, solid line, respectively) are characteristic of nickel in an octahedral geometry, indicating that these species also coor- dinate to the nickel. The absence of a pre-edge shoulder at -8338 eV characteristic of 5-coordinate nickel(I1) complexes (21) precludes the possibility that significant concentrations of 5-coordinate complexes of the diepimer are formed under these conditions.

From analysis of the EXAFS spectra of the coordinated derivatives of F430 we can determine metal-ligand bond dis- tances and the identity of axial ligands. However, in order to obtain dependable information, the concentration of nickel in the sample should be -2-5 mM. This criterion is met only with aqueous solutions (with or without added ligands) of F430 and its configurational isomers (21) and with solutions of the diepimer in neat 1-methylimidazole or neat P-mercaptoetha- nol. In P-mercaptoethanol the diepimer is 4-coordinate, yield- ing edge and EXAFS spectra (not shown) and bond distances (Table I, fit 1-2) nearly identical to those of the monoepimer and diepimer in aqueous solution (fits 1-1 and 1-3, respec- tively).

When dissolved in neat 1-methylimidazole, the EXAFS spectrum of the diepimer is drastically different from the spectrum observed in aqueous solution (Fig. 3) and closely resembles the EXAFS spectrum of the native 6-coordinate F430 in H20 (21). The frequency of the EXAFS oscillations is significantly greater for the 6-coordinate diepimer in l-meth- ylimidazole due to an increase in nickel-ligand bond distances

relative to the square-planar species. The increased bond distances are indicative of an increase in the macrocycle core size, which we attribute to the change from low to high spin that typically accompanies the conversion from square-planar to octahedral geometry in Ni(I1) complexes. The core expan- sion of the diepimer in 1-methylimidazole is also evident in the Fourier transform (FT) of the EXAFS data (Fig. 3). The strong first-shell peak of the FT is shifted -0.2 A to longer distance in 1-methylimidazole, relative to the square-planar complex in water. Similar shifts are seen in the second- and third-shell peaks of the FT. These latter peaks are due to scattering from the a- and P-carbons, respectively, of the pyrrole rings (as well as the axial imidazole rings), and their shift to longer distance is consistent with core expansion.

Curve-fitting analysis of the back-transformed filtered first- shell peak of the diepimer in 1-methylimidazole demonstrates t n increase in the average nickel-ligand distances from 1.89 A in the square-planar complex (fit 1-3) to -2.1 8, in the six- coordinate bis(1-methylimidazole) complex (fits 1-4 and 1-5). Satisfactory fits of the filtered data for the latter complex were obtained with either a single shell of 6 nitrogens at 2.08 8, (fit 1-4) or a two-shell fit (fit 1-5) with 4 (presumably equatorial) nitrogens at 2.05 8, and 2 (axial) nitrogens at 2.15 A. The goodness of fit parameter f’ is significantly better in the two-shell fit.

Further evidence that this 6-coordinate spkcies is actually the bis( 1-methylimidazole) complex comes from curve-fitting analysis of the back-transformed filtered first- and second- shell peaks. These fits are performed by isolating the first- and second-shell FT peaks in combination and back-trans- forming to k-space. These filtered EXAFS data (Fig. 4) are then fit using as invariant parameters the structural details determined from the first-shell fit (Table I) and a second- shell Debye-Waller factor from [Ni(corphin)SCN]z, and al- lowing the number and distance of second-shell atoms to vary. Although multiple-scattering contributions are not explicitly treated in this procedure, the use of structurally similar models for the extraction of second-shell scattering functions (e.g. Ni(OEpc),[Ni(corphin)SCN]~) largely corrects for this omission. .

In the case of the square-planar diepimer, the second-shell contribution to the filtered EXAFS can be fit in this manner with 8.0 carbons at 2.90 8, (Fig. 4a, Table 11, fit 2-1), corre- sponding to the eight a-carbons of the pyrrole rings. The FT of this fit closely approximates the FT of the unfiltered EXAFS in the region from R’ = 1.3-3.2 8, (Fig. 4a), giving us confidence in the results of the fitting procedure. In the same manner, the filtered first- and second-shell EXAFS of the diepimer in 1-methylimidazole can be fit with a single shell of 12.1 carbons at 3.05 8, (fit 2-2), corresponding to the eight a-carbons of the pyrrole rings plus the two a-carbons from each of the two axial 1-methylimidazole ligands. Improved fits can be obtained by splitting the second shell of carbon scatterers into two shells, analogous to the fit of the first- shell EXAFS (fit 1-5). A fit with ei ht carbons at 3.02 8, (the pyrrole carbons) and four at 3.15 1 (the imidazole carbons) gives a significant improvement in f’ (Fig. 46, fit 2-3), relative to the single-shell fit (fit 2-2). Also, the FT of fit 2-3 more closely matches the FT of the unfiltered EXAFS in the region of the second-shell scatterer peak from R‘ = 2.1-3.2 8, (Fig. 4b). Although comparable fits of these data can be obtained with axial bond distances shorter than the equatorial bonds, the fit with long axial bonds to the 1-methylimidazole ligands (fit 2-3) makes more sense chemically and implies a structure very similar to that of the bis-imidazole complex of Ni[4-N- methylpyridyl)4porphine]4+, which has Ni-N(pyrro1e) bond

Exogenous Ligand Binding to Factor F430 7279

TABLE I Curve-fitting results for the first shell of F4, and related complexes

Filter gives the portion of the FT (k = 3-12.5 A-', k3 weighted) that was isolated (using a hajf-Gaussian function of width 0.10 A) and back-transformed to k-space for curve fitting over the k = 3-12.5 A" range; N. is the coordination number/nickel; R is the nickel-scatterer distance; Au2 is a relative mean square deviation in R , Au* = u'(samp1e) - u2(reference), where the reference compound is Ni(OEpc), unless otherwise noted.

Sample Solvent conditions Fit Filter Ni-(C,N,O)

N." R A 2 f b

A Monoepimer 10 mM Pi, pH 7 1-1 1.15-1.88 Diepimer 0-Mercaptoethanol 1-2 1.18-1.84

10 mM Pi, pH 7 1-3 1.17-1.83 1-Methylimidazole 1-4 1.29-2.12 1-Methylimidazole 1-5 1.29-2.12

5 M CN-, pH 12 1-6 1.28-2.08

F430 10 mM Pi, pH 7 1-7 1.29-2.12

A 1.89 1.89 1.89 2.08 2.05 2.15 2.11 2.01 2.10

A' -0.0002 0.060 -0.0012 0.029 0.0001 0.032 0.0013' 0.029

-0.0008d 0.015 0.0019'

-0.0026 0.029 0.0011' 0.0031 0.039

Coordination numbers were not varied during optimization. * f' is a relative goodness-of-fit statistic,

where N is the number of data points to be fit. The reference compound for this shell of scatterers is [Ni(c~rphin)SCN]~. The reference compound for this shell is Ni[2,2'-bis(2-imidaz01yl)biphenyl]~(C10~)~.

e The reference compound for this sub-shell of scatterers is K2Ni(CN)4.

20(

"0 1.5 3.0R (Ap.5 6.0 7.5

FIG. 3. EXAFS (top) and Fourier transforms (bottom) of the 12,13-&epimer in neat 1-methylimidazole (-) and in 10 m~ Pi, pH 7 (---). The data were collected at 10 K, and the aqueous sample contained 20% (w/w) glycerol.

distances of 2.04 A and axial Ni-N(imidazo1e) distances of 2.16 A (30). However, the lack of a unique best fit to the data permits only the conclusion that static disorder is present in the metal-ligand bond distances of the bis( 1-methylimidazole) complex of the diepimer.

Analysis of the EXAFS data for the diepimer in a concen- trated aqueous solution of cyanide provides similar evidence of coordination of the added ligand. An increase in the fre-

3.0 5.0 7.0 9.0 11.0 13.0 k

FIG. 4. Left, Fourier-filtered first- and second-shell EXAFS (-) and best fit to the filtered EXAFS (---); right, the Fourier transforms of the raw unfiltered EXAFS (-) and the Fourier transforms of the best fits (---) for the 12,13- diepimer in (a) 10 mM Pi, pH 7 (fit 2 - l ) , (b ) neat l-methylim- idazole (fit 2-3), and (c) 5 M cyanide, pH 12 (fit 2-4).

quency of the EXAFS oscillations and a shift of the first-shell FT peak to longer distance in the presence of cyanide are observed (Fig. 4c), again indicative of macrocycle core expan- sion. Curve fitting of the data from the cyanide complex required using two shells of scatterers in the first coordination sphere (fit 1-61, indicating that the axial cyanide ligands coordinate at a shorter distance than the 1-methylimidazole ligands (2.01 and 2.15 A, respectively). This shorter bond distance could be due to either less steric interaction between the axial ligands and macrocycle or to greater p7r-d7r interac- tion in the cyanide complex.

The cyanide complex also exhibits a large increase in the intensity of the second-shell FT peak (at R' = 2.8 A, Fig. 4c) relative to the 4-coordinate parent complex (Fig. 4a) or the 6-coordinate bis(1-methylimidazole) complex (Fig. 4b). This increase in intensity is due to the so-called focusing effect

7280 Exogenous Ligand Binding to Factor F430 TABLE I1

Curve-fitting results for the second shell of the 12,13-diepimer of Fan and its ligated complexes All symbols are defined as in Table I.

Sample Solvent conditions Fit" Ni-(C,N)

Filter N. R A 2

A A A'

Pb

Diepimer 10 mM Pi, pH 7 2-1 1.17-3.10 8.0 2.90 0.0050 0.042 1-Methylimidazole 2-2 1.29-3.14 12.1 3.05 0.0050 0.031 1-Methylimidazole 2-3' 1.29-3.14 8 3.02 0.0009 0.015

5 M CN-, pH 12 2-4' 1.28-3.21 8 3.13 0.0031 0.022 4 3.15 0.0010

2 3.23 -0.002gd a The first-shell parameters used in these fits are those shown in fits 1-3, 1-5, and 1-6 for the sample in 10 mM

Pi, 1-methylimidazole, and 5 M cyanide, respectively. f is defined as in Table I.

e In these fits the coordination numbers were fixed while R and A 2 were allowed to vary. For the other fits, Au' was fixed while N, and R were allowed to vary.

The reference compound for this sub-shell of scatterers is K'Ni(CN),.

TABLE I11 Summary of visible spectral parameters for complexes of Fa0

and its configurational isomers Spectra were obtained at ambient temperature. C.N. is the nickel

coordination number; X,,, is the wavelength of the peak at maximal absorbance; FWHM is the full width of the absorption peak measured at half-maximum.

Sample Solvent conditions C.N. Amax FWHM nrn nrn

Monoepimer 10 mM Pi, pH 7 4 429 41 Diepimer 10 mM Pi, pH 7 4 430 43

Pyridine 6 441 51

5 M CN-,pH 12 6 434 66 1-Methylimidazole 6 438 59

F43o 10 mM Pi, pH 7 4,6" 430 58 Pyridine 6 439 61 1-Methylimidazole 6 438 63 5 M CN-,pH 12 6 433 69

''F430 in dilute aqueous buffer is an equilibrium mixture of four- and six-coordinate species (36, Footnote 3).

(31), whereby the back scattering from the second-shell nitro- gens of the cyanide ligands is focused by the intervening carbon atom. Such effects occur when the absorbing atom and the two scattering atoms are collinear (or nearly so). Focused second-shell scattering is well known in the case of cyanide and carbonyl complexes of metals (32-34). The filtered first- and second-shell FT peaks of the bis-cyanide complex of the diepimer can be fit with the first-shell parameters of fit 1-6 and a second shell consisting of two nitrogens from axial cyanide ligands plus the eight a-carbons from the pyrrole rings (Fig. 4c, fit 2-4). The FT of fit 2-4 affords a very good match to the FT of the unfiltered EXAFS in the region of the second-shell FT peak (Fig. 4c). Attempts to fit this region with a single shell of scatterers required -18 carbon atoms.

Optical Spectral Evidence of Ligand Binding-Characteris- tic differences are observed in the optical spectra of F430 and its 12,13-diepimer, a much narrower main absorption peak and distinct shoulders at -500, -360, and -290 nm in the spectrum of the square-planar diepimer (21). The optical spectrum of the 13-monoepimer is nearly identical to that of the diepimer, consistent with our conclusion that the optical spectral differences between FdS0 and these configurational isomers stem from different macrocycle conformations related to differences in the metal coordination state.

Addition of a large excess of cyanide to an aqueous solution of the square-planar diepimer or dissolution of lyophilized diepimer in neat pyridine or 1-methylimidazole produces characteristic changes in the optical spectrum (not shown).

1

1 5,50

1280 1360 1440 1520 1600 1680

FREQUENCY (crn'')

FIG. 5. Resonance Raman spectra of the 12,13-diepimer in (a) 10 mM Pi, pH 7, (b) neat pyridine, (c) neat l-methylimi- dazole, and (d) 5 M aqueous cyanide, pH 12. Spectra were obtained at ambient temperature (-295 K) with 441.6-nm excitation. Dashed lines indicate features due to solvent.

The most noticeable change is a 5-10-nm red shift in and a broadening of the visible absorption peak (Table 111). In addition, the shoulders at -500 and -360 nm are lost in the presence of these ligands. Coordination of cyanide, pyridine, or 1-methylimidazole to the 13-monoepimer induces optical spectra that are nearly identical to those of the diepimer complexes. Addition of other anionic species (e.g. SCN-, OH-, acetate) or dissolution in weakly coordinating solvents (meth- anol, ethanol, 8-mercaptoethanol) or other nitrogenous bases (piperidine, pyrazine) causes little or no change in the optical spectrum of the diepimer.

Resonance Raman Evidence of Ligand Binding-Informa- tion on the coordination state of Fds0 and its configurational isomers can also be obtained from resonance Raman spectra.

Exogenous Ligand Binding to Factor F430 7281

TABLE IV Raman frequencies and separation of the two strong high frequency

lines in complexes of F430 and its isomers Spectra were obtained with 441.6-nm excitation at ambient tem-

perature. C.N. is the nickel coordination number; Au is the separation between v1 and VP.

Sample Solvent conditions C.N. v1 vz Av cm" cm" cm"

Monoepimer 10 mM Pi, pH 7 4 1528 1621 93 1-Methylimidazole 6 1545 1617 72 5 M CN-,pH 12 6 1560 1630 70

Diepimer 10 m M Pi, pH 7 4 1529 1623 94 Pyridine 6 1542 1619 77 1-Methylimidazole 6 1550 1619 69 5~ CN-,pH 12 6 1562 1625 63

FUO 10 mM Pi, pH 7" 4 1534 -1627 -93 6 1556 1629 73

Pyridine 6 1545 1620 75 1-Methylimidazole 6 1551 1617 66 5 M CN-,pH 12 6 1566 1628 62

"F4, in dilute aqueous buffer is an equilibrium mixture of four- and six-coordinate species (36, Footnote 3).

a I . , .

8300 0320 8340 8360 8380 8400 Energy (ev)

FIG. 6. Nickel K-edge x-ray absorption spectra of F4so (a) in 10 mM Pi, pH 7, and (b) in Hz0 with 7 M imidazole, pH adjusted to 12 with solid KOH. The spectra were obtained at 10 K, and samples contained 20% (w/w) glycerol.

The RR spectrum of a structurally similar Ni(corphin) com- plex (26) indicates that the separation of the two strongest bands in the high-frequency region of the spectrum (at v1 = 1530 cm" and up = 1620 cm", due to vibrations of the macrocycle) is related to coordination number (35, 36). The RR spectra of the F430 derivatives are similar to those of the model and, as in the model spectra, a separation of -90 cm" is observed for the high-frequency peaks of the 4-coordinate diepimer in aqueous buffer solutions (Fig. 5, Table IV).

The RR spectrum of the diepimer changes dramatically in the presence of strongly coordinating ligands, as shown in Fig. 5. In the presence of excess cyanide or in neat pyridine or 1-methylimidazole the strong 1529-cm" peak of the square- planar species shifts 15-30 cm-' to higher frequency, resulting in a peak separation of -70 cm". This is very similar to the results for the F430 model (35, 36) where the separation de- creases to -70 cm-' in the 6-coordinate complexes. Noncoor- dinating solvents (e.g. methanol, ethanol, or ,f3-mercaptoeth- anol) have no effect on the RR spectrum of the diepimer.

Evidence of Exogenous Ligand Binding to Natiue F430"

Direct XAS evidence for axial coordination of added ligands to native F430 is much harder to acquire. This is due principally

t

1556 lb.

1280 1360 1400 1520 1600 16

FREQUENCY (crn-1)

BO

FIG. 7. Resonance Raman spectra of Flso in (a) 10 mM Pi, pH 7, (b ) neat pyridine, (c) neat 1-methylimidazole, and ( d ) 6 M aqueous cyanide, pH 12. Spectra were obtained at ambient temperature (-295 K) with 441.6-nm excitation. Dashed lines indicate features due to solvent.

to the low solubility of the native factor in solvents other than water. Furthermore, the concentrated solutions of F430 required for the XAS technique (>1 mM) are more rapidly oxidized to F6w than are more dilute solutions, and this oxidation appears to be further accelerated by addition of exogenous bases. XAS data for F430 coordinated by added base have been obtained only for the factor in concentrated aqueous solution of imidazole. The x-ray absorption edge spectrum of this sample is shown in Fig. 6, together with the edge spectrum of F430 in dilute aqueous buffer. Both spectra are characteristic of 6-coordinate octahedral nickel. However, the edge changes slightly in the presence of imidazole, becom- ing very similar to the edge spectrum of the diepimer in neat 1-methylimidazole (Fig. 2b), thereby suggesting that imidazole is capable of coordinating to native F430.

Our results demonstrate that cyanide, pyridine, and 1- methylimidazole are capable of coordinating to the nickel of the 12,13-diepimer of F430. Coordination of these three strong- field ligands to native (nonepimerized) F430 gives optical spec- tra very similar to the spectra of the 6-coordinate diepimer complexes. Comparison between the optical spectra of F430

and the diepimer with a particular axial ligand shows only slight differences in the energy and full-width-at-half-maxi- mum (FWHM) of the main absorption band (Table 111), suggesting that structurally similar bis-ligand complexes are formed. In the case of F430 the changes between the optical spectrum in the absence and in the presence of added ligands are more subtle than for the diepimer.

7282 Exogenous Ligand Binding to Factor F4~o

Preliminary optical titrations of native F430 in ethanol with 2-methylimidazole yield an effective binding constant of 8.4 M" (298 K), assuming formation of a mixed ligand (ethanol, 2-methylimidazole) complex and correcting for the known binding constant for ethanol? In ethanol, the 12,13-diepimer shows no affinity for 2-methylimidazole, even at the highest concentrations attainable, allowing calculation of an upper limit of its binding constant as C0.2 M-'. Thus, native F430 shows an enhanced affinity for 2-methylimidazole ligands by at least a factor of -40.

The RR spectra of native F430 in the presence of these strongly coordinating ligands changes dramatically relative to the spectrum in aqueous solution (Fig. 7). The 1534 cm" peak (Fig. 7a), which is due to the 4-coordinate form of the factor (36); disappears in the presence of pyridine, l-methylimida- zole, or cyanide, leaving only a single peak at 1545-1566 cm" (Fig. 7, b-d), characteristic of six-coordinate F430 complexes. These six-coordinate F430 complexes exhibit a -70 cm" sep- aration of the two high-frequency bands (Fig. 7, Table IV) much like the six-coordinate diepimer complexes (Table IV). In fact the RR spectra of the diepimer and F430 with the same axial ligands are nearly identical (Figs. 5 and 7, Table IV).

DISCUSSION

Spectral Characterization of the Exogenous Ligand Ad- ducts-The XAS results presented above for the 12,13-diepi- mer of F430 clearly indicate that cyanide, pyridine, and 1- methylimidazole are capable of coordinating to the nickel in this configurational isomer of F430. This conclusion is based on the dramatic changes in the x-ray absorption edge spec- trum of the diepimer in the presence of these ligands (Fig. 2). Furthermore, analysis of the EXAFS data for the cyanide and 1-methylimidazole complexes of the diepimer (Figs. 3 and 4) demonstrates macrocycle core expansion and the contribution of these exogenous ligands to both the first- and second- coordination sphere nickel-ligand scattering (Tables I and 11).

Direct demonstration of exogenous ligand binding to native F430 via XAS has not been possible, with the exception of the bis-imidazole complex (Fig. 6). This difficulty is due to the instability of the native factor toward epimerization as well as oxidation and its low solubility in pyridine and l-methylim- idazole (as well as all other nonaqueous solvents we have examined). However, optical and resonance Raman spectra provide strong evidence for coordination of added ligands to the isolated native factor, especially through comparison with the spectra of the exogenous ligand complexes of the diepimer. The optical spectral parameters (Amax, peak width, and peak shape) of the F430 complexes with cyanide, pyridine, and 1- methylimidazole are nearly identical to those of the analogous diepimer complexes (Table 111). The resonance Raman spec- tra of the exogenous ligand complexes of F430 and the 12,13- diepimer are also very similar. For instance, the peak frequen- cies of the bis-pyridine complex of F430 differ by less than 3 cm" from the frequencies of the bis-pyridine complex of the diepimer (Table IV, Figs. 5 and 7). Similar small frequency differences are observed in comparisons of the RR spectra of the cyanide and 1-methylimidazole complexes of F430 with the analogous complexes of the diepimer. This is very strong evidence that these two configurational isomers of F430 form identical complexes with these three strong-field ligands.

We have examined the affinity of many ligands for the nickel in the 12,13-diepimer of F430, including anions (acetate, hydroxide, chloride), nitrogenous bases (piperidine, pyrazine), pseudohalogens (azide, thiocyanate, cyanate), thiols (P-mer-

A. K. Shiemke, J. A. Shelnutt, and R. A. Scott, manuscript in preparation.

captoethanol, dithiocarbamates), and thioethers (thiophene). Based on optical and RR spectra, we find that only cyanide, pyridine, and substituted imidazoles coordinate strongly. It appears that in order to get complete ligation to the diepimer, the ligand must be a relatively strong base and have an unsaturated site capable of coordinating to the nickel. Strong bases with only saturated nitrogens (piperidine, piperazine) bind weakly or not at all to the diepimer. The same is true for species that have unsaturated nitrogens but that are only weakly basic (pyrazine, azide). This suggests that r-bonding between the metal and axial ligands is important in the formation of these complexes.

In the case of native F430, axial ligation reactions can be driven to completion only with cyanide, pyridine, and substi- tuted imidazoles, but partial ligation appears to take place with weakly coordinating solvents such as water, methanol, and ethanol (36).3 This suggests a higher affinity of native F430 for axial ligation compared to the diepimer (the 2-meth- ylimidazole titrations in ethanol indicate that this difference in affinity is >40-fold for imidazole-like ligands). Partial coordination of other bases (e.g. pyrazine, piperazine) may also occur, but in these cases addition of a large excess of the ligand causes rapid oxidation of F430 to FW, making the axial ligation reaction difficult to study. Anions such as azide and thiocyanate appear to coordinate more readily to the native isomer of F430 than to the diepimer (not shown). However, the native factor shows much less affinity for azide or thiocyanate than for cyanide; only small optical spectral changes are observed in the presence of -3 M azide or thiocyanate, whereas similar concentrations of cyanide will drive the ligation reac- tion nearly to completion.

The low-frequency region of the Raman spectra of these axially coordinated F430 derivatives is devoid of any features attributable to axial metal-ligand vibrations. Only very weak features are observed below 900 cm" (37), and they do not change appreciably between four- and six-coordinate species. This is analogous to the results for most other metal tetra- pyrroles, although probably for different reasons. For porphy- rins and chlorins, the electronic transitions which affect the resonance condition for RR spectroscopy are in-plane polar- ized (38), hence the out-of-plane metal-ligand vibrations are rarely resonance-enhanced. In the case of F430 the chromo- phore is likely to be much less planar than the porphyrins, and the electronic transitions responsible for the visible spec- trum should have considerable out-of-plane character. How- ever, the conjugation pathway is so small in F430 that there appears to be little or no interaction between the metal orbitals and those of the conjugated part of the macrocycle, hence the relatively subtle optical spectral differences between four- and six-coordinate complexes ( d e supra) and the ab- sence of resonance-enhanced metal-ligand vibrations.

Implications for the Enzymatic Mechanism-Given the demonstration that the nickel of isolated F430 is capable of binding certain axial ligands, it is possible that this factor could participate in substrate binding during enzymatic turn- over. XAS results indicate that F430 bound to the methyl reductase contains six-coordinate nickel (18). However, we find that the axial coordination to the isolated factor is easily reversible and the axial ligands are quite labile, i.e. removal of the added exogenous ligands by chromatography or ultra- filtration yields the original factor (this is true for the other configurational isomers, as well). This implies that one (or more) of the F430 ligands in the resting state of the enzyme could be readily replaced by substrate during the enzymatic reaction.

It is significant that we have been unable to find any

Exogenous Ligand Binding to Factor F430 7283

evidence for the existence of a five-coordinate form of F430 having only a single axial ligand. This is typical of sterically unhindered macrocyclic nickel complexes; addition of the first axial ligand usually causes the change in spin state making addition of the second axial ligand more favorable. In the case of enzyme-bound F430, one may imagine a situation where one strongly coordinating ligand is supplied by the protein, allow- ing facile exchange at the sixth coordination position. The affinity for various species at the sixth position would then be controlled by the protein structure in the ligand binding pocket, much as with exogenous ligand binding to hemoglobin.

As with other macrocyclic nickel complexes (39, 40), the affinity of F430 for axial ligands is rather low; very large excess of ligand must be added to the isolated factor to drive the ligation reaction to completion. However, relative to nickel porphyrins, F430 has a greater affinity for weak-field ligands such as water, methanol, and ethanol, and a smaller affinity for saturated nitrogenous bases, such as piperidine. It is possible that design of the F430 binding pocket in the methyl reductase could compensate for the small ligand affinity of F430 by placing one or more ligating groups in an orientation favorable for axial coordination to the nickel. This could lead to a very large effective ligand activity and promote axial coordination. Such an effect appears to be important in the structure of nickel-protoporphyrin IX-substituted myoglobin (41,42), where the nickel is five-coordinate, with the proximal histidine as the apical ligand.

The preference of F430 for unsaturated nitrogenous bases as axial ligands implicates imidazole as a potential ligand to the enzyme-bound factor. We intend to explore this possibility via XAS using the bis( 1-methylimidazole) diepimer complex discussed above as a model for imidazole ligation in the enzyme-bound F430. However, as stated above the factor bind- ing site may be such that a more weakly coordinating group is oriented to compensate for the low affinity. Thus, coordi- nation by solvent, alcohol-, or carboxylate-oxygen in the holoenzyme cannot be ruled out on the basis of these results, and the native isomer of isolated F430 does show some affinity for these types of ligands? Significantly, the affinity of the F430 nickel for sulfur ligands is extremely low; no axial coor- dination occurs either in neat ,&mercaptoethanol or in aqueous solutions saturated in dialkyldithiocarbamates. Thus, if CH3SCoM binds to F430 in the holoenzyme it may be through the sulfonate oxygens rather than the sulfur of this substrate. Our preliminary XAS studies of the enzyme-bound F430 indi- cate a six-coordinate nickel with only nitrogen, carbon, or oxygen ligation?

However, the relative affinity of F430 for axial ligands may depend on the oxidation state of the nickel. Several EPR signals have been attributed to Ni(1) forms of enzyme-bound F430 in whole-cell preparations of M. therrnoautotrophicurn, and some of these Ni(1) species are involved in the in uiuo production of methane (17). Similar EPR signals have been observed in some preparations of purified methyl reductase (43), and the F430 penta-methyl ester has also been reduced to the Ni(1) state (44). Proposed mechanisms for methyl reduc- tase (45) suggest substrate binding to F430 in the Ni(1) oxida- tion state. The axial coordination chemistry of F430 could be greatly affected by such a metal-based reduction. It has been speculated recently (17) that the multiple Ni(1) forms of enzyme-bound F430 observed in whole-cell EPR spectra differ in their axial coordination geometry, with one presumably active form having a Ni-S bond to the anion of HSHTP.

Conch.sio?w"ix-coordinate complexes of F430 and its 12,13-diepimer can be formed with cyanide, pyridine, and 1- methylimidazole. From our optical and RR spectroscopic

characterization of these complexes we can conclude that the structures of native and epimeric complexes with a particular ligand are practically identical. Yet, these two configurational isomers of the factor exhibit a marked difference (at least a factor of 40) in affinity for certain axial ligands. This differ- ence most likely stems from the greater stability of the four- coordinate form of the diepimer, relative to the four-coordi- nate (or six-coordinate) form of the native factor. This sta- bilization of the diepimer may be due to unfavorable steric interactions in F430 that are removed by the epimerization, as suggested by Eschenmoser and co-workers (26, 27,46).

None of these exogenous-ligand complexes of F430 closely approximate the spectral properties of the factor bound to methyl reductase. The RR spectrum of the holoenzyme is particularly anomalous (relative to the isolated factor), with the two strong high-frequency bands at 1575 and 1652 cm" (36), some 15-20 cm" higher than in any form of the isolated factor we have studied (Table IV). Also, the separation of these peaks if 78 cm", halfway between the values observed for four- and six-coordinate forms of isolated F430, despite the XAS evidence that the enzyme-bound factor is six-coordinate (18). The optical spectrum of methyl reductase also differs markedly from the spectra of the exogenously ligated com- plexes of isolated F430 that we have characterized. The holo- enzyme has a peak at -420 nm (values reported in the literature range from 418 to 425 nm), with a shoulder at 445 nm (5). This spectrum is unique in that all the six-coordinate complexes of F430 that we have generated have peak maxima at wavelengths >430 nm (Table 111). The nickel x-ray absorp- tion edge spectrum of methyl reductase differs subtly from that of the exogenous-ligand complexes of isolated F430 in that the peak in the enzyme spectrum is broader, with less well defined features above the initial sharp edge rise.3

These spectral differences between methyl reductase and isolated F430 may indicate that the axial ligands to the enzyme- bound factor are different from those present in the complexes of isolated F430 that we have studied to date. Analysis of the methyl reductase nickel EXAFS appears to rule out the possibility of sulfur ligation to F430 in the resting state of the enzyme. Alternatively, the enzyme-bound factor could have two dissimilar axial ligands, whereas the complexes we can form with the isolated factor all have identical fifth and sixth ligands. We have not been able to generate any complexes of isolated F430 having dissimilar axial ligands or sulfur ligation, so the effects of such structural features on the spectroscopic properties of F430 remain untested. The unique spectroscopic features of the enzyme-bound F430 could alternatively indicate that the enzyme-bound factor differs from isolated F430 in some respect other than the nature of the axial ligands. The ease with which the factor is extracted from methyl reductase (5) would argue against the disruption of any covalent bonds; thus the isolated factor probably retains all the structural components of the enzyme-bound F430. However, it is possible that upon binding to the enzyme, F430 adopts a macrocycle conformation different from any conformation accessible to the isolated factor. As we have shown in this work, the RR and optical spectra are very sensitive to alterations in macro- cycle conformation. The coordination chemistry of the nickel (Le. the ease of axial-ligand binding and substitution) would be expected to be sensitive to macrocycle conformation as well.

Acknowledgements-We would like to thank Dr. Ralph S. Wolfe for the generous gift of the Methanobacterium cells. We also thank Dr. Marly K. Eidsness for contributing to the early stages of this investigation.

7284 Exogenous Ligand Binding to Factor F430

REFERENCES 22. Gunsalus, R. P., Romesser. J. A., and Wolfe, R. S. (1978) Bio- 1. Daniels, L., Sparling, R., and Sprott, G. D. (1984) Biochim.

2. Wolfe, R. S. (1985) Trends Biochem. Sci. 10, 396-399 3. Rouviere, P. E., and Wolfe, R. S. (1988) J. Biol. Chem. 263,

7913-7916 4. Ellefson, W. L., Whitman, W. B., and Wolfe, R. S. (1982) Proc.

Natl. Acad. Sci. U. S. A. 79, 3707-3710 5. Hausinger, R. P., Orme-Johnson, W. H., and Walsh, C. (1984)

Biochemistry 23,801-804 6. Noll, K. M., and Wolfe, R. S. (1986) Biochem. Biophys. Res.

Commun. 139,889-895 7. Pfaltz, A,, Jaun, B., Fassler, A., Eschenmoser, A., Jaenchen, R.,

Gilles, H. H. Diekert, G., and Thauer, R. (1982) Helu. Chim. Acta 65,828-865

8. Livingston, D. A., Pfaltz, A,, Schreiber, J., Eschenmoser, A., Ankel-Fuchs, D., Moll, J., Jaenchen, R., and Thauer, R. K. (1984) Helu. Chim. Acta 67, 334-351

9. Bible, K. C., Buytendorp, M., Zierath, P. D., and Rinehart, K. L. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 4582-4586

10. Noll, K. M., and Wolfe, R. S. (1987) Biochem. Biophys. Res.

11. Ellerman, J., Hedderich, R., Bocher, R., and Thauer, R. K. (1988)

12. Noll, K. M., Rinehart, K. L., Tanner, R. S., and Wolfe, R. S.

13. Ankel-Fuchs, D., Bocher, R., Thauer, R. K., Noll, K. M., and

14. Bobik, T. A., Olson, K. D., Noll, K. M., and Wolfe, R. S. (1987)

15. Bobik, T. A., and Wolfe, R. S. (1988) Proc. Natl. Acad. Sci.

16. Sauer, F. D., Blackwell, B. A., and Kramer, J. K. G. (1987) Biochem. Biophys. Res. Commun. 147, 1021-1026

17. Albracht, S. P. J., Ankel-Fuchs, D., Bocher, R., Ellerman, J., Moll, J., van der Zwaan, J. W., and Thauer, R. K. (1988) Biochim. Biophys. Acta 955,86-102

18. Eidsness, M. K., Sullivan, R. J., Schwartz, J. R., Hartzell, P. L., Wolfe, R. S., Flank, A. M., Cramer, S. P., and Scott, R. A. (1986) J. Am. Chem. SOC. 108, 3120-3121

19. Pfaltz, A,, Livingston, D. A., Jaun, B., Diekert, G., Thauer, R. K., and Eschenmoser, A. (1985) Helu. Chim. Acta 68, 1338- 1358

20. Diekert, G., Konheiser, U., Piechulla, K., and Thauer, R. K. (1981) J. Bacteriol. 148,459-464

21. Shiemke, A. K., Hamilton, C. L., and Scott, R. A. (1988) J. Biol. Chem. 263,5611-5616

Biophys. Acta 768, 113-163

Commun. 145, 204-210

Eur. J. Biochem. 172,669-677

(1986) Proc. Natl. Acad. Sci. U. S. A. 83,4238-4242

Wolfe, R. S. (1987) FEBS Lett. 213, 123-127

Biochem. Biophys. Res. Commun. 149,455-460

U. S. A . 85,60-63

23. 24. 25.

26.

27.

28.

29.

30.

31. 32.

33.

34.

35. 36.

37.

chemis& 17,'2374-2377. Shelnutt, J. A. (1983) J. Phys. Chem. 87,605-616 Scott, R. A. (1985) Methods Enzymol. 117,414-459 Scott, R. A., Schwartz, J. R., and Cramer, S. P. (1986) Biochem-

istry 25,5546-5555 Kratky, C., Fassler, A., Pfaltz, A., Krautler, B., Jaun, B., and

Eschenmoser, A. (1984) J. Chem. SOC. Chem. Commun. 1368- 1371

Kratky, C., Waditschatka, R., Angst, C., Johansen, J . E., Plaque- vant. J. C.. Schreiber. J.. and Eschenmoser. A. (1985) Helu. . . . Chin . Ac ta '68 , 1312-1337

Knam. S.. Keenan. T. P.. Zhang. X.. Fikar. R.. Potenza. J. A., an'd'schugar, H. j. (1986) J. A&. c b m . SOC. ios, 1882-1883

Basson, S. S., Bok, L. D. C., and Leipoldt, J. G. (1969) Acta Crystallogr. Sect. B Struct. Crystallogr. Cryst. Chem. 25, 579- 583

Kirner, J. F., Garofalo, J., Jr., and Scheidt, W. R. (1975) Znorg. N u l . Chem. Lett. 11, 107-112

Teo, B. K. (1981) J. Am. Chem. SOC. 103,3990-4001 Scott, R. A., Schwartz, J. R., and Cramer, S. P. (1985) J. Znorg.

Biochem. 23, 199-205 Smith, D. A., Heeg, M. J., Heineman, W. R., and Elder, R. C. (1984) J. Am. Chem. SOC. 106, 3053-3054

Bein, T., McLain, S. J., Corbin, D. R., Farlee, R. D., Moller, K., Stuckv. G. D.. Wollerv, G.. and Sayers, D. (1988) J. Am. Chem. - . soc. i io , ~ ~ o ~ - ~ ~ ~ o ~ ' '

Shelnutt. J . A. (1987) J. Am. Chem. SOC. 109, 4169-4173 Shiemke; A. K.; Scott, R. A., and Shelnutt, J. A. (1988) J. Am.

Shiemke, A. K., Eirich, L. D., and Loehr, T. M. (1983) Biochim. Chem. SOC. 110, 1645-1646

B ~ O D ~ V S . Acta 748. 143-147 38. Gouterman, M. (1978) in The Porphyrins (Dolphin, D., ed) Vol.

111, Part A, pp. 1-165, Academic Press, New York 39. Walker, F. A., Hui, E., and Walker, J. M. (1975) J. Am. Chem.

SOC. 97, 2390-2397 40. Ihara, Y., Fakuda, Y., and Sone, K. (1987) Znorg. Chem. 26,

3745-3750 41. Findsen, E. W., Alston, K., Shelnutt, J. A., and Ondrias, M. R.

(1986) J. Am. Chem. SOC. 108,4009-4017 42. Shelnutt, J . A., Alston, K., Ho, J.-Y., Yu, N.-T., Yamamoto, T.,

and Rifkind, J. M. (1986) Biochemistry 25,620-627 43. Albracht, S. P. J., Ankel-Fuchs, D., Van der Zwaan, J. W.,

Fontijn, R. D., and Thauer, R. K. (1986) Biochim. Biophys. Acta 870,50-57

44. Jaun, B., and Pfaltz, A. (1986) J. Chem. SOC. Chem. Commun.

45. Walsh, C. T., and Orme-Johnson, W. H. (1987) Biochemistry 26,

46. Eschenmoser, A. (1986) Ann. N. Y. Acad. Sci. 471, 108-129

1327-1329

4901-4906