Proc. Natl. Acad. Sci. USA Vol. 73, No. lO, pp. 3338-3342, October 1976 Chemistry

Carbon monoxide binding to pentacoordinate mercaptide-heme complexes: Kinetic study on models for cytochrome P-450

(kinetics /thermodynamics / alkoxide-heme)

C. K. CHANG AND D. DOLPHIN

Department of Chemistry, University of British Columbia, Vancouver, B.G Canada V6T 1 W5

Communicated by E. Margoltash, July 16, 1976

ABSTRACf Mercaptide anions form exclusively penta-coordinate heme complexes fRS--hemel in polar and nonpolar solution over a wide range of mercaptide concentration. These complexes have a Soret peak at 408 om and a formation constant of about 2.5 X 1()4 M-l, and combine with CO to give a CO-cytochrome P-450 type spectrum. Kinetics of CO binding to mercaptide-heme complexes [RS--heme] have been studied by the flash photolysis method. Characteristic constants for this reaction suggest close similarities betwen [CH:r(CH2h-S--heme] and cytochrome P-450. The reaction of alkoxide anion with heme has also been examined but no evidence was found for the existence of the [RO--hemeoCO] species.

Cytochromes P-450 are a unique class of hemeproteins that catalyze the hydroxylation of a wide variety of organic com-pounds through the activation of molecular oxygen (1, 2). Al-though the primary locus of such enzyme systems is in the mi-crosomal fraction of mammalian liver and adrenal cortex much of our present knowledge concerning the properties or'P-450 has been obta~ned from studies on the more readily available soluble P-450 of bacterial origin. A substantial amount of in-formation is now available on the functional aspects of the catalytic reaction but comparatively little is known concerning the structure of the active site of the enzyme itself. Specifically, the nature and roles of axial ligands coordinated to the pros-thetic group, protoheme, are yet to be elucidated. In the absence of definitive x-ray crystal structures of P-450, we have under-taken the approach of synthesizing model compounds which mimic P-450, in order that its mechanism of enzymic activities can be elucidated.

The absorption spectrum with the unusual red-shifted Soret band of CO·P-450 complex has already been successfully duplicated with simple iron(II) porphyrins by having a mer-captide ion ligated trans to CO, at the fifth coordination site of heme (3-6). This evidence, coupled with earlier electron spin resonance experiments on both enzymes and model compounds (7-9), strongly implies that a mercaptide ligand is coordiDated to all the ferric stages and at some stages of the ferrous complex in the P-450 catalytic cycle. In the absence of CO, our model exhibited a spectrum with absorption bands at wavelengths identical to those of reduced P-450 (3, 4). Since the unligated form of P-450 is assumed to be a pentacoordinate heme with subsequent ligation of CO or O2 taking place at the sixth site, we have examined the reaction between mercaptide ion and heme and studied the properties of the "mono" mercaptide-heme species. We wish to report here the thermodynamics and kinetics of the interactions between mercaptide ion, heme, and CO, and to report some observations on the obligatory nature

Abbreviations: P-450, cytochrome P-450; DMA, N,N-dimethylacet-amide; Me2SO, dimethyl sulfoxide; Heme, iron(II) protoporphyrin IX dimethyl ester; Hemin, iron(III) protoporphyrin IX chloride dimethyl ester.

3338

of the mercaptide ligand in bringing about the long-wavelength Soret peak of the CO complexes.

MATERIALS AND METHODS Protohemin chloride dimethyl ester was prepared by insertion of iron into protoporphyrin IX dimethyl ester, which was syn-thesized from protoporphyrin via its acid chloride (10). Since hemin is reduced by mercaptide, reduction of Fe(III) proto-porphyrin IX chloride dimethyl ester prior to reaction was not necessary. In other cases, it was reduced by the PdjCaH2 method previously described (11). Dimethyl sulfoxide (MezSO) and N,N-dimethylacetamide (DMA) were purified by distil-lation over calcium hydride under reduced pressure.

Mercaptide-Crown Ether Complex. (a) Anhydrous solu-tions: KH (0.3 g; 50% in oil, Alfa) was washed twice with pen-tane and dried under vacuum. About 0.3 g of butanethiol (Eastman) in 2 ml of benzene was added under argon, and ex-cess thiol and solvent were removed by vacuum after 1 hr. The resultant white powder was then dissolved in either deaerated Me2S0 or toluene along with an equivalent amount of di-benzo-l8-crown-6 (3). The solubility of the mercaptide salt in toluene even in the presence of crown ether was rather poor. A saturated solution has a concentration of about 2 mM. The exact mercaptide concentration was determined by titration with HCI. Care was exercised to exclude oxygen, since mer-captide solution underwent rapid oxidation to give disulfide.

(b) DMA and H20 system: KOH pellets (0.2 g) were dissolved in a minimum volume of water. The amount of water was scaled so that the final solution had less than 5% water. As long as the water content was within this limit ( 450 nm) and monitored at shorter wavelengths (A < 500 nm). The duration of the flash was adjusted between 50 and 500 p.sec with a decay time of about 40 !J,sec. The output signal from the photomul-tiplier was processed electronically through two log amplifier

100

80

i 5 60

i ::!! E '" 40

20

Chemistry: Chang and Dolphin

350

408 I

i\ 451 j:

I \

450 550 A (nm)

25

20

5

650

FIG. 1. Optical absorption spectra of Fe(1I) protoporphyrin IX dimethyl ester. and mercaptide complexes in toluene at 23°: (-), without carbon monoxide; ( .... ), in the presence of 1 atom of CO. Background absorption of dibenzo-l8-crown-6 present in the solution has been compensated.

circuits to give direct display of absorbancy change as well as the first-order rate plots (14).

RESULTS

The formation and spectral properties of pentacoordinate [RS-·Heme]

Reactions between iron(II) porphyrin and various ligands (L) can be generally divided into two categories, as depicted in Eq. 1.

K, K, Heme + 2L ~ L·Heme + L ~ L·Heme·L [1)

There are ligands such as pyridine and imidazole with which K2 is larger than K 1, such that the formation of hexacoordinate "hemochromes" is favored over the pentacoordinate species and isolation of the intermediate is difficult unless special ste-reochemical designs are incorporated into the heme molecule (12, 15, 16). On the other hand, a ligand like CO has a larger K 1, and preparation of the pentacoordinate heme, in solution, is possible by controlling the ligand concentration (17).

When protoheme was reacted with the mercaptide-crown ether complex only a single spectrum reflecting a single species was observed over a wide range of mercaptide concentration (1-450 mM). This spectrum (unbroken curve in Figs. 1 and 2) bears close resemblance to that of unligated P-450 (2). Although direct titration of heme with mercaptide solution resulted in spectral transformation with isosbestic points, and stoichio-metric calculations indicated that only one mercaptide ligand was coordinate to each heme, we have alternatively employed a flash photolysis technique to establish that the species which exhibited the 408 run Soret peak W!lS indeed a monomercaptide heme.

It is well known that all CO·heme complexes are photodis-sociable, that is, Ll in Eq. 2 is greatly enhanced by light

h_1 L'Heme-c0 ~ L·Heme + CO (2)

hi

while the chelation of L and Heme is little affected. Chang and Traylor have constructed a flash photoly,sis apparatus equipped with cross light-filtering devices and sharp cut-off edge for the

Proc. Natl. Acad. Sci. USA 73 (1976) 3339

··jA - -- --- ---

I 1.4 1

1.2

g 1.0 ~ o ..

.l:)

tV 0.8

0.6

0.4

0 .2

: :

t

t

\ ....

400 500 600 A (nm)

FIG. 2. Optical absorption spectra of [RS-.heme] (-) and (RS-.Heme-CO] ( .••. ) in DMAat23°. The 0 and. areabsorbancies read from flash photolysis relaxation traces at time zero and at equilibrium, respectively.

light pulse (12). This set-up is capable of precisely monitoring the absorbancy change of the heme solution at the very moment wben CO is flashed off. By performing this procedure at various wavelengths the whole spectrum of the CO-stiipped penta-coordinate heme can be obt.1lined.

When the [RS-·Heme-CO] complex (dotted curve, Figs. 1 and 2) was studied by this flash method under pseudo-first-order conditions, clean exponential relaxation curves were observed (Fig. 3). The apparent on rates for CO recombination were independent of wavelength or the concentration of

1.4

3 '" u 1.0 c '" ..0 ~ g 2 0 >

..0 0.6

3340 Chemistry: Chang and Dolphin Proc. Natl. Acad. Sci. USA 73 (1976)

Table 1. Characteristic constants for binding of carbon monoxide by Fe(lI) protoporphyrin IX dimethyl ester·mercaptide complex and hemeproteins

Heme compound

Fe(U) protopor-phyrin IX di-methyl ester

Bacterial P-450, camphor freec

Bacterial P-450, with camphorc

Liver microsomal P-450d

Horse myoglobin, pH 7.0e

Solvent

DMA DMA DMA DMA DMA Toluene Toluene Toluene

Concen-tration

of BuS® (mM)

1.8 1.8

100 300

1.8 2 2 2

Tem-pera-ture (C)

2

23} 23 23 40

2 23 40

4

4

4

20

1.1 X 105

1.2 X 105

5.1 X 10'

3.7 X 10' 0.14

4.5 X lOS 0.63

5 X lOS

2.5 19 17 18 65

0.5 5

42

8.9 X 10' 9.8 X 103 1.0 X 104 1:0 X 104 2.2 X 103 2.1 X lOS 2.0 X 10' 2.4 X 103

2.2 X.10'

2.6 X lOS

3-7 X 105

4 X 107

i:1H (kcal

mol-I)

-16

-19

o

-10.5

i:1S (cal mol-I

degree-I)

-35

-48

BuS@is butyl mercaptide potassium crown ether. P1 / 2 is CO pressure at half saturation. 1 cal = 4.184 J. a Estimated error ± 10%. Solubilities of CO per torr of CO in DMA: at 2°, 4.5/lM; 23°, 5.3/lM; 40°, 7.0 /lM (ref. 19, also B. R. James and D.

Wang, unpublished results) and in toluene: 2°, 9.5 I'M, 23° and 40°, 9.9 I'M (extrapolated from data given in ref. 20). b Calculated using Eq. 3. c See ref. 21. d Refs. 22 and 23. e See pp. 224-226 ofref. 18.

mercaptide in solution. The absorbancy of the sample at the beginning and the equilibrated state of the relaxation process was read directly from the trace and reconstructed in Fig. 2. It is significant that the CO-dissociated spectrum is identical to the unligated spectrum, thereby substantiating the postulate that the unligated form of mercaptide.heme (and P-450) is a pentacoordinate heme. Since the variation of both the mer-captide concentration and the duration of the flash pulse (50-500 p.sec) has absolutely no effect on the absorbancy of the CO-stripped intermediate, the possibility that a second mer-captide with extremely fast on rate binds to the heme during the flashing process is unlikely. Furthermore, the monophasic nature of the CO relaxation curve plus the fact that both kl and Klare independent of mercaptide concentration (vide in jra ) invalidates such argument.

Equilibria and kinetics of CO binding by pentacoordinate [RS-·Heme]

As shown by Eq. 3, the combination of CO and heme can be expressed by a rate equation:

[3]

The use of different CO concentrations should allow for the determination of both kl and Ll (p. 194 of ref. 18). In practice, the value of Ll derived by this extrapolation is usually subject to greater error than klo and should thus be checked against the value calculated from static equilibrium measurement of K = kdk- 1• The observed on rats are plotted versus CO concen-tration in Fig. 4. The small perturbations of rate observed at different mercaptide concentrations are inSignificant with re-spect to the 45-fold difference in mercaptide concentration; and the slight decrease of kobs at higher concentration is believed

due to the corresponding depression of CO solubility in the liquid phase by ionic salt. * The kinetic data and equilibrium constants measured by direct CO titrations are listed in Table l. The binding of mercaptide by heme and carbomonoxyheme have also been measured by titrating mercaptide with heme and carbomonoxyheme, respectively. These equilibrium values are used to construct the following cycle.

Heme

K, 2.5 X 10<

(toluene) llX 10'

(DMA)

5~::enejl 12'~£:e) 5XIO< LOX 10'

(DMA) (DMA)

Heme-CO ~(=::;:=.; RS-'Heme-CO K,

5X10'

(toluene)

27 X 10" (DMA)

Reactions of alkoxide with heme

When a carefully degassed Fe(lU) protoporphyrin IX chloride

• The previous erroneous assignment of the unligated form of mer-captide.heme complex as a hexacoordinate species (3) based on CO affinity data could also have resulted from this fluctuation in CO solubility.

Chemistry: Chang and Dolphin

200 Bu§ -Fl I

i g ..!!

.. 100 ..c ~o

100 200 Carbon monoxide {torr}

300

FIG. 4. Relationship between carbon monoxide concentration and the observed CO on rate (k 1) for the formation of the [RS-' Heme·CO] complex. One torr = 133 pascals. Concentrations of butyl mercaptide were: e, 0.01 M; c, 0.05 M; .... , 0.1 M; 0,0.45 M.

dimethyl ester solution in either DMA or Me2S0 was mixed with a dilute solution of methoxide (NaOMe or KOMe) or the Me2S0 anion

( KOO(CH3)

CH2

in Me2S0, an interesting auto-reduction of hemin took place instantaneously. The resulting ferrous complex exhibited an absorption spectrum with the Soret band centered at 435 nm and an additional shoulder around 405 nm. This species was quite stable when oxygen had been completely excluded. The same spectrum was also observed when reduced heme was ti-trated with the base; in such cases, clean isosbestic points were observed. Crown ether did not have any marked influence on the spectra, and the close spectral resemblance between these complexes and dihydroxyheme (24) strongly suggests a [RO-· Heme·-OR] structure. We have not been able to identify an intermediate corresponding to the pentacoordinate [RO-· Heme] species. An interesting property that should be em-phasized is that at high concentration of the base (base to heme ratio> 100:1), no spectral change could be detected when these complexes were exposed to 1 atmosphere of CO. Similar ob-servations have been reported for dihydroxyheme (24). At low alkoxide concentration, however, addition of CO resulted in the formation of a different spectrum, which was identical in every respect to that of a normal heme-CO complex (Soret peak at 412 nm). This alkoxide-heme system has been examined by our flash photolysis technique and the result indicated that the CO-dissociated intermediate has a spectrum different from that of the original unligated alkoxide-heme (435 nm), the disso-ciated spectrum being similar to that of a reduced heme (422 nm). This implies that [RO-·Heme--OR] or [RO-·Heme] must dissociate RO- ligands before the formation of a pentacoordi-nate [Heme-CO] can take place. Except for the unlikely coin-cidences that the intermediate [RO- ·Heme] has a spectrum identical to [Heme], and [RO-·Heme-CO] a spectrum identical to [Heme-CO], the formation of the complex [RO-·Heme-CO] must be an extremely unfavorable process.

DISCUSSION

Comparisons between the Model System and P-450. The coordination of mercaptide anion to ferrous heme represents an unusual case of heme chemistry in that no evidence for the formation of the di-mercaptide-heme can be found (K2 = 0 in Eq. I). The exclusion of the hexacoordinate species in solution,

Proc. Natl. Acad. Sci. USA 73 (1976) 3341

Table 2. Spectral characteristics of CO complexes

Compound

Bacterial P-450* In solution Single crystal*

[RS-·Heme·CO] in DMA [RS-.Heme.CO] in toluene

.. See Ref. 25.

Wavelengths (nm)

365 447 363 446 376 460 370 451

Absorbance ratio

("450" nm: "370" nm)

2.15 2.28 0.98 1.74

over a wide range of mercaptide concentration, greatly simplifies the preparation of the pentacoordinate [~-•. Heme], which could be a suitable model for the P-450 active site. The cause of this inability to bind two mercaptide ligands is not obvious but may be related to the repulsive force. engendered by the soft shell mercaptide and the elec~on-rich ferr?us iron. In fact, the binding of just one mercaptide to heme IS rather weak if the availability of the anion is not optimized (3) and a substantial reduction of affinity has been observed when the mercaptide salt itself, instead of its crown ether complex, was used in this reaction.

As shown in Figs. 1 and 2, the polarity of the heme environ-ment appears to playa significant role in the absorption spec-trum of the [RS-·Heme-CO] complex. Both the wavelength of the "Soret" peaks and the ratio of absorbances between the "450" band and the "370" band can be varied by changing solvent polarity. Splitting of a single Soret band into two bands ("hyper" type spectrum) observed in our model systems and in the CO·P-450 complex has been interpreted, as a charge transfer from a mer~ptide sulfur orbital to the porphyrin eg (11'*) coupled to the normal porphyrin 11' -11'* transition (25). The comparison of wavelength and absorbance ratio between models and the enzyme (Table 2) suggests that the heme en-vironment in the protein is most likely nonpolar. This idea is congruous with the observation that during enzymic reactions nonpolar hydrocarbon substrates are closely associated with the P-450 active site (2) and a hydrophobic active site would favor such substrate binding.

Kinetic and equilibr.ium data for the formation of both [RS-·Heme] and [RS-·Heme-CO] complexes also show a slight solvent effect. The studies of Rougee and Brault on CO binding by deuteroheme in various solvents (17, 26, 27) indicate that solvents such as substituted formamides are slightly coordinated to heme, and although the affinity of these solvents towards heme is low, they are capable of modulating the binding be-tween heme and other strong coordinating ligands. As shown in Scheme L there is a 2-fold decrease of the formation constants when measured in DMA, and this could be viewed as a solvent coordination effect. With respect to the enthalpy change of the CO binding by mercaptide-heme, it should be noted that even in this coordinating solvent, tl.H for CO binding is -16 kcal/ mol, a magnitude comparable to other coordination reactions of heme with small ligands. The previous literature report that no enthalpy change was observed for liver microsomal P-450 (23) is therefore very surprising, and could imply that an en-dogenous or exogeous ligand was already strongly coordinating at the sixth site of the reduced P-450, thereby compensating the heat of reaction. In this context, enthalpy data for the pu-rified bacterial P -450, with and without camphor, are needed to compare with the model systems. The kinetic and equilib-rium data also reveal the major difference between our model systems and P-450 lies in the CO off rate. The smaller k-l of

3342 Chemistry: Chang and Dolphin

CO·P-450 could mean that the protein imparts some stabilizing effect upon the CO complex.

The nature of the Fe-CO bond has been probed by infrared spectroscopy. We have found the carbonyl stretching frequency of [RS-.Heme-CO].in DMA is 1923 cm-I , the lowest Vco re-ported for CQ.heme complexes (5, 28, 29). Since Vco is supposed to reflect the extent of 1r back-bonding to CO, and thus reflect the 1r-electron density on the iron atom, this finding corrobo-rates the suggestion that mercaptide p orbitals extensively overlap with heme orbitals (4). It should be borne in mind, however, that the strength of the iron-CO bond per se (as re-lated to vco) apparently has no direct correlation with the overall affinity for CO (29). This is evidenced by the fact that despite the very "strong" bonding of the [rs-·Heme.CO) complex the CO affinity for our models or even the enzyme are rather poor when compared to other CO complexes of the general type [L·Heme-COj.

Mercaptide Versus Other Ligands. The presence of a sulfur ligand in P-450 has been repeatedly suggested in the recent literature (1-9). Although the analogy and correlation between our model systems and the enzyme presented here is a strong one, we still have to confront the ultimate, yet inevitable, queStion of whether a mercaptide is indeed coordinated to heme in P-450, and one must ask the question: could there be any other ligands responsible for the "hyper" type spectrum? Indeed, the mercaptide postulate recently has been questioned by Hager's observations with chloroperoxidase (30-32). The striking optical and MOssbauer spectral similarities found be-tween chloroperoxidase and P-450· suggest that an identical axial ligand is present in both enzymes, yet there is evidence that only one disulfide cystine linkage is found in both the ox-idized and reduced form of chloroperoxidase (32). In view of the striking difference in coordination properties between mercaptide and mercaptan we have examined a number of negatively charged ligands such as alkoxide and imidazole anions to see if they can produce a hyper spectrum.

No appreciable differences were observed between imida-zolide alkali salt and neutral imidazole in terms of both reduced and CO spectra, and the same was found for amine anions. The behavior of alkoxide as well as hydroxide anion should prove to be theoretically interesting in that they form a different type of spectrum for the heme complex, presumably of the struc-ture

[RO-'Heme--OR] (R'" lI,CHa, _S(CH3); CHa

yet they show no tendency to form [RO-·Heme-CO) complexes. It is not clear why this should be such an unfavored process when contrasted to the coordination of CO found in [RS-. HemeoCOj and (NC-·Heme-COj complexes (33), and of course, no P-450 type spectrum was observed with alkoxide. These results combined with our previous observation (3) of the in-ability of disulfide, carboxylate, or phenoxide ligands to bring about the "450" band prompts us to suggest that at this stage, mercaptide is solely responsible for the unusual spectra of the unligated and the CO coordinated forms of P-450. These conclusions call for further investigation into the properties of both enzymes, especially the kinetic and thermodynamic as-pects of the CO binding reaction, since we now have on hand an authentic pentacoordinate [RS-.Heme] by which direct comparisons with the enzyme can be made.

The authors wish to acknowledge the generosity of Prof. T. G. Traylor in allowing us the use of his facilities while performing the kinetic portion of this work. e.K.e. thanks John Geibel for his kind hospitality and assistance. This work is a contribution from the

Proc. Natl. Acad. Sd. USA 78 (1976)

Bioinorganic Chemistry Group and is supported by operating and negotiated development grants from the National Research Council of Canada and the United States National Institutes of Health (AM-17989).

1. Cooper, D. Y., Rosenthal, 0., Snyder, R. & Witmer, C. (1975) Cytochromes P-450 and bs (Plenum Press, New York), pp. 1-334.

2. Gunsalus, I. C., Meek, J. R., Lipscomb, J. D., Debrunner, P. & Munk, E. (1974) in Molecular Mechanisms of Oxygen Activa-tion, ed. Hayashi, O. (Academic Press, New York), pp 559-613.

3. Chang, c. K. & Dolphin, D. (1975) J. Am. Chern. Soc. 97, 5948-5950.

4. Chang, e. K. & Dolphin, D. (1976) }. Am. Chern. Soc. 98, 1607-1609.

5. Collman, J. P. & Sorrell, T. N. (1975) J. Am. Chern. Soc. 97, 4133-4134.

6. Stern, J. O. & Peisach, J. (1914) J. Bioi. Chern. 249, 7495-7498.

7. Tsai, R, Yu, C. A., Gunsalus, I. e., Peisach, J., Blumberg, W., Orme-Johnson, W. H. & Beinert, H. (1970) Proc. Natl. Acad. Set. USA 66,1157-1161.

8. Collman, J. P., Sorrell, T. N. & Hoffman, B. M. (1975) J. Am. Chern. Soc. 97,913-914.

9. Koch, S., Tang, S. C., Holm, R. H., Frankel, R. B. & Ibers, J. A. (1975) J. Am. Chern. Soc. 97,916-918.

10. Chang, C. K. & Traylor, T. G. (1973) Proc. Natl. Acad. Set. USA 70, 2647-2650.

11. Brinigar, W. S. & Chang, C. K. (1974) }. Am. Chern. Soc. 96, 5595-5597.

12. Chang, e. K. & Traylor, T. G. (1975) Proc. Natl. Acad. Sci. USA 72,1166-1170.

13. Chang, C. K. & Traylor, T. G. (1975) Biochem. Biophys. Res. Commun. 62,729-735.

14. Chang, C. K. (1976) Applied Spectroscopy, 30,364-365. 15. Collman, J. P., Gagne, R. R., Reed, e. A., Halbert, T. R., Lang,

G. & Robinson, W. T. (1975) J. Am. Chern. Soc. 97, 1427-1439.

16. Almog, J., Baldwin, J. E. Huff, J. (1975) J. Am. Chem. Soc. 97, 226-228.

17. Rougee, M. & Brault, D. (1973) Biochem. Biophys. Res. Com-man. 55,1364-1369.

18. Antonini, E. & Brunori, M. (1971) Hemoglobin and Myoglobin in their Reactions with Ligands (North Holland Publishiitg Co., Amsterdam).

19. Rempel, G. L. (1968) Doctoral dissertation, University of British Columbia, p. 170.

20. Seidell, A. (1940) Solubilities of Inorganic and Metal-Organic Compounds (VanNostrand, Princeton, N.J.), 3rd ed., Vol. I, pp 217-219.

21. Peterson, J. A. & Griffin, B. W. (1972) Arch. Biochern. Biophys. 151,427-433.

22. Omura, T. & Sato, R (1964) }. Bioi. Chem. 239,2370-2385. 23. Debey, P., Hui Bon Hoa, G. & Douzou, P. (1973) FEBS Lett. 32,

227-230. 24. Keilin, J. (1949) Biochem. J. 45,448-455. 25. Hanson, L. K., Eaton, W. A., Sligar,S. G., Gunsalus, I. C., Gout-

erman, M. & Connell, C. R (1976) }. Am. Chem. Soc. 98, 2672-2674.

26. Rougee, M. & Brault, D. (1975) Biochemistry 14,4100-4106. 27. Brault, D. & Rougee, M. (1974) Biochernistry 13,4591-4602. 28. Caughey, W. S. (1970) Ann. N.Y. Acad. Set. 174,148-153. 29. Caughey, W. S., Alben, J. 0., McCoy, S., Boyer, S. H., Charache,

S. & Hathaway, P. (1969) Biochemistry 8,59-62. 30. Champion, P. M., Munck, E., Debrunner, P. G., Hollenberg, P.

F. & Hager, L. P. (1973) Biochemistry 12, 426-435. 31. Hollenberg, P. F. & Hager, L. P. (1973) }. Bioi. Chern. 248,

2630-2633. 32. Chiang, R., Makino, R., Spomer, W. E. & Hager, L. P. (1975)

Biochemistry 14,4166-4171. 33. Keilin, J. (1949) Biochem.}. 45,440-448.