There is a strong temperature dependence to the EPR spectra of many NO-heme ...... Townes Dailey theory (34, 35) predicts a principal direction of positive ...
THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 258, No. 12, Issue of June 25, pp. 7437-7448,1983 Printed in U.S.A.
Electron Nuclear Double Resonance and Electron Paramagnetic Resonance Study on the Structure of the NO-ligated Heme us in Cytochrome c Oxidase* (Received
Russell
LoBruttoS,
From the Department Bioenergetics, State
Yau-Heui of Physics
University
Wei&
and Center of New York
Rita
for publication,
Mascarenhas,
Charles
Biological Macromolecdes at Albany, Albany, New York for
May
3, 1982, and
P. Scholes7,
and the Department 12222
in revisedform, January 6,1983) and Tsoo E. King of Chemistry
and Laboratory
of
be observed by the EPR technique, although some later products of O2 catalysis may be observed by EPR (2). Nitric oxide binds to heme protein systems, and the stereochemistry of NO binding to heme systems (3-6), with a characteristically bent Fe-NO bond, is simlar to that of 0, binding (7-9). The unpaired electron spin of NO adds to the even electron ferrous heme system and causes the overall system to become EPR visible. Attention was first focused on NO-hemoglobin (10,11) and NO-myoglobin (12, 13) complexes. From single crystal EPR studies, the electronic g value nearest 2.00 was found to lie near the heme normal. Hyperfine coupling to 57Fe, NO, and histidine nitrogen showed that the unpaired electron was delocalized from NO to Fe and to the proximal histidine (12, 13). The directions of the NO hyperfine tensor relative to the heme normal and the electronic g tensor gave evidence for a bent Fe-NO bond. Compared to the NO complexes of other heme proteins (14) and model heme systems (15-17), the NO adduct of the ferrocytochrome a3 yielded well resolved frozen solution EPR spectra and hyperfine couplings (18). Recent studies with isotopically enriched I”NJhistidine have shown histidine to be the nitrogenous ligand of the u3 heme (19). Single crystals of cytochrome c oxidase are not available, but the NO-ligated protein was prepared in oriented membranes (20) and the EPR orientation behavior then was measured with respect to the membrane normal. There is a strong temperature dependence to the EPR spectra of many NO-heme complexes such as myoglobin (21, 22) and NO-protohemin-imidazole (21). The dependence indicates a thermal equilibrium between conformations which have differing amounts of hyperfine coupling to the sixth The respiratory chain located in the inner mitochondrial nitrogenous Iigand and which may have differing bond angles membrane ends at cytochrome oxidase, whose cytochrome a3 and bond distances. Until low temperature x-ray work is done center catalyzes the reduction of diatomic oxygen to water on these complexes, or unless there are internal structural (1). The reaction has been known for more than half a century, checks (such as might come from ENDOR’J, there is a funbut the catalytic mechanism is still not understood. Studies damental problem in correlating low temperature EPR inforof electronic structure and stereochemistry can certainly mation with room temperature x-ray information. Fortuthrow some light on the elusive mechanism. A problem is that nately, the nitrosylferrocytochrome a3 complex happens to the true 0, adducts of the reduced form of the enzyme cannot yield, with proper preparative methods, only one type of EPR spectrum (21). Conventional EPR of NO-ligated heme proteins has not * This work was supported by National Institutes of Health Grants determined several important magnetic parameters or has GM 16767, HLB 12576, AM 17884, and 2507RR0712213. The costs of publication of this article were defrayed in part by the payment of determined them only with poor resolution. Since EPR from page charges. This article must therefore be hereby marked “adverfrozen material resolves only one set of large hyperfine coutisement” in accordance with 18 U.S.C. Section 1734 solely to indicate plings near g = 2.00, the smaller components of hyperfine The techniques of EPR and electron nuclear double resonance (ENDOR) were used to probe structure and electronic distribution at the nitric oxide (NO)-ligated heme a3 in the nitrosylferrocytochrome a3 moiety of fully reduced cytochrome c oxidase. Hyperfine and quadrupole couplings to NO (in both “NO and 14N0 forms), to hi&dine nitrogens, and to protons near the heme site were obtained. Parallel studies were also performed on NO-ligated myoglobin and model NOheme-imidazole systems. The major findings and interpretations on nitrosylferrocytochrome a3 were: 1) compared to other NO-heme-imidazole systems, the nitrosylferrocytochrome as gave better resolution of EPR and ENDOR signals; 2) at the maximal g value (gX = 2.09), particularly well resolved NO nitrogen hyperfine and quadrupole couplings and mesoproton hyperfine couplings were seen. These hyperfine and quadrupole couplings gave information on the electronic distribution on the NO, on the orientation of the g tensor with respect to the heme, and possibly on the orientation of the FeNO plane; 3) a combination of experimental EPR-ENDOR results and EPR spectral simulations evidenced a rotation of the NO hyperfine tensor with respect to the electronic g tensor; this implied a bent Fe-NO bond; 4) ENDOR showed a unique proton not seen in the other NO heme systems studied. The magnitude of this proton’s hyperfine coupling was consistent with this proton being part of a nearby protein side chain that perturbs an axial ligand like NO or OZ.
this
fact. $ Current address, Johnson Research Foundation, G4, University of Pennsylvania, Philadelphia, PA 19104. 3 Current address, Department of Biochemistry, National YangMing Medical School, Shih-Pai, Taipei, Taiwan. 7 Recipient of National Institutes of Health Research Career Development Award AM 00274.
‘The abbreviations used are: ENDOR, electron nuclear double resonance; p.t.p., peak to peak; TPP, tetraphenylporphyrin; OEP, octaethylporphyrin; PMS, phenazine methosulfate; DMF, N,N-dimethyl formamide; Mb, myoglobin; Hb, hemoglobin; i.d., inside diameter; o.d., outside diameter; RF, radio frequency.
7437
7438
ENDOR-EPR of NO-ligated Heme a3 in Cytochrome c Oxidase
couplings to the bound NO and to the sixth ligand need to be determined. Nitrogen quadrupole couplings would yield important information on theelectronic distribution. It may be possible to inferrelative orientation of magnetic axes and the hemeenvironmentfrom hyperfinecouplings, particularly from the anisotropic dipolar couplings to certain assignable protons. MATERIALS AND METHODS
Instrumentation-EPR spectra were obtained with a Bruker ER420 spectrometer equipped for low power dispersion operation and equipped with a home-built low temperature microwave head and liquid helium double Dewar. Details of the ENDOR apparatus are given in Ref. 23. In the 5-70 K range, a Linear Research model LR130 temperature controller wasused. To avoid EPR lineshape distortion, standard absorption mode first derivative spectra (dx"/ dH)were taken at 77 K with 100 kHz modulation (of 1.8 G p.t.p.1 at powers of 10 milliwatts. For spin counts, EPR spectra were digitized and double integrated, and the results were compared to a 1 mM Cu(11) perchlorate standard. EPR frequencies and magnetic field strengths for g value measurements were obtained as described in Ref. 24. For NO. heme systems the g values were empirically determined from spectral features as follows: The maximum g value, g,, was determined from the lowGeld derivative maximum; the intermediate g value, g,, (=: 2.00), was measured a t the first derivative zero crossing; the minimum g value, g,, was measured approximately at the high field derivative minimum. The minimum g value thus measured may be in some error because of its spectral proximity to the intermediate g value and because of the large outlying hyperfine couplings associated with the nearby intermediate g value. ENDOR was done exclusively in the dispersion (x') mode to optimize spectral resolution and signal to noise. Optimum 14N and 15N ENDOR was observed at 6-8 K with 100 kHz field modulation of about 7 G p.t.p. and ENDOR RF fields of about 0.5 G p.t.p. Optimum proton ENDOR was found at 4.2 K with field modulation of about 0.5 G p.t.p. Microwavepowers of 1-10 microwatts were employed. EPR Spectral Simulations- We used Program SIM (25) to simulate the EPRspectra of nitrosylferrocytochrome a3 having either I4NOor I5NO as a ligand This program incorporated the rhombic g tensor and included both NO and histidine hyperfine couplings, as estimated from EPR andENDOR. The line width associated with theg, features could be made different from that associated with g, and g,. We modified SIM in order to include by methods of Ref. 26 the rotation of the NO hyperfine tensor with respect to theelectron g tensor. Sample Preparation-Bovine cardiac cytochrome c oxidase was prepared by the method of Ref. 27 with modifications as described in Ref. 28. Samples contained typically 100 mgof protein/ml with a copper-heme a ratio in the 1.0-1.1 range, with 11 nmol of heme a/mg of protein, and with about 20% by weight phospholipid. The enzymatic activity was about 12 nmol of O,/s/nmol of heme a. The sample was dissolved in 1%sodium cholate, 50 mM phosphate buffer, pH 7.4. Prior to checking for exchangeable protons by ENDOR, a sample of oxidase was solubilized in the above cholate-containing buffer which had been prepared with DzO (99.8% deuterated, Bio-Rad Laboratories). The oxidase was dialyzed against this deuterated buffer for 1.5 h. (Overall deuteration of this buffer was slightly less than 99.8% because the cholate and phosphate were not deuterated.) NO. heme.imidazole complexes wereprepared with standard, fully protonated, triply crystallized imidazole (Sigma), fully protonated 1methylimidazole (Sigma), or 98% deuterated imidazole (Stohler ISOtopes, Waltham, MA). Heme a was prepared by the method of Ref. 28. Protohemin (Sigma, equine type III), FeTPP (Strem, Newburyport, MA) and FeOEP (Strem) were usedwithout further purification. Sperm whale metmyoglobin (Sigma, type 11) was chromatographically purified on a DEAE-cellulose (DE52, Whatman) column. NO derivatives of heme proteins and model hemes were prepared under strictly anaerobic conditions in a Thunbergcell containing two side arms and a main compartment. The cell was connected to a 9mm o.d., 7 mm i d . quartz EPR-ENDOR tube at thebottom and toa vacuum line at the top. About 1.5 ml of cytochrome oxidase solution was placed in the main compartment with 10-12 pl of 0.01% PMS (Sigma) and 40 ell of 200 mM reduced NADH (Sigma) in the side arm. After 10-12 cycles of evacuation and flushing with high purity argon and gentle agitation over ice, the oxidase was mixed with the reductants in the side arm. The reduction was allowed to continue for 2030 min overice. The NO gas, scrubbed with NaOH, was then
introduced by way of a three-way stopcock to a partial pressure of about 200 mm and reacted with the reduced oxidase for 10-15 min over ice.' (14N0 was from Linde and "NO from Prochem Isotopes, Summit, NJ). Excess NO was removed by evacuating, flushing with argon 3-4 times. The sample was drained into the attached EPR tuhe, frozen by plunging into liquid nitrogen, and stored under liquid nitrogen. The product of these reactions was fullyreduced cytochrome c oxidase exhibiting the EPRspectrum of nitrosylferrocytochrome a3 (18-21). There was no EPR evidence for oxidized cytochrome a with its low spin ferric heme a and Cu. signals. A similar procedure was used for the preparation of NO-Mb derivatives, except that for reduction of Mb a 1.5 molar excess (relative to heme) of sodium dithionite (Hardman& Holden, Ltd., Manchester, England) was used. NO-heme-imidazolecomplexes were prepared by a method similar to that of Scholler et at. (29), which initially created the NO-heme complex by reductive nitrosylation. The resultant NO. hemewas anaerobically dissolved in DMF (99.5%deuterated, Merck) and a 60fold molor excess ofimidazole added under anaerobic conditions. The resultant six-coordinate NO-heme-imidazole complex (as readily noted from its EPR spectrum) was tipped into the EPR tuhe and quickly frozen in liquid N2. RESULTS
EPR Results-The spectra in Fig. 1 offer a comparison of 14NO-and "NO-ligated nitrosylferrocytochrome a3 and NOMb. (The NO-ligatedoxidase referred to henceforth isexplicitly nitrosylferrocytochrome u3 with NO ligated to heme ~3.) As reported elsewhere (18),the 14NO-and "NO-ligated oxidase spectra are respectively dominated by nine and six hyperfine line patterns whose centers lie at the first derivative zero crossing (gz= 2.006). The larger hyperfine coupling in the pattern (Table IB) is assigned to NO and the smaller to the histidine nitrogen (19). There are less well resolved hyperfine structures associated with the low field (gX)and high field (g,) g values. The g values of the NO-ligated oxidase were experimentally determined as g,, g,, g, = 2.091 f 0.002, 2.006 +- 0.002, and 1.980 f 0.003. NO-Mb has less well resolved hyperfine structure than the NO-ligated oxidase. Wealso noted the feature of NO-Mb near g = 2.045 which indicatesanothertype of NO complex, possibly the more stable one at high temperature (21, 22). TheEPRspectra of model NO-imidazolecomplexes of FeTPP, FeOEP, heme a, and protohemin were not as well resolved as thatof the NO-ligated oxidase. They more closely resembled the spectrum of NO-Mb in their overall appearance, more poorlyresolved hyperfine structure, and somewhat in theirg values, especiallytheir smallervalue ofg, compared to NO-ligated oxidase (2.07 uersw 2.09). The similarityofthe model complexes to NO-Mb and to each other, rather than to NO oxidase, was especially interesting for the hemea model because this meant that the uniqueness of the NO-ligated oxidase EPR spectrum is not solely due to hemea. Nitrogen ENDOR-ENDOR spectra and ENDOR frequencies are usually interpretable by first order expressions resulting from hyperfine, quadrupole, and nuclear Zeeman interactions (23). To first order, a 14N( I = 1) nucleus gives four ENDOR transitions at 14
vENWR
?4
1 I4Aeff J +- % Qes
f "VN
(la)
and a lSN ( I = %) nucleus gives two ENDOR transitions at 15
~ N D O R=
M 1 15Am I
*
15un
(lb)
In the presence of 1-atm NO and excess reductant, cytochrome oxidase was reported slowly to reduce NO to N20, to cycle through a non-EPR-detectable state and then to a fully oxidized form (30). Our method of preparation yielded the reduced nitrosylferrocytochrome a3 with its characteristic spectrum in Fig. 1. Double integration revealed that our EPR signal accounted for one-half the total heme a concentration in the sample, as was expected. No fully oxidized form was present.
ENDOR-EPR of NO-ligated Heme a3 in Cytochrome c Oxidase
7439
where Aeff is the effective first order hyperfine coupling, Q.rr is the effective quadrupole coupling, and U N is the Larmor T = 77'K frequency (the Zeeman term) for the nucleus itself in the gr = 2.092 Ye ' 9.306 GHz applied external magnetic field. Predicted pairs of ENDOR I A frequencies occur separated to first order by 2vN = % ~ g , f l n H , where g, is the nuclear g value, f l n is the nuclear Bohr magneton, and H is the applied magnetic field. Information-rich, well resolved, single crystal-like ENDOR spectra canoften be obtained from frozen solutions at g value extrema if a principalg, a principal hyperfine, and a principal quadrupole axis coincide (23). Well resolved ENDOR was obtained at g, (Table IA), at which Fig. 2 compares the 1-31MHz spectra of 14N0 and15N0ligated to heme a3 in oxidase. When the ligand was changed from 14N0to 15N0,one set of ENDOR peaks clearly changed; this set was from NO bound to heme u3.The 15N0 hyperfine coupling was larger than that of the 14N0by the expected ratio of nuclear g values, which I I I I 1 I is 1.40. The ENDOR of 14N0 bound to heme a3 occurred as 3300 3200 the expected quadrupole-split and Zeeman-split quartet. Quadrupole couplings can be determined from the frequencies b FERROCYTOCHROME ISNO 0, of this quartet and the resolution of this quartet, shown in I gx = 2.092 T ~77'K the inset to Fig. 2, was good enough to show that thesplitting 1 va 9.352GHz of its higher frequency Zeeman pair is slightly larger than the 0 1 x splitting of its lower frequency pair. Although the best reV solved ENDOR from bound NO was found at g,, ENDOR spectra of bound NO were obtainable at g values all across the EPR spectrum. In observing ENDOR across the EPR J spectrum from g = 2.09 to g = 1.97, we found that therewere U z ENDOR resonances (Table I, B and C) of NO which stayed 2 cn essentially at the same minimal frequencies found at g, ( i e . a near 21 MHz for ' W O and near 15 MHz for I4NO).QuadruL W SMALL COUPLING - 6.9G pole couplings wereresolved only at g, for the 14N0. The LARGE COUPLING- 29.86 ENDOR intensities from both 14N0 and 15N0 at g, and g, were smaller relative to adjacent histidine nitrogen ENDOR than atg,, and therewas a broad shoulder extending to higher frequencies from the 14N0 and 15N0 ENDOR resonances. At I I I 3400 3200 3300 g, = 2.006, a broad ENDOR resonance was found stretching from the minimal frequencies up to a higher frequency which "NO MYOGLOBIN was about 30 MHz for I4NO and 40 MHz for I5NO. These respective higher ENDOR frequencies corresponded to the T = 77'K large hyperfine coupling which EPR had directly resolved at ve 9.323GHz g,. EPR was the better technique for resolving the larger NO hyperfine couplings observed at g, because the ENDOR resolution of these large couplings was poor; the ENDOR resonances due to them were broad, featureless shoulders. (In Table IA, the experimental NO nitrogen A values obtained at g, are called A,; in Table IB, the minimal nitrogen coupling observed at g, by ENDOR is called Amin. In Table IC, the nitrogen coupling found at g, is called A,. In fact, A,, A,, and Aminare virtually identical and in future discussion are often referred to asAmin.In TableIB, the maximal nitrogen hyperfine coupling which was obtained by EPR atg, is called Amax.) From NO-ligated oxidase, another setof nitrogen ENDOR resonances was observed whose frequencies were unchanged 3200 when the bound NO was changed from 14N0 to 15N0. This 3300 3400 set was a quartet at g,, it had nearly the same magnitude of MAGNETIC T I E L D hyperfine couplings as the EPR-resolved NHlsdoes at g,, and lGAUSSl FIG. 1. This is a comparison of frozen solution EPR spectra this set was not observed in five-coordinate, imidazole-less of nitrosylferrocytochrome as ligated with (a)14N0 or (b) NO complexes. Thus, this setof ENDOR resonances, occurI6NO and of ( c ) "NO-ligated myoglobin. These spectra were ring in the 6-10-MHz region was assigned to NHisrand its obtained at 77 K with nonsaturating microwave powers of about 10 behavior was followedover the entireg value range from g, to milliwatts. The field modulation amplitudes of 1.8 G p.t.p. for the g,. At g,, the ENDOR signal from NHiswas a group of overlapoxidase and 3.0 G p.t.p. for the NO-Mb were chosen so as not to distort the EPR line shape. The NO and Nnis hyperfine couplings of ping resonances. From this group, it was possible to extract one larger NHis hyperfine coupling that agreed with the hythe NO-ligated oxidase are noted. Spectrum c reveals little direct hyperfine coupling and also shows evidence near g = 2.045 for a perfine coupling measure directly by EPR at g, and another different species. smaller hyperfine coupling similar to the coupling measured 0
"NO FERROCYTOCHROME 9,
J
-
I "
Y
I
-
h -
7440
ENDOR-EPR of NO-Ligated Heme a3 in Cytochrome c Oxidase TABLE I Nitrosylferrocytochromew Nucleus
MHz
MHz
A. Nitrogen couplings near g, 1 5 ~ 0
2.086
1 4 ~ 0
2.086
14
2.086
NHis
B. Nitrogen couplings at g, I~NO
2.000
14N0
2.000
I4
2.000
NHis
2.000
22.42 f 0.15 19.93 f 0.15 12.76 f 0.10 14.69 f 0.05 15.70 f 0.05 17.96 f 0.05 6.57 f 0.05 8.02 f 0.05 8.51 f 0.07 9.92 f 0.07
1l5AX]= 42.36 f 0.25' 2g&H = 2.75 I"AJ = 30.56 f 0.10' 1.zQ4'1 = 1.03 f 0.05' 2gnPnH= 1.97
20.0 f 0.3' 22.6 f 0.3 -
]*5Am,nl = 42.6 2g&H = 2.89
7.23 f 0.06 9.33 f 0.06 10.06 f 0.06 12.49 f 0.20 6.00 f 0.06 8.13 f 0.06 8.13 f 0.06 10.06 f 0.06
(A,I = 16.51 +- 0.10' lQul = 0.45 f 0.05 2g,&H = 1.97
* O.lbrd
-
[Amin/
lQzzl
= 19.55 -+ 0.1''
= 1.00 f 0.05
1 15Amel]= 29.8 f 0.8 Gd = 83.4 f 2 MHzd 114Ams,I= 21.4 f 0.4 G = +59.9 f 1 MHzd
lAmexl =
6.9 f 0.4 G
= 19.3 -+ 1 MHz'
2g&H = 2.06 lAminJ = 16.2 f 0.1" IQI = 0.65 2 0.05 2g&H = 2.06
C. Nitrogen couplings near g, 1 5 ~ 0
1.97 1.97 1.97
18.9 f 0.4 22.4 f 0.4 5.95 f 0.3 8.09 f 0.10 9.99 f 0.10
1l5AY]= 42.2 f 0.6' 2gnP,,H= 2.92 I
(I4A,1 = 16.1 k 0.2' = 0.67 + 0.1 2 n p , , ~= 2.08 ENDOR frequencies from the average of increasing and decreasing frequency sweeps and the quoted errors from the noise-induced uncertainty in the exact peak position. See text and Footnote 5 for detailed justification of signs of nitrogen hyperfine terms. For 15N0,signs are most likely negative and for I4NOand N H positive. ~ ~ The A, and A, values of hyperfine are the minimal (AL)values of hyperfine coupling mentioned in the textwhile the Amaxhyperfine coupling is the maximal (A,) value. Positive sign determined from second order effects discussed in Section A (Miniprint). The minimal hyperfine couplings (Amin)from 15N0 are resolved at g, by ENDOR, while EPR directly resolves the maximal hyperfine couplings (Amax)from ISNOand 14N0. e -, at least two of the expected four I4NO ENDOR lines are observed at approximately 14 and 16 MHz, but additional lines near the free proton frequency may be obscured. 'As described in text, the ENDOR from the "NH,~ g, at is a set of overlapping resonances, one set of which has the large hyperfine couplings A,, seen by EPR and theother set of which has a smaller coupling similar to that seen at g, and g,. The latter coupling probably is the coupling corresponding to those sites which have g = 2.00 but lie in the g,, gy plane, while the former is along the g, direction.
at g, and g., Finally, we noted a broad and very weak ENDOR signal in the 3-4-MHz region which may conceivably have been from weakly hyperfine coupled heme nitrogens. Fig. 3 offers a comparison of ENDOR from "NO-Mb and "NO-Mb taken near the maximum g value of NO-Mb (gJ. The ENDOR resolution was not as good as that of the NOligated oxidase, but from these data we can at least determine which ENDOR signalisfromheme-bound NO. The NO hyperfine coupling, which can best be determinedfrom 15N0, is 10-15% less than the corresponding coupling to I5NO in l5NO-1igatedcytochrome oxidase. The I5NO coupling at g, is only about one-half that predicted for the minimal NO hyperfine coupling, althoughnotactually resolved: by single Values for the nitrogen NO hyperfine coupling tensor have been reported from single crystal EPR of NO-Hb and NO-Mb at 77 K (1113, 22). The hyperfine structure from which the components of the tensor were derived was not experimentally resolved at a large number of orientations; in particular, thesmaller components of NO hyperfine coupling were lost in the EPRline width. The reported smaller values of the NO hyperfine tensor and thedirections for these smaller values
IQJ
crystal EPR of NO-Mb (22). NO ENDOR peaks from NOMb were about four times broader than those seen in NOligated oxidase, and this meant that NO Zeeman and quadrupole splittings were not resolved. The histidine nitrogen ENDOR from NO-Mb was resolved at frequencies similar to those found inNO-ligated oxidase. Nitrogen ENDOR results from NO-Mb and NO-FeTPP-imidazole are given in Table I11 (see Miniprint4). Thehyperfine coupling of 15N0 in 15NOwere inferred either from a fitting procedure (12) or from theoretical estimates (22), rather than from explicitly resolved hyperfine structure. We also note, in light of the temperature dependence of NOMb EPR spectra (19). that the EPR measurements were performed a t 77 K while our ENDOR measurements were performed a t 10 K. Supplementary material (including Fig. 7 and Table 111) is presented in miniprint at theend of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. Request Document No. 82 "1151, cite the authors, and include a check or money order for $3.20 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal thatis available from Waverly Press.
-
ENDOR-EPR of NO-ligated Heme a3 in Cytochrome c Oxidase
io Y
1
4
FERROCYTOCHROME ~ ~ g3 g = 2.086 9.32 GHz H = 3.19 k Gauss
I
I
U
25 0
8 n
(b"
I7.0
90
l5NO FERROLYTOCHROME a3
IO
I
z w
9:
Ve = H z
~~~
33 0
7441
ferrocytochrome a3 complex, occurred at all g values with a hyperfine coupling of about 2 MHz. At g,, the peaks labeled 3, that have a hyperfine coupling of about 1.4 MHz, and 5, with a coupling of about 0.9 MHz, were best resolved. At g,, peaks labeled 4 withsimilar hyperfine splitting to peaks labeled 3 a t g, were found. No proton ENDOR signals were seen to changewhen deuterated buffer was used. Model complexes were useful in assigning several of the proton ENDOR signals.Weassigned the peaks labeled 3 observed at g, (Fig. 4a) to heme mesoprotons. NO-Mb and NO-FeTPP-imidazole, which have mesoprotons, exhibited protonENDORpeaks (Fig. 5, a and b ) whose hyperfine coupling of 1.35 MHz was within 5% of the coupling of the peaks labeled 3 from NO-ligated oxidase. Such peaks were not evident from NO-FeTPP-imidazole, which lacks mesoprotons, as indicated in Fig. 5c. From NO-FeTPP-imidazole prepared with protonated imidazole, protonresonances were found (Fig. 5d) withvery similar hyperfinecouplings (3.5-4.0 MHz) and similar appearance to those of resonances labeled 1 from NO-ligated oxidase. When deuterated imidazole was used (Fig. 5e), such resonances disappeared. The use of I-methyl (i.e. N-methyl) imidazole caused no change in the resonancesobserved with protonated imidazole. Peaks with hyperfine coupling of 3.54.0 MHz from our model NO-FeTPP-imidazole are thus assigned to the protons of imidazole which are on the carbons adjacent to the Fe-liganding imidazole nitrogen. It is thus consistent toassign peaks labeled 1 from NO-ligated oxidase
~
25.0
170
-FREQUENCY [MHz 1
90
IO
1 4 ~ 0 MYOGLOBIN
FIG.2. ENDOR spectra near g, showing hyperfine couplings to (a) 14N0 and (b) 16N0 and to NHb in NO-ligated oxidase. Arrows indicate position of Zeeman pairs. More details of Zeeman splittings of "NO were revealed by sweeps over a narrower frequency range, as shown in the inset, where the higher frequency Zeeman splitting was larger and thelower frequency Zeeman splitting smaller than the expected first order Zeeman splitting of 2 ~ ,= 1.97 MHz (see Section A, Miniprint). Conditions for ENDOR are as noted in text. Approximately 10-min signal averaging at 2 s/sweep was used to obtain each spectrum.
Mb atg, was somewhat higher in frequency than thecoupling measured at g,. Proton ENDOR-First orderENDOR frequencies for a proton, whose interaction with the applied magnetic field (Zeeman interaction) is larger than itshyperfine interaction, are given by:
14
x3
2
31.0
z
W
N HISTIDINE
I
I
I
I
25.0
19.0
13.0
7.0
z c3 cn
g n
14
NO
1.0
I ~ N OMYOGLOBIN
b
I
I
I
I
g = 2.086 H z 3.20kG T.6 K
AH/^ (2) where v N M R is the free proton NMR frequency and AH is the proton hyperfine coupling. T o first order, proton ENDOR spectra center at thefree proton NMR frequency. Proton ENDORresolved from NO-ligated oxidaseis shown in Fig. 4, a-d, and the ENDOR frequencies are detailed in Table 11. Proton ENDOR occurred at all g values of the NOligatedoxidase E P R signal;excellent resolution of proton i 5 ~ ~ "N HISTIDINE ENDOR was observed at g, and good resolution a t g,. HyperI I I I fine couplings to several different types of proton were evi31 0 25 0 19 0 130 70 IO dent. Peaks labeled l were found with a maximum hyperfine " E N D O R FREQUENCY (MHz) coupling of 3.5-4.0 MHz at intermediate g values. The hyperFIG.3. ENDOR spectra at gx from (a)14N0 and (b) "NOfine couplingof peaks labeled 1 diminished at g, so that they became lost under ENDOR signals from protons with hyper-ligated Mb showing overallpoorer resolution than that of NONevertheless, it is obvious that the resonances above 15 fine couplings of less than 1.5 MHz. Although still apparent oxidase. MHz are from NO and that thelower frequency resonances assigned in Fig. 4d at g = 1.979, the peakslabeled 1 disappeared at g = to histidine occur at about the same frequency in both NO-Mb and 1.970. Peaks labeled 2, which appear unique to the nitrosyl- NO-ligated oxidase. VPrm.
=
VNMR
f
-
7442
ENDOR-EPR of NO-ligated Hemea3 in Cytochrome c Oxidase
FIG.4. Proton ENDOR spectra of nitrosylferrocytochromeastaken at g values of (a)2.090, (b) 2.042, (e) 2.006, and (d)1.979. Conditions for ENDOR are as given under "Materials
..
T
and Methods," and assignments of peaks (0L n are discussed inthe text. This particular z sample was not deuterated, butENDOR W of a deuterated sample was identical.
id
g = 2 006
C
H=327kG
I 12I
I
I
I
I
I
1
l
I
I
g
I979
H = 332kG
I
16
14
I
12
ENDOR FREQUENCY [MHz]
I
I
-
14
I
I
1
16
TABLE I1 Proton hyperfine couplings in nitrosylferrocytochromea3 g value
Magnitude of hyperfine couplings" Peaks 1
Peaks 2
Peaks 3
Peaks 4
Peaks 5
MHz 1.970 1.979 1.990 2.006
2.042 2.067 2.090
3.51 rt 0.18 3.42 f 0.20 3.52 t 0.02 4.25 & 0.30 (weak shoulder)
1.43 k 0.09 1.43 +- 0.03
2.10 f 0.09 2.00 +- 0.09 1.99 f 0.09 1.99 f 0.09
2.20 f 0.14 2.18 f 0.09 2.23 f 0.20
1.30 f 0.14 (weak shoulder) 1.14 f 0.09 1.43 k 0.03
0.82 k 0.04
The couplings were determined from the frequency difference between the two peaks corresponding to the same proton. The quoted errorswere estimated from the noise-induced uncertaintiesin the exact peak positions. to these imidazole protons. No proton ENDOR signals like those labeled 2 from NO-ligated oxidase were observed from any of the models, including the ones prepared from heme a.
histidine was more isotropic, but this coupling was also observed by EPR to have its maximal value at g,, implying that the 2p orbital of NHiacontaining the unpaired electron points close to theg, direction. The hyperfine coupling to a nitrogen DISCUSSION will have an isotropic part, Apermi, that is proportional to Nitrogen HyperfineInteraction-ByENDOR, we have unpaired nitrogen s electron density and ananisotropic part, found an entirerange of magnetic fields stretching from g, to A,, that is proportional to unpaired electron density in the g, for which there exists the sameminimal NO nitrogen valence 2p orbital. It will be found that A, = Apermi+ 2 A p and hyperfine coupling. There is thus a planeof minimal hyperfine Al = AFermi- A,. Both AFermiand A, are proportional to the coupling which contains g, and is close, at least, to containing nuclearg value ('*g,, = +0.403, 15gn= -0.565). When hyperfine g,. The minimal hyperfine coupling would be Ami,,= AL. The contributions arise from unpaired valence 2s and Zp electron maximal hyperfine coupling, Amax= AN,appears at g,. Un- density, Apermiand A, will be positive for 14Nand negative for '. paired electron on Fe and NO oxygen is proposed to cause " AFermiand A, can be computed from the data in Table I, nonaxial dipolar couplings to NO nitrogen (31), but the NO hyperfine interaction experimentally appears axial. The ma- and 2s and 2p electron densities can be estimated. In making jority of a nitrogen's hyperfine interaction should arise from such estimates, we have taken the Fermi interaction of a unpaired electron on the nitrogen itself. If the unpaired elec- single 2s electron as 1450 MHz and -2200 MHz for '*N and tron at nitrogen has 2s and 2p character, then parallel to the 15N, respectively, and the dipolar interaction, A,, as 47 and direction of the 2p orbital the hyperfine coupling would be AB, -66 MHz for 14Nand 15N,respectively (32).We caution that and perpendicular to it the hyperfine coupling would be Al. first order ENDOR expressions (Equations l a and Ib) yield Our findings imply that thenitrogen 2p orbital containingthe only the magnitude of hyperfine couplings, not signs; care unpairedelectron on NO nitrogen points closer to the g, must be exercised in choosing the signs of A! and A A . Fortudirection. The nitrogen hyperfine coupling to heme-bound nately, the wrong choice of sign often yields chemically un-
ENDOR-EPR of NO-ligated Heme
a3 in Cytochrome c Oxidase
7443 g:2.071 H ~3.16kG
I c
I
.-u \ x -u
u
-
J
Q
z '3
\
I
I
I
12
I ,
I
14
d '
v,
I
J
I
I
12
16 I
I
I
I
I
14 e.
1
I
I
I
I
16
12
I
9: 2.007
I
16 I
I
I
I
12 1 I g = 2.007 H.3.28kG
14
I
i
16
LL 0 0
z W
I
I
12
I
I
14
I
ENDOR FREQUENCY
I
14
[MHz]
-
I
I
16
I
FIG. 5. Proton ENDOR of model complexes. a, NO-Mb, b, NO-FeOEP-imidazole; c, NO-FeTPP-NOimidazole. All were taken at their respective maximal g value. Note, on comparing a, b, and c, the disappearance of the peaks labeled 3 in the FeTPP complex which lacks mesoprotons. d is the spectrum from NO-FeTPPimidazole (protonated;ZM(PR0T)) and e is from NO-FeTPP-imidazole (deuterated;ZM(DEUT)). The purpose of d and e is to show that the outlying resonances, with approximately 3.5 MHz coupling, can be from imidazole protons. reasonable spin densities: and may not predict the observed second order correctionto theZeeman splittings. (see Section A, Miniprint) For 14N,the choice of 14Aw = +59.9 and l4AL= +30.6 MHz yielded 14A~ermi = +40.2, 14A, = +9.9. For 15N0, 1 5I ~ -83.4 and 15AL= -42.4 MHz yielded, l5Averrni = -56.0, I5A, -13.7 MHz. From these values of Avermi and A,, we computed that the unpaired spin on NO nitrogen of nitrosylferrocytochrome u3 has 2.8% 2s characterand 21% 2p character.In free NO x* orbital has 65% molecule, the unpaired electron in the nitrogen 2p character and 2.5% 2s (32). For NO-ligated oxidase, the large percentage of unpaired 2p electron density implies that NO still has a considerable remnant of its ?r* parentage, but unpaired electron density has been covalently redistributed away from the NO. For the 14N of histidine, A B= +19.1 and AL = +16.3 MHz, andthisleadsto Arerrni= 17.4 and A, = 0.9 MHz. The percentages of 2s and 2p electron are thus 1.2% 2s and 1.9% 2p. Thesepercentages suggest thatthehistidinenitrogen binds to heme iron through an sp2 orbital directed along or near theg, direction. In NO-Mb, a larger asymmetry of NO hyperfine coupling was found (Table IIIa, Miniprint); the NO hyperfine coupling was about 30%larger at g, than atg,. A more polar or distorted The choice of both AI and Al positive for 15N0 or both AI and Al negative for"NO leads to negative spin densityin the nitrogen valence Zp orbital of an NO systemwhichshouldhavepositive electron spin density inits highest partly occupied molecular orbital. If we were to choose opposite signs like l4A1 = +59.9 and "A, = -30.6 MHz for oxidase-ligated "NO or "AI = -83.4 and 15AL= +42.4 MHz
NO-binding environment in myoglobin (as created by distal histidine) may give the NO nitrogen hyperfine tensor more nonaxial character and/orlead to greater noncolinearityof g and hyperfine tensors. With NO and NHishyperfine couplings known from EPR and ENDOR,we simulated the EPR spectrumof nitrosylferrocytochrome u3. As long as colinear g and NO hyperfine tensors were used, simulations of the EPR signal near g, were not entirely satisfactory? The poorer resolution of the NO ENDOR signal near g, had led us tobelieve that g, was near, but not in, the minimal hyperfine coupling plane. Simulations were thus performed where the NO hyperfine tensor was rotated in the g,-g, plane using g, as the rotation axis. A rotation of A, and Amin by 15-20" away from g, and g,, respectively, yielded a better simulation to the EPR features. The histidine tensor was already so isotropic that rotating it had little effect on the simulation.Fig. 6 offers a comparison of experimental EPR spectra to thesesimulations. Rotation of the NO hyperfine tensor with respect to the electronic g tensor has been observed from single crystal EPR studies (1113, 22) and taken as evidence for a bent Fe-NO bond (see Section D, Miniprint). An additional rotation of Amaxtoward g, would make g, no longer a principal direction for minimal hyperfine coupling and thus would be inconsistent with the excellent resolution of minimal NO hyperfine couplings near g,. In simulations, a rotation of Amaxtoward g, by as little as 15"distorted the spectral features at g, so that the detail seen
'Another spectral simulation for the NO-ligated oxidase was reportedinRef. 33. Colinearhyperfineand g axes wereused. The histidine and "NO hyperfine couplings and g values were similar to for oxidase-ligated"NO, the resultant spin densitieson the nitrogen ours except that the value of A- was 20% larger. Small additional would be -0.17% for 2s and +64% for 2p. Free NO molecule has an couplings to pyrrole nitrogens, which are not resolved by EPR or estimated 65% 2p spin density. Thus, a 64% Zp density would seem ENDOR,wereincluded.Although the resultant simulationbears unreasonably highfor an NO systemwhich now mustcovalently general resemblanceto the "NO-ligated oxidase spectrum,it does not share its unpaired nitrogen2p electron with heme iron and histidine. match the details of the EPR spectrum in the g, or g, region.
E N D O R - E P R of NO-ligated Heme a3 in Cytochrome c Oxidase !
I
similarity to the molecule, NO the g, direction would definitely not be parallel to the N-0 bond direction. In past ENDOR work, excellent resolution of hyperfine and quadrupole couplings has arisen fromfrozen solutions when there was colinearity of a g axis, a hyperfine axis, and a quadrupole axis. Here theresolution of 14N0 quadrupolarcouplings was better a t g, than anywhere else, implying g, is colinear with a 14N0 quadrupolar axis. If themajority of nitrogen 2p electron density on the NO nitrogen were confined to the FeNO plane, then a simple argument based on 2p valence population and Townes Dailey theory (34, 35) predicts a principal direction of positive quadrupole coupling perpendicular to the FeNO plane (see Section B, Miniprint). It is thus plausible to take g, perpendicular to the FeNO plane (see Section D, Mini3200 3300 3400 print). Proton Hyperfine Interaction-The ENDOR of protons 3, which are heme mesoprotons, and the protons 1, which are more tentatively assigned as CH protons of imidazole, appeared with best resolutiononly at certain g values; their hyperfine interaction was anisotropic and had different values along different gaxes. These protons had relatively small hyperfine couplings relativeto their nuclear Zeeman coupling, so that little of the wave function of the unpaired electron must extend directly to them. In NO-heme complexes previously studied (11-13, 22), the unpaired electron is largely confined to the Fe andNO, which are relatively distant from protons. The localization of electron spin at a distance from I I I a large fraction of proton the protons in question means that 3200 3300 3400 hyperfine coupling should be from anisotropic dipolar couMAGNETIC FIELD pling which may be calculable from geometric considerations. [GAUSS] The first order expression for dipolar coupling between a i, and an electron localized in position FlG. 6. Simulations of the EPR spectra from (a)lsNO-li- proton nuclear moment, j is gatedferrocytochrome as and (6) “NO-ligatedferrocyto0
EPR Data --Simulation
I
-
chrome as. The purpose of the insets is to show in detail the effect in the g, = 1.98 region from varying the angle between the maximal (Amax)and minimal (Amin)hyperfine tensor directions and theg, and g, directions, respectively. For a, the values of the relevant EPR parameters were g,, g,, g, = 2.092, 2.006, and 1.980; 15NO:A,, A,, A, = 14.5, 30.5, and 15.2 G; 14N& A,, A,, A, = 5.6, 6.9, and 5.8 G. p.t.p. Gaussian line widths used for a were 8.0 G at g, and g, and 4.0 G at g,. For b, the relevant values of the EPR parameters were g,, g,, g, = 2.092,2.006, and 1.9795; 14NO:A,, A,, A, = 10.4,22.0 and 10.9 G; “ N H ~A,, ~ : A,, A, = 5.6,6.9, and 5.8 G p.t.p. Gaussian line widths used for b were 9.0 G a t g, and g, and 4.5 G at g,.
= g,&gHP, (3 Cos2 0
- l)/(R?,) = A d
(3
COS%
- 1)
(3)
where be and Pn are the electron and nuclear Bohr magnetons, g, and gH are the electron and proton g values, R,j is the electron-to-proton vector, a?d 8 is the anglebetween the magnetic field and vector R+ A d isthe dipolarcoupling constant. Equation 3 can be used to demonstrate what happens if the magnetic field is near the principal 8 = 0” hyperfine axis. At 8 = O ” , proton ENDOR resonanceswill be split by a maximal for the experimentally resolved features was lost. Although value of u d . In frozen solution, even at an externalg value nospectralsimulations were performed for NO-Mb,the like g,, there will be some random misorientation in the angle broadening of its hyperfine features, compared to thoseof the between the hyperfineaxis, g axis, and magnetic field. A oxidase, may indicate a larger noncolinearity of g tensor and distribution of slightly different molecular site conformations NO hyperfine tensor aswell as larger intrinsic line widths. could contribute to such misorientation.As long as the prinNitrogen Quadrupole Interaction-This interaction reflects d Adipole = 0, the change 8 = O”, where the overall symmetry of the electronic distribution at a nitro- cipal g axis centers at d8 gen. The resolutionof 14N0 quadrupole and Zeeman couplings in ENDOR splittings will be second order in misorientation at g, was so good that small second order changes in the will be minimal. It is Zeeman splittings could be observed, and from these we angle and the ENDOR peak widths useful to note that the ENDOR splittings for the mesoprotons determined that the quadrupole coupling constant along the of NO-ligated oxidase were maximal and their line widths g, direction was positive (see Section A, Miniprint). In free were minimal a t g,, implying that g, points close to a principal NO molecule, thequadrupoleinteraction along the N-0 direction for mesoproton coupling. For NO-Mb and model bond is negative, while it is positive perpendicular to the NO-heme-imidazole systems, the mesoproton couplings at g, N-0 bond (32).7 If the oxidase-ligated NO retainselectronic were about 0.05 MHz less and the line widths were 1.5 to 2 in the oxidase. Perhaps in these ’The nuclear quadrupole coupling strength is frequently given by times broader than thosemore misorientation between the g, the nuclear quadrupole coupling constant e2qz2Q,where eq, is the z latter systems there is component of the electric field gradient at the nucleus and eQ is the direction and the mesoprotonhyperfine axis. nuclear quadrupole moment (35). When the quadrupolar spin HamIf the unpaired electron is spread over several atoms, we iltonian is written as I . Q.Z, then Qzz = H (e’q,Q). For free NO, e2q,Q would have to sum contributions of the form of Equation 3 is in the -1.6- to -2.2-MHz range (32). Thus, for free NO, the coupling Qzz, where z is the N-0 bond direction, is in the -0.8- to over the different locationsof the electron. Even SO, there is coupling axis the -1.1-MHz range, and Qu and Q, are in the +0.4- to +0.55-MHz adipolar tensor and near its maximum angular behavior is similar to that described by Equation 3 range.
ENDOR-EPR of NO-ligated Heme a3 in Cytochrome c Oxidase near 0 = 0". To compute mesoproton hyperfine couplings, we have employed various spin distributions: and by varying the position of the magnetic field with respect to heme coordinates have computed both the direction and magnitude of maximal mesoproton dipolar coupling. With our various schemes of spin distribution: we have computed that thelargest principal dipolar coupling, 2 A d , for mesoprotons has a value in the 1.551.88-MHz range and points within 8" of the Fe-mesoproton vector. If the entire magnitude of the 1.42-MHz mesoproton coupling measured at g, were dipolar, then Equation 3 and the computed estimates of 2 A d predict that theangle between the g, axis and the axis for largest mesoproton hyperfine coupling would be less than 24". Both the maximal nature of the mesoproton hyperfine coupling and theminimal ENDOR line width at g, imply that thetwo axes are even more nearly coincident. In NO-Mb, the g, direction and Fe to mesoproton direction can be shown to lie 6" away from one another (22, 36). The imidazole protons labeled 1 have a maximal hyperfine coupling of 3.5-4.0 MHz and the ENDOR resonances from them are broad. We take this to mean that our principal g axes do not coincide with any of their principal hyperfine axes. The hyperfine coupling to these protonsdiminishes near g, and g,, and the largest coupling occurs closer to g,. With our various schemes for spin distribution: we have computed that the largest principal dipolar coupling 2 A d has a value in the 2.5-4.0-MHz range. Its computed direction lies about 35" from the Fe-NHi, axis. The unique protons 2 have approximately the same 2.0MHz splitting at all g values. The simplest explanation is a direct 2.0-MHz Fermiinteraction of electron spinwitha proton not having a large, anisotropic dipolar interaction. We do not believe that the NH proton on the imidazole ring of histidine would be the proton 2 because a proton with the characteristics of proton 2 is not seen from NO-Mb or the NO-heme-imidazole models which have the NH proton. Further, the proton 2 does not exchange with deuterium. The proton 2 could be part of an amino acid side chain that might weakly and covalently interact with the NO (or 0,) ligand. An example of such a proton would be a CH, proton on the p carbon adjacent to a cysteine sulfur. A'cysteine sulfur has been proposed as a bridging ligand between Fe and Cu of the a3 center (37). The /3 protons of cysteine are certainly able to give an isotropic contact interaction if there is unpaired spin density on their adjacent sulfur (38, 39). g Tensor-The electronic interaction with NO T * , K, and u ligand orbitals will greatly contribute to removing the 4-fold electronic symmetry of heme iron d orbitals; this symmetry loss is shown simply by the rhombic nature of the g tensor. Because excited electronic states aremixed into thehighest occupied molecular orbital that contains the unpaired electron spin, some components of the g tensor, notably g, and g,, may differ fromthe free electron g value of 2.0023. The differences are Proportional to the strengthof spin-orbit coupling, which is larger for metal d electrons than for N or 0, and inversely In using Equation 3 to compute dipolar couplings to protons, we have tried several schemes for the distribution of unpaired electron. 1) The simplest was to put 100%of the unpaired electron spin on Fe. 2) The next,which agrees with our ENDOR findings, was to put 3.5% of an unpaired electron on Nws, 73% on the Fe, and 23.5% on NO nitrogen. 3) The last, using numbers derivedfrom NO-Mb single, on Fe, and 50% on crystal EPR (12,13), was to put 5% on N H ~45% NO nitrogen. For both mesoprotons and the CH protons of imidazole, the magnitude of the largest dipolar coupling diminished in going from scheme 1 to scheme 3. Schemes 2 and 3 tilted the principal axis for largest hyperfine coupling slightly out of the heme plane away from the Fe to mesoproton direction.
7445
proportional to energy differences between the highest OCCUpied molecular orbital and the excited states (40). Ab initio molecular orbital computationsshow many molecular orbitals with some dcharacter to existnear the highest occupied molecular orbital (41, 42), and numerous schemes have been proposed to account for the g values (12,21,40,43). We have tried not to become firmly attached to anyparticular molecular orbital scheme or set of energy levels, but for reasons given in SectionC of the Miniprint, ascheme may bepreferred 2 that puts considerable electron density into a d orbital and z into a NO x* orbital that overlaps with it. Conclusion and Summary-The most striking qualitative feature thatdistinguished the NO-ligated oxidase from other NO-heme systems studied here was the better spectral resolution. ENDOR and EPR showed a substantial difference at NO-binding sites between heme proteins. The better resolution may show that the NO-binding site at the a3 center has a greater restriction on available conformations and/or there is a greater colinearity of hyperfine and g tensor directions for nitrosylferrocytochrome a3. From the extra resolution, we obtain insight on the direction of g axes with respect to heme coordinates, on the bending of the Fe-NO bond, on how the FeNO plane may lie, and on the electronic distribution on the NO. Since the unpaired electron resides in the highest occupied electronic orbital, the spectroscopic information thus obtained is from a valence electron likely to be involved in electron transfer reactions. Even though at present we may lack a complete underlying physical model, the ENDOR-EPR information should be useful in future probes of small electronic changes at the a3 center, such as may be brought on when neighboring cytochrome a changes valence state or cytochrome c binds to theoxidase. It is tempting to propose a detailed model ofthe NO-binding site as inferred from our magnetic resonance data. For this model, we refer to the Miniprint, Section D. We consider the model there tentativebecause the inferences used in obtaining it, although physically plausible, do rely on making approximations like the dipolar approximation andthe TownesDailey approximation, or on assuming symmetry properties of the electronic wave functions and g tensor. Obviously, if our study had been on a single crystal of NO-ligated oxidase, we would have more confidence in our model. To ensure that the inferences drawn in obtaining our model are sound, it wouldbewell to have at least one complete, orientationdependent, single-crystal EPR-ENDOR study (as has been done with high spin ferric heme in aquometmyoglobin (44)) on an NO-heme system whose physical structure is known from x-ray. Acknowledgments-We are grateful to Dr. Satoshi Yoshida for his help in our initial work with NO heme systems and to Professor D. C. Doetschman for a critical reading of this manuscript and for a preprint of his single crystal NO-Hb study (45). Addendum-A recent very thorough single-crystal EPR study of human nitrosylhemoglobin yielded somewhat different g and NO A tensors from those previously reported for NO-Mb and NO-Hb (1113, 22). Notably, the lesser values of the NO hyperfine tensor were in the range of 25-45 MHz and so were more consistent with values measuredhere by ENDOR. A special EPR cavity goniometer was used to give two independent rotational degrees of freedom and to make more crystal orientations accessible (45). REFERENCES 1. Wikstrom, M., Krab, K., and Saraste, M. (1981) Cytochrome Oxidase, a Synthesis, Chapters 4-6, Academic Press, London 2. Brudvig, G. W., Stevens, T. H., Morse, R. H., and Chan, S. I. (1981) Biochemistry 20, 3912-3921
7446
ENDOR-EPR of NO-ligated Hemea3 in Cytochrome c Oxidase
3. Piciulo, P. L., Rupprecht, G., and Scheidt, W. R. (1974) J. Am. Chem. SOC.96, 5293-5295 4. Scheidt, W. R., and Piciulo, P. L. (1976) J. Am. Chem. Soc. 98, 1913-1919 5. Scheidt, W. R., Brinegar, A. C., Ferro, E. B., and Kirner, J. F. (1977)J. Am. Chem. SOC.99, 7315-7322 6. Deatherage, J. F.,and Moffat, K. (1979) J. Mol. Biol. 134, 401417 7. Jameson, G. B., Rodley, G. A., Robinson, W. T., Gagne, R. R., Reed, C. A., and Collman, J. P. (1978) Inorg. Chem. 17, 850857 8. Jameson, G.B., Molinaro, F. S., Ibers, J. A., Collman, J. P., Brauman, J. I., Rose, E., and Suslick, K. S. (1978)J. Am. Chem. SOC.100,6769-6770 9. Phillips, S. E. V. (1978) Nature ( L o n d . ) 273, 247-248 10. Kon, H. (1968) J. Biol. Chem. 243,4350-4357 11. Chien, J. C. W. (1969) J. Chem. Phys. 51, 4220-4227 12. Dickinson, L. C.,and Chien, J. C. W. (1971) J. Am. Chem. SOC. 93,5036-5040 13. Dickinson, L. C., and Chein, J. C. W. (1974) Biochem. Biophys. Res. Commun. 59,1292-1297 14. Yonetani, T., Yamamoto, H., Erman, J. E., Leigh, J. S., Jr., and Reed, G. H.(1972) J. Biol. Chem. 247, 2447-2455 15. Kon, H., and Kataoka, N. (1969) Biochemistry 8,4757-4762 16. Kon, H.(1975) Biochim. Biophys. Acta 379, 103-113 17. Wayland, B. B., and Olson, L. W. (1974) J. Am. Chem. SOC.96, 6037-6041 18. Blokzijl-Homan, M. F. J., and Van Gelder, B. F. (1971) Biochim. Biophys. Acta 234.493-498 19. Stevens, T. H., and Chan, S. I. (1981) J. Biol. Chem. 256,10691071 20. Barlow, C., and Erecinska, M. (1979) FEBS Lett. 98,9-12 21. Morse, R. H., and Chan, S. I. (1980) J. Biol. Chem. 255, 78767882 22. Hori, H., Ikeda-Saito, M., and Yonetani, T. (1981) J. Biol. Chem. 256,7849-7855 23. Scholes, C. P.(1979) in Multiple Electron Resonance Spectroscopy (Dorio, M., and Freed, J. H., eds) pp. 297-329, Plenum Press, New York 24. Mulks, C.F., Scholes, C. P., Dickinson, L. C., and Lapidot, A. (1979) J. Am. Chem. SOC.101, 1645-1654
25. Lozos, G. P., Hoffman, B. M., and Franz, C. G., PROGRAM SIM, Quantum Chemistty Program Exchange, Indiana University, Bloomington, IN, Program No. 265 26. Gordy, W. (1980) Theory and Applications of Electron Spin Resonance, pp. 66 and 188, Wiley, New York 27. Kuboyama, M., Yong, F. C.,andKing, T. E. (1972)J. Bwl. Chem. 247,6375-6383 28. Yu, C., Yu, L., and King. T. E. (1975) J. Biol. Chem. 250, 13831392 29. Scholler, D. M., Wang, M.-Y. R., and Hoffmann, B. M. (1979) J. Biol. Chem. 254,4072-4078 30. Brudvig. G. W., Stevens. T. H.. and Chan. S. I. (1980) Biochemistry19.5275-5285 31. Doetschman. D. C.. Schwartz. S. A.. and Utterback, S. G. (1980) Chem. Phys. 49,’l-8 32. Dousmanis, C. (1955) Phys. Rev. 97, 967-970 33. Twilfer, H.,Gersonde, K., and Christahl, M. (1981)J. Magn. Res. 44,470-478 34. Townes, C. H., and Dailey, B. P. (1949) J. Chem. Phys. 17, 782796 35. Lucken, E. A. C. (1969) Nuclear Quadrupole Coupling Constants, Chapters 7 and 11, Academic Press, New York 36. Takano, T. (1977) J. Mol. Biol. 110,569-584 37. Powers, L., Chance, B., Ching, Y., and Angiolillo, P. (1981) Biophys. J. 34,465-498 38. Saxebel, G., and Herskedal, 0.(1975) Radiat. Res. 6 2 , 395-406 39. Stevens, T. H.,Martin, C. T., Wang, H., Brudvig, G. W., Scholes, C. P., and Chan, S. I. (1982) J. Biol. Chem. 257, 12106-12113 40. Doetschman, D. C. (1980) Chem. Phys. 48,307-314 41. Mun, S. K., Chang, J. C., and Das, T. P. (1979) Proc. Natl. Acad. Sci. U.S . A. 76,4842-4846 42. Mun, S. K. (1979) Ph.D. thesis, State University of New York at Albany 43. Trittelvitz, E., Gersonde, K., and Winterhalter, K. H.(1975) Eur. J. Biochem. 51, 33-42 44. Scholes, C. P.,Lapidot, A., Mascarenhas, R., Inubushi, T., Isaacson, R. A., and Feher, G. (1982) J. Am. Chem. Soc. 104,27242735 45. Utterback, S. G., Doetschman, D. C., Szumowski, J., and Rizos, A. K. (1983) J . Chem. Phys., in press
7447
ENDOR-EPR of NO-ligated Heme a3 i n Cytochrome c Oxidase S U P P L E m A L MATERIAL TU EPR STUDY ON TRB STRUCTURB OF TEE NO-LIUTED 3 3 IN cyMcBR0llE c OXIDASE By Charles P. Scholes, Russell LoBr'Ytto, Yau-Huei Wei, Rita Mascarenhas. -R
AND
and T.00 E. Xing A. E ~ mequencies R Second-Order Effecte. If we take the electronic 2e-n interaction as the largeat magnetic interaction, it is quite generallyto use first-order perturbation theoryto obtain expressions like equation 1. Second-order effects may give additional information On the signa of hyperfine and quadrupole couplings. The second-order COZIBCtion to the hyperfine term occurs because the external magnetic field IS
-
not Infinitely large compared to the hyperfine field. In the came here we take themagnetic field along thex direction and perpendicular to the Y and z directions. 1n the terminology used in the text, Ax would be AA = and A would be A 1 %in. The exprs8siOnS AI would be for 15No E N w R frequencies become
sin.
1\11
-
-
15v ENDOR
-
112 jl5AX/f (15vn
-
l5AZ 15\/(lu,)l
Eq.
4
where y e 1 e the m i c r a a v e frequency. If l5AZ and 15A have the same sign, then the tern in l5AZ 15A decrease. the EeSplitEing from 2l5V,, to 2l5Vn l5Az 15A 12(ue)i zo the Ze-n splitting for 15N0 should from 2.75 to 2.56 KEz at H = 3.19 k Gaues (Table la). The meaaured splitting wa* 2.49 MHz, I.=., definitely less than 215Vn. This result, in the context of the present experiments, ill consistent with 15A1and 15A1 both negative. If 1 5 A 1 and l5A11 were both positive, then it follorn that the valence 2p electrons would have negative spin density rather than theexpected positive apin density for unpaired spin in a valence orbital.
-
For 1%
14"- ENWR
-
two
-
ofmeman pairs are expected3 these are 14\1+mm8
sets
( 1 1 2 l4AX [14vN
-
+ (azz
-
I1'Az
- Q,,,,12/14(14Ax f l4vN)I - 312 Qu
f
Eq.
14Ay12/8vell
5b
The second-order termm in i1'AZ f "\)21i8v,l are important because they may help UB to deternine the sign of the quadrupole interaction along the x direction. The sign Of the hyperfine coupling itself depend. On the sign of the nuoleair g-value SO that when l5AX, 15$, and l5AZ are all negative, than I4Ax, 'I$, and l4AZ will all he positive. When l4AX is positive, then thehigher frequency Eapair will have a larger splitting than the lover frequency Z e B M Pair if Q,,>O. In particular, the upper Eeeman pair would have a predicted splitting of 2?'vnI + i1'AZ + 14Ay)2/ I4uJ 2.19 KEz, while the l m a r fzequency pair would have a predicted Bplitting Of ZIl4un) (l4AZ 14Ay)21(4ve) 1.95 KEE. The exPeriuntal Zeeman splitting. were found to be 2.26 f 0.07 and 1.93 f 0.11 M H z . respectively, and theae results are consistent with Qu>O. 8. guadropole Interaction Emlanation for its PositiveS i g n P e r ~ o n dicular to the Few Plane. An approximate method lending itself t o c h a ically intuitive bonding schemes is the Tames-Daileyapproach 134, 35). We use the arthod simply to provide a rationale for the sign of the quadrupole interaction along the direction perpendicular to thePeNO plane, a direction which may be parallel to 4,. The Tomen-Dailey approximation is unlikely to provrde a c m p l e t e quantitative explanation of the overall ni-
-
-
-
-
-
tragen quadrupole mteraction for so cmplicated a system as NO-ligated heme. We leave this task t o the theoretician. The T m e s - D a i l e y method ascribe. the electric field gradient at the nltrogen to inequality in the total populations of valence m , ,. p and& electrons. FOZ example, the quadrupole interaction, Q,,, is related to the populations NX. Ny. Nz in e,, pY, p Z Orbitals by the telation Q,, OL (NX flNy + NZ)l and cycllcally for y and z. The pr0pOrtionality constant is found from the estimate of the quadrupolecoupling for a single electron ~n a 2p orbital. When the quadrupole Hamiltonian is written as 2.Q.L. the
Another scheme has been proposed, where the unpaired electron is puLinto a H O W that has metal dyrcharactel and 1 NO character perpendicular tO the PaNO plane ( 4 3 ) . As pointed out by Morse et al. (21). this scheme Will account for a g-value less than the free electron 2.0023, but requires a momt unusual and unlikely ordering Of metal Orbitals to account for the other 9-valuas, notably the OM close to 2.0023. It .ems unlikely that this scheme would aocount for the observed hyperfine coupling to histidine because the sigma-bonding dz2 Orbital ia not part Of the proposed HOnD. The eristence Of other schemes does point Out the need for additional spectroscopic data toestablish the ordering of metal d and NO electronic orbitals in NOheme SYStemB. D. A Model for the NOBinding Site at NitxO.ylferrOCYtOChrome a3. In the Dimsussion we have inferred variortll aspects Of the physical structure at the NO binding eite from OYI magnetic resonance data. In this Section W e bring theas and other inferences on the structure together in a tentative rodel, e h n m in Pigvre 7. At gx the well-resolved character and minimal magnitude of NO hyperfine U that themagnetic field was along a principal magnetic m u p l i n g meant to . hyperfine axis for the Ix) nitrogen. The minimal NO nitrogen hyperfine coupl i n g at 9, would indicate that gx points perpendicular to an eleotronlc orbital on NO nitrogen whish contains the unpaired 2p electron spin. The excellent resolution of meso proton hyperfine couplings occurred with maximal hyperfine coupling and minimal linewidth at gx, and such couplings were in reasonable agreaent withc w u t e d dipolar couplings if g , points near the Pe-tole-proton direction. we thus inferred that the g , direction points near the Pe-to-meso-proton direction in the heme plane. At g, the excellent resolution of nitrogen quadruple muplings f r m was evidence that gx i1 a .0 pointing along a principal nitrogen qusdrupole axis. The positive Sign for tk. nitrogen quadrupole coupling is consistent with the g, direction h i n g pe,rpandicular to a plof large electron density centered at the NO nitrogen8 tk. likelihoad is thusthat gx is perpendicular to the P a l l plans. To naka theperpendicular to the PeNO plane point near the heme r . 0 protons, the PeNO plane itself would have to lie so as to bisect an N-Fa-N 90. angle created by pyrrole nitrogens, and the PeNO plane would have to intersect the heme plane at approximately a right angle. Thus the FeNO plane would have minimal repulsive overlap with heme nitrogens. Such a di-
' k o
IeCtiOn for the orlentation of the F a 0 plane is saen in some models ( 3 , I ) , but the exact orientation of theFeNo plane, as well as the PeNO bond angle. may be affected by near-lying amino acid side chains on which could reside the proton(s) that BcEOunt for the unique proton E m R s ~ g n a l"2'.
Figure 7. m-1iaat.d a
our
model of the ferrocvtochr-
PERTURBING
ceitter as obtained frcm
-&R-EPR evidence. Salient point. are: The maximalgva1w i g ) point. near the heme res: protone and perpendicular to theF m plane SO that theP a w plbi.et. tha 90. angle between heme nitroqens. There is a rotation of the and 'Lpl,, cmponantm 0 the NO h m r f i n e tsnmor with resw t to g and g directions SO that ii 9. ie'taken as the le-NO bond direction, the PaNO bond angle is in the 105-110. ram=. A perturblng m i n o acid aidechain is s h n m that maya=m u n t for the unique proton B N m R signal seen for NCligated oxidase.
+
-
-
proportionality constant is -4.5 ~ z l e l a s t r o n 135). Footnote 7 explains the known details of thequadrupole Interaction for free NO. when electron density is concentrated for d u e c t e d Fa-NO and PeN-0 bondzng in II,T * , and o orbitals in the PeNO plane. then the prlnclpal quadm p o l e coupling in the direction perpendicular to thePeNO plane will be poliLtive. The explanatlon is that when the x direction is taken perpendicular to theFeNO plane and when the y and z directions in the FeNO plane contain greater electron denslty than the x direction, then flN + Nz) > NX. Slnce the prOprt10nallty constant relating Q,, to Nx,Ny, Nz LS negative, Q,, is positive. C. Metal Character of the Unpaired Electron and it8 Effect on q-values. Our results are most compatible with the unpaired electron lylng in a Hono that has considerable metal dz2 character where the d Orbltal p x n t s near the normal to the heme. The dz2, orbltal serves to transfer s p m v ~ a 814ma bond from a n * Orbltal on the NO to the trans histidine. Svch a transfer accounts for the hyperfrne Coupling to the hlstldlne nltrogen, whose hyperfme covplrng IS mainly iaotropic, but with the largeat component (indlcative Of sigma b o n d m g ) along the z a x l s . ~ e c a u s ethe $ component o f Orbital angular m m e n t m does not connect dz2 to other d Orbitals, the value Of gz would be the one closest to the free electron 9-value (2.0023), and the gz direction would be Closest to the heme normal a s I D found. TO account for one of the otherg-values being le86 than 2.0023, It 1s necessary to include another metal orbital In the HOMO. An Orbital Whlch has been p r o w s e d (31, 40) 1s the dXz orbital. whlch I" combination with dz2 would enhance covalent overlap I" the PeNO plane wlth a r* Orbital of NO dlrected toward the Fe.
-
we next conaidered evldence for a bent PeNO bond. Our EPR aimulatione (Figure 6) indicated that the -ha1 component Of NO hyperfine coupling lies close to gz but 1s rotated from It in the g -g plane by 15-20*, and a mlnimmal component of NO nitrogen hyperfine cou~li& lies near g y in the g -9 plane but is rotated from g by 15-20'. The nltrogen 2p Orbital contiin:ng the vnpaird electron spl: thns lies perpendicular to gx and in the g -g plane, close to, but tllted away from gz. If gx is perpendicular to Y " the PeNO plane, we can then take g and 9, as lying in the FeNO plane. If we follov the line Of reasoning used In previous EPR alngle CryBtal Work
122) in which gz was assumed to lie along the le-NO bond and the minimal NO hyperfine Coupling was asslmed to lie along the FBN-0bond in the PeNO plane, the angle Of FeNO bending would turn Out to he 90° + 1 1 5 O to 20') = 105- to 110.. The recent single Crystal EPR study Of NOHb by Utterback et al.(footnote 4 ) hae glven us cause to reconalder the above rather B y m e t r i c m d e l for NO binding in nltm8ylferrocytOChrome a.. The NO nitrorren hwerfine tensor for 0 and B Hb BubUnlts Was subatantlally rotated away from the 9 tensor. The maximal 9-value lgx) and the nearest hyperfine axis to it were i 40. from each other. This particular hyperfme axis, which was belleved perpendicular to the N-0 bond, was 25' Out Of the heme plane. The g Z = 2.00 direction wae found to be wlthin 10' of the heme n o m 1 but the maxlmbl hyperfine Coupling a x l e was i 40. rotated from Utterback et gz. al.auggested that a tight, distorted heme p c k e t ~n hemoglobln causes Offset of theNO nitrogen from the heme normal and c a w e s the n f Orbital Of NO that c o n t a m s the unpaired electron to twlst aut of the FeNO plane. The resultant non-colinearity of hyperfme and g tensors is a ponalble reason why frozen NOHb EPR Spectra (llke those of NO&) do not show the rasolYtiOn Of nltra~Ylfer~OcYtOchrorne*3. The greater collnearlty Of hyperfine and 9 tenSOrB I" nitT08ylfeT~OcytOchTOme*3 may m p l y a less distorted NO b m d l n g slte ~n nlt=OSylfe==OCytOCh~m=*3.
ENDOR-EPR of NO-ligated Heme a3 in Cytochrome c Oxidase
7448
Table 3a NO& NITROGEN C O W L I N G S IMzl
Complex Myoglobin NO
N U C ~ Y L I q-value 15N0 2.086
Table 3b FeTPP-NO-IMIDAZOLE NITROGEN COUPLINGS I M Z )
COupllngs f r m E N M R IHHz) l X 5 A x / =36.6 f 0.6
Hyperfine coupllng from EPR (Gauss b M z l
NO-FeTPP-IMID
27.7 2 0 . 3 =
“NO
2.086
[‘‘Ax/=
l4NhLs
2.086
1
I4Nhls
2.086
IOxx/ =
15N0
2.005
Axl= 15.9
2 0.2
0.6 f 0.1 )15A11= 28.0
2
0.5Gb
= 78.4 5 1.5
I4NO
2.005
1I4Az/= 20.0
I42.005 Nhls
15N0
1.969
1ISA
Complex NUCleYS
I=
47.0
=
56.0
lAzl=
6.7
=
18.5
nnz
0.5Gb 1.5 HHz
2 0.5G 1.5 MHz
2.0
‘Computed from the z solved “NO lines (Figure 3 ) neal 13 and 15 M r : there may be addltional lsNO lmes near 11-12 MHz which are not well resolved from HbNO. ’EPR d221dH2.
spllttlngs in NOMb were resolved by second harmonic detectlo” Of
q-value 14N0
2.07
Couplings from E N W R (MHz) 114Ax/= 31.7
0.3
Hyperfine coupling f r w EPR IGau88 L MHz1