J . Phys. Chem. 1992,96,7917-7922
7917
Probing the Iron Center of the Low-Spin Cyanide Adduct of Transferrin by ESEEM Spectroscopy Penny A. Snetsinger? N. Dennis Chasteen,*,+Jeffrey B. Cornelius,i and David J. Singel*.* Department of Chemistry, University of New Hampshire, Durham, New Hampshire 03824; Department of Chemistry, Haruard University, Cambridge, Massachusetts 02138; and Department of Molecular Pharmacology, Albert Einstein College of Medicine, Bronx, New York 10461 (Received: July 26, 1991; In Final Form: May 7 , 1992)
Electron spin echo envelope modulation (ESEEM) spectroscopy was used to investigate electron-nuclear interactions in a cyanide adduct of the iron transport protein transferrin in order to better understand the structure of the metal site in this tricyano adduct is formed only in the C-terminal iron(II1) binding site of the protein unusual complex. The low-spin, S = and is characterized by a rhombic EPR spectrum having principal g factors of g, = 2.34, gyy= 2.15, and g,, = 1.92. Twoand three-pulse ESEEM spectra were taken at the canonical positions in the EPR powder pattern at two microwave frequencies (8.8 and 10.5 GHz) at 4 K. ESEEM patterns of the adduct made with I3CN-show no "C modulations, presumably because the magnitudes of the carbon hyperfine coupling and Zeeman interactions are mismatched. The adduct made with CI5Nshowed peaks assignable to lSN. A detailed analysis of the ISNspectra is presented from which the I5N nuclear hyperfine coupling is determined. In conjunction with I3C ENDOR data, the I5N ESEEM data indicate that two cyanides are coordinated in a bent configuration trans to one another along the direction belonging to gxx. No evidence for histidine coordination to the iron was obtained from the ESEEM data, raising the possibility that this protein ligand has been displaced by cyanide in the adduct. Studies with D20 indicate that the iron center is accessible to, but not directly coordinated by, the solvent, a result consistent with no change in metal site solvation in the adduct compared to the native protein and with the expectation that CN- would be a preferred ligand relative to water.
Introduction Cyanide anion is often used as a probe of the exogenous ligand sites in iron-protein complexes since it is kinetically inert and produces spectroscopicallydistinct low-spin complexes upon interaction with the high-spin metal center. Low-spin cyanide complexes of hemoglobin and ferrihemes'*2 have been well characterized, but recently there has been interest in a number of non-heme iron(II1) proteins and model complexes which also form low-spin cyanide The serum protein transferrin consists of two independent lobes, N-terminal and C-terminal, each containing one site which binds iron or other metals. The X-ray structure of rabbit serum transferrinSand the related protein lactoferrin6 indicates that the ligands are two tyrosines, one histidine, and one aspartate with the remaining two coordination positions filled by (bi)carbonate. The addition of cyanide to transferrin results in the formation of a low-spin adduct at only one of the two iron(II1) binding sites, the C-terminal site, which exhibits a rhombic EPR spectrum with principal g values g,, = 2.34,g = 2.15,andgz, = 1.92.' ENDOR spectroscopy has been succ&hy used to investigate the orientationdependent "C nuclei interactions of the isotopically substituted I3CN- adduct of the proteinsg Electron spin echo envelope modulation (ESEEM) is particularly well-suited to the study of weak hyperfine interactions of low Zeeman frequency nuclei and often proves successful in studying the low-frequency region which is problematic in ENDOR. The complementary nature of the two spectrampies makes it advantageous to use S E E M to supplement the information obtained from ENDOR about the ligand environment of the iron. We report here ESEEM measurements on the cyano adduct of transferrin. Our focus is to further elucidate the hyperfine interactions sustained by the cyano ligands, which are specifically probed via 13Cand ISN isotopic substitution. In particular, by combining orientation-selective and frequency-tracking techniques18the IsN hyperfine interaction is determined in detail. To gain some additional insight into the nature of the protein-metal complex, the ESEEM patterns are also examined for modulation effects ascribable to nitrogen nuclei in endogenous ligands and Authors to whom correspondence. should be addressed. 'University of New Hampshire. Harvard University. #Albert Einstein College of Medicine.
*
0022-3654/92/2096-79 17$03.00/0
to deuterium in samples prepared in deuterium oxide.
Experimental Section The samples were prepared as described for the ENDOR experiments? ESEEM spectra were taken on the ESE spectrometer constructed at the Biotechnology Resource in Pulsed EPR Spectroscopy at the Albert Einstein College of Medicine. The spectrometer is described in detail elsewhere.'O Spectra were taken at liquid helium temperatures using a folded stripline reflection cavity with a specially designed head tmembly.11J2Electron spin echo envelopes were obtained using both two-pulse and threepulse sequences. The time domain data were transformed to the frequency domain by use of the algorithm described by Mims." Threepulse echo envelopes were generally recarded with a range of times ( 7 ) between the first and second pulses to prevent accidental suppression of frequencies. Spectra were taken at two spectrometer frequencies, 8.8 and 10.5 GHz. To obtain orientation-dependent information, spectra were taken at field values corresponding to each of the three principal g values.
Results and Discussion Cyanide Ligand ESEEM. Three cyanide groups have been previously shown to be involved in the formation of the transferrin-cyanide a d d ~ c t .However, ~ this earlier work was unable to rule out the possibility that one or more of the cyanide groups were bonded to cationic sites on the protein and not to the metal center d i r e ~ t l y .ENDOR ~ spectroscopy was only able to locate definitively one pair of "C resonance lines when isotopically enriched I3C cyanide was used. Since ENDOR cannot be used quantitatively to determine the number of ligands and since the quality of the low-frequency ends of the ENDOR spectra was poor, ESEEM was used to look for additional interactions from the cyano groups. To probe selectively for interactions with the cyanide carbons, time domain patterns for both 13C- and l2C-1abeled cyanide samples were obtained, and their ratio was computed in the manner first suggested by Zweier et al.I4 Such quotient patterns, obtained at field values corresponding to gxx,g and gZ1(showed no modulation (data not shown), indicating t g t there is no ob servable interaction due to I3C along any of these axes. The absence of I3Cpeaks in the ESEEM spectra is not entirely unexpected since the hyperfine coupling of I3C observed in the ENDOR spectrum of transferrin was quite large, on the order 0 1992 American Chemical Society
7918 The Journal of Physical Chemistry, Vol. 96, No. 20, 1992
0.20
5.20
10.20
Snetsinger et al.
0.30
TIME bs)
0.20
5.30
10.30
TIME @s)
0.30
5.30
10.30
TIME @(s)
Figure 1. ( a ) Three-pulse ESEEM spectrum of C"N adduct along gxx (8.78 GHz, Ho5 2677 G, T = 175 ns). (b) C"N spectrum (8.76 GHz, Ho5 2677 G, T = 175 ns).
of 30 MHz.9 Couplings of this magnitude are not typically observed with X-band S E E M spectroscopy since A/2 >> v, = 4 MHz.I5 Moreover, this result suggests the absence of any additional 13Cnuclei with smaller couplings in the range 3-10 MHz that might have escaped detection by ENDOR. PreviousENDOR work showed weak signals in the 1-5-MHz range probably due to I4N nuclei and possibly from some radiofrequency intcrference.l6 In contrast to ENDOR, the low-frequency peaks expected for weakly coupled nitrogen nuclei are well suited for observations by ESEEM. The time domain patterns of the C14N and C15N adducts are shown in Figures 1,2, and 3 for the fields corresponding to gxx,gyy,and g,,, respectively. All of the echo envelopes are substantially modulated, and stark differences are evident between the pairs of patterns of the I4Nand 15N-labeledcomplexes obtained at corresponding field positions; evidently, the ESEEM patterns are dominated by modulations arising from the cyanide nitrogens. Since ISN has a nuclear spin of the ESEEM spectra of the C15N adduct were employed to determine the hyperfine coupling parameters; the C14N adduct exhibits considerably more complicated spectra (vide infra). The spectra that correspond to the ESEEM patterns of the CI5N adducts shown in Figures 1-3 are shown in Figure 4. At the field position belonging to gxx(Figure 4a) two narrow intense peaks at 0.59 and 1.59 MHz approximately centered around the free precessional frequency of 1.15 for ISN are observed. Likewise, at the field value corresponding to g,, (Figure 4c) there are two sharp peaks at 1 . 1 1 and 1.76 MHz centered about the Zeeman frequency of 1.39 MHz. There are also broader and less intense peaks at 2.70 and 3.53 MHz. At the gyyirradiation position (Figure 4b), an intense peak at the Zeeman frequency 1.36 MHz is observed with less intense peaks at 2.75 and 4.09 MHz. A single sharp peak at the proton Zeeman frequency is observed in all spectra but is not shown in the figures. Low-amplitude components at twice the frequency of intense fundamentah are evident in these spectra and in many of the other
Figure 2. (a) Threc-pulse ESEEM spectrum of C1'N adduct along gyy (8.78 GHz, Ho= 2906 G, T = 242 ns). (b) Ci5N spectrum (8.76 GHz, Ho= 2906 G, 7 = 323 ns).
two- and three-pulse ESEEM spectra. The presence of the combination and harmonic peaks in the threepulse spectra makes it apparent that there are at least two equivalent CN- groups coordinated to the iron center; such peaks are observed when more than a single nucleus contributes to the modulation." The 15N ESEEM observed with excitation at the gxx,gyy,and g,, EPR positions can be analyzed (vide infra) as deriving from a single, magnetically equivalent set of nuclei, presumably from two CNgroups coordinated trans to one another. Although there have been reports of ESE at frequencies from 1 to 34.2 typically ESE has been done at 9.5 GHz, X-band. The importance and advantages of taking data at several spectrometer frequencies have only recently been recognized. To facilitate analysis of the nitrogen interaction, modulation envelopes were taken at two frequencies, 8.8 and 10.5 GHz. With data at more than one frequency, the isotropic and anisotropic contributions to the observed hyperfine couplings can be determined.8J2 The observed ESEEM frequency, yo&, is given by (vOb)* = (v,
+ tm,A)2 + mS2Bz
(1) where v, is the free precessional frequency of IsN, t is the sign of the isotropic hyperfine coupling, i.e. 4 = i l , m, = and A and B are the secular and nonsecular parts of the hyperfine coupling, respectively. The equation can be separated into field-dependent (gauged by v,) and -independent terms: 4(vOb2- v,') = 4,5A~,+ ( A 2 + B2) (2) Thus, a graph of 4(vOb2- v,2) vs v, for the different spectrometer frequencies will result in a line with slope 44A and intercept of (Az + Bz). Note that the intercept cannot be less than zero. Figures 5 and 6 show these graphs for the gxxand g,, data. There is some uncertainty in peak position since different values of T cause the frequency peak to shift position slightly. The peak positions for IsN at all T values are listed in Table I. From these graphs the couplings when the field is set at gxxare A = 0.99 f
The Journal of Physical Chemistry, Vol. 96, No. 20, 1992 7919
Iron Center of the Cyanide Adduct of Transferrin
1.59
10.0
5.0
0.0
FREQUENCY (MHz) 5.30
0.30
10.30
TIME Ocs)
5.0
0.0
10.0
FREQUENCY (MHz)
5.30
0.30
10.30
u
1
1.76
"IbfE ots)
Figure 3. (a) Thrcc-pulse ESEEM spectrum of CI4N adduct along g, (8.78 GHz, Ho= 3226 G, T = 290 ns). (b) CI5Nspectrum (8.76 GHz, Ho= 3226 G, T = 290 ns).
TABLE I: ESEEM Peak Positions of CI5N with Various Values of T at Spectrometer Frequmcies of 8.76 and 10.34 CHz f r q (GHz) 8.78
10.34
(ns) 175 263 162 242 323 145 218 290 207 415 191 381 257 342
T
g,
gyy g,, g,
gyv g,,
peak position (MHz)' 0.59, 1.59, 3.58 0.71, 1.60 1.36, 2.75, 4.09 1.32, 2.73, 4.02 1.35, 2.73, 4.06 0.48, 1.09, 1.75, 2.50, 3.60 1.10, 1.75, 2.70, 3.58 1.11, 1.76, 2.70, 3.53 0.97, 1.74, 3.62 0.99, 2.19, 3.62 1.53, 3.19 1.59, 3.28 1.36, 1.98, 2.78, 3.90 1.24, 1.98, 2.79, 3.97
The proton precessional frequencies, 1 1.4-16.6 MHz, observed in the various spectra are not listed.
0.16 and B = 0.75 f 0.30 MHz and at g,, are A = 0.67 f 0.03 and B = 0.44 f 0.14 MHz (standard errors).
The A values obtained with excitation at the gxxand g,, EPR field positions, at which orientation selection is essentially complete! provide direct measures of the magnitudes of A , and A,,, respectively (elements of the A matrix in the electron g principal axis system). In the case of excitation at gyy, the orientation selection is not complete and the analysis is less straightforward. The spectrum at gyyis a (ESEEM amplitude weighted) powder pattern involving, nominally, all orientations in the plane jointly spscified by the principal axis of g that belongs to gvvand a vector lying between the axes belonging to gxxand g,. Inasmuch as the observed spectrum consists of a single line of modest width at the
C 0.0
'
"
'
'
"
"
l
5.0
10.0
FREQUENCY (MHz)
Figure 4. (a) Frequency spectrum of Figure lb. (b) Frequency spectrum of Figure 2b. (c) Frequency spectrum of Figure 3b. The deep troughs at frequencies below 1 MHz stem from overlapping side lobes of the strong peaks at 1 MHz and at -0 frequency associated with the echo decay. The peak at 4.09 MHz occurs at 3 times the frequency of the 1.36-MHzpeak and exists, like the fundamental and sccond harmonic, at all values of T surveyed, regardless of windowing schemes used prior to Fourier transformation.
-
free lSNLarmor frequency, the secular hyperfine coupling must maintain a value approximately equal to zero over the domain of sampled orientations. Accordingly, we take Ayy = 0, and inasmuch as the hyperfine coupling must remain close to zero at those orientations in the xz plane included in this restricted powder average, we deduce that A , and A,, must be of opposite sign. The B values obtained by graphical analysis of the 15NESEEM data establish parametric relations among the magnitudes of the offdiagonal elements of A. With A assumed symmetric, the valuea obtained with excitation at g, and g, are measures of An2 + AV2 and Ax: + A,,:, respectively. Unfortunately, the data obtained at the guyfield do not give a significant measure of Ax: + Ay:.2s Accordingly, the off-diagonal elements of A cannot be fully determined; however, we can delimit them to a compact neighborhood specified by the correlated inequalities (values in MHz) 0.55 i lAxyl 5 0.87, 0.66 1 IAxzI 1 0, 0 5 lAYJ50.66
(3)
Snetsinger et al.
7920 The Journal of Physical Chemistry, Vol. 96, No. 20, 1992
T
I
frcq (GHz) 8.78
I
6 1
\
4vn
TABLE 11: ESEEM Peak Positions of C"N with Variow Values of T at Spectrometer Freauencies of 8.78 .ad 10.52 CHz
I
I
1 I
.6
-4
.2
I
.6
.lo
\
I
I
I
2
4
10.52
i
tI
text.
t
1 Figure 6. Graph of IJNESEEM data along g, as described by eq 2 in .4
text.
The limits within which we have been able to fix the elements of A are sufficiently narrow that it is fruitful to examine the properties of A. In particular, we consider the eigenvalues and eigenvectors of the A matrices that correspond to the outer limits of the neighborhood, where the properties are most succinct: the eigenvalues (apart from overall sign) and the magnitudes of the eigcnvactor oompollglte are uniquely determined by the magnitude alone of the off-diagonal elements owing to the presence of an off-diagonal element of zero magnitude. By decomposing the into the comeigenvalues XI, X2, and X3, with IX,( I IX,l I ponents A,oa Tr A/3, A? (A3 - AO0)/2,and A,* E (Az - Xl)/2, we determine, for Ay2= 0 Aoo * fO.ll MHz, Azo 10.66 MHz, A2' = 10.35 MHz (4) while for A, = 0 A$
fO.ll MHz, Azo
(ns) 175 263 350 162 242 323 145 218 290 242 415 191 381 257 342
peak position (MHz)O 2.54, 3.02, 1.68, 2.16, 2.54 gyy 0.55, 2.79, 2.23, 2.80, 0.59, 1.37, g, 2.54, 3.88, 1.95, 2.18, 2.54, 3.81, g, 0.27, 0.74, 0.57, 2.19, gyy 0.76, 2.82, 0.64, 2.38, g, 0.66, 2.46, 0.66, 2.54 g,
4.35 2.84 3.44, 3.48, 2.14, 4.67 2.74, 4.69 1.14, 3.91 3.68 3.70 4.90
6.26 4.30 2.63, 3.51, 4.39, 5.37, 6.05 3.89, 4.69, 6.94 2.98, 3.30
"The proton precessional frequencies, 11.4-16.6 MHz, observed in the various spectra are not listed.
Figure 5. Graph of I5NESEEM data along g, as described by eq 2 in
.3
T
10.71 MHz, A'' = f0.58 MHz (5)
These sets of coupling constants are clearly very similar. We may conclude that the 15Nhyperfine interaction is dominated by the axial coupling (pj/A$I- 6.1-6.6) with appreciable rhombicity (p?/Azqn 0.61-0.82). Moreover, the major axis of A maintains an orientation close to the direction associated with g., The angle between these vectors is 25O for the case of Ay2= 0 and 32O for A,, = 0, angles not very different from one another. In principle, the ESEEM data obtained with the CI4N adducts afford a means to check the analysis of the 15Ncyanide ESEEM. In practice, this implied process is complicated, no general analysis of I4N ESEEM powder patterns (including those powder averages restricted by orientation selection) for a nucleus with a dominantly anisotropic hyperfine interaction has yet been published. The ESEEM data on the CI4N adduct of transferrin provide a body of experimental results that both motivates and facilitates such an undertaking-which is now underway in our laboratories. Endogenous and Solvent Ligand HEEM. In addition to modulations from cyano ligands, the ESEEM patterns shown in Figures 1-3 might also contain modulations from other sources. Of particular interest are nitrogen modulations arising from a possible protein histidine ligand-as observed in Cu2+and V@+ complexes of transferrin with carboxylate Inasmuch as the electronic structure of the VOz+ and iron-cyano complexes of transferrin are thought to be similar, equivalent histidine ligation should produce comparable modulation effects in the two systems. The ESEEM patterns of Figures 1-3 were therefore further examined. Since any peaks arising from endogenous I4N would be present in both the CI4N and CI5N ESEEM spectra, a search for peaks common to both was made. Table I1 summarizes the frequency results for each of the different 7 values at both spectrometer frequencies. As can be seen, there is some variation in peak location as 7 is changed since the line shape is affected by 7 suppreasion effects." Exemplary spectra are shown in Figure 7. In all cases a number of peaks in the 2-7-MHz region typical of 14Nmodulations are present. In the spectrum obtained at fields corresponding tog, there are peaks at 2.54, 3.054.35, and 11-40 MHz (Figure 7a). The data obtained at gw include peaks at 0.59 and 12.40 MHz and an archipelago of weak peaks extending from 1.37 to 6.05 MHz (Figure 7b, Table 11), which is almost certainly a remnant of a broad peak riddled with suppression gaps as we have observed on occasion with I4N nuclei having highly anisotropic interactions. The spectrum obtained at gZ2(Figure 7c) has a prominent peak at 2.54 MHz and lesser peaks at 3.81, 4.69, and 13.79 MHz. Comparison of the data in Tables I and I1 for CI5N and C14N reveals few common frequencies. In the data obtained at 10 GHz a single common component appears with excitation of the field corresponding tog, and with 7 = 415 ns. This peak, like all other common frequencies, has a definite assignment as a multinuclear combination peak in the C15N ESEEM spectra. We conclude that no compelling evidence exists in the ESEEM data far histidine
The Journal of Physical Chemistry, Vol. 96, No. 20, 1992 7921
Iron Center of the Cyanide Adduct of Transferrin '
'2.54'
'
'
'
'
'
-
g,
4.35
0.00
7.50
15.0
FREQUENCY (MHz)
12.40
I ,
I
1
1
0.00
I
7.50
'
'
'
15.0
FREQUENCY (MHz)
0.00
7.50
15.0
FREQUENCY (MHz)
Figure 7. Frequency spectrum of I4CN adduct for field along (a) gxx (8.78 GHz, Ho= 2677 G, T = 175 ns), (b) gYy(8.78 GHz, H,,= 2906 G,7 = 323 ns), and (e) grr (8.78 GHz, Ho = 3226 G,T = 290 ns).
coordination in the cyano complex of iron transferrin. The accessibility of the metal site to solvent was qualitatively investigated by using D 2 0 as a solvent. The spectra obtained in D 2 0 were divided by the H 2 0 spectra to remove interfering modulation components from the 2H spectral region.14 The resulting quotient ESEEM patterns obtained at field values corresponding to g,, gyy,and g,, exhibit an oscillation at the deuterium Larmor frequency with an amplitude (quotient) of about one-half, similar to that observed by Eaton et al. in 2H20-exchanged V 0 2 + complexes of t r a n ~ f e r r i n . ~ ~ The lack of resolved hyperfine splitting in any of the 2 H 2 0 spectra-in particular in those obtained at the g, and g,, field positions where there is essentially complete orientation selection along two orthogonal axes-strongly suggests that the hyperfine coupling is less than the 25-kHz peak width, thus establishing a minimum F t 2 H (pointdipole) distance in excess of 2.9 A. Simulations of the 2H quotient ESEEM patterns are also consistent with a minimum distance of 3 A.23 The spectral line probably therefore belongs to the deuterium nuclei of ambient water
-
molecules which are not directly coordinated to the iron and/or possibly to exchangeable deuterons on the protein in the vicinity of the paramagnetic center. ESEEM studies of solvent exchange in vanadyl transferrin have also shown that solvent does not coordinate directly to the metal center of that com~lex.2~ The D E E M results with the cyano-iron transferrin complex are consistent with those obtained from ENDOR spectra which show some reduction in the matrix proton peak upon H2D-D20solvent exchange but no evidence for directly coordinated water.I6 Cyanide Hyperflne lntencbioas a d Structure. A simple picture of the electronic structure of the iron center has emerged from the analysis of the electron g factors and the ENDOR determined 13Chyperfine couplings of the transferrin complex with ~yanide.79~ The principal axes of & define the ligand field axis system, with the coordinated atoms situated essentially along the coordinate axes. Within this framework, the unpaired electron spin density is largely located in the 3d, orbital of the iron(II1)' with a p proximately 1% delocalization onto the cyanide carbon atoms? The present IsN ESEEM results are amenable to analysis within this picture and provide further insight into the nature of the cyanide coordination at the iron binding site in transferrin. The small isotropic lSN coupling of 10.11 MHz reflects the scant delocalization of the spin density onto the ligands, in this instance -0.005% (= 0.1 1/2160 X 100, unit spin density giving a coupling of 2160 MHz for 15N2s).The anisotropic hyperfine coupling can be modeled, albeit somewhat crudely, as a point-dipole interaction between the electron and nuclear point magnetic dipole situated respectively at the iron and I5NpositiomB In this model the value of A t provides a measure of the distance between the iron and the IsN nucleus, while the major axis of the traceless portion of A indicates the iron-nitrogen line of centers. With the above assumptions in mind, we obtain from the experimental value of A t an iron-nitrogen distance of -2.3 A and an Fe-C-N angle of -72'. While this analysis indicates cyanide coordination along g, direction, in a bent configuration its limitations are abundantly clear. The iron-nitrogen distance appears to be too smallg0and the angle too large; more fundamentally, the rhombic character of A cannot be accounted for by this simple model. It is worth noting, however, that more reasonable bond lengths and angles would be obtained if the interaction between the ISNand the carbon-centered electron spin density were also included in this point-dipole calculation. It is difficult to carry out such refinements in the absence of detailed insight into the distribution of the unpaired electron spin density within the u and x orbitals of the ligands. The IsN ESEEM data are rather easily accommodated within the simple model of the electronic structure and hyperfine coupling that has been successfully applied to earlier EPR and ENDOR With x-coordination of the CI5N- ligands that give rise to the ISN ESEEM, it is apparent that we observe the same cyanide species as observed by 13C ENDOR of 13CN- adducts of transferrin. The multinuclear combination peaks indicate more than a single nucleus contributes to the ESEEM, and adhering to the minimal assignment above, we determine that a pair of cyanide ligands are coordinated at opposite ends of the x axis. As noted previously? such a configuration of cyanide ligands requires reorganization of the protein coordination, prehaps with displacement of the histidine ligand since no ESEEM evidence was obtained for its presence. If a third CN ligand is also coordinated to the metal ion, it is extremely elusive, exhibiting neither "C ENDOR nor ESEEM effects and, accepting the assignment above, no ISN ESEEM. While these results cannot prove the absence of a third cyanide ligand, they begin to establish such constraints on the possible value of its hyperfine couplings, that its absence from the coordination sphere becomes an increasingly appealing explanation of the results. Finally, it is interesting to compare these results with those of Wang and de Boer for Fe(CN)63-in a KCl lattice.26 In uspect, the couplings that we determine for lSN in the cyanide complex of transferrin are remarkably similar to those reported by Wang = 7.1 and IA22/A201= 0.59 vs 6.1-6.6 and de Boer (IA20/Aoo1)
7922 The Journal of Physical Chemistry, Vol. 96, No. 20, I992 and 0.61-0.82, respectively, for transferrin. In scale, however, the I5Ncouplings are 4-5 times smaller in transferrin than in the Fe(cN)6+, despite the fact that the 13Ccouplings are very similar in the two system^.^,^^ The distinction may be related to the different geometric arrangement of the F d - N fragment in the two systems: linear in Fe(CN)6” and bent in transferrin. In the linear system, r-conjugation could lead to the large hyperfine interaction with a I5N nucleus approximately 3 A from the metal center. It should also be noted that the electron g factors and the g axis orientations are very different in the system studied by Wang and de Boer from those deduced for the cyanide complex of transferrin.
coacluriorrs The results from the present ESEEM study along with those derived from ultraviolet-visible absorption, EPR, and ENDOR spectroscopic s t ~ d i e s ’ *have ~ ~ ’ ~provided further insight into the structure of the cyanide adduct of transferrin. These results are consistent with a model of the metal site in which the iron is coordinated by two phenolate groups of tyrosine residues and two trans cyano groups coordinated in a bent configuration. The identities of the presumed fifth and sixth ligands are unclear. The data argue against the ligation of histidine, water, or the third cyanide group. A lack of histidine coordination would be consistent with recent X-ray absorption fine structure data of the low-spin cyanide complex of transferrin which suggests that the complex may be formed at a new binding site for iron on the protein.*’ Acknowledgment. We thank Prof. Jack Peisach for helpful comments on the manuscript. This work was supported by NIH Grants GM20194 and RR-02583. Regi~tryNo. Fe, 7439-89-6.
References and Notes (1) Scholes, C. P.; Van Camp, H. L. Eiochim. Eiophys. Acra 1976,434, 290. ( 2 ) LoBrutto, R. L.; Wei, Y. H.; Mascarenhas, R.; Scholes, C. P.; King, T. E. J. Biol. Chem. 1983, 258,7437. (3) (a) Wittaker, J. W.; Lipscomb, J. D. J . Biol. Chem. 1984,259,4487. (b) Orville, A. M.; Lipcomb, J. D. J. Biol. Chem. 1989, 264,8791. (4)(a) Spartalian, K.; Carrano, C. J. Inorg. Chem. 1989, 28, 19. (b) Carrano, C.J.; Carrano, M. W.; Sharma, K.; Backes, G.; Sanders-Loehr, J. Inorg. Chem. 1990, 29, 1865. (c) McDevitt, M. R.;Addison, A. W.; Sinn, E.; Thompson, L. K. Inorg. Chem. 1990, 29, 3425.
Snetsinger et al. (5) (a) Sarra, R.; Garratt, R.;Gorinski, B.; Jhoti, H.; Lindley, P. Acra Crysrallogr. 1990, E46, 763. (b) Bailey, S.;Evans, R. W.; Garratt, R. C.; Gorinsky, B.; Hasnain, S.;Horsburgh, C.; Jhoti, H.; Lindley, P. F.; Mydin, A.; Sarra, R.; Watson, J. L. Biochemistry 1988, 27,5804. (6)(a) Anderson, B. F.; Baker, ff. M.; Dodson, E. J.; Norris, G. E.; Rumbull, S.V.; Waters, J. M.; Baker, E. N.Proc. Narl. Acad. Sci. U.S.A. 1987,84,1769. (b) Baker, E.N.; Anderson, B. F.; Baker, H. M.; Haridas, M.; Norris, G. E.; Rumball, S.V.; Smith, C. A. Pure Appl. Chem. 1990,62, 1062. (7)Swope, S.K.; Chasteen, N. D.; Weber, K. E.; Harris, D. C. J. Am. Chem. Soc. 1988, 110,3835. (8) Gerfen, G. J.; Singel, D. J. J . Chem. Phys. 1990, 93,4571. (9)Snetsinger, P. A.; Chasteen, N. D.; VanWilligen, H. J. Am. Chem. Soc. 1990,112,8155. (10) McCracken, J.; Peisach, J.; Dooley, D. M. J. Am. Chem. Soc. 1987, 109,4054. (1 1) Lin, C. P.; Bowman, M. K.; Norris, J. R. J. Mugn. Reson. 1985,65, 369. (12)Britt, R. D.; Klein, M. P. J . Mugn. Reson. 1987,74, 535. ( I 3) Mims, W. B. J. Magn. Reson. 1984,59,29 1. (14)Zweier, J.; Aisen, P.; Peisach, J.; Mims, W. B. J . Biol. Chem. 1979, 254, 3512. (15)Lai, A.; Flanagan, H. L.; Singel, D. J. J . Chem. Phys. 1988,89,7161. (16)Snetsinger, P. A. Ph.D. Dissertation, University of New Hampshire, Durham, NH, 1990;p 64. (17) McCracken, J.; Pember, S. F.; Benkovic, S . J.; Villafranca, J. J.; Miller, R. J.; Peisach, J. J . Am. Chem. Soc. 1988, 110, 1069. (18) Crookham, H.; Brown, D.; Belford, R. L.; Clarkson, R. B. Presented at the International Electron Paramagnetic Resonance Symposium, 31st Rocky Mountain Conference, Denver, 1989. (19)Schmidt, J. Chem. Phys. Luff.1972, 14,411. (20) Mims, W. B.; Nassau, K.; McGee, J. D. Phys. Reu. 1961, 123,2059. (21) Liao, P. F.; Hartman, S. R. Phys. Rev. 1973,E8,69. (22)Cosgrove, S.A.; Singel, D. J. J. Phys. Chem. 1990, 94,2619. (23)Eaton, S.S.;Dubach, J.; Kundalika, M. M.; Eaton, G. R.;Thurman, G.; Ambruso, D. R. J. Biol. Chem. 1989,264,4776. (24)Cosgrove, S.A.; Singel, D. J. J . Phys. Chem. 1990, 94,8393. (25)Wertz, J. E.; Bolton, J. R. Electron Spin Resonance; McGraw Hill: New York, 1972;Table C. (26)Wang, D. M.; de Boer, E. J . Chem. Phys. 1990, 92,4698. (27)Evans, R.; Garratt, R. C.; Hasnain, S.S.; Lindley, P. F.;Neu, M.; Strange, R. W.; Sarra, R. 10th International Conference on Iron and Iron Proteins, Oxford, UK, July 27, 1991;paper P39. (28)With A ii:0 and Y, >> IEI, the ESEEM frequency may be expressed as Y = Y, + E2/8u,. Because the impact of the nonsecular hyperfine coupling on the sptctrum (a shift of Y from Y,, or dispersion of Y about Y, in a powder average) is scaled by the factor 8u,, the value of E is difficult to determine along y . If E were roughly 0.5 MHz, as in the excitation at x and y , then the spectral effects of E would amount to less than 25 kHz. (29)Such a treatment ignores factors such as electron g factor anisotropy and covalency in the metal ligand bonds. (30)Typical bond lengths (1.8 A, Fe-C; 1.O A, C-N) are assumed (ref 9 and references therein).