ESR and ENDOR of copper(II) complexes with nitrogen donors


Marco Bortolus , Marco Bisaglia , Alfonso Zoleo , Maria Fittipaldi , Maurizio ... Christian Remenyi, Roman Reviakine, Alexei V. Arbuznikov, Juha Vaara...
0 downloads 0 Views 913KB Size


J . Phys. Chem. 1992, 96,9684-9691

9684

efficiency with mass up to at least m / z 12 360. For molecular weight determination by matrix-assisted laser desorption, the results give some hints for future instrumental design of linear or reflectron time-of-flight instruments, regarding vacuum conditions, instrumental geometry and others. Furthermore, assuming that ion decay predominantly is characterized by fragment ion formation rather than by charge stripping, the obtained reaction cross sections can be taken as being valid for the efficiency of daughter ion formation as well. If this assumption can be confirmed in future investigations, then reflectron-time-of-flight mass spectrometry of peptides or proteins employing postsource decay6v2' will turn out to be a valuable alternative to collision-induced or photoinduced decomposition for structure analysis. CID experiments might then have to be reevaluated with respect to the question whether the observed loss of fragment yield above m / z = 1000 is really due to physical properties of the collision process or predominantly due to instrumental transmission effects. Acknowledgment. Support by the Bennigsen-Foerder-Program of the Ministry of Science and Research, NRW, Germany, is gratefully acknowledged.

References and Notes (1) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2288. (2) Karas, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. Inr. J . Mass Spectrom. Ion Processes 1981, 78, 53. (3) Egge, H.; Katalinic, J. P.; Karas, M.; Stahl, B. Pure Appl. Chem. 1991, 63, 491. (4) Schar, M.;Bornsen, K. 0.; Gassmann, E. Rapid Commun. Mass Spectrom. 1991, 5, 319.

(5) Spengler, B.; Kaufmann, R.; Kirsch, D. Rapid Commun. Mass Specfrom. 1991, 5, 198. (6) Spengler, B.; Kirsch, D.; Kaufmann, R.; Jaeger, E. Rapid Commun. Mass Spectrom. 1992, 6 , 105.

(7) Aberth, W. Anal. Chem. 1986, 58, 1221. (8) Dahl, D. A.; Delmore, J. E. Idaho National Engineering Laboratory,

Idaho Falls, Idaho. (9) Beavis. R. C.: Chait. B. T. Raoid Commun. Mass Soecrrom. 1989.3,

432. ' (10) Haddon, W. F.; McLafferty, F. W. Anal. Chem. 1969, 41, 31. (1 1) Geno, P. W.; Macfarlane, R. D. Inr. J. Mass Spectrom. Ion Processes 1989, 92, 195. (121 Takano, R.; Kallei, 0. B.; Swanson, R.; Dickerson, R. E. J . Biol. Chem.' 1973, 248, 5244. (13) Neumann, G. M.; Sheil, M. M.; Derrick, P. J. Z . Naturforsch. 1984, 39a. 584. (14) Strupat, K.; Karas, M.; Hillenkamp, F. Int. J . Mass Specrrom. Ion Processes 1991, 1 1 1 , 89. (15) Davis, S. C.; Derrick, P. J.; Ottinger, C. Z . Naturforsch. 1990,450, 1151. (16) Bricker, D. L.; Russell, D. H. J . Am. Chem. Soc. 1986, 108, 6174. (17) Spengler, B.; Bahr, U.; Karas, M.; Hillenkamp, F. Anal. Instrum. 1988, 17, 173. (18) Bahr, U.; Spengler, B.; Karas, M.; Hillenkamp, F. In Ion Formation

From Organic Solids, IFOS IV; Benninghoven, A., Ed.; Wiley: New York, 1989; pp 143-148. (19) Spengler, B.; Bahr, U.; Hillenkamp, F. Insr. Phys. Conf. Ser. 1988, 94, 137. (20) Speir, J. P.; Gorman, C. S.; Cornett, D. S.; Amster, I. J. Anal. Chem. 1991, 63,-65. (21) Glish, G.L.; Todd,P. J. Anal, Chem. 1982, 54, 842. 122) SDender. B.: Kirsch. D.: Kaufmann. R.: Karas. M.; Hillenkamp, . . F.:. Giessmain, fi. Rapid Commun. Mass Spectrom. 1990, 4; 301. (23) Kaufmann, R.;Spengler, B.; Kirsch, D. In Merhods and Mechanisms for Producing Ions From Large Molecules; Standing, K. G., Ens, W., Eds.; Plenum Press: New York, 1991; pp 235-245. (24) Spengler, B.; Kirsch, D.; Kaufmann, R.; Karas, M.; Hillenkamp, F.; Giessmann, U. Proc. 38th ASMS ConJ Mass Spectrom. Allied Top., June 3-6, 1990, Tucson, AZ 1990, 162. (25) Kaufmann, R.; Kirsch, D.; Rood,H.-A,; Spengler, B. Rapid Commun. Mass Spectrom. 1992, 6, 98. (26) Spengler, B.; Kaufmann, R. Analusis 1992, 20, 11. (27) Tang, X.;Beavis, R.; Ens, W.; Lafortune, F.; Schueler, B.; Standing, K.G. Inr. J . Mass Spectrom. Ion Processes 1988, 85, 43-67.

ESR and ENDOR of Cu(I1) Complexes with Nitrogen Donors: Probing Parameters for Prosthetic Group Modeling of Superoxide Dismutase H.-J. Scholl and J. Hiittermann* Fachrichtung Biophysik und Physikalische Grundlagen der Medizin, Universitiit des Saarlandes, 6650 HomburglSaar, Germany (Received: March 31, 1992; In Final Form: August 5, 1992)

The coordination properties of square-planar complexesof Cu(I1) with N4or N202ligation using imidazole, pyridine, glycine, ethylenediamine, diethylenetriamineimidazole,and ammonia as ligands were investigated by electron spin resonance (ESR) and electron nuclear double resonance (ENDOR) spectroscopies at 77 K and about 5 K, respectively. Combining isotopic substitution in metal, ligand, and solvent with ESR measurements at X-(9.5 GHz) and Q-band (34 GHz) and second-order spectra simulation, a comprehensive set of ESR parameters was obtained. In ENDOR, emphasis was given to extracting the nitrogen ligand couplings for all compounds and the proton interactions for imidazole and pyridine. The assignments were substantiated by powder-type ENDOR simulations. The data are discussed in terms of nitrogen anisotropy parameters as a function of ligand binding modes. In relation to the 'type 2" copper protein superoxidedismutase (SOD), the tetraimidazole compound was found to have model character for the proton but not the nitrogen couplings, whereas tetrapyridine could not model either of the interactions.

1. Introduction There is extensive literature on the electron spin resonance (ESR) spectroscopic properties of complexes of copper(I1) with N4 or NzX2(X = 0, S) ligation due to the relevance of this coordination with respect to type 1 or type 2 copper proteins (for a review see, e.g., refs. 1 and 2). In addition there are several electron nuclear double resonance (ENDOR) and pulsed ESR studies concentrating on diverse aspects like nitrogen and remote nitrogen interaction in tetraimidazole~~,~ and a series of related compounds,5as well as proton interactions in tetraimida~oles~ and related complexes.6 In spite of this coverage there does not appear to exist a single report in which both ESR and a high-resolution

technique like ENDOR were applied to proton and nitrogen interactions in a systematic way for a series of Cu(I1) complexes. We have recently reported on 'Hand I4N interactions obtained by ENDOR from the Cu(I1) site in Cu,Zn2-superoxide dismutase (SOD) from bovine and human plasma sources.' It turned out that most of the resonance lines had to be treated in a phenomenological way since the available knowledge about imidazole ligation to copper3v4was not as complete as to allow for a detailed transfer to the protein. As a consequence, the effects of binding of N< and CN- were even less amenable to interpretation in terms of structural changes of SOD. Although the proposed change in I4N couplings upon CN- binding given by van Camp et ala8was

0022-365419212096-9684%03.00/0 0 1992 American Chemical Society

The Journal of Physical Chemistry, Vol. 96, No. 24, 1992 9685

ESR and ENDOR of Cu(I1) Complexes SCHEME I

-*

I %

supported, no secure interpretation could be derived for this interaction after N3- addition nor was the proton interaction part for both anions understood in any detail. We therefore felt that there was a need for a more thorough study of potential model compounds combining ESR and ENDOR techniques, together with spectra simulation methods. The intent is to apply this knowledge to the native protein ENDOR data and subsequently approach the structural changes upon anion binding. Another question to be studied was Yokoi's proposal, that imidazole coordination to copper should be distinguishable from that of other nitrogens by a strong isotropy, yielding AII/A, values close to 1.9 Consequently, the same range of model compounds was chosen as used in ref 9, Le., square-planar complexes of Cu(I1) with imidazole (Cu(Im)4), pyridine (C U ( P ~ ) ~glycine ), (Cu(gl)J, ethylenediamine (Cu(et),), and ammonia (Cu(NH,),). In addition, diethylenetriamineimidazole (Cu(Det1)) was included as potential model for SOD. The chemical structures of the complexes are shown in Scheme I. 2. Experimental Section The following list gives the suppliers and the compounds used: Merck ammonia (solution, 25%); ethylene glycol; ethylene glycol-d, (99% deuteration); D 2 0 (99.8%); disodium hydrogen phosphate; ethylenediamine; glycerol (free of water); glycine; potassium hydrogen phosphate; copper (metallic powder); copper(11) nitrate trihydrate; copper(I1) sulfate pentahydrate; sodium nitrate; pyridine-d5 (99.8% deuteration); H N 0 3 (65%). Sigma: glycine-d5(98% deuteration); copper(I1) acetate monohydrate; imidazole; D N 0 3 (68% in D 2 0 (98%)). Riedel-de-Haen: pyridine. Medgenix: copper-63 (metallic powder, 97% enrichment). IC Chemicals: imidazole-d4 (98.8%); pyridineJ5N. CEA, France: imidazole-I5N2. The substances were used as obtained. The diethylenetriamineimidazole sample was a gift from L. Banci (University of Florence). For preparation of complexes containing the natural Cu isotopic composition (69.2% 63Cu/30.8%65Cu)the Cu(1I) salts like nitrate or acetate in crystalline form were dissolved in a 60% v/v glycerol or ethylene glycol mixture with water, to produce a 10 mM concentration in Cu(I1). The mixture contained either 0.4 M pyridine, ammonia, and imidazole, respectively, or 21 mM ethylenediamine or glycine, following the recipes given previo u ~ l y . ~After , ~ , ~complete dissolving of the salt, the pH was adjusted by adding KOH and HCI to give values of 10.0 for ammonia, ethylenediamine,and glycine. For imidazole the pH was set to 8.0, for pyridine to 7.0. Sample volumes of about 0.15 mL for X-band and 0.05 mL for Q-band ESR spectroscopy were transferred to the respective sample tubes and were frozen immediately to 77 K. For room temperature measurements, freshly prepared samples (0.15 mL) were filled into a flat cell and measured directly.

,

10 mT I

w Figure 1. ESR spectra (X-band, first derivative) of Cu(I1) complexes with ammonia (first trace, 63Cu/65Cu; second trace,Wu), diethylene-

triamineimidazole (third trace, 63Cu/6sCu),ethylenediamine (fourth trace, 63Cu/65Cu;fifth trace, 63Cu),and glycine (63Cu/65Cu and W u , respectively) at 77 K in glycerol/water. Samples containing enriched 63Cuwere prepared for metallic powder dissolved in 65% H N 0 3 such as to achieve 0.08 M concentration in copper. From this about 50 pL was added to the solution containing the ligand, to yield about 0.4 mL of the sample after pH adjustment. Deuterated specimens were p r o d u d by the respective preparation employing the deuterated compounds. ESR data acquisitions at 77 K on a Bruker ESP 380 instrument (X-band, -9.5 GHz) and on a Q-band (-34 GHz) ER 220 spectrometer as well as spectra analysis using Atari MEGA ST computers and the HOMER program package was performed as described recently.1o At X-band, the magnetic field was measured with a NMR gaussmeter (Bruker ER035M) and the microwave frequency was measured with a Hewlett-Packard counter 5350 B. ESR simulation involved the first-order algorithm contained in the HOMER package as well as the second-order programs SIM 14 (QCPE 265) and MSPOWD, a modified version of the former Magnspec 3 (QCPE 150). ENDOR spectra were taken on a Bruker ER 420 ESR spectrometer at about 5 K using the home-built ENDOR setup described previously.lOJl Simulation of 'H interactions in ENDOR following the principles outlined by Henderson6was performed using the HEROS program', also employed recently.1°

3. Results 3.1. ESR. Figure 1 gives the X-band ESR spectra of Cu(I1) complexes with ammonia (first two traces), with diethylenetriamineimidazole (third recording), with ethylenediamine,and with glycine (last four traces), respectively. Where two spectra are shown for one substance, the lower one is for the pure 63Cuisotope and the upper one is for the natural composition. The spectra have a close interrelation in that they all show the typical pattern of square-planar complexes in which three of the expected four hyperfine transitions in the parallel region (low-field range) are visible while the fourth component is masked by overlap with A,

Scholl and Htittermann

9686 The Journal of Physical Chemistry, Vol. 96, No. 24, 1992

I

,Al

1

reconstructed

*

I

I\

J

Figure 3. Second derivative ESR spectra of the m, = +3/2 transition for Cu-tetrapyridine employing 63/65 mixed isotopes (top), 63Cuonly (second trace), 63Cutogether with pyridineJ5N (third trace) as well as reconstruction of top spectrum.

of copper-tetrapyridine (top three traces) and copper-tetraimidazole (bottom three traces) giving for each the influence. of exchange of 63Cu/65Cu (top) against 63Cu (middle) and I4N against "N together with 63copper (bottom). The letters on the spectra reflect the ENDOR field positions. Figure 2. ESR spectra (X-band, first derivative)

"Cu let-dg)

features and with "extra absorption" lines" toward the high-field end of the spectra. It is apparent that the resolution of the g , / A , part is influenced by the isotopic composition. Also, the line width of the A,, transitions b m a smaller when 63Cuonly is used. The situation is similar for pyridine and imidazole as ligands for which the corresponding spectra are shown in Figure 2 (upper and lower three traces, respectively). Even more simplification of spectra is achieved when 63Cu is combined with 15Nas a ligand. This is shown for pyridine in the third trace and for imidazole in the bottom spectrum of Figure 2. It is of interest to analyze the outermost All line (mi = +3/2) for the equivalence of the interacting nitrogen nuclei. Figure 3 gives the componding experimental spectra in second-derivative display for resolution enhancement of the mixed isotope and the 63Cusystem, as well as for the 63Cusystem together with 15Nsubstituted pyridine. For the last combination it is quite apparent that the 1:46:41 quintet line group derives from four equivalent ISNnuclei each having nuclear spin I = 1/2. The transfer to the I4Nnucleus (I= 1) requires nine lines for equivalent nuclei which is in fact observed. Finally, the 11 lines in the sample with 63Cu/65Cuare the result of the different nuclear g factors of the two isotopes which, for the large A,,values of copper brings about sizeable shifts. The reconstruction of two nine-line trees for each of the isotopes (taking into account this difference and the ratio of isotopic weights) clearly supports this picture, as shown in the bottom spectrum of Figure 3. Similar treatments were performed with ethylenediamineand imidazole as ligands, which all show equivalence of the nitrogen interactions. Other parameters of interest in the ESR spectra are the values of g, as well as of A, for the copper and for the nitrogen interaction. As mentioned before, this is usually a problem at X-band frequency, since the combination of parallel and perpendicular parameters leads to strong overlap of the g, region with the m, = -312 transition of

& Figure 4. Q-band (34 GHz) ESR spectra (first derivative) of the Cu(I1) complexes under study; lower trams show a blow-up of the g,, region

(W. the parallel part and to "extra absorption" lines. This makes a safe parameter extraction nearly impossible. A partial solution is measurement at Qband to determine the perpendicular copper parameters. Figure 4 shows the results for the compounds under study. Apparently, the assumption of axial symmetry derived from X-band holds for most of the complexes. Ethylenediamine and diethylenetriamine-imidazoleshow distortions in the perpendicular region which could result from either rhombicity in g or from a single g, together with an A, value of copper larger than the line width. In a series of simulations using SIM 14 and MSPOWD

The Journal of Physical Chemistry, Vol. 96, No. 24, 1992 9687

ESR and ENDOR of Cu(I1) Complexes

TABLE I: ESR Parameters of Copper Model Complexes As Determined Experimentally (First Row) Together with Parameters Derived from Second-Order Simulation at X-Band (SX) and at Q-Band Frequencies (SQ)" gll

%1(1m),

SQ

sx sx

63cu(PY)4

SQ

sx sx

"CU(NH~)~

SQ

sx

W~(et)~

SQ

sx sx

63/65C~(DetI)

SQ

sx

63cu(g1)2

SQ

sx

2.262 2.262 2.262 2.262 2.263 2.267 2.263 2.263 2.241 2.241 2.241 2.206 2.205 2.205 2.205 2.212 2.21 1 2.21 1 2.267 2.267 2.267

c u All

g1

17.4 17.8 17.8 17.8 17.5 17.8 17.8 17.8 18.4 18.7 18.7 19.5 20.0 20.0 20.0 19.1 19.0 19.0 16.3 16.7 16.7

2.047 2.047 2.05 1 2.053 2.053 2.053 2.047 2.047 2.041 2.04 I 2.041 2.040 2.040 2.052 2.052

Cu A,

N A, 1.17 (1.88)

2.2 2.2 2.2 1.7 1.7 1.7

N Ail 1.48 (2.14)

(1.88) 1.17 1.05 (1.51)

(2.14) 1.48 1.4 (1.91)

1.05 (1.51)

1.4 (1.91) 1.33

0.89

1.18

0.9

1.2

2.4 2.4

(21.4)

2.9 2.9 2.9 (3.1)

(20.3) (20.3)

2.9 (3.1) 2.9 (3.1)

(17.9)

1.5 (1.6) 1.5

"HFvalues in mT; (I5N A values); (65CuA values). Error limits:

gll, Ag = 0.0008; Cu(A),

AH = 0.2 mT; N(A), AH = 0.05 mT.

exp.

sim.

Figure 5. Experimental (upper) and simulated (lower) spectra (second derivative) emphasizing the g, part of the Q-band (34 GHz) patterns of Cu-tetrapyridine (top half) and diethylenediamine showing influence of A, in the latter sample.

the influence of the two possible factors was studied comparing tetrapyridine and ethylenediamine. Both programs give essentially identical results. Figure 5 shows the comparison of the experimental together with the simulated spectrum for tetrapyridine (top two spectra) using, in the simulation, the optimized g, A, and line width tensors. A similar match in line shapes in ethylenediamine can only be obtained by keeping the axial symmetry of the complex for both g and A but letting A, become as large as 2.9 mT (bottom two recordings). Using two different g factors gives unrealistic line shapes. Application of these parameters to the corresponding spectral part at X-band frequencies leads to an estimate of the All value for the nitrogen interaction in those cases for which the resolution is sufficiently well developed. This condition is fulfilled, for example, in the case of tetrapyridine with IsN substitution. The upper part of Figure 6 gives the experimental spectrum together with a second-order simulation. The agreement is convincing evidence for the correct assignment of the IsN hyperfine coupling for the perpendicular region. The bottom part of Figure 6 deals with the situation for diethylenediamine for which only some indications for 14N lines are seen

,

10 m T i )

Figure 6. Total spectrum simulation (X-band) for tetrapyridine (top half, first derivative) and diethylenediamine (bottom half) and comparison with experimental spectrum for I5N (top) and I4N ligands (bottom).

in the perpendicular region of the mixed-isotope specimen, which are not sufficient to be included in the simulation. For comparison with a simulation (bottom spectrum), the experimental pattern was therefore convoluted with a Gaussian line shape function of 0.7-mT width (second trace) which yields good agreement. The ESR spectra parameters of all compounds under study are listed in Table I; the monoisotopic composition (63Cu) is given when available, as the resolution was best for this composition. All spectra could be simulated with a high degree of agreement with experimental recordings using one data set only for X- and Q-bands. The only adjustable parameter used was an axial line width tensor. This, however, has to be different between X-and Q-bands anyway since resolution of subsplittings in Q-band is lost experimentally. Therefore, no attempt to estimate g and A distributions ('strain") was made. Control measurements of the isotropic coupling at 300 K in solution gave values which, when applied to calculate the g, values, showed very good agreement with those derived from simulation. The quadrupolar interaction

9688 The Journal of Physical Chemistry, Vol. 96, No. 24, 1992

Scholl and HUttermann 63c" lPYlq

u

23 7 RHz

field h

Figure 7. ENDOR spectra (10-20 MHz, first derivative) of "weakly coupled" proton and of nitrogen interactions in Cu(I1)-tetrapyridine demonstrating effects of microwave frequency shift (top two traces), deuterated solvent (third spectrum), and perdeuterated pyridine for natural copper isotope mixture and for 63Cuonly (bottom two recordings). Microwave frequency is 9.51 GHz except for second spectrum (9.22 GHz). Field position (cf. Figure 2): h.

term of copper was not included in the simulation. Tests using SIM14 showed that its influence is very small in agreement with the expectation for square-planar comple~es.'~ 3.2. ENDOR. It is one of the characteristics of the complexes studied that the ENDOR resonanca of the coordinated nitrogens overlap, at X-band ESR frequencies, with the corresponding 'weakly coupled" proton resonances (10-20 MHz), centered around Y,, the nuclear Zeeman frequency (14 MHz at g 2). Both types of spectra require, however, different recording conditions due to their different line widths (ca.100 lcHz vs 1 MHz). Thus, in frequency modulation (FM) mode, the amplitude of the frequency modulation must be small to preserve resolution for proton lines but becomes large for optimizing the nitrogen pattern intensities. When amplitude modulation (AM) is used instead, the resolution often is reduced considerably so that quadrupolar interaction is not detected.*s9 We therefore prefer to present FM-modulated spectra. Figure 7 presents a series of ENDOR spectra for Cu(I1)tetrapyridine along the working point h indicated in Figure 2. The top two spectra are for the natural isotopic mixture taken at two microwave frequencies to show the frequency shift of the two outermost proton lines with the nuclear Zeeman terms, indicating their origin from proton interactions. The high-frequency branch of this proton pair clearly overlaps with some broad lines. Their origin as due to I4N interactions is verified in the bottom two spectra in which perdeuterated pyridine is used to clear the "proton window". The lack of change in spectra with changing copper isotopes leaves no other pcmibility than an assignment to 14N. On first glance the pattern appears to comprise a doublet only centered at about 18.6 MHz, but close inspection shows that its contribution is still extending into the free proton range at 14.6 MHz. Figure 8 gives the features for the parallel region (top two recordings) in which the corresponding 14Ninteraction yields a

Figure 8. ENDOR spectra (first derivative) of Cu(I1)-tetrapyridine showing 14N couplings at field position a (cf. Figure 2) (top two recordings) and I5Ninteractions in substituted pyridine at both g! extreme (a) and at field position h. Noiseless spectra are simulations (see text). (*: proton line.)

triplet pattern due to a quadrupolar interaction having twice the value of the nuclear h m a n term. Again, the use of perdeuterated pyridine clears the overlap between proton and nitrogen lines aiding the assignment of the latter. Another proof comes from the 15N-substitutedpyridine for which a doublet due to I = 1/2 should be observed at any of the working points. Although we did not expect the lines to be even broader than those of the I4N isotope, the corresponding spectra shown in the bottom two recordings of Figure 8 unmistakably derive from the ISN interactions, their coupling values reflecting precisely the difference in magnetic moments of the two nuclei. The simulations also match the experimental spectra. The corresponding results for tetraimidazole are displayed in Figure 9. The top two spectra give the range of 14Ninteractions and the extent of their overlap with the proton resonances by comparing, at the high-field position h, the influence of ligand perdeuteration with the protonated imidazole, both measured in a deuterated solvent. The bottom part of the figure emphasizes the 14Ninteractions at both fields c and h giving a triplet centered around 19.1 MHz at the former, near single-crystal type position. The peculiar quartet pattern at field h is a superpositionof parallel and perpendicular features. Both experimental curves are reproduced with good agreement by simulations. Table I1 lists the parameters of the I4N interactions. No data are given for diethylenetriamineimidazole, since no discernible lines were detected perhaps due to strong overlap with protons together with low intensity. Reference is made to the work of Van Camp et a1.4 and of Yokoi? The ratio R = AII/A, introduced by this latter author in order to distinguish imidazole from other nitrogen coordination is also given. Other data referenced are from the work of I w a i z ~ m iin , ~ which the A, values only were measured. When these data are compared with those of the ESR (Table I), the necessary g correction should be applied. Proton ENDOR interactions were emphasized for pyridine and imidazole coordination since a major background of this study relates to the identification of ENDOR in SOD.' The different spectra in Figure 7 for tetrapyridine establish first of all, by comparing the influence of the perdeuterated solvent (third re-

The Journal of Physical Chemistry, Vol. 96, No. 24, 1992 9689

ESR and ENDOR of Cu(I1) Complexes

*

I 14.5 MHz

1 1

,

*Hi+ 1 2 ’ 5 YHz Figure 10. Experimental and simulated proton interactions for field positions g and e in Cu(Im), in deuterated solvent in which NH interaction is missing. 14 1 E ’

*

I 19.1 Mz

I 20.3 Mz

Figure 9. ENDOR spectra (first derivative, 10-20 MHz) showing range and shape of proton and nitrogen interactions separated by perdeuteration in Cu(I1)-tetraimidazole (top two recordings) for field position h (cf. Figure 2). Bottom four spectra experimental and simulated nitrogen interactions for field positions c and h. TABLE II: Nitrogen Hyperhe Couplinep from ENDOR ligand AN, QY A; QY R

Im-I4N Im-ISN

Py-I4N Py-ISN

NH3 et gl

38.2 40.3 39.8 53.5 33.8 34.8 47.4 30.4 31.7 26.8 27.6 27 28.4 29.3

0.7

0.0 0.5 0.0

40.6 42.3 41.6 56.8 37.4 41.3 52.4 37.2 39.1

-1.4

0.0 -1.0 0.0

1.06 1.05 1.05 1.06 1.11 1.19 1.11 1.22 1.23

a e

b c e e

b e

b

39.4

1.43

b d

35.8

1.22

b

“All coupling values in MHz; R = A ; / A y . bMedium H20; ref 9. Medium H20/glycerol; ref 4. Medium H,O/glycerol; ref 5 . cParameters used in simulation; this work.

cording), that all proton lines observed should come from the ligand rather than from interacting solvent protons. With this provision, the interactions of the ortho, meta, and para protons can be suggested as indicated by the letters A, B, (C), and D. A comparison with simulations is not given since the crystal structure is not available to our knowledge. For imidazole the two top spectra of Figure 9, which give the situation along the g, extreme ESR point h, have been obtained under conditions which slightly overmodulate the proton lines in order to enhance the 14Ninteractions. Even so it becomes apparent that, as for pyridine, the resonances observed in the protonated imidazole are all from the ligand since the solvent used was deuterated. Apart from this, the situation is more complex. The top spectrum of Figure 9 indicates that the intensities of proton lines show asymmetric relations between high- and low-frequency branches of the resonances. This asymmetrical behavior becomes stronger at field positions closer to g,,. Figure 10 gives two experimental spectra (top) for positions g (left) and e (right) which were optimized for resolution in the proton region. Apparently, for the latter, two low-frequency branch lines are missing in the

,

: > 114.4 MHz Figure 11. Experimental spectra for Cu(Im), (top) and perdeuterated Cu(Im-d,), (middle) both in H20solvent and simulation (bottom) of N, proton spectrum at field position g. 2 MHz

experimental spectrum whereas in the spectrum from g only the intensity increases with increasing frequency. The simulations for both field positions given beneath both curves cannot reflect the intensity difference which probably arises from the nuclear relaxation Tln15but otherwise show a fair agreement with the experimental line positions. The data given in Figure 10 did not contain the exchangeable proton at N1since the solvent used was deuterated. The proton can be studied selectively by comparing the spectra from Cu(Im), in protonated solvent with those from perdeuterated imidazole in protonated solvent. The results are shown in Figure 11 together with a respective simulation of the proton in question. It becomes apparent that again there is a strong intensity asymmetry, specifically for this proton. Also, the simulations are seen to reproduce at least the total extent of the spectrum of this proton reliably well. The parameters used in the simulations of the proton ENDOR spectra are given in Table 111. The relevant input parameters are distances from the copper center to the protons, the copper point spin density, and the isotropic couplings deriving from delocalized spin density to the imidazole ring positions. The assignment of specific protons to resonance lines in the spectra can be visualized from the chemical structures given in Scheme 11. 4. Discussion

One of the goals of the present work was to obtain sound ESR parameters of the compounds. This is, among others, a prerequisite for the application of ENDOR simulation routines. Specifically, the enormous influence of distances and metal center as well as ligand spin densities on the simulated proton spectra requires a high degree of reliability of ESR parameters in order to enable

9690 The Journal of Physical Chemistry, Vol. 96, No. 24, I992 TABLE III: Parameters Used for Proton ENDOR Simulation in

WImh HC2 Im 1 Im 2 Im 3 Im 4

(a) Distances r (Cu/Protons):” Irl (A) H N l Id (A) HC5 3.12 Im 1 5.05 Im 1 Im 2 3.25 5.12 Im 2 3.22 Im 3 5.12 Im 3 3.18 Im 4 5.11 Im 4

d C u ) = 0.72 Irl (A) HC4 5.28 Im 1 5.15 Im 2 5.14 Im 3 5.16 Im 4

Irl (A) 3.31 3.30 3.27 3.24

(b) Isotropic Coupling Ai, type Aim (MHz)

HC2 0.8

HNl 1.8

HC5 1.7

HC4 1.8

(c) Distances r(Cu/N,,)”

“ From ref

16.

SCHEME I1 Im 3 Im 2

Py

I

- Cu - Im 4

Py-cu

-Py

I

I

cfc3\

BB H

H

cc x x DD EE

YY

I

BB

I n

16S.i

comparison with experimental results at corresponding field positions in the experimental powder ESR spectrum. Having secured this way that all complexes are strictly axially symmetrical, we can now apply the well-known empirical relations for classifying coordination by means of glland Allfor the copper together with their rati0.”3~~For the latter, values (recalculating All into cm-I) of 120 (imidazole, pyridine), 115 (NH,), 113 (diethylenetriamineimidazole), 107 (ethylenediamine), and 128 cm (glycine) are obtained. All are in the range expected for square-planar complexes. Examination of the individual parameters finds for all CuN, complexes the expected drop in g and corresponding increase in All when going from strictly spl-hybridized ligands (imidazole and pyridine) to sp3-ligated nitrogens like NH3 and amines. The only N202-typeligation, that with glycine, again behaves as expected in the replacement of two nitrogens with less electron rich oxygens increases again the gl, and decreases All, respectively. The values of the I4N interactions can be used to derive the characteristic parameter n2 of hybridization following the expressions given, e.g., in refs 19 and 20 based on McGarvey’s Taking the ENDOR values for A,Iand A, for the I4N interactions from Table 11, we find 0.64 for imidazole, 0.68 for pyridine, and 0.77 for ammonia, each of the values falling close to the theoretically expected ones of 0.66 for spz and 0.75 for sp3 hybridization. The derivation of the nitrogen s and p spin densities using the average values for unit spin of ref 20 gives values of total spin densities of 6% in imidazole (3.8% p; 2.2% s), 6.1% in pyridine (4.7% p; 1.9% s), and 7.8% in ammonia (6% p; 1.8% s), respectively. In the simulationsof the proton spectra of the imidazole compound we used 7% delocalized in the nitrogens. For the complementary spin density on the copper, the simplified MO calculuszogives for all three compounds the value of a2 = 0.77. The I4N interactions obtained in this work by ENDOR compare well, in general terms, with those reported earlier.4+s+8 Iwaiwmi and colleagues discuss a tendency for the Allvalues to drop in the order N4 to N202,comparable to the behavior of the Cu A,,values discussed above. In addition, the hybridization state was considered to control the coupling value giving planar amide and aromatic nitrogens larger values than nitrogens with tetrahedral conformation^.^ Except for glycine, all compounds measured belong to the CuN,-type coordination. Clearly, in the Cu-N4

Scholl and Htittermann group, the planar nitrogens of imidazole and pyridine comply with the expectation since their values are significantly larger than those of the two tetrahedral ligands ammonia and ethylenediamine. Also, there is a clear drop in values toward N2OZcoordination (glycine). Yokoi did not employ the absolute couplings but rather the ratio R = AII/A, in order to distinguish imidazole coordination from that of other nitrogens; values of R I 1.10 were considered as indicative for imidazole coordination? The quantitative analysis of the data presented in this work, however, suggests a different border. Our ratio R is about the same for imidazole but considerably smaller (1.1 1 vs 1.19) for pyridine. Specifically, the All value for pyridine differs by about 4 MHz yielding the reduction in R. We would therefore rather draw a division line between R values of about 1.10 and those of 1.20 and above, thus placing pyridine and imidazole into the same category. Then, nitrogens with planar hybridization as in pyridine and imidazole have different R values from those with tetrahedral hybridization, indicating a connection between the R value and the corresponding classification proposed by Iwaizumi on account of the observed A ,-coupling value.5 We mention that extracting a reliable value for the All interaction, as is needed to calculate the R value, is always a problem, even in ENDOR. Using AM modulation, as was done by Yokoi? smears most resolution and makes this task even more difficult. In diethylenetriamineimidazolewe could not detect very well the 14Nresonances, let alone the difference between imidazole and amine couplings. This finding may be related to different couplings for the different nitrogen types, which could wash out the resolution by strong overlap of broad lines. In the context of SOD, the analysis of the nitrogen ENDOR couplings in tetraimidazole gives no clue to the curious behavior of the respective interactions at the high-field positions (g,) in the protein.8 The model compound shows completely “normal” line shapes and spacings. The same holds for tetrapyridine. Both have equivalent couplings of approximately the same size which is also seen in ESR. We note that tetrahistidine coordination also gives comparable, equivalent ESR-detected couplings.22 Thus we have to conclude that it is the rhombic symmetry of the complex in the protein which gives rise to the observed ENDOR effects at the extreme high-field part of the ESR spectrum.a Considering the proton couplings, we note that there is a surprisingly large comparability between the tetraimidazole results and SOD. For one thing, the conspicuous asymmetry in intensity of the low- and high-frequency branches is reproduced by the imidazole but not the pyridine ligand. Furthermore, the resonances produced by the N1 (exchangeable) and C5 as well as those of the two close protons, C2and C4of the tetraimidazole, agree very well with the lines from corresponding protons in the protein when considering the “in-plane” ENDOR working position. One can at the Cu ion as origin. overlay the two crystal s t r u c t ~ r e s ’fured ~*~~ In comparison to the strictly symmetrical tetraimidazole, which has all ring normals in the plane through copper and the four nitrogens, two of the histidines of SOD (44and 46) can be made approximately coplanar with the imidazoles, whereas the other two (61 and 118) then have their ring normals nearly under 90’ to those of the other pair. This means that the correlation between, for example, HE1 protons of SOD and C2 protons of tetraimidazob, holds only for His-61 and His-1 18. This feature is nicely reproduced by simulations. Similar considerations apply for the agreement at the high-field position between the N, proton and the exchangeable proton in SOD, which then should involve His-1 18 (HD1). This will be detailed in a future publicati~n.~~ For the present it suffices to conclude that C ~ ( 1 m is, ) ~ when corrected for comparable geometry, an excellent model compound for the proton range, whereas pyridine as a ligand gives very different results. This is surprising, since the two complexes hardly differ in any other of the parameters, in ESR and ENDOR, except for the size of the proton couplings. Acknowledgment. This work was supported by a grant from the Deutsche Forschungsgemeinschaft. The valuable assistence

J. Phys. Chem. 1992,96,9691-9696 of G. P. Diiges, H. Reinhard, and J. Ohlmann in providing for simulation programs is gratefully acknowledged. We thank Prof. Dr. L. Banci (Florence) for the gift of a sample of diethylenetriamineimidazole.

References and Notes (1) Solomon, E.I.; Penfield, K. W.; Wilcox, D. E. Srrucr. Bonding 1983, 53, 1. (2) Solomon, E. I.; Gewirth, A. A.; Westmoreland, T. D. In Aduanced EPR; Hoff, A. J., Ed.; Elsevier: Amsterdam, 1989; pp 865-908. (3) Mims, W. B.; Peisach, J. J . Chem. Phys. 1978, 69, 4921. (4) Van Camp, H. L.; Sands, R. H.; Fee, J. A. J . Chem. Phys. 1981, 75, 2098. (5) Iwaizumi, M.; Kudo, T.; Kita, S.Inorg. Chem. 1986, 25, 1546. (6) Henderson, T. A.; Hurst, G. C.; Kreilick, R. W. J . Am. Chem. SOC. 1985. 107. 7299. (7) Hkermann, J.; Kappl, R.; Banci, L.; Bertini, I. Biochim. Biophys. Acra 1988, 956, 173. (8) Van Camp, H. L.; Sands, R. H.; Fee, J. A. Biochim. Biophys. Acra 1982, 704, 75. (9) Yokoi, H. Biochem. Biophys. Res. Commun. 1982, 108, 1278. (IO) DBges, G. P.; Hiittermann, J. J . Phys. Chem. 1992, 96, 4787.

9691

(1 1) Hiittermann. J.; Kappl, R. In Metal ions in Biological Sysrem; Sigel, H.1 Ed.;Marcel Dekker: New York, 1987, Val. 22, pp 1-80. (12) Reinhard, H. Diploma-Thesis. University of Saarland, FRG, 1992. (13) Ovchinnikov, I. V.; Konstantinov, V. N. J . Magn. Reson. 1978, 32, 179. (14) Belford, R. L.; Duan, D. C. J . Magn. Reson. 1978, 29, 293. (15) Kevan, L.; Kispert, L. D. Elecrron Spin Double Resonance; Wiley: New York, 1976; p 18. (16) McFadden, D. L.; McPhail, A. T.; Garner, C. D.; Mabbs, F. E. J . Chem. Soc., Dalron Trans. 1976, 47. (17) Peisach, J.; Blumberg, W. E.Arch. Biochem. Biophys. 1974,165,691. (18) Sakaguchi, U.; Addison, A. W. J . Chem. SOC.,Dalron Trans. 1979, 600. (19) Basosi, R. J. Phys. Chem. 1988, 92, 992. (20) Yordanov, N. D.; Stankova, M.; Shopov, D. Chem. Phys. Lerr. 1976, 39, 174. (21) McGarvey, B. R. J . Phys. Chem. 1967, 71, 51. (22) Basosi, R.; Valensin, G.; Gaggelli, E.;Froncisz, W.; PasenkiewiczGierula, M.; Antholine, W. E.; Hyde, J. S . Inorg. Chem. 1986, 25, 3006. (23) Tainer, J. A.; Getzoff, E.D.; Beem, K. M.; Richardson, J. S.;Richardson, D. C. J. Mol. Biol. 1982, 160, 181. (24).Reinhard, H.; Kappl, R.; Hiittermann, J.; Bertini, I.; Banci, L.; Viezzoh, M. S. Manuscript in preparation.

Use of Rotated Jacobi Coordinates to Calculate Vibrational Levels of HCN Jo& Ziiiiiga,* Adolfo Bastida, and Albert0 Requena* Departamento de Qujmica Fisica, Universidad de Murcia, 301 00 Murcia, Spain (Received: April 1 , 1992)

A study is made of the use of rotated Jacobi coordinate systems to compute vibrational energy levels of HCN. The coordinate systems are derived by making an orthogonal rotation of the radial Jacobi coordinates, and the three possible arrangements of the Jacobi variables are comparatively considered. The vibrational self-consistent-field (VSCF) method is used to check the quality of the different coordinate systems. It is found that the Jacobi coordinates associated with the N-CH arrangement and rotated to transform into a system of curvilinear normal coordinates provide the best SCF energies and are the most useful to calculate the exact vibrational energy levels of HCN by the variational configuration interaction (CI) method.

I. Introduction In recent years, accurate theoretical determination of highly excited vibrational states of small polyatomics has become an area of central interest due essentially to the great advances in experimental techniques to probe such states.l-' In the high-energy regime, the vibrational motions are very anharmonic and coupled, and solution of the vibrational Schrodinger equation turns into a difficult task, even for triatomics. As a consequence, a variety of approximate and variational methods have been developed to deal with this problem?-'8 Part of this theoretical research has been focused on finding optimal sets of vibrational coordinates in which to represent the Hamiltonian and the wavefunctions. The reason for that is clear. Optimal coordinate systems will lead to small couplings between the vibrational modes, thus allowing not only to simplify the calculation of the vibrational states but also to obtain a better physical insight into the vibrational behavior of the molecule. In this context, different coordinate systems, including as examples local, Jacobi, hyperspherical, ellipsoidal, etc., have been studied, as well as techniques to improve their quality. 10~13,17-33 Jacobi coordinates and different variants of them have been successfully used to describe large-amplitude vibrational motions in triatomics along the last d e c a d ~ . ~ JThese ~ J ~ coordinates are naturally defined for treating atom-molecule collisions and their use is, in principle, well indicated for describing atom-diatom complexes and for isomerization ~ y s t e m s . ~Reently, J~ however, Jacobi coordinates have been shown to be also useful in typical bound state calculations.20~21~28~29~34~36 In fact, for triatomic molecules of C, and D3h symmetry and for those with linear equilibrium geometries, these coordinates can be transformed into

curvilinear normal coordinates by making a proper rotation of the Jacobi radial distance^.^^^^^.^^ Jacobi coordinates have also the advantage of being orthogonal; Le., the kinetic energy operator for them is diagonal. In this paper, we present a study of the quality of different rotated Jacobi coordinate systems to calculate highly excited vibrational energy levels of HCN. Concretely, we consider the systems associated with the three possible arrangements of the Jacobi variables. We usc the vibrational self-consistent-field (SCF') approa~h'~.~',~* to assess the suitability of the rotated systems in describing the vibrational motions of the molecule. We find that, for each Jacobi arrangement, the optimum value for the rotation angle is that for which the potential energy function is totally decoupled up to second order, so that the rotated system can be identified as a curvilinear normal system. Moreover, the best SCF energies are those obtained by using the rotated coordinate system corresponing to the N-CH Jacobi arrangement. Consequently, this coordinate system is shown to be clearly superior to the others in computing exact vibrational energies for HCN using the variational configuration interaction (CI) method. In section I1 we discuss the rotated Jacobi coordinate systems and the SCF and CI methods, and section I11 presents the results obtained and the analysis thereof. 11. Theory The internal Jacobi coordinates R , r, and 0 for a triatomic molecule A-BC are defined as the distance of atom A to the center of mass of BC (R), the length of BC diatom bond ( r ) , and the angle between rand R (e). The vibrational Hamiltonian in these coordinates can be written asz8

0022-3654/92/2096-969 1$03.00/0 0 1992 American Chemical Society