вход по аккаунту


How N2 Might Be Activated by the FeMo-Cofactor in Nitrogenase.

код для вставкиСкачать
[l] a) H. 0. House. Modern Synthetic Rracrions, 2nd ed., W. A. Benjamin,
Menlo Park, 1972. p. 492; b) J. March, Advanced Organic Chemistry, 4th
ed.. Wiley. New York, 1992.
[2] Reviews: a) D. Caine. Alkylutiuns of Ends and Enolures. in Comprehensive
Orgunic Synthesis. Vol. 3. (Eds.: B. M. Trost. I. Fleming, G. Pattersen).
Pergamon, Oxford, 1991. p. 54, b) C. M. Thompson, D. L. C. Green. Titruhvdrun 1991, 47,4223; .*,fi-unsaturated 2-alkoxy ketones: c) G. Stork.
R. C. Danhelser, J. Org. Chem. 1973.33.1775: d) G. Stork, R. L. Ddnheiser. B. Ganem, J. Am. Chem. Soc. 1973,95,3414; 1,B-unsdturated~-dialkylamino ketones: e) T. A. Bryson, R. B. Gammill. Etruhedrori Lrri. 1974.
3963; f) E. Telshow, W. Reusch, J. Org. Chem. 1975. 40, 862; [(-enamino
ketones: g) G. Bartoh. M. Bosco. C. Cimarelli, R. Dalpozzo, G. Palmieri,
J. Chem. Soc., Perkin Truns. 1 1992, 2095; h) G. Bartoli, C. Cimarelli. G.
Palmieri, G. Rafaiani. Tvrruhedrun: A s p m e t r y 1992, 3, 719.
131 a)Y. L. Chen, P. S . Mariano, G. M. Little, D. OBrien, P. L. Huesmdnn,
J. Org. Chem. 1981.46,4643; theoretical calculations (HF/6-31 G/3-21G)
have been executed on mono- and dianions of acyclic fi-enamino ketones:
b j G . Bartoli, M. Bosco, C. Cimarelli. R. Dalpozzo, M. Guerra, G.
Palmieri. J. Churn. Suc. Perkin Truns. 2 1992, 649.
[4] Similar chelate intermediates have been invoked to explain the stereoselective reduction of ketones bearing a P(O)R, functionality in the fi position:
J. Elliott, S . Warren, Telruhedrun Lett. 1986. 27. 645.
[5] For comprehensive reviews on this topic see: a ) G. A. Molander, Chem.
Rev. 1992,92,29; b) H. B. Kagdn, J. L. Namy, T&uhedron 1986,42,6573;
for a leading paper on organocerium addition to carbonyl substrates see:
c) T. Imamoto, N. Takiydma, K. Ndkamura, T. Hatajimd, Y Kamiya, J.
Am. Chem. Suc. 1989, 111,4392.
[6] These reagents can be efficiently prepared simply by adding the corresponding Grignard reagent to the CeCI, suspension in THF [Sc].
[7] It is known that tris(cyclopentadieny1)ytterbium is able to deprotonate
2.4-pentadione giving a chelate complex: G. Bielang, R. D. Fischer, Inurg.
Chim. Actu 1979, 36. L389.
[El The low reactivity of secondary and tertiary Grignard reagents may be also
ascribed to the increase of basicity observed moving from primary to
tertiary organomagnesium reagents.
[9] Organolithium reagents d o no react analogously, probably because they are
more basic than Grignard reagents.
[lo] 2a: D. P. Curran. J. A m . Chem. Suc. 1983, 102, 5826; 2b: C. Najera, M.
Yus, D. Seebach, Helv. Chrm. Acra 1984, 67, 289; 2 c : S . Fukuzdwa, T.
Tsuruta. T. Fujinami, S . Sakai. J. Chem. Suc., Perkin Truns. I 1987, 1473;
2d: T. J. Leitereg, D . J. Cram, J. Am. Chem. Soc. 1968. 90, 4019; Ze: E.
Hasegawa, K. Ishiyama, T. Horaguchi, T. Shimizu, J. Org. Chem. 1991,56.
removing one of the S atoms of the latter. The six inner iron
atoms of the cofactor (Fe2-Fe7 see 2) form an approximate
trigonal prism. The coordination environment around Mo is
approximately octahedral and that around Fel tetrahedral.
Fe2-Fe7 have quite distorted trigonal-planar coordination.
Two of the three bridging ligands between the two clusters
are sulfide ions S 2 - , while the third ligand Y is probably a
nitrogen or oxygen atom, or a less ordered sulfur.[']
We first examine the electronic structure and the bonding
in the cofactor model.[21To simplify the analysis and yet not
lose essential chemical information we have replaced the uncertain Y atom with S 2 - , the ligating atoms from the cysteine, histidine, and homocitrate with the simplest ligands,
H-. Finally we symmetrize the structure to C,, point sym~netry.'~]
The bond lengths are set at 2.75 (Fe-Fe (all)), 2.85
(Fe-Mo), 2.3 (Fe-S), 2.35 (Mo-S), 1.6 (Fe-H), 1.70 A (MoH). In much previous work we have found it useful to set
distances equal and then allow the overlap populations
How N, Might Be Activated by the
(OPs) to tell us which bonds are stronger. This is why we
FeMo-Cofactor in Nitrogenase**
have equalized the Fe-Fe separations in 2.
What about the oxidation states of the metal centers in l ?
By Huibin Deng and Rould Hoffmunn*
Mo ENDOR and EXAFS studies indicate that the most
We finally have a structural model for the nitrogenase
likely formal oxidation state for Mo is +4;14] and
FeMo-cofactor, based on 2.7 8, resolution crystal structure
Mossbauer investigations show that Mo does not seem to
of the Azotobacter vinelandii MoFe-protein by J. Kim and
change oxidation state when the MoFe-protein is reduced.[51
D. C. Rees.['] Here we examine the electronic structure of
We assign the four iron atoms in the 4Fe: 3s cluster together
this model, and its possible N, binding modes.
a + I 0 charge, as in the cubane clusters [Fe,S,(SR),]2-.[61
The FeMo-cofactor (I) contains two clusters of composiThe three iron atoms in the 1Mo:3Fe:3S cluster are given a
tion 4Fe: 3s and IMo: 3Fe:3S, bridged by three non-protein
total $7 charge, to be in line with the spin state of the
ligands. The cofactor is linked to the protein through the Fel
FeMo-cofactor ( S = 3/2).["' Thus the actual molecule calcuand Mo metal atoms (see 2). The sulfur atom of the C ~ S " ~ ' ~lated, a model of a model, is [HFe4S,(p-S),Fe,S,MoH3]residue coordinates Fel, whereas the Mo atom is ligated by
the imidazole group of the Hisa442 residue, as well as a
The given charge assignment is by no means unique, or
homocitrate HC through hydroxyl and carboxyl oxygen
even correct. But we must start somewhere. Thorneleyatoms."] The 4Fe:3S and lMo:3Fe:3S clusters may be
Lowe kinetic models['] for dinitrogen reduction indicate that
structurally related to common 4Fe :4 s cubane clusters by
dinitrogen binds to more reduced forms of the cofactor.
Thus we will also examine the consequence of adding up to
[*I Prof. R. Hoffmann, Dr. H. Deng
three electrons to the cofactor model 2 (i.e., the reduction of
Department of Chemistry
all Fe atoms to Fe").
Cornell University
Ithaca, NY 14853-1301 (USA)
An energy level diagram for 2 is shown in Figure la. The
Telefax: Int. code + (607)255-5707
HOMO is a singly occupied orbital. There is another orbital
[**I We are grateful to the National Science Foundation for its support of this
only 0.03 eV above it; in fact there are nine orbitals within
work through CHE8912070. Dr. Douglas Rees was kind enough t o trdnshalf an eV around the HOMO. Thus the system should be a
mit to us his refinement of the MoFe-protein structure. We are also gratehigh-spin one. The electron coupling in the cofactor 1 is
ful to him for discussions of the N, binding.
Verla~sgesrllschqftmbH. 0-69451 Weinheim. 1993
$ 10 00+ .2SjO
Angew. Chem. Inr. Ed. Engl. 1993. 32, Nu. 7
7 Fe
-21 !
= ri
- 10
- 14
- 12
- 14
Fig. 1. a ) Energy level diagram for 2. Fe 4s/4p stands for hybrid orbitals, the striped block at the bottom represents ligand-based (S, H) orbitals. b) and c) “Density
of states’. (DOS) diagrams for 2; the dotted line IS the total DOS of 2, and the contributions of seven Fe atoms and the Mo atom to the total DOS are shown by the
shaded areas in b) and c). respectively.
likely to be complicated, so our results on 2 are not inconsistent with S = 3/2 for 1. We can also show the energy levels
in a “density of states” (DOS) fashion,[*] as in Figures 1 b
and 1 c. This is convenient when the contributions ofindividual atomic orbitals or fragment molecular orbitals are desired. Figures 1 b and 1 c illustrate such contributions from
the Fe and Mo atoms, respectively.
We can analyze these orbitals in some detail. The orbitals
below - 13.5 eV are mainly ligand-based (S and H) orbitals,
which are metal-ligand bonding. A block of 35 levels, mainly Fe 3d orbitals, is concentrated between - 13.5 and -9 eV.
The bottom of this block is Fe-Fe bonding, while the top
part is Fe-Fe antibonding. The HOMO is at about the center
of the Fe 3d block. The Mo 4d orbitals are higher in energy
than Fe 3d orbitals; the Mo 4d orbitals split into “eg” and
“t,,” sets, typical for an octahedral transition metal complex.Iy1The six empty levels (above -7 eV) are composed
mainly of combinations of 4s and 4p orbitals on three-coordinate Fe atoms. The resulting hybrids are approximately
perpendicular to the local coordination planes of these Fe
atoms. One such MO is drawn in Scheme 1. In these orbitals
resides the coordinative “unsaturation” of these Fe atoms
and their reactivity toward bases. The presence of these acceptor orbitals and the “open” cavity in the trigonal prism
Schemc 1 One of the 4s/4p hybrid orbitals of 2.
Angm Clwm. hi. E d Erg/.
1993.32, N o . 7
formed by Fe2-Fe7 have made people think about occuption of the cavity, even though there is no structural evidence
for such a species. The model 3 has the geometry in question;
when one does a calculation with a main group element in
the center, for example, 0 2 -(Fe-0 = 2.1 A), four of the six
Fe acceptor orbitals are indeed pushed up. Thus, there is
nothing wrong electronically with a central atom in 2.
If we now return to the symmetrized cofactor model 2, the
overlap populations (OPs) between three-coordinate Fe
atoms (e.g., Fe2-Fe3, Fe2-Fe5, Fe5-Fe6) are substantial
(0.14-0.15). This is actually equivalent to something like a
single bond. For instance, unbridged [Fe,(CO),]*- at the
same Fe-Fe distance has an Fe-Fe OP of 0.21. There definitely is metal-metal bonding everywhere in 2, and across
the trigonal prismatic cavity in particular.[’ O1 This is reflected by the shorter Fe2-Fe5, Fe3-Fe6, and Fe4-Fe7 distances (2.5-2.6 A) in the structurally refined cofactor model.
We assume that the first step in nitrogen fixation is the
binding of N, to the FeMo-cofactor.“ ‘I Various models with
different N, binding sites are shown in Scheme 2. Model 4
has also been suggested by Orme-Johnson.1’21We set all the
Fe-N or Mo-N distances at 2.0 A, and the N-N distance at
1.1 A, the value in free dinitrogen. We then compared the
overlap populations for the N-N bonds and the net charges
on the N atoms (Table 1) for various binding modes and for
two different redox states of the cofactor (the “reduced”
state has three more electrons than the “native” state). As
the N-N bond has to be weakened and protonation at N has
to occur before the coordinated N, can be reduced to NH,,
Sg VCH ~ r l ~ i g . ~ g e s e l l r c hmhH.
q f i D-6945/Weinheim.1993
$ 10.00+ .2S10
quence of the occupation of 7c: orbitals of N,. The contribution of this fragment MO to all the composite orbitals of 4
is shown in Figure 2. The n: orbital is occupied to the extent
of 20 %. The models 6 and 9 are also possible binding modes,
as the N-N bond is highly polarized and the terminal N atom
carries a substantial net negative charge.
E Ievl
Scheme 2. Theoretically studied complexes from N, and 2.
- 10 .............................
....,. .....
. ....:....+
. .......
,,,.::,.:... ...........
:I:;: ...
a small value of N-N OP and a net negative charge on the N
atoms should activate an N, molecule for reduction.
In models 4-6 the Y ligand is replaced by dinitrogen,
while in 7-9 dinitrogen approaches the cofactor at a face of
the Fe, trigonal prism. Models 7-9 give too close N-S contacts; we would like to reject them for that reason, but hesitate because the cofactor may expand its dimension upon
reduction or N, binding. In 10, the simpIe insertion of N,
into the Fe, trigonal prism gives rise to very short Fe-N
distances (ca. 1.8 A). To obtain an Fe-N separation of 2 A,
a more reasonable value, the Fe, trigonal prism has to be
expanded until the Fe-Fe distances, originally 2.75 A, reach
3 A. Table 1 includes data for both geometries of 10. Two
models involving end-on binding of N, are considered in 11
and 12. One of the ligands on Mo has to be dissociated in
order for Mo to bind N, in 12.
Table 1. Overlap populations (OPs) of N-N bonds and net atomic charges on
the N atoms for the N,-FeMo-cofactor models shown in Scheme 2.
“Native” State
OP(N-N) Charge on N [a]
Free N, 1.73
“Reduced” State
(three more electrons)
OP(N-N) Charge on N [a]
As shown in Table 1 , centered 10 gives the smallest OP for
the N-N bond; however, the net charges on N are quite
positive, which is undesirable for protonation. Model 4
seems to be the best compromise-it has two ingredients
important for N, reduction: a lowered N-N OP, and negative net charge on N. The reduced OP is primarily a conse-
Verlagsgesellschafi mhH, 0-69451 Weinheim, 1993
Fig. 2. The contribution of the N, ilp*fragment MO (shaded area) to the total
DOS of 4 (dotted line). The integration line indicates the percentage of il; filled,
on a scale from 0 to 100%.
The “reduced” state (with three more electrons) gives OPs
for the N-N bond and net charges on the N atoms comparable to the “native” state. This is because the three additional electrons enter one or two levels immediately above the
HOMO in the Fe 3d block, where there is relatively little N,
orbital contribution. The net charges on the Fe or Mo atoms
that bind N, do become more
In conclusion, our initial study finds that metal-metal
bonding in the FeMo-cofactor model of nitrogenase is important; the Fe-Fe bonds are especially strong among threecoordinate iron atoms. N, coordinated models 4, and possibly 6 and 9 are, we think, best activated for reduction.
Received: January 29,1993 [Z 5841IEl
German version: Angen. Chem. 1993, 10S, 1125
0.08, 0.06
0.46, -0.43
0.26, 0.32
0.57, -0.39
0.41, 0.28
0.24, 0.16
0.62, -0.17
0.44, -0.14
[a] When two N atoms are not equivalent, the first value is for the “left” (or the
bound) nitrogen atom in the structures in Scheme 2. [b] Fe-N = 1.79, FeFe = 2.75 .& (for all edges of the Fe, trigonal prism). [c] Fe-N = 2.0, FeFe = 3.03 A (for all edges of the Fe, trigonal prism).
[I] J. Kim, D. C. Rees, Science, 1992, 257, 1677-1682; J. Kim, D. C. Rees,
Nature, 1992,360,553-560; see also D. Sellman, Angew. Chem. 1993,105,
67; Angew. Chem. Int. Ed. Engl. 1993, 32, 64. A 2.2 resolution refinement has just been published: M. K. Chan, J. Kim, D. C. Rees, Science
1993, 260, 792-794. There are no major changes in the model.
[2] Our calculations are of extended Hiickel type: R. Hoffmann, J. Chem.
Phys. 1963,39,1397- 1412; R. Hoffmann, W. N. Lipscomb, ihid. 1962,36,
2179-2188,3489-3493. The parameters for Fe, Mo, and S are taken from
earlier work: J. Silvestre, R. Hoffmann, Inorg. Chem. 1985.24.4108-4119;
R. H. Summerville, R. Hoffmann, J. Am. Chem. Soc. 1976,98,7240-7254.
Two other parameter sets were tested - in one the sulfur Hi, parameter was
raised by 2.3 eV, in the other the Mo 4d H,, parameter was lowered to the
energy of the Fe 3d orbitals. The Important results on the N-N bond
weakening and bound N, polarity remain unaffected by these parameter
[3] We also performed calculations on the cofactor model with real hgating
atoms and the geometry from the X-ray crystallographic analysis, and
obtained similar results.
[4] B. K. Burgess, Chem. Rev. 1990, 90, 1377-1406.
[5] R. Zimmermann, A. X. Trautwein in Nitrogen Fixation (Eds.: A. Miiller,
W. E. Newton), Plenum, New York, 1983, pp. 63-81.
[6] S . Harris. Pol-vhedron, 1989, 8, 2843-2882.
0570-0833193~0707-lO643 10.00+ .2S/0
Angew. Chem. Int. Ed. Engl. 1993, 32, No. 7
[7] R . N. F. Thorneley. D . J. Lowe in Molybrienum Enzymes (Ed.: T. Spiro),
Wiley. New York. 1985. pp. 221 -284.
[S] R. Hoffmann. Angen.. Chrm. 1987, 99. 871-906; Angew. Chem. Inr. Ed.
D i g / . 1987.26.846-878; SolidcinriSurfuces, VCH, New York. 1988; G. F.
Holland, D. E. Ellis, W. C. Trogler. J. Am. Chem. Suc. 1986. 108, 18841894.
[9] When the Mo 4d orbitals are put in with an energy of the Fe 3d (see end
of ref. [2]). the M o "f,,"set is filled and the M o reduced. The results of the
N 2 binding are not affected.
[lo] In a hypothetical model 3 with a central 0 atom, all the Fe-Fe (and
Fe Mo) interactions are weak, but definitely bonding (OPs range from
0.04 to 0.09). similar to those i n a cubane cluster [Fe,S,(SR),]*(OP = 0.07). I n addition there IS reasonable Fe-O bonding.
[ l l ] One or two protons may also be bound t o the MoFe-protein 171. The
protons may not necessarily bind to thecofactor: but if they d o they would
interact with the donor levels of the cofactor (e.g., the lone pairs of sulfide
ions. or the filled Fe 3d orbitals), and should have little effect on the
chemical bonding and the N,-binding properties of the cofactor. We confirmed this by repeating the N,-Binding studies with two protons bound to
two bridging sulfide ions, and to Fe2 and Fe5 along the perpendicular
direction of the coordination planes. respectively.
[12] W. H. Orme-Johnson. Science. 1992. 257, 1639- 1640.
[13] We also performed calculations on N, binding with the 02--centered
model 3. but d o not present details here. Binding modes 7-9 give unreasonably short N - 0 distances. For 5. 6. 11. 12, the central 0'- has little
intluence on the O P of the N-N bond. Only for 4 are both the O P for the
N N bond and the net atomic charge on the N atom substantially reduced.
to wlues of 1.39 and -0.32. respectively. This is because the energy levels
within h a l f a n eV above the H O M O have some contributions from the rc*
state of the N Lfragment. and they are filled when the central 0 2raises
~ the
energy of the HOMO by 0.4 eV. Of course. 10 is impossible in this case.
Structural Evidence of the Aromaticity of
Borepins: A Comparison of 1-Chloroborepin and
Tricarbonyl( 1-ch1oroborepin)molybdenum**
By Arthur J: Ashe, III,* Jeff U.: Kampf; Wolfram Hein,
structure of 1 b by using crystals grown from the neat liquid
and sealed in capillaries. Selected bond lengths are listed in
Table 1, while Figure 1 illustrates the structure, showing a
completely planar borepin ring as had been anticipated from
The B-C bond length (1.51 A) is dis-
Table 1 Selected bond lengths
[A] of the borepins 1 b and 3 b
1 b (exp.)
1 b (6-31G')
3 b (exp 1
C-C range
k 0.058
tinctly shorter than those which have been found for nonconjugated B-C bonds (typical range 1.55-1.59 A),['*]indicating n bonding in the borepin ring. The ring C-C bond
lengths range from 1.37 to 1.42 A; the formal single bonds
are somewhat longer than the formal double bonds. However, this range of bond lengths (* 0.058 A) is considerably
smaller than that found for cycloheptatrienes (1.33 1.46 A)r131and other cyclic polyenes. Indeed, the range of
C-C bond lengths of l b is comparable with that of
napthalene (1.36- 1.42 A), and is consistent with an aromatic
and Roger Rousseau
Dedicated to Professor Hans Bock
on the occasion of his 65th birthday
Borepins 1 have been intensely investigated since they are
the neutral isoelectronic analogues of tropylium 2.I' -91 However, theoretical work has suggested that the aromaticity of
1 H-borepin (1 a) is more limited than that of 2.['01The ready
availability of minimally substituted borepins offers the opportunity to obtain experimental data on borepin aromaticity.[8.91 We report here on the structures of I-chloroborepin
(1 b) and tricarbonyl(1-ch1oroborepin)molybdenum (3 b).["l
At room temperature 1b is a moisture- and air-sensitive
liquid (m.p. - 37 "C). We have determined the X-ray crystal
Fig. 1. Crystal structure of 1 b (ORTEP). Bond angles ["I: CIl-Bl-Cl 116.76(6),
C1-B1-C1 a 126.5, Bl-Cl-C2 127.42(9), CI-CZ-C3 128.96(9). CZ-C3-C3a
For purposes of comparison we have performed a Hartree
Fock level ab initio MO calculation of 1 b with the 6-31 G *
basis set." The optimized structural parameters listed in
Table 1 do not differ significantly from those reported for 1 a
calculated at the same
While the calculated and
experimental structure for 1 b are in reasonable agreement
(corresponding bond lengths are i 0.02 A), the calculated
structure has a larger range of C-C interatomic distances
(i0.093 A); thus the bond alternation is overestimated.
The reaction of 1 b with tricarbonyltris(pyridine)molybdenum and BF, OEt, gave the complex 3 b, which serves as
a useful precursor for a variety of borepin complexes. The
reduction of 3 b with lithium triethylborate afforded I Hborepin(tricarbony1)molybdenum (3a) as stable crystals.
This complex cannot be prepared directly from the very
labile free 1 H-borepin.
It was of particular interest to obtain a crystal structure of
3 b and to compare it with that of 1 b. Structural data are
known for 3d9]and 4[2'1but not for the corresponding free
ligands. The structure of 3 b (Fig. 2) consists of a nearly
planar chloroborepin ring that is q7-bound to the Mo(CO),
unit. The B-Mo distance (Table 1) is slightly longer than the
3a, R = H
3b, R = CI
3c,R = CeH5
Prof. Dr. A. J. Ashe, 111, Dr. J. W. Kampf. Dr. W. Klein, R. Rousseau
Department of Chemistry, The University of Michigan
Ann Arbor, MI 48109-1055 (USA)
Telefax: Int. code f(313) 747-4865.
This work was supported by the donors of the Petroleum Research Fund,
administered by the American Chemical Society. W K. thanks the
Deutsche Forschungsgemeinschaft for a postdoctoral fellowship.
A n g w C'hem. Inr. Ed. EngI. 1993, 32, No. 7
il", VCH Verlug.~grsell.~chufi
mhH, 0-69451 Weinhelm. I993
0570-0833193jO707-1065J 10.00+ ,2510
Без категории
Размер файла
450 Кб
cofactor, nitrogenand, femo, might, activated
Пожаловаться на содержимое документа