Catalytic unfolding and proteolysis of cytochrome c induced by synthetic binding agents

Article abstract of DOI:10.1021/ja0317731

A class of polyanionic copper porphyrin dimers is shown to selectively increase the susceptibility of cytochrome c to proteolysis through binding-induced disruption of tertiary and secondary structure. The free energy of the protein conformation leading to proteolytic attack is stabilized by about 2.4 kcal/mol in the bound state. The proteolytic acceleration is catalytic in nature, requiring only a fraction of an equivalent of metalloporphyrin to effect complete, rapid digestion in the presence of a protease.

Full text of DOI:10.1021/ja0317731

Published on Web 09/18/2004
Catalytic Unfolding and Proteolysis of Cytochrome c Induced
by Synthetic Binding Agents
Kevin Groves, Andrew J. Wilson, and Andrew D. Hamilton*
Contribution from the Department of Chemistry, Yale UniVersity,
New HaVen, Connecticut 06511
Received December 17, 2003; E-mail: andrew.hamilton@yale.edu
Abstract: A class of polyanionic copper porphyrin dimers is shown to selectively increase the susceptibility
of cytochrome c to proteolysis through binding-induced disruption of tertiary and secondary structure. The
free energy of the protein conformation leading to proteolytic attack is stabilized by about 2.4 kcal/mol in
the bound state. The proteolytic acceleration is catalytic in nature, requiring only a fraction of an equivalent
of metalloporphyrin to effect complete, rapid digestion in the presence of a protease.1
Introduction
protease. In such cases, the folded structure must be disrupted
in order for efficient proteolysis to occur. In the laboratory, this
The binding of synthetic agents to large areas on a protein
surface represents a relatively new approach to protein recogni-
tion and enzyme inhibition. In contrast to active site inhibitors,
such allosteric receptors do not target the active site of the
protein directly, but rather a large number of different functional
groups for selective binding interactions.2-4
can be accomplished by the use of relatively high concentrations
of chemical denaturants or high temperatures.13 However,
denaturation by such drastic measures is inherently nonselective.
In Nature, selective degradation of proteins is accomplished
under physiological conditions. As an example, in the natural
proteolytic degradation system of ClpA/ClpP, part of the Clp/
Hsp100 family, tagged proteins are first selectively unfolded
by ClpA through an ATP dependent denaturation step prior to
being fed into a proteolytic chamber of ClpP for degradation.1
Selective degradation of target proteins by synthetic agents
is an attractive goal, but one that has been difficult to
achieve.16,17 Targeted protein degradation offers the possibility
of catalytic antagonists that could deactivate numerous equiva-
lents of the target enzyme. Davis et al. have demonstrated
selective and catalytic protein degradation by covalently at-
taching active site binders of targeted proteins to a protease.17
In their system, the binding group recognizes the target protein,
bringing the protease into close vicinity for proteolytic attack.
Much earlier, Wilchek et al. attempted to use biotinylated
proteases to enhance the proteolysis of avidin and streptavidin,
biotin binding proteins that show high thermodynamic stability
and resistance to proteolytic degradation.16 However, the activity
of the biotinylated proteases was inhibited by complexation with
avidin. Thus, simply bringing a protease into close proximity
with a protein by means of a binding group does not guarantee
But beyond surface recognition of native protein conforma-
tions, there lies the potential for recognition of non-native
conformations that are induced by a binding event.5,6 Binding-
induced conformational changes or even significant unfolding
of an active protein with subsequent loss of function represents
an emerging strategy for the modulation of protein function.7,8
Recently, we have reported a family of substituted porphyrin
derivatives that strongly bind to and induce unfolding of
cytochrome c (cyt. c), selectively and stoichiometrically below
room temperature and under nearly physiological conditions (pH
7.4, 50 mM NaCl).6,9,10
In addition to the disruption of protein function, protein
unfolding plays a key role in the process of protein disposal
through proteolytic digestion. Proteolysis requires conforma-
tional mobility of segments of a protein in order for the peptide
bonds to reach the active site of a protease for cleavage.11-15
In their native folded forms, many proteins are resistant to
proteolytic degradation because stable secondary and tertiary
structures effectively shield the peptide backbone from the
(1) Weber-Ban, E. U.; Reld, B. G.; Miranker, A. D.; Horwich, A. L. Nature
(London) 1999, 401, 90-93.
(11) Fontana, A.; Fassina, G.; Vita, C.; Dalzoppo, D.; Zamai, M.; Zambonin,
M. Biochemistry 1986, 25, 1847-1851.
(2) Jain, R. K.; Hamilton, A. D. Org. Lett. 2000, 2, 1721-1723.
(3) Peczuh, M. W.; Hamilton, A. D. Chem. ReV. 2000, 100, 2479-2493.
(4) Park, H. S.; Lin, Q.; Hamilton, A. D. J. Am. Chem. Soc. 1999, 121, 1,
8-13.
(12) Hubbard, S. J.; Eisenmenger, F.; Thornton, J. M. Protein Sci. 1994, 3, 757-
768.
(13) Bark, S. J.; Muster, N.; Yates, J. R. I.; Siuzdak, G. J. Am. Chem. Soc.
2001, 123, 1774-1775.
(14) Hu, Y.; Fenwick, C.; English, A. M. Inorg. Chim. Acta 1996, 242, 261-
269.
(5) Shi, L.; Palleros, D. R.; Fink, A. L. Biochemistry 1994, 33, 7536-7546.
(6) Jain, R. K.; Hamilton, A. D. Angew. Chem., Int. Ed. 2002, 41, 641-643.
(7) Peterson, J. R.; Lokey, R. S.; Mitchison, T. J.; Kirschner, M. W. Proc.
Natl. Acad. Sci. U.S.A. 2001, 98, 10 624-10 629.
(15) Wang, L.; Chen, R. X.; Kallenbach, N. R. Proteins: Struct., Funct., Genet.
1998, 30, 435-441.
(8) Leung, D. K.; Yang, Z.; Breslow, R. Proc. Natl. Acad. Sci. U.S.A. 2000,
97, 5050-5053.
(16) Bayer, E. A.; Grootjans, J.; Alon, R.; Wilchek, M. Biochemistry 1990, 29,
11 274-11 279.
(9) Wilson, A. J.; Groves, K.; Jain, R. K.; Park, H. S.; Hamilton, A. D. J. Am.
Chem. Soc. 2003, 125, 4420-4421.
(17) Davis, B. G.; Sala, R. F.; Hodgson, D. R. W.; Ullman, A.; Khumtaveeporn,
K.; Estell, D. A.; Sanford, K.; Bott, R. R.; Jones, J. B. ChemBioChem 2003,
4, 533-537.
(10) Aya, T.; Hamilton, A. D. Bioorg. Med. Chem. Lett. 2003, 13, 2651-2654.
9
10.1021/ja0317731 CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 12833-12842
12833

A R T I C L E S
Groves et al.
that a favorable conformation for proteolytic attack can easily
be reached. We describe here a multicomponent catalytic system
for the rapid, efficient and catalytic digestion of cyt. c that
involves conformational activation through selective binding and
unfolding of the protein by substituted metalloporphyrins. The
porphyrins target proteolytically resistant R-helical structures
for unfolding, allowing facile proteolysis by trypsin or other
proteases that leads to release of the activating binder back into
solution to continue a catalytic cycle.
Figure 1. (a) CD spectra of cyt. c (black), 3a (red) and their 1:1 com-
plex (blue). (b) Titration of induced CD signal at 425 nm versus equivalents
of 3a.
wavelengths, 222 nm corresponding to the secondary structure
of cyt. c, and at 425 nm, which represents an induced CD
(presumably a coupling between the heme and the porphyrin)
that is present in the 1:1 complex of folded cyt. c with 3a, but
which vanishes upon thermal unfolding (Figure 1).
With less than two equivalents of 3a complete loss (corre-
sponding to a final state equivalent to that at > 90 °C) of the
induced CD signal centered at 425 nm occurred at a lower
temperature than complete thermal denaturation of the protein
secondary structure, as judged by the CD signal at 222 nm.
Furthermore, the fraction of protein that was unfolded at the
completion of the loss of induced CD was approximately half
an equivalent relative to 3a (Figure 2). However, when two or
more equivalents of 3a were present, the melting transitions
observed for the two wavelengths coincided (Figure 3). These
results indicate that there is a change in stoichiometry upon
denaturation of the protein from a 1:1 complex with the native
form to a 2:1 complex with the denatured state. With 1.2
equivalents of 3a, only 60% of the protein can exist as a
denatured 2:1 complex. The balance of protein remains in its
native folded state, but uncomplexed.
At just below 85 °C (Tm of cyt. c), a single equivalent of 3a
can unfold the protein, while at lower temperatures (60-70 °C)
and less than 1 equivalent of 3a a disproportionation of the 1:1
complex with the native form into uncomplexed native protein
and doubly bound unfolded protein occurs. Such disproportion-
ation points to a cooperative role between the two binding events
of 3a to the denatured state that effects unfolding of cyt. c below
its Tm. The change in stoichiometry also explains the increase
in ∆Tm with increased 3a, since higher concentrations will favor
the doubly bound complex with the denatured state over the
singly bound complex with the native conformation. The
cooperativity of the transition can clearly be seen by plotting
the fraction of unfolded cyt. c versus equivalents of 3a at 60
Results and Discussion
One of our first generation of protein binding porphyrins,
1a, substituted with four tyrosine-aspartic acid dipeptides, binds
to natively folded cyt. c as a 1:1 complex with a dissociation
constant of 20 nM.2 The simple derivative 2a, first reported to
bind to cyt. c by Fisher,18 binds more weakly (Kd ) 950 nM)
and has only a very small effect on the thermal stability of cyt.
c. In contrast, free base porphyrin 1a was found to lower the
melting point of cyt. c by about 20 °C at a mole ratio of 1.2:1,
and even further at higher equivalents.6 Similar behavior to 1a
was found with achiral free base porphyrin 3a, indicating that
the precise configuration of hydrophobic and charged residues
is not crucial for unfolding activity.9 The melting transitions of
cyt. c with 1.2 equivalents of 1a or 3a were found to be
considerably broader than that of native cyt. c, suggesting that
a second or third equivalent of the porphyrin may be binding
to the unfolded protein, stabilizing it relative to the 1:1 complex
with the native form. To investigate the stoichiometry of binding
to unfolded cyt. c, the thermal melting behavior over a
concentration range of porphyrin was monitored by CD at two
(18) Clark-Ferris, K. K.; Fisher, J. J. Am. Chem. Soc. 1985, 107, 5007-5008.
9
12834 J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004

Catalytic Unfolding and Proteolysis of Cytochrome c
A R T I C L E S
Figure 4. Thermal unfolding by CD of cyt. c at 222 nm (black), cyt. c 2.4
equiv. 3b at 222 nm (red) and cyt. c + 2.4 equiv. 3b at 410 nm (blue).
only 2:1 complexes with cyt. c and partially or fully unfold the
protein below room temperature. The greater unfolding ability
of the copper porphyrin dimers may result in part from the less
favorable binding to the natively folded form, which would
require breaking up of the dimer in order to form a preferred
1:1 complex observed with the free base porphyrins. Instead,
the copper complexes unfold the protein even at low temper-
atures to allow 2:1 binding to the preformed dimer. Compound
1b unfolds cyt. c essentially completely at room temperature,
showing a CD spectrum that resembles the thermally unfolded
protein (T > 90 °C).9 For cyt. c bound by dimeric 3b, two
distinct conformational transitions are observed.
Figure 2. Thermal unfolding by CD of cyt. c at 222 nm (black), cyt. c +
3a at 222 nm (red) and cyt. c + 3a at 425 nm (blue) with (a) 0.5 equiv. 3a
and (b) 1.2 equiv. 3a. Approximately half an equivalent of protein relative
to 3a is unfolded (red curve) at the temperature corresponding to complete
loss of the 1:1 complex (blue curve).
In addition to the loss of R-helical structure observed in the
CD spectrum at 222 nm on addition of 3b, the native
coordination of methionine 80 to the heme is lost. A second
transition, with a melting point of about 35 °C is also observed,
and is characterized by further loss of R-helical character as
determined by the CD signal at 222 nm as well as a distinct
transition observable at 410 nm (Figure 4). The loss of the
Met80 heme coordination is observable in the UV at 695 nm.21
Heme coordination is lost upon 2:1 binding of copper complex
3b at room temperature, observable by the band at 695 nm in
the difference spectrum, but not upon 1:1 binding of free base
3a, even in the presence of 2.2 equiv, further supporting the
Figure 3. (a) Thermal unfolding by CD of cyt. c at 222 nm (black), cyt.
c + 2.4 equiv. 3a at 222 nm (red) and cyt. c + 2.4 equiv. 3a at 425 nm
(blue). (b) Titration of unfolding cyt. c by CD at 222 nm versus equivalents
of 3a at 60 °C.
°C, which shows a characteristic shape of a cooperative
transition (Figure 3b).
Effects of Copper and RT Denaturation. Consistent with
the observation of a preference for 2:1 binding to unfolded cyt.
c, we have reported that the copper complexes of 1a and 3a,
1b and 3b, which exist largely as preformed dimers,9,19,20 form
(19) Pasternack, R. F.; Francesconi, L.; Raff, D.; Spiro, E. Inorg. Chem. 1973,
12, 2606-2611.
(20) Pasternack, R. F.; Huber, P. R.; Boyd, P.; Engasser, G.; Francesconi, L.;
Gibbs, E.; Fasella, P.; Cerio Venturo, G.; Hinds, L. d. J. Am. Chem. Soc.
1972, 94, 4511-4517.
(21) Schejter, A.; Luntz, T. L.; Koshy, T. I.; Margoliash, E. Biochemistry 1992,
31, 8336-8343.
9
J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004 12835

A R T I C L E S
Groves et al.
Phthalocyanine tetrasulfonate 8a and its copper complex 8b
induced small but identical melting point depressions, whereas
the cationic tetra(4-methylpyridinium)porphyrin 9a, which
shows no affinity for cyt. c, and its copper complex 9b had no
effect on the structure of cyt. c. Free copper ions lower the
melting point of cyt. c by about 5 °C, but when complexed
with 1,4,8,11-Tetraaza-cyclotetradecane (cyclam) show no ef-
fect, suggesting the small melting point decrease is related to
nonspecific interaction of the free copper ions with various side
chains that are buried in the native form and exposed upon
thermal denaturation. Thus, porphyrin bound copper ions
themselves do not appear to play a significant role in cyt. c
denaturation outside of increasing the stacking propensity of
the porphyrin surface.
To further investigate the role of charged copper tetraphen-
ylporphyrin dimers, we looked at the activity of the heterodi-
mer 10 between 2b and 9b,22 which is water soluble but
charge neutral. This heterodimer showed essentially no effect
on the structure of cyt. c, reinforcing the necessity of negative
charge to complement the positive charge on the protein. Indeed,
complexation of 2b with 9b completely eliminated the activity
of 2b.
A dimeric porphyrin, or even two equivalents, is not always
necessary to lower the melting point of cyt. c, however.
Dimerization can be prevented by substituting the porphyrin
with more negative charge, as in compound 5a with 16
carboxylates which does not dimerize due to the electrostatic
repulsion of the carboxylates. 5a binds cyt. c very strongly (Kd
) 4 nM) and induces a melting point depression of about 20
°C. The same melting point depression was observed with 1 or
2 equivalents of 5a as well as with its copper complex 5b,
confirming 1:1 binding stoichiometry and further implicating
the role of copper as a means of inducing dimerization and 2:1
binding in the cases of porphyrins with less negative charge.
Similarly, compound 6, also with 16 carboxylates but with
additional hydrophobic surface, was reported to have an even
greater effect on the structure of cyt. c, showing a melting point
of about 35 °C.10 The behavior of 6 was also cleanly 1:1, due
again to the electrostatic repulsion of 16 carboxylates. But the
effects of the dimers are not just to form an entity that has double
the negative charge of the monomer. Not one of 5a, 5b or 6
unfolded cyt. c at room temperature like the copper tetraphen-
ylporphyrin dimers, even those with only 4 carboxylates.
Therefore it must be a critical aspect of the anionic tetraphe-
nylporphyrin dimer that imparts the greatest denaturing effect
on cyt. c.
It is important to note that strong binding of this class of
compounds to cyt. c, although necessary for denaturation at low
concentrations, is not a general predictor for denaturing activity.
As an example, compound 5a, which binds with 4 nM affinity,
only induces a 20 °C melting point depression, whereas 6,10
with slightly better affinity, showed a 50 °C depression. This
is because the molecules can bind to both the natively folded
as well as any of a collection of various unfolded states. The
stronger the binding to the natively folded form, the weaker
any denaturation effect will be. Thus, it is selectivity for the
denatured state over the native state that drives the denaturation
process for a particular porphyrin. Charge matching can increase
the general affinity of the receptor for the protein, but does not
Figure 5. UV spectra of (a) cyt. c (black), cyanocyt. c (red), and cyt.
c-cyanocyt. c (blue) and (b) cyt. c + 2 equiv. 3b in separate tandem cells
(black), cyt. c + 2.4 equiv. 3b bound (red) their difference spectrum
(separate-bound) (blue) and the corresponding difference spectrum for 3a
(green).
preference of 2:1 binding to unfolded forms of cyt. c in contrast
to 1:1 binding to the native protein (Figure 5).
The disruption of the Met80 heme coordination is not criti-
cal for stability of the R-helical structure of cyt. c, however,
since the cyanide complex of cyt. c, cyano-cyt. c, has a melting
point (observed by CD at 222 nm) slightly higher than that of
native cyt. c. Furthermore, the binding of 3b to cyanocyt. c is
equally strong as that to native cyt. c, suggesting that the
porphyrins do not stack face-to-face with the heme, but rather
interact with other regions of the protein to interfere with the
stability of the secondary structure. Carey et al. have shown
that the helical character of cyt. c requires the presence of
N-terminal and C-terminal peptide regions (but not a significant
segment in the middle) and the heme, even noncovalently bound.
Thus, the stability of the secondary structure of the C-terminal
fragment of cyt. c depends on tertiary interactions, which can
be disrupted by binding a porphyrin.
The denaturation activity of copper porphyrins 3b and 1b
results from the preferential binding of two equivalents of the
porphyrins to a denatured state of cyt. c. The substituents of
the copper porphyrin dimers, in contrast to the free bases, are
not essential to the denaturation effect. Rather, the substituents
of the copper porphyrins act to increase the binding affinity,
hence allowing denaturation to occur at lower concentrations.
At concentrations above µM, dimeric19 copper porphyrins 2b
and 4b also effected considerable denaturation of cyt. c at room
temperature, while the corresponding monomeric free base
porphyrins 2a and 4a showed only modest melting point
depressions of about 15 °C.
Tetraanionic coproporphyrin-I 7a, which lacks meso phenyl
groups and binds weakly to cyt. c, as well as its copper complex
7b had no effect on the stability of the protein (Figure 6).
(22) Erdem, S. S. J. Porphyrins Phthalocyanines 1998, 2, 61-68.
9
12836 J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004

Catalytic Unfolding and Proteolysis of Cytochrome c
A R T I C L E S
Figure 6. CD titration showing unfolding of cyt. c with 0 to 8.8 equivalents of (a) 3b, (b) 2b, (c) 4b, and (d) 7b.
necessarily translate into greater denaturing activity. The design
of the next generation of targeted denaturants should focus on
this difference, perhaps targeting disparate functionality on a
folded protein that could better interact with a synthetic receptor
in a less rigid unfolded state.
Catalytic Digestion of Cyt. c. Cytochrome c is a highly stable
protein (Tm ) 85 °C) that is relatively resistant to proteolysis.
Even after 20 h in the presence of trypsin (0.02 µg/µL) at 21
°C and pH 7.4, minimal digestion was detectable by SDS-PAGE
analysis. In contrast, with the addition of two equivalents of
dimeric 3b under the same conditions, the bulk of the protein
was digested within 15 min and the digestion was essentially
complete within 30 min (Figure 7). Furthermore, when the
assisted digestions were complete, 3b and trypsin remained
active, since the digested fragments do not associate with the
porphyrin. Thus, the system should be catalytic, as the porphyrin
will be free to bind and activate additional protein after each
digestion cycle.
Kinetic measurements of relative digestion rates were ob-
tained by monitoring the CD signal at 222 nm, which reflects
the degree of R-helical character of the folded protein in the
sample, in a manner similar to that of Wang and Kallenbach.23
The CD signal of cyt. c was unaffected by catalytic amounts of
achiral 3b under experimental conditions, and the signal from
trypsin was negligible. MALDI-MS and SDS-PAGE analyses
revealed that under the reaction conditions there was no
significant accumulation of large peptide fragments during the
course of the digestions. Instead, after the first cut was made
on the protein, the remaining thermodynamically less stable
pieces were digested to small peptide fragments quickly, thus
allowing a continuous quantitative measure of the concentration
of undigested cytochrome c during the course of the digestion
by CD.
Figure 7. SDS-PAGE analysis of the tryptic digestion of cytochrome c
(a) alone and (b) in the presence of 4 equiv. 3b. Digestion time shown in
minutes.
imately 4 h at 30 °C and pH 7.4, whereas cyt. c and trypsin
without 3b resulted in less than 15% digestion in the same time.
Just 0.025 equivalents of dimeric 3b effected completion within
20 h, while the digest with trypsin alone had not even reached
one-third completion in that time, representing more than 25
catalytic turnovers relative to dimeric 3b (Figure 8).
Using this method, 0.1 equivalents of dimeric 3b resulted in
complete digestion of cyt. c in the presence of trypsin in approx-
(23) Wang, L.; Kallenbach, N. R. Protein Sci. 1998, 7, 2460-2464.
9
J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004 12837

A R T I C L E S
Groves et al.
Table 1. Relative Initial Rates of Tryptic Digestion of Cyt. c in the
Presence of Catalytic Amounts of Various Porphyrins
initial rate
relative
initial ratea
1
(
µ
M min-
)
cyt. c
cyanocyt. c
0.023
0.081
0.053
0.076
0.14
0.058
0.44
0.030
0.035
0.15
1
3.5
2.3
3.3
6.1
2.5
cyt. c + 0.05 equiv. 3b
cyt. c + 0.10 equiv. 3b
cyt. c + 0.20 equiv. 3b
cyt. c + 0.20 equiv. 3a
cyanocyt. c + 0.20 equiv. 3b
cyt. c + 0.20 equiv. 5a
cyt. c + 0.20 equiv. 2a
cyt. c + 0.02 equiv. 2b
19 (5.4b)
1.3
1.5
6.5
a Rates were obtained by CD at 222 nm in 5 mM sodium phosphate, pH
7.4, 30 °C under initial rate conditions. b Rate relative to cyanocyt. c.
Figure 8. Kinetics of the tryptic proteolysis of cytochrome c (20 µM) as
measured by CD at 222 nm, 30 °C, 5 mM sodium phosphate, 50 mM NaCl,
0.02 µg/µL trypsin with 0.0 (black), 0.025 (red), and 0.10 (blue) equivalents
of dimeric 3b.
are high enough to ensure saturated 2:1 binding. Relative rates
of proteolysis have been used previously to estimate free energy
differences, relative to the conformation for proteolytic attack,
between cyt. c in the oxidized and reduced forms as well as
the azide complex.23 The 50-fold acceleration for tryptic
proteolysis observed for the complex of cytochrome c with 3b
at 30 °C corresponds to a lowering of the free energy necessary
to reach the conformation(s) for initial proteolytic attack by ∼2.4
kcal/mol.
Other cyt. c binders can also increase the susceptibility of
the protein to proteolysissheme ligands such as cyanide or
azide, which disrupt native heme coordination of the protein
by displacing Met80, also accelerate proteolysis.14,23 Indeed,
heme ligation of cyt. c has been shown to be critical for its
resistance to proteolytic attack.14 English et al. examined various
metalloderivatives of cytochrome c, and correlated the strength
of Met80 coordination with resistance to proteolysis. The
coordination of Met80 must be broken in order for the protein
to reach conformations for efficient peptide bond cleavage. The
observed disruption of the native heme ligation of cyt. c by
binding of 3b likely contributes to the increase in susceptibility
of cyt. c to trypsin.
Displacement of Met80 coordination cannot be the only effect
contributing to the activation of cyt. c to digestion, however.
The rate of proteolysis of the cyanide complex of cyt. c is still
25 times slower than that of the complex with 3b. Furthermore,
in the presence of 0.1 equivalents of dimeric 3b, digestion of
the cyanide complex was accelerated about the same amount
as native cytochrome c, suggesting an alternative mechanism
of action (Table 1). The disruption of the R-helical secondary
structure, which is particularly resistant to proteolysis11 and is
observable by the thermal behaviors of the CD signal at 222
nm of the bound complex, contributes greatly to the marked
rate enhancement of digestion.
To probe the specific regions of cyt. c that are made vul-
nerable to proteolytic attack in the presence of 3b,24 limited
digestions were carried out and the results subjected to MALDI-
TOF analysis. The peaks in the MALDI spectrum that cor-
respond to initial cleavage fragments early in the digestion were
identified, and locations of those cuts within the protein were
determined. Sites of initial cleavage were determined only when
Figure 9. (a) initial rate kinetics for tryptic digestion of cyt. c with 0.0,
0.025, 0.05, and 0.1 equiv. of dimeric 3b at 35 °C and (b) plot of relative
rate of initial digestion versus catalytic equivalents of dimeric 3b.
Under the same conditions, the benchmark hydrolysis of
N-benzoyl arginine ethyl ester (BAEE) by trypsin was actually
inhibited, proceeding at only 75% in the presence of 2 µM
dimeric 3b, presumably from weak binding of the porphyrin to
trypsin. Figure 9a shows the initial rates of cyt. c digestion by
trypsin with varying catalytic concentrations of 3b. By plotting
the initial rates of tryptic digestion with varying equivalents of
dimeric 3b (shown in Figure 9b) and extrapolating to 1 full
equivalent, we found an acceleration of more than 50-fold for
the tryptic digestion of the complex of cyt. c with dimeric 3b
over the native protein. Despite the fact that 3b binds cyt. c as
a 2:1 complex (by Job analysis),9 the apparent kinetic order of
3b in the catalyzed digestion was found to be 1. This result
reflects that the porphyrins exist as dimers under the reaction
conditions, so the reaction is effectively first order in dimer,
but also that the concentrations of the protein and porphyrin
(24) With 2.2 equivalents the steady-state concentration of partially digested
fragments is high enough to identify in the MALDI spectrum, whereas
with catalytic amounts few large fragments are observed. For the digestion
of cyt. c alone, higher concentrations (50 µM) were used to allow
observation of low concentrations of intermediate fragments.
9
12838 J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004

Catalytic Unfolding and Proteolysis of Cytochrome c
A R T I C L E S
Figure 10. Location of unassisted (green) and 3b assisted cleavage sites (red) for digestion of cytochrome c with trypsin, proteinase k, and thermolysin.
R-Helical regions of the text sequence of the protein are shown in red.
both fragments of a single cut on the intact protein could be
detected.
the distribution of initial assisted cleavage sites indicates that
the proteolytic acceleration is not a result of making just a single
residue or region vulnerable, but a more general disruption in
the stability of the secondary structure of the protein. Under
saturation conditions, where concentrations of porphyrin are far
above the binding constant, unsubstituted copper porphyrin 2b
induced cleavage at essentially the same locations as substituted
compound 3b, reaffirming the role of the copper porphyrin
dimer core as the active component for denaturation.
Three different proteases, trypsin, proteinase K and thermol-
ysin, with varying residue selectivity and electrostatic properties
were chosen. To observe the greatest increase in proteolytic
sensitivity, digestions were carried out at 25 °C, below the
optimal temperature for the proteases. Under these conditions,
each protease shows significant initial digestion at only one
location in the absence of any porphyrin, based on observable
fragments. Although all three proteases cleave different residues,
each cleavage site lies along loop regions in the same area of
the protein around the heme crevice (Figure 10). This result is
consistent with the observations by English14 that the strength
of heme ligation is a predictor of the rate of proteolysis.
As with denaturation effects, the acceleration of digestion of
cytochrome c by the copper porphyrin dimers is selective. Cy-
tochrome c551, which shares major structural features with
cytochrome c but lacks much of the surface positive charge
shows no proteolytic acceleration in the presence of 3b. Myo-
globin and R-lactalbumin were similarly unaffected by 1b9 and
the tryptic hydrolysis of BAEE was actually slightly inhibited
by 3b, presumably through weak binding to trypsin, which also
carries significant positive charge.
In the presence of 3b, however, fragments from a multitude
of different initial cleavage sites were observed with all proteases
over a fairly large distribution throughout the protein (Figure
10). The cleavage occurred irrespective of secondary structure
along loop and R-helical regions (Glu61 through Glu69 and
Lys88 through Ala101, red in Figure 10), but with a distinct
preference for the C-terminal half of the protein. No initial
cleavage was observed around the heme nor on the N-terminal
R-helix (Val3 through Lys13). The general locations of assisted
cleavages were consistent across the three different proteases,
indicating that the effects are not specific to any one protease,
but a result of the propterties of the porphyrin/cyt. c complex.
The porphyrins apparently do not function by relieving charge
repulsion between a cationic protease (trypsin) and cationic
substrate (cyt. c). Although 0.20 equivalents of essentially
monomeric 3a showed about 40% of the catalytic acceleration
of tryptic digestion observed with 2b or 3b, hexadecacarboxylate
5a had almost no effect on the rate of proteolysis, despite a
low dissociation constant of 4 nM (Table 1). Additionally, initial
rates of proteolysis by thermolysin and proteinase K, which carry
less positive charge than trypsin, were accelerated to a similar
extent by 3b as found with trypsin. However, the accumulation
of intermediate fragments precludes quantitative measurements
of digestion rates of these proteases by CD spectroscopy. Thus,
the nature and charge of the protease is not important to achieve
accelerated digestion.
In order for a residue inside an R-helix to be cleaved, the
helix must unwind so that the peptide bond can reach the active
site of the protease. The presence of initial cuts within R-helices
that do not occur in the absence of 3b corroborates the CD
evidence that the R-helical secondary structure of cytochrome
c is being disrupted by the presence of the binders. Furthermore,
9
J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004 12839

A R T I C L E S
Groves et al.
standardized spectrophotometrically using ꢀ550 ) 2.95 × 104 M-1 cm-1
after reduction with sodium dithionite (1 mg/mL) in 5 mM sodium
phosphate, pH 7.4. Cyanocytochrome c was prepared by incubating
cytochrome c (500 µM) in 5 mM sodium phosphate, pH 7.4 with 2
mM sodium cyanide at 37 °C for 30 min. Incorporation of cyanide
was confirmed by loss of the band at 695 nm in the visible spectrum.21
Stock solutions of proteins and porphyrins were prepared in 5 mM
sodium phosphate buffer and the pH adjusted to 7.4. Protease solutions
were prepared by dissolving the enzymes 1 µg/µL in 1.0 mM HCl and
used fresh.
Figure 11. Proposed catalytic cycle for the accelerated digestion of cyt. c
by a porphyrin dimer.
The copper porphyrins described here are conformational
activators of the substrate protein that assist peptide bonds that
are otherwise shielded inside R-helices or other folded structures
in accessing the active site of a protease. The apparent catalytic
activity toward digestion of cyt. c, depicted in Figure 11, arises
from the recycling back into solution of the binder upon diges-
tion of the bound protein. The kinetic acceleration for the di-
gestion stems from the binding-induced unfolding of the pro-
tein, where all the necessary energy to effect denaturation is
supplied in the form of excess binding energy. The catalytic
cycle is driven forward entropically by the complete digestion
of the protein and concomitant release of the binder back into
solution.
Such a catalytic cycle, where synthetic porphyrins act as
cocatalysts by conformationally activating a protein to cleavage
by a protease represents a new strategy for directed protein
degradation. Previously, a process of this type had only been
seen in members of the proteasome family (such as ClpA/ClpP).1
In the present case, the driving force for unfolding, rather than
ATP, is the favorable thermodynamics of binding to the
unfolded state of the protein.
Proteolysis Experiments. For CD kinetic studies, 400 µL of 20
µM protein in 5 mM sodium phosphate buffer at pH 7.4 and various
equivalents of porphyrins were equilibrated to the temperature of the
reaction. Protease solution (4 to 8 µL of 1 µg/ µL protease) was added
and the samples were vortexed for 15 s before being placed in the 2
mm quartz CD cells pre-equilibrated to the reaction temperature inside
the CD sample holder. The CD signal at 222 nm was recorded with an
averaging time of 15 s for 30 min to 20 h, depending on the rate of
digestion.
For MALDI-TOF-MS analysis, 10 µL aliquots of digestion samples
taken between 15 s and 30 min of addition of protease, depending on
the rate of digestion, were quenched in 20 µL of a 10 mg/mL solution
of sinapinic acid matrix in 50:50 H2O/acetonitrile with 2% TFA. 0.5
to 1 µL of the quenched samples were spotted onto a 100 well stainless
steel MALDI plate for analysis. Spectra were calibrated against the
parent protein peak (M+1 ) 12361.2). Sites of initial cleavage were
determined by identification of both fragments that result from the
cleavage.
For SDS-PAGE analyses, 60 µL of cytochrome c (20 µM) and
porphyrin (80 µM) were incubated for 10 min. (5 mM phosphate, 50
mM NaCl, pH 7.4). Trypsin (2 µg in 2 µL of 0.1 N HCl) was added
to the reaction mixture and at appropriate intervals, 10 µL of the solution
was removed from the reaction mixture and quenched by addition to
20 µL tris sample buffer (Biorad 2%) and heating at 95 °C for 5 min.
At the end of the reaction, the samples were loaded onto 16.5% tris/
tricine gels and constant voltage of 100 V was applied for 2 h (running
buffer 10x SDS tris/tricine). The gels were washed with water (15 min)
10% acetic acid 0.025% Commasie blue staining solution (2 h) and
water (2 h). The gels were dried in the standard manner and imaged
using a PC.
Conclusions
In conclusion, we have demonstrated a selective and catalytic
system for the efficient proteolytic digestion of cytochrome c.
The effect is achieved by inducing substantial conformational
changes, including disruption of native heme ligation and
unraveling of R-helical secondary structure upon binding of
synthetic porphyrin receptors to the protein. This results in a
lowering of the free energy of the conformation for proteolytic
attack by trypsin by ca. 2.4 kcal/mol, corresponding to more
than 50-fold acceleration of proteolysis. We have observed up
to 25 catalytic turnovers relative to the dimeric porphyrin entity.
The sites of assisted cleavage are broadly distributed throughout
the C-terminal half of the protein, indicating a general disruption
of protein structure.
Syntheses. Stock solutions of tetra-(4-sufonatophenyl)porphine were
standardized by UV using ꢀ414 ) 5.33 × 105 M-1 cm-1 25
.
Stock
solutions of coproporphyrin-I were standardized by UV using ꢀ372
)
1.16 × 105 M-1 cm-1 for dimeric form at 200 mM NaCl.26 Stock
solutions of tetra-(4-N-methylpyridinium) porphyrin were standardized
by UV using ꢀ372 ) 2.26 × 105 M-1 cm-1 27
.
Copper(II) Tetra-(4-carboxyphenyl) Porphine 2b. A 10-fold
excess (w/w) of copper acetate was added to a solution of the free
base tetra-(4-carboxyphenyl)porphine in methanol. After 12 h at room
temperature, the insoluble copper porphyrin was then filtered off and
washed with 1 mM HCl, distilled water, ethanol then ether and dried
under vacuum. Incorporation of copper was confirmed by UV and stock
Experimental Section
General Details. Horse heart cytochrome c, proteinase K, thermol-
ysin and cytochrome c551 were purchased from Sigma. Trypsin,
modified, sequencing grade was purchased from Roche. Coproporphy-
rin-I, meso tetra-(4-carboxyphenyl)porphine and meso tetra-(4-sufonato-
phenyl) porphine were purchased from Strem. Phthalocyanine tetra-
sulfonate and meso tetra-(4-N-methylpyridinium) porphyrin were pur-
chased from Calbiochem. Copper(II) phthalocyanine tetrasulfonate was
purchased from Avocado. All other reagents were purchased from
Aldrich and used without further purification. NMR spectra were ob-
tained using a Bru¨cker 400 DPX spectrometer. CD spectra were mea-
sured on an Aviv 62DS spectrapolarimeter. UV-vis spectra were mea-
sured using an Agilent A453 spectrometer. MALDI-MS spectra were
recorded using an Applied Biosystems Voyager-DE Pro MALDI-TOF
mass spectrometer. Fluorescence spectra were recorded using a Hitachi
F-4500 fluorescence spectrophotometer. Horse heart cytochrome c was
solutions were standardized by UV using ꢀ415 ) 3.46 × 105 M-1 cm-1 19
.
Copper(II) Tetra-(4-sulfonatophenyl) Porphine 4b. A 10-fold
excess of copper(II) acetate was added to a solution of the free base
tetra-(4-sufonatophenyl) porphine in 95:5 methanol/water. After 12 h,
the copper complex was precipitated with acetone and filtered. The
filtrate was washed with acetone then washed off the filter with distilled
water. Incorporation of copper was confirmed and stock solutions were
standardized by UV using ꢀ413 ) 4.16 × 105 M-1 cm-1 28
.
(25) Larsen, R. W.; Omdal, D. H.; Jasuja, R.; Niu, S. L.; Jameson, D. M. J.
Phys. Chem. B 1997, 101, 8012-8020.
(26) Giovannetti, R.; Bartocci, V.; Ferraro, S.; Gusteri, M.; Passamonti, P.
Talanta 1995, 42, 1913-1918.
(27) Lu, M.; Guo, Q.; Pasternack, R. F.; Wink, D. J.; Seeman, N. C.; Kallenbach,
N. R. Biochemistry 1990, 29, 1614-1624.
9
12840 J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004

Catalytic Unfolding and Proteolysis of Cytochrome c
A R T I C L E S
Copper(II) Coproporphyrin-I 7b. To a standardized aqueous stock
solution of the free base coproporphyrin-I 7a was added 1 equivalent
of copper (II) acetate. Incorporation of copper was rapid requiring only
about 5 min to reach completion as confirmed by UV-vis (5 mM
phosphate, pH 7.4) λ 379, 530, 563 nm, which matches that reported
in the literature.26
temperature, after which time a red precipitate had formed and all green
color had vanished from the solution. The red solid was filtered and
washed with methanol, 0.01 M HCl, and water. The solid was dissolved
off the filter with DMF and evaporated to dryness yielding 12 mg of
product (90%); MALDI-TOF-MS m/e 1844.5 [M+H]+ calcd for
C96H76N12O24Cu(+H), found 1844.4; UV-vis (H2O, 5 mM phosphate,
pH ) 7.4) λ 416, 541, 576 nm.
Copper(II) Tetra-(4-N-methylpyridinium) Porphyrin 9b. To a
standardized aqueous stock solution of the free base tetra-(4-N-meth-
ylpyridinium) porphyrin 9a was added 1 equiv of copper (II) acetate
and the solution was allowed to stand at room-temperature overnight.
Completion of metalation was confirmed by UV-vis (5 mM phosphate,
pH 7.4) λ 425, 548 nm, which matches that reported in the literature.22
meso-Tetrakis-[(4-carboxyphenylamidomethyl-isophthaloyl)-
bis-iminodiacetate] Porphyrin 5a. 5-[(Benzyloxycarbonyl)amino-
methyl]isophthalic acid29 (250 mg, 0.76 mmol), di-tert-butyl-iminodi-
acetate (465 mg, 1.9 mmol) and 1-[3-(dimethylamino)propyl]-3-
ethylcarbodiimide hydrochloride (EDC) (306 mg, 1.6 mmol) were
placed in 25 mL of dry dichloromethane and stirred at room temperature
for 6 h. The solution was extracted successively with 30 mL each of
1 M HCl, saturated sodium bicarbonate and brine, dried over sodium
sulfate, and evaporated to dryness. The product was purified on a silica
column eluting with 1:1 hexanes/ethyl acetate yielding 375 mg (63%)
of the Z-protected product as a white solid. The product (375 mg, 0.73
mmol) was deprotected by catalytic hydrogenation in 30 mL of
vigorously stirred methanol with 10% Pd/C (25 mg, 0.023 mmol)
catalyst under 1 atm of hydrogen for 24 h. The solution was filtered
through diatomaceous earth and the methanol evaporated. The depro-
tected (aminomethyl-isophthaloyl)-bis-iminodiacetate tetra tert-butyl
ester was purified on a silica column eluting with dichloromethane with
0.5% triethylamine and an increasing fraction of methanol from 2% to
5% yielding the pure deprotected product as a colorless, sticky, solid
(265 mg, 88%): 1H NMR (400 MHz, CDCl3) δ ) 1.43 (s, 18H), 1.47
(s, 18H), 3.85 (s, 2H), 3.91 (s, 4H), 4.16 (s, 4H) 7.32 (s, 1H), 7.46 (s,
2H).
meso-Tetrakis-(4-carboxyphenyl-amidomethyl-benzoyl)-iminodi-
acetate) Porphyrin 3a. To a vigorously stirred solution of 4-(ami-
nomethyl)benzoic acid (1.0 g, 6.6 mmol) in 50 mL of water at 0 °C
was added, slowly, benzylchloroformate (1.1 mL, 7.3 mmol) and the
solution was warmed to room temperature and stirred for 16 h. The
solution was acidified to pH 3 with 1 M HCl and the product filtered
off as a pure white solid and washed with 50 mL each of water, ethanol
and diethyl ether yielding 1.8 g (95%). The Z-protected 4-(aminomethyl)
benzoic acid (260 mg, 0.91 mmol), di-tert-butyl-iminodiacetate (250
mg, 1.0 mmol) and 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide
hydrochloride (EDC) (190 mg, 0.98 mmol) were placed in 25 mL of
dry dichloromethane and stirred at room temperature for 4 h. The
solution was extracted successively with 30 mL each of 1 M HCl,
saturated sodium bicarbonate and brine, dried over sodium sulfate and
evaporated to dryness. The product was purified on a silica column
eluting with 2:1 hexanes/ethyl acetate yielding 375 mg (80%) of the
Z-protected product as a white solid. The product (375 mg, 0.73 mmol)
was deprotected by catalytic hydrogenation in 12 mL of vigorously
stirred methanol with 10% Pd/C (42 mg, 0.04 mmol) catalyst under 1
atm of hydrogen for 4 h. The solution was filtered through Celite and
the methanol was evaporated. The final product, di-tert-butyl-N-(4-
aminomethyl-benzoyl)-iminodiacetate, was purified on a silica col-
umn eluting with dichloromethane with 0.5% triethylamine and an
increasing fraction of methanol from 2.5% to 10% yielding the pure
deprotected product as a colorless, sticky, solid (250 mg, 69% overall
yield): 1H NMR (400 MHz, CDCl3) δ 7.39 (d, J ) 8.1 Hz, 2H), 7.34
(d, J ) 8.1 Hz, 2H), 4.16 (s, 2H), 3.95 (s, 2H), 3.90 (s, 2H), 1.47 (s,
9H), 1.44 (s, 9H).
To a stirred solution of meso-tetrakis-(4-carboxyphenyl) porphyrin
(0.10 mg, 0.126 mmol) in anhydrous dichloromethane (25 mL) was
added oxalyl chloride (390 uL, 4.4 mmol) and a catalytic amount of
DMF (1µL). The reaction mixture was then stirred overnight under an
inert atmosphere and dried for 8 h under high vacuum. The resultant
acid chloride was then dissolved in 8 mL of freshly distilled tetrahy-
drofuran to which was added a solution of (aminomethyl-isophthaloyl)-
bis-iminodiacetate tetra tert-butyl ester (0.27 g, 0.41 mmol) and
diisopropylethylamine (0.7 mL, 4.1 mmol) in 10 mL of dichlo-
romethane. The reaction mixture was allowed to stir under an inert
atmosphere for 15 min then evaporated to dryness. The residue was
dissolved in 40 mL of dichloromethane, washed successively with 10%
citric acid, 1 M NaOH, and saturated NaCl, dried over sodium sulfate
and evaporated to give 290 mg (98%) of crude product. The tert-butyl
protected product was purified in small batches (∼40 mg) on a column
of Brockman grade III basic alumina eluting with dichloromethane with
1% methanol resulting in a yield of 92%. 1HNMR (400 MHz, CDCl3)
δ ) -2.87 (bs, 2H), 1.47 (s, 72H), 1.50 (s, 72H), 4.07 (s, 16H), 4.17
(s, 16H), 4.74 (bs, 8H), 7.20 (s, 4H), 7.60 (s, 8H), 8.40 (s, 16H), 8.91
(s, 8H), 9.56 (bs, 4H).
To a solution of meso-tetrakis-(4-carboxyphenyl) porphyrin (63 mg,
0.08 mmol) in 10 mL of dry dichloromethane was added di-tert-butyl-
N-(4-aminomethyl-benzoyl)-iminodiacetate (120 mg, 0.32 mmol) and
1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide hydrochloride (EDC)
(86 mg, 0.35 mmol), and the mixture was stirred at room temperature
under nitrogen overnight. The reaction mixture was extracted succes-
sively with saturated sodium bicarbonate, 1 M sodium hydroxide and
brine, dried over sodium sulfate and evaporated to dryness. The tert-
butyl protected product was purified on a silica column eluting with
dichloromethane with an increasing fraction of methanol from 0% to
5%. Treatment with 95% TFA/H2O (5 mL) at room temperature for 4
h followed by evaporation afforded the triflate salt of the free acid
product as a dark green solid (104 mg, 65%): 1H NMR (400 MHz,
DMSO-d6) δ 9.40 (t, J ) 5.0 Hz, 4H), 8.80 (s, 8H), 8.29 (d, J ) 8.5
Hz, 8H), 8.26 (d, J ) 8.5 Hz, 8H), 7.47 (d, J ) 8.1 Hz, 8H), 7.29 (d,
J ) 8.1 Hz, 8H), 4.62 (d, J ) 4.7 Hz, 8H) 4.08 (s, 8H), 3.97 (s, 8H),
-3.00 (s, b, 2H); MALDI-TOF MS m/e 1783.5 [M+H]+ calcd for
C96H78N12O24(+H), found 1783.4; UV-vis (H2O, 5 mM phosphate,
pH ) 7.4) λ 416, 517, 554, 581, 635 nm.
Treatment of the tert-butyl protected compound (35 mg, 0.011 mmol)
with 95% TFA/H2O (5 mL) at room temperature for 4h followed by
evaporation afforded the triflate salt of the free acid 5a as a dark green
solid (28 mg, 100%). 1HNMR (400 MHz, CDCl3) δ ) -2.93 (bs, 2H),
4.07 (s, 16H), 4.17 (s, 16H), 4.68 (s, 8H), 7.18 (s, 4H), 7.54 (s, 8H),
8.33 (s, 16H), 8.33 (s, 16H), 8.88 (s, 8H), 9.49 (bs, 4H). MALDI-
TOF-MS m/e 2419.6 [M+H]+ calcd for C116H98N16O44(+H), found
2419.4. UV/vis (H2O, 5 mM phosphate, pH ) 7.4) λ 415, 518, 555,
581, 643 nm.
meso-Tetrakis-[(4-carboxyphenyl amidomethyl-isophthaloyl)-bis-
iminodiacetate] Copper(II) Porphyrin 5b. Compound 5a (20 mg,
0.0076 mmol) was dissolved in 4 mL of methanol and copper(II)
chloride (2.0 mg, 0.015 mmol) was added. The solution was stirred
for 24 h at room temperature, after which time a red product was
precipitated by addition of 15 mL of 0.1 M HCl. The red solid was
Copper(II) meso-Tetrakis-(4-carboxyphenyl amidomethyl-ben-
zoyl)-iminodiacetate) Porphyrin 3b. Compound 3a (15 mg, 0.0075
mmol) was dissolved in 5 mL of methanol and copper(II) chloride (2.5
mg, 0.019 mmol) was added. The solution was stirred for 3 h at room
(28) Gibbs, E.; Skowronek, W. R., Jr.; Morgan, W. T.; Muller-Eberhard, U.;
Pasternack, R. F. J. Am. Chem. Soc. 1980, 102, 3939-3944.
(29) Ritzen, A.; Frejd, T. Eur. J. Org. Chem. 2000, 3771-3782.
9
J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004 12841

A R T I C L E S
Groves et al.
filtered and washed with 0.01 M HCl and distilled water. The solid
was then washed through the filter with methanol and evaporated to
dryness yielding 16 mg of product (87%); MALDI-TOF-MS m/e
2480.5 [M+H]+ calcd for C116H96CuN16O44(+H), found 2480.4.
UV/vis (H2O, 5 mM phosphate, pH ) 7.4) λ 414, 541 nm.
Acknowledgment. We thank the National Institutes of Health,
Institute of General Medicine (GM35208) for financial support
of this work.
JA0317731
9
12842 J. AM. CHEM. SOC. VOL. 126, NO. 40, 2004