Kinetics and mechanism for CO2 scrambling in a N-carboxyimidazolidone analogue for N1-carboxybiotin

Article abstract of DOI:10.1021/ja026753y

The N-carboxyimidazolidone anion, 2^-, was prepared as an analogue for N¹-carboxybiotin, and the kinetics of its CO₂-dependent chemistry were studied in polar aprotic media. The objective was to assess the viability of unimolecular CO₂ elimination from N¹-carboxybiotin as a microscopic step in biotin-dependent carboxyl transfer enzymes. The anionic 2^- was prepared as its lithium salt by first deprotonating 2-imidazolidone with phenyllithium, followed by direct reaction with carbon dioxide. This procedure also permitted isolation of the ¹³C enriched derivative 2^-{¹³C} by reaction with ¹³ CO₂. Proton and ¹³C NMR and isotope-dependent FTIR measurements confirmed that carboxylation had occurred at the nitrogen atom of 2-imidazolidone to give 2^-. Time-dependent FTIR spectroscopy showed that Li2 undergoes carboxyl exchange with free carbon dioxide, with kinetics indicative of rate-limiting unimolecular dissociation of the N-CO₂ bond. Under these conditions, the weak Lewis acid Mg2+ catalyses the exchange of 2^- with free CO₂, which appears to be related to the ability of the metal ion to coordinate to 2^-. Reaction of Li2 with carboxylic acids in DMSO results in acid-dependent decarboxylation of 2^- with a rate that is dependent on the concentration of the acid and its pK_a. A common mechanistic framework is presented for both Lewis acid catalyzed carboxyl exchange and acid-dependent decarboxylation that involves initial interaction at the carbonyl oxygen and which has the effect of polarizing the nitrogen lone pair toward the imidazolidone ring rather than the carboxyl group. Lewis acid interaction with the carbonyl oxygen thus weakens the N-CO₂^- bond and functions as a trigger for dissociation of CO₂. In the context of biotin-dependent enzymes, this suggests a means by which the kinetically stable N¹-carboxybiotin cofactor intermediate might be triggered for dissociation of CO₂.

Full text of DOI:10.1021/ja026753y

Published on Web 08/31/2002
2
Kinetics and Mechanism for CO Scrambling in a
1
N-Carboxyimidazolidone Analogue for N -Carboxybiotin
Fiona J. Lihs and M. Tyler Caudle*
Contribution from the Department of Chemistry and Biochemistry, Arizona State UniVersity,
Tempe, Arizona 85287-1604
Received May 1, 2002
-
1
Abstract: The N-carboxyimidazolidone anion, 2 , was prepared as an analogue for N -carboxybiotin, and
the kinetics of its CO
assess the viability of unimolecular CO
dependent carboxyl transfer enzymes. The anionic 2 was prepared as its lithium salt by first deprotonating
2
-dependent chemistry were studied in polar aprotic media. The objective was to
1
2
elimination from N -carboxybiotin as a microscopic step in biotin-
-
2
-imidazolidone with phenyllithium, followed by direct reaction with carbon dioxide. This procedure also
13
-
13
13
13
2
permitted isolation of the C enriched derivative 2 { C} by reaction with CO . Proton and C NMR and
isotope-dependent FTIR measurements confirmed that carboxylation had occurred at the nitrogen atom of
-
2
-imidazolidone to give 2 . Time-dependent FTIR spectroscopy showed that Li2 undergoes carboxyl
exchange with free carbon dioxide, with kinetics indicative of rate-limiting unimolecular dissociation of the
+
-
N-CO
CO
2
bond. Under these conditions, the weak Lewis acid Mg2 catalyses the exchange of 2 with free
-
2
, which appears to be related to the ability of the metal ion to coordinate to 2 . Reaction of Li2 with
carboxylic acids in DMSO results in acid-dependent decarboxylation of 2 with a rate that is dependent on
-
a
the concentration of the acid and its pK . A common mechanistic framework is presented for both Lewis
acid catalyzed carboxyl exchange and acid-dependent decarboxylation that involves initial interaction at
the carbonyl oxygen and which has the effect of polarizing the nitrogen lone pair toward the imidazolidone
ring rather than the carboxyl group. Lewis acid interaction with the carbonyl oxygen thus weakens the
-
N-CO
2
bond and functions as a trigger for dissociation of CO
2
. In the context of biotin-dependent enzymes,
1
this suggests a means by which the kinetically stable N -carboxybiotin cofactor intermediate might be
2
triggered for dissociation of CO .
Introduction
a CO -equivalent, bicarbonate, to a carbon nucleophile, forming
2
a new carbon-carbon bond in the process. As a consequence,
Biotin is the critical organic cofactor in biotin-dependent
carboxyl transfer enzymes (BDCs), which are important in
mammalian and avian fatty acid biosynthesis.1-3 Biotin, when
model studies have sought to mimic fundamental chemical steps
4-6
believed to occur in the enzyme. Many of these have centered
on 2-imidazolidone, 1, and related derivatives as a model for
the biotin headgroup. The general consensus arising from these
chemical studies is (1) that biotin itself is a poor nucleophile
covalently bound to the BDC enzyme complex, functions to
shuttle a carboxyl group between two independent enzyme
active sites. In the first active site, called biotin carboxylase,
the biotin cofactor is carboxylated at the N1 position by
accepting a carboxyl group from bicarbonate. The functionalized
7
and requires preactivation, probably by proton transfer, in order
1
to accept the carboxyl group and (2) that N -carboxybiotin is a
1
N -carboxybiotin is then translocated to the second active site,
kinetically poor carboxyl donor and probably requires enzymatic
called carboxyl transferase, where it transfers the carboxyl group
to a terminal acyl substrate. This latter reaction has been
(4) Kluger, R. A.; Tsao, B. J. Am. Chem. Soc. 1993, 115, 2089-2090. Musashi,
Y.; Hamada, T.; Sakaki, S. J. Am. Chem. Soc. 1995, 117, 11320-11326.
Cravey, M. J.; Kohn, H. J. Am. Chem. Soc. 1980, 102, 3928-3939. Kohn,
H. J. Am. Chem. Soc. 1976, 98, 3690-3694. Kondo, H.; Miura, K.;
Sunamoto, J. Tetrahedron Lett. 1982, 23, 659-662. Botella, J. M.; Klaebe,
A.; Perie, J.; Monnier, E. Tetrahedron 1992, 48, 5111-5122. Caplow, M.
J. Am. Chem. Soc. 1965, 87, 5774-5785. Bruice, T. C.; Hegarty, A. F.
Proc. Natl. Acad. Sci. U.S.A. 1970, 65, 805-809.
1
postulated to occur via initial elimination of CO2 from N -
carboxybiotin in the carboxyl transferase active site,2 Scheme
,3
1
, but there is presently no chemical basis for this hypothesis.
Thus, the gross mechanistic features of this enzymatic process
1
have been characterized, but many chemical details of N -
(
5) Caplow, M.; Yager, M. J. Am. Chem. Soc. 1967, 89, 4513-4521.
(
6) Mori, H.; Makino, Y.; Yamamoto, H.; Kwan, T. Chem. Pharm. Bull. 1973,
carboxybiotin formation, translocation, and decarboxylation in
BDC enzymes are still unresolved.
The fundamental chemistry of BDC enzymes is of interest
since these enzymes affect a net transfer of carbon dioxide from
2
1, 915-916. Otsuji, Y.; Arakawa, M.; Matsumura, N.; Haruki, E. Chem.
Lett. 1973, 1193-1196. Sakurai, H.; Shirahata, A.; Hosomi, A. Tetrahedron
Lett. 1980, 221, 1967-1970. Matsumura, N.; Asai, N.; Yoneda, S. J. Chem.
Soc., Chem. Commun. 1983, 1487-1488.
(7) Attwood, P. V.; Tipton, P. A.; Cleland, W. W. Biochemistry 1986, 25,
8197-8205. Tipton, P. A.; Cleland, W. W. Biochemistry 1988, 27, 4317-
To whom correspondence should be addressed. E-mail: tcaudle@asu.edu.
4325. Fry, D. C.; Fox, T. L.; Lane, M. D.; Mildvan, A. S. J. Am. Chem.
Soc. 1985, 107, 7659-7665. Perrin, C. L.; Dwyer, T. J. J. Am. Chem. Soc.
1987, 109, 5163-5167. Levert, K. L.; Lloyd, R. B.; Waldrop, G. L.
Biochemistry 2000, 39, 4122-4128.
(
(
(
1) Jitrapakdee, S.; Wallace, J. C. Biochem. J. 1999, 340, 1-16.
2) Knowles, J. R. Annu. ReV. Biochem. 1989, 58, 195-221.
3) Attwood, P. V.; Wallace, J. C. Acc. Chem. Res. 2002, 35, 113-120.
11334
9
J. AM. CHEM. SOC. 2002, 124, 11334-11341
10.1021/ja026753y CCC: $22.00 © 2002 American Chemical Society

Kinetics and Mechanism for CO
2
Scrambling
A R T I C L E S
Scheme 1
Scheme 2
Lithium N-Carboxy-2-imidazolidone. (Li2) 1.0 g (11.6 mmol)
-imidazolidone, 1, was suspended in 50 mL anhydrous THF. One
2
equivalent of 2.0 M phenyllithium solution (5.8 mL, 11.6 mmol) was
added dropwise while stirring. Once all of the phenyllithium had been
added, the reaction was stirred under one atm of dry CO
2
gas for 12 h.
The resulting white precipitate was collected by filtration using Schlenk
-
1
techniques. IR (KBr, cm ) 1716, 1633, 1340, 1255, 1172, 810, 730.
1
H NMR (DMSO, ppm) δ 3.16(triplet), 3.6(triplet), 6.1(singlet). The
activation to facilitate N-CO2 bond cleavage. It is therefore
1
H NMR spectrum also shows the presence of unreacted 1 and about
equiv water associated with Li2 in the product. Analysis found (calc)%
for (Li2)(1)0.14(H O) : C, 31.1(31.6); H, 4.2(4.8); N, 18.7(19.0).
1
important to understand factors affecting the reactivity of N -
1
carboxybiotin, which may be relevant to its enzymatic catalysis.
2
1.1
1
13
13
The aqueous chemistry of N -carboxybiotin decarboxylation
Lithium N-{ C}Carboxy-2-imidazolidone. (Li2{ C}) 0.10 g (1.16
mmol) 2-imidazolidone, 1, was suspended in 10 mL anhydrous THF.
One equivalent of 2.0 M phenyllithium solution (0.58 mL, 1.16 mmol)
shows a dependence on pH indicative of both acid-dependent
8
and acid-independent pathways. We have sought to characterize
was added dropwise while stirring. Once all of the phenyllithium had
these pathways in nonaqueous media for several reasons. First,
nonaqueous solvents have been shown to increase the rate for
decarboxylation of N-carboxyimidazolidone, suggesting desol-
been added, the reaction was stirred under one atm dry 1 CO
3
gas for
2
1
0 min. The resulting white precipitate was collected by filtration using
13
1
Schlenk techniques. For Li2{ C}: H NMR (DMSO, ppm) δ 3.16, 3.6,
1
9
vation of N -carboxybiotin may play a role in carboxyl transfer.
-1
6
.1 IR (KBr, cm ) 1716, 1600, 1340, 1255, 1172, 805, 730.
Second, Lewis acid interactions with the carboxyimidazolidone
functional group, including proton transfer and metal binding,
can be more directly controlled in aprotic media, enabling more
detailed study of these interactions.
2
CO Exchange Kinetics. Kinetics were measured using a two-
syringe rapid mixing apparatus connected to an infrared-transparent
flow cell with ZnS windows. The flow cell was mounted inside a
Nicolet Avatar 360 Fourier transform infrared spectrometer. Reaction
was initiated when the two reagent solutions were rapidly mixed and
directed into the flow cell. The time-dependent infrared spectrum was
In this paper, we report on the preparation of N-carboxyimi-
-
dazolidone, 2 , directly from 2-imidazolidone, 1, Scheme 2,
-
1
2
monitored between 2380 and 2230 cm , the CO region of the
using carbon dioxide as the CO2 source. This molecule is a direct
1
spectrum. Kinetic decay curves were extracted from the time-dependent
spectra and fit to a first-order exponential decay to derive the observed
analogue for the headgroup of the N -carboxybiotin intermediate
-
in BDC enzymes. We show that 2 undergoes carboxyl
rate constant, kobs
.
exchange with free CO2 and that the kinetics of this reaction
A sealed gas control apparatus was used to prepare a solution of
clearly indicate rate-limiting N-CO2 bond cleavage. Compound
1
3
DMSO preequilibrated with approximately 760 mmHg of CO
2
, which
-
2
can be protonated by enzymatically relevant Brønsted acids
10
equates to 0.130 M CO
2
in solution. This solution served as a stock
resulting in the decarboxylation of Li2 two orders of magnitude
solution for the preparation of reagent solutions of varying concentra-
-
more rapidly than carboxyl exchange on 2 . Catalysis of
tions of 13CO . The desired reagent solution was mixed with a solution
2
2
+
-
carboxyl exchange by the weak Lewis acid Mg is also
examined as a general model for noncovalent interactions with
of 2 and delivered to the IR cell using the apparatus described above.
The same procedure was used for the magnesium-dependent experiment,
1
except that the solution of 2
-
was premixed with varying concentrations
N -carboxybiotin that may be important in CO2 transfer by BDC
of Mg2 . For decarboxylation kinetic experiments, the sample of 2 in
+
-
enzymes.
DMSO was reacted with the appropriate carboxylic acid and the time-
Experimental Section
dependent IR spectrum in the CO
2
region was monitored. In this case,
in solution was observed.
only the formation of free 12CO
2
Materials. Phenyllithium as a 2 M solution in cyclohexane and ether
was obtained from Acros. The 2-imidazolidone precursor was obtained
from Aldrich, recrystallized from tetrahydrofuran (THF), and stored
under nitrogen. THF was distilled from sodium benzophenone ketyl
and stored under nitrogen. Dimethyl sulfoxide (DMSO) and DMSO-
Results
The reaction of lithium imidazolidonate, Li3, with gaseous
carbon dioxide in THF results in precipitation of a white solid,
Li2, Scheme 2, which evolves carbon dioxide upon treatment
6
d were vacuum-distilled from calcium hydride and stored under
1
3
1
13
nitrogen. CO
2
gas (99%) was obtained from Cambridge Isotope
with strong acid. CHN and H and C NMR spectroscopy
suggested that the isolated product is an approximately 88:12
mixture of Li2 and unreacted 1, which could not be separated.
Laboratories. All preparative and analytical manipulations were carried
out under nitrogen either in a nitrogen-filled glovebox or using standard
Schlenk apparatus.
(
10) Welford, P. J.; Brookes, B. A.; Wadhawan, J. D.; McPeak, H. B.; Hahn,
C. E. W.; Compton, R. G. J. Phys. Chem. B 2001, 105, 5253-5261.
Gennaro, A.; Isse, A. A.; Vianello, E. J. Electroanal. Chem. Interfacial
Electrochem. 1990, 289, 203-215.
(
8) Tipton, P. A.; Cleland, W. W. J. Am. Chem. Soc. 1988, 110, 5866-5869.
9) Rahil, J.; You, S.; Kluger, R. J. Am. Chem. Soc. 1996, 118, 12495-12498.
Gao, D.; Pan, Y.-K. J. Org. Chem. 1999, 64, 1151-1159.
(
J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002 11335

A R T I C L E S
Lihs and Caudle
Figure 1. (a) Infrared spectra of 1 (dashed) and 2 (solid), showing
-
1
appearance of the -CO2 bands (*) at 1633 and 810 cm . (b) Infrared
1
3
13
spectra of Li2 (solid) and Li2{ C} (dashed) showing shifts in the CO2
13
sensitive bands. (c) Difference spectrum (Li2 - Li2{ C}). Infrared spectra
were measured in the solid state as KBR pellets.
Washing with hot THF reduced the amount of 1 somewhat but
did not result in a homogeneous sample of Li2. The infrared
spectrum of the product shows formation of two important new
-
1
bands at 1625 and 803 cm that are absent in 1, Figure 1a.
These are characteristic of the asymmetric stretching and
-
-
scissoring modes, respectively, of the -CO2 moiety in 2 . In
13
13
the spectrum of 2{ C} prepared from CO2, these modes shift
to lower energy as demonstrated in Figure 1b and c, confirming
that the carboxyl group arises from CO2 gas and that the IR
1
Figure 2. 300 MHz H NMR spectrum of Li2 in DMSO-d6.
-
1
assignments are correct. The carbonyl band at 1700 cm
remains nearly unchanged by isotope labeling at the carboxyl
carbon, demonstrating the carboxylation does not directly
involve the carbonyl group but rather occurs at the nitrogen
center. These isotope-dependent vibrational assignments are
1
1
consistent with those made by Clarkson and Carey.
The proton NMR spectrum of Li2 in DMSO-d6, Figure 2,
also demonstrates carboxylation at the nitrogen center. The
-
triplet signals from 2 at 3.16 and 3.60 ppm show that the
methylene protons are inequivalent on the NMR time scale, but
are equivalent in 1. We also see a new resonance at 6.85 ppm
that is not present in 1 and that integrates to a single proton
-
when compared with the ethylene signals of 2 , and which we
-
assign to the -NH proton of 2 . There is no evidence from
1
3
either NMR data or from compositional analysis for the
formation of any dicarboxylated complex. We therefore for-
mulate Li2 as shown in Scheme 2.
Figure 3.
C NMR spectra of Li2 in DMSO-d6 containing natural
abundance 13C (bottom) and after exposure to CO2 (top).
13
To address this problem, we used Fourier transform infrared
spectroscopy to track the CO / CO exchange. This technique
2 2
took advantage of the fact that CO2 and CO2 exhibit separate
12
13
At the limit of solubility in DMSO, the 13C NMR signal
arising from the carbamato carbon in unenriched Li2 is not
12
13
-
1
1
3
infrared bands at 2337 and 2271 cm , respectively, which are
both within the IR solvent window of DMSO. For 13CO2
absorption, we should observe a simple decrease in the intensity
observable. However, when the solution is exposed to CO2
1
3
gas, a C NMR signal at 152 ppm is clearly observed, Figure
. This indicates that the carboxyl group in Li2 is exchanged
3
1
3
13
2
of the free CO band whereas for carboxyl exchange, loss of
with free CO2 on this time scale, resulting in enrichment of
the carboxyl carbon in 13C. Time-dependent NMR spectra
1
3
2
free CO will be accompanied by a concomitant increase in
CO2. FTIR had the additional advantage of permitting us to
collect real-time data on the 1-h time scale, so we were able
obtain kinetic information about the reaction as well.
When a solution of Li2 is rapidly mixed with a solution
containing dissolved CO , the time-dependent FTIR spectra
in Figure 4 are obtained. These data show a decrease in the
intensity of the CO2 band at 2271 cm and a concomitant
increase in the CO2 band at 2337 cm . When the intensity at
1
2
showed that this reaction was complete within about 30-60
13
min, making C NMR unsuitable for measurement of rate data
1
3
since each C NMR spectrum took at least 6 min to collect.
Furthermore, the NMR experiment does not distinguish a
carboxyl exchange process from a simple 13CO2 absorption
1
3
2
1
2
process since we cannot directly observe formation of CO2.
1
3
-1
1
2
-1
(11) Clarkson, J.; Carey, P. R. J. Phys. Chem. 1999, 103, 2851-2856.
1
1336 J. AM. CHEM. SOC. VOL. 124, NO. 38, 2002
9

Kinetics and Mechanism for CO
2
Scrambling
A R T I C L E S
Figure 4. Time-dependent FTIR spectra in the free CO2 region upon mixing
1
3
Li2 with CO2 in DMSO. The arrows show decrease in absorbance at 2271
Figure 6. Observed first-order rate constant for CO exchange as a function
2
-
1
13
-1 12
2+
+
cm ( CO2) and increase in absorbance at 2337 cm ( CO2) with time.
of [Mg ], (b), and [TBA ], (9). Both cations were added as their triflate
2+
salts. Solid line shows fit to kobs ) 2kf′K[Mg ]o. [Li2] ) 0.04 M, [CO2]
)
0.035 M, T ) 25 °C.
-
to zero. The CO2 scrambling reaction is first-order in [2 ] and
zero-order in [CO2]. Because NMR spectra show the presence
-
of uncarboxylated 1 in DMSO solutions of 2 , the dependence
of the CO2 scrambling rate on the presence of excess 1 was
measured. When one or two additional equivalents of excess 1
were added to the reaction mixture, the rate of carboxyl
exchange was unchanged. This shows that the rate is indepen-
dent of the presence of 1, consistent with the general observation
that there is no direct reaction between 1 and CO2.
The countercation affects the CO2 exchange rate as shown
by measuring the carboxyl exchange rates in the presence of
-
Mg(OTf)2 (OTf ) trifluoromethanesulfonate ) triflate) and
2+
tetrabutylammonium triflate. For [Mg ] between zero and 0.10
M, we find that the carboxyl exchange rate is increased linearly
Figure 5. Observed first-order rate constants kobs for carboxyl exchange
by Li2 in DMSO as a function of total [CO2]. The solid line shows a linear
fit to the data, giving kobs ) 0.083(5) min + 0.0 M min [CO2]. [Li2]
2
+
as [Mg ] increases, Figure 6. This clearly results from the
-
1
-1
-1
2+
addition of the Mg cation since the tetrabutylammonium
)
0.040 M, T ) 25 °C. (inset) Decay curves showing time-dependent
-
1
-1
triflate salt had no effect on the CO2 exchange rate. Magnesium-
catalyzed CO2 exchange is also independent of the concentration
of CO2 as expected. The exchange reaction was also carried
out in the presence of Mg(OTf)2 and an excess amount of TBA-
(OTf). These catalyzed rates remained unchanged compared
2
decrease at 2271 cm (0) and increase at 2337 cm (O). Solid lines are
-
kobst
fit to the equation (A - A∞) ) (Ao - A∞)e
to obtain kobs.
these two wavelengths is plotted against time, the decay curves
in Figure 5(inset) are obtained. The decrease in intensity at 2271
-
1
-1
cm is mirrored by the increase at 2337 cm . When the data
are modeled as a first-order exponential decay, an excellent fit
to the experimental data is obtained and essentially identical
observed rate constants, kobs, are obtained for loss of free
with those containing only Mg(OTf) , except at very high
-
concentrations of OTf , which slightly inhibited magnesium
2
+
catalysis. Our interpretation of this result is that the Mg
-
catalysis requires that 2 bind to it and that the inhibitory effect
1
3
12
CO2 and gain of CO2. These data taken together confirm
of high concentrations of triflate is a result of competition for
2+
that we are observing a CO2 exchange process in which the
the Mg . This hypothesis is also consistent with the observation
that polydentate magnesium chelators such as 15-crown-5
(1,4,7,10,13-pentaoxacyclopentadecane) also inhibit catalysis of
carboxyl exchange by the magnesium ion. A substantial aging
effect in these experiments was also observed. When the
carboxyl scrambling rate was measured immediately after
-
carboxyl group in 2 is scrambled with free dissolved CO2. The
time scale for equilibration of the system measured by FTIR is
consistent with the NMR data in showing that the reaction is
complete in about 1 h.
Using the gas transfer device described in the Experimental
Section, we prepared reagent solutions of varying concentration
of dissolved 13CO2, which were used in studying the dependence
of kobs on total CO2 concentration, [CO2]o. These experiments
showed that the time-dependent loss of 13CO2 or gain in 12CO2
consistently fit a first-order exponential decay over the range 0
2
+
mixing Li2 with Mg , the reaction was rapid. However, if
2
+
samples of Li2 and Mg were permitted to age longer than
thirty minutes before kinetic measurements, no catalytic effect
was observed. This suggests a magnesium-dependent aggrega-
-
tion reaction of 2 in solution that yields a product inert to
carboxyl exchange. Hence all magnesium-dependent kinetics
were measured immediately after mixing Li2 and Mg .
<
[CO2]o < 0.065 M. Moreover, when kobs was correlated with
2+
total [CO2]o, Figure 5, we found that the reaction is zero order
in [CO2]o. Fitting a straight line to the data in Figure 5 gave a
y-intercept of 0.083(5) min-1 and a slope that is essentially equal
-
The kinetics for scrambling of the carboxyl group in 2 with
free CO2 suggests a unimolecular pathway for decarboxylation
J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002 11337

A R T I C L E S
Lihs and Caudle
Scheme 3
protonation is then to be expected since the free carbamic acid
H2 is unstable with respect to decarboxylation and rapidly loses
CO2 to give 1. The observation that carboxyl transfer is
2
+
catalyzed by Mg and that the decarboxylation rate is proton-
-
dependent for 2 in aprotic media suggests that a common
mechanistic framework may be operative in both reactions and
that this might be important in the carboxyl transfer reactions
catalyzed by biotin enzymes.
Discussion
The mechanism for uncatalyzed carboxyl exchange on 2-
must be consistent with the kinetics that shows that the observed
first-order rate constant is independent of [CO2]. CO2 scrambling
therefore proceeds via a pathway in which the entering CO2
molecule inserts after the rate-limiting step. This precludes any
mechanism involving a dicarboxylated intermediate or a con-
certed interchange pathway. The kinetic data is therefore most
consistent with a rate-determining step involving a unimolecular
Figure 7. Observed first-order rate constant for decarboxylation of Li2 in
DMSO as a function of acid concentration for benzoic acid (b) and acetic
acid (9). Solid lines show fits to eq 5. [Li2] ) 0.040 M, T ) 25 °C.
-
-
dissociation of 2 , generating free CO2 and the ureido anion
of 2 in aprotic media. On the basis of this, we predicted that
-
-
3
as steady-state intermediates, as shown Scheme 3a. This
decarboxylation of 2 by acids should be independent of the
-
pathway leads to carboxyl scrambling because reaction of 3
acid concentration and should occur at a rate similar to that for
carboxyl exchange. To probe this mechanistic issue, acid-
promoted decarboxylation of Li2 by acetic and benzoic acid
was studied. Rapidly mixing a DMSO solution of Li2 with a
solution of either acid produces CO2, whose formation was
monitored by FTIR. The time-dependent absorption increase
1
2
13
-
occurs with either CO2 or CO2 to regenerate 2 or give
-
13
2
{ C}. Analysis of Scheme 3a yields the basic rate expression
in eq 1.
d[CO2]
-
)
k ([2 ] - [CO ]) - k [CO ]
(1)
f
o
2
r
2
-
1
dt
at 2337 cm exhibits a first-order exponential rise, indicating
that the reaction is first-order in Li2. However, the first-order
rate constant is acid-dependent, Figure 7, and increases as the
Neglecting the small 13C isotope effect, we can assume that kf
)
kr, which gives eq 2.
-
concentration of acid is increased. Decarboxylation of 2 by
benzoic is rapid and exhibits saturation behavior. The acetic
acid promoted decarboxylation is slower and exhibits a nearly
linear dependence on [HA], consistent with the higher pKa of
acetic acid. Both of these observations are inconsistent with
simple acid-independent decarboxylation followed by protona-
tion of the ureido anion and instead suggests that protonation
occurs prior to C-N bond cleavage. The rapid loss of CO2 upon
d[CO2]
-
)
k [2 ] - 2k [CO ]
(2)
f
o
f
2
dt
Equation 2 is a linear inhomogeneous differential equation which
yields a first-order exponential decay as its solution with kobs
) 2kf. Scheme 3a therefore gives a rate law consistent with the
observed [CO2]-independence of kobs. The average value for kobs
11338 J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002

Kinetics and Mechanism for CO
2
Scrambling
A R T I C L E S
over all of the studied CO2 concentrations is 0.083(2) min-1 so
that eq 2 gives a first-order rate constant for cleavage of the
reaction mechanism involving the Mg ion interacting with
2
2+
-
prior to the dissociation of CO2, Scheme 3b. Catalysis is
then achieved by the preferential stabilization of the transition
state leading to Mg3 , which could result from the ability of
the magnesium ion to relieve the buildup of negative charge
on the nitrogen atom as the CO2 dissociates. The magnesium
dependent pathway would then control the rate of carboxyl
exchange in the presence of Mg . Scheme 3b yields the rate
law in eq 3, which assumes that magnesium binding is weak,
-
1
C-N bond of kf ) 0.041(2) min .
1
+
Unimolecular dissociation of free CO2 from N -carboxybiotin
has been suggested as a critical step in carboxyl transfer in biotin
enzymes.2 This has the consequence of generating the highly
,3,8
-
basic biotinate anion, which is analogous to 3 . The biotinate
2
+
anion could then function as the base for generating the
carbanionic substrate intermediate, leading to a mechanism for
CO2 transfer as shown in Scheme 1. This would be consistent
1
3
that 2{ C} ) [CO2], and that kf′ ) kr′.
-
with the work of Stubbe et al. showing that loss of F from the
d[CO2]
substrate analogue â-fluoropropionyl-CoA is catalyzed by
)
k ′[2] [Mg] K - 2k ′K[Mg] [CO ]
(3)
f
o
o
f
o
2
propionyl-CoA carboxylase, but only under conditions where
dt
1
12
N -carboxybiotin is generated by the biotin carboxylase. It has
Equation 3 gives kobs ) 2kf′K[Mg]o which is consistent with
also been shown that biotin-dependent transcarboxylase cata-
2
+
1
3
the observed linear dependence on [Mg ]. A value of 2kf′K )
lyzes proton exchange between pyruvate and propionyl-CoA,
-1
-1
2+
4
2(2) M min is obtained from the slope of the plot of [Mg ]
suggesting that biotin carries protons between the active subsites.
2
+
1
against kobs. We know that Mg binding is weak since the plot
does not exhibit saturation behavior, and we are unable to
calculate K from the kinetic data. Therefore, kf′ cannot be
determined quantitatively, but it must be considerably greater
than kf if eq 3 is to hold. The catalytic effect of magnesium ion
in nonaqueous solution contrasts with the inhibitory effect of
divalent metal ions on decarboxylation of N-carboxyimidazoli-
Thus, N -carboxybiotin appears to be required for generation
of the substrate enolate. Critical to the validity of the enzymatic
mechanism in Scheme 1 is that unimolecular CO2 elimination
be equal to or faster than the steady-state turnover rate for
-
1
product formation, which is as high as 660 min for pyruvate
1
4
carboxylase. In as much as the unimolecular dissociation rate
for N-carboxyimidazolidone is reflective of that expected for
5
1
done in aqueous solution. In the latter case, the decreased
N -carboxybiotin, the uncatalyzed reaction is 4 orders of
2
+
2+
decarboxylation rates in the presence of Mn and Cu may
be more related to their competition for protonation sites, and
hence they inhibit decarboxylation by inhibiting protonation of
N-carboxyimidazolidone.
The free acid of N -carboxybiotin is unstable to decarboxy-
lation in aqueous solution, and this occurs at a rate substantially
faster than acid-independent carboxyl exchange. Therefore, an
alternative mechanism for the carboxyl transferase would
involve protonation of N -carboxybiotin, followed by elimination
magnitude too slow to be kinetically viable. Hence, an important
function of the carboxyltransferase component of the carboxy-
1
lase enzymes may be to catalyze the dissociation of N -
1
carboxybiotin. Indeed, N -carboxybiotin is unstable in the
1
carboxyl transferase subsite and is decarboxylated in a hydrolytic
leak when no bound substrate is available.15 Recent measure-
8
ments on the lifetime of the carboxylated 1.3S subunit of
16
transcarboxylase give t1/2 ≈ 8 h. This is not competitive with
the steady-state reaction turnover and suggests that the hydrolytic
leak is probably not the result of spontaneous decarboxylation
1
of CO2. This process is distinguished from Scheme 1 by the
fact that proton transfer to biotin occurs prior to CO2 elimination.
This latter hypothesis is chemically viable as shown by the fact
1
of N -carboxybiotin before it reaches the second subsite.
2+
We observed that the weak Lewis acid Mg catalyzed the
-
-
that 2 reacts with carboxylic acids to eliminate CO2 at a rate
exchange of 2 with free CO2, as shown by the data in Figure
that is more rapid than the unimolecular dissociation reaction.
The overall acid-dependent decarboxylation rate in DMSO is
dependent on the concentration of the acid and on its pKa as
shown in Figure 7. This suggests that the acid-promoted
6. On the basis of control experiments, this phenomenon would
appear to be related to the ability of the metal ion to coordinate
-
-
to 2 . Given the ionic nature of the interactions between 2
2+
+
2+
and Mg or Li , it is to be expected that coordination of Mg
-
decarboxylation of 2 occurs via an acid-dependent proton-
would be favored by its higher charge/radius ratio. This
experiment was partly inspired by the previously reported
transfer preequilibrium. Fast proton transfer gives H2, followed
by rate-limiting dissociation of CO2, Scheme 3c. The rate law
for this reaction is given in eq 4.
-
observation that metal cations stabilized 2 to decarboxylation
5
in aqueous solution. However, we find that metal ions such as
2+
-
Zn result in decarboxylation of 2 in DMSO, which probably
d[CO2]
arises from its greater Brønsted acidity in the presence of trace
)
2+
dt
water when compared to Mg under similar conditions. Thus,
the interaction of 2 with Mg2+ is more easily interpreted in
the context of the uncatalyzed carboxyl exchange reaction.
The rate of carboxyl exchange exhibits a linear dependence
on the concentration of the metal ion present and suggests a
-
K (1 + x) - (K (1 + x) - 4(K - 1)K x)
H
x
H
H
H
k2
([2] -
o
[
]
2(K
- 1)
H
[
CO ]) (4)
2
(
(
(
(
(
12) Stubbe, J.; Fish, S.; Abeles, R. H. J. Biol. Chem. 1980, 255, 236-242.
Stubbe, J.; Abeles, R. H. J. Biol. Chem. 1977, 252, 8338-8340.
-
The expression in large brackets gives the fraction of 2 in its
protonated form and was derived on the basis of the proton-
dependent preequilibrium for formation of H2, in which x )
[HA]o/[2]o. Since all of the quantities in large brackets are time-
independent, eq 4 is a first-order differential equation and gives
an observed first-order rate constant of
13) Rose, I. A.; O’Connell, E. L.; Solomon, F. J. Biol. Chem. 1976, 251, 902-
9
04.
14) Legge, G. B.; Branson, J. P.; Attwood, P. V. Biochemistry 1996, 35, 3849-
3
856.
15) Wallace, J. C.; Phillips, N. B.; Snoswell, M. A.; Goodall, G. J.; Attwood,
P. V.; Keech, D. B. Ann. N.Y. Acad. Sci. 1985, 447, 169-188.
16) Rivera-Hainaj, R. E.; Pusztai-Carey, M.; Reddy, D. V.; Choowongkomon,
K.; S o¨ nnichsen, F. D.; Carey, P. R. Biochemistry 2002, 41, 2191-2197.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002 11339

A R T I C L E S
Lihs and Caudle
simple aliphatic carbamates, C-N bond rotation has the effect
of localizing the lone pair onto the nitrogen atom to a greater
degree, making the nitrogen atom more nucleophilic and thus
subject to protonation and decarboxylation. Indeed, such zwit-
terionic species are believed to be important in the acid-
K (1 + x) - (K (1 + x) - 4(K - 1)K x)
H
x
H
H
H
kobs ) k2
(5)
[
]
2
(K - 1)
H
Equation 5 results in saturation behavior for kobs since the
expression in the brackets increases from 0 to 1 as [HA]o
increases. Equation 5 was fit to the data in Figure 7 to obtain
20
dependent decarboxylation of simple aliphatic carbamates.
1
However, the nitrogen lone pair in N -carboxybiotin is still partly
-
1
a value for the terminal rate constant k2 ) 23(2) min , which
is the rate constant for unimolecular elimination of CO2 from
H2. Assuming that this value is the same whether proton transfer
occurs from acetic or benzoic acid, the slower rates for acetic
acid dependent decarboxylation result from a smaller KH value
for acetic acid (KH ) 0.022(5) M ) versus benzoic acid (KH
15(4) M ). This is in turn consistent with the higher pKa
for acetic acid compared with benzoic acid in DMSO.
delocalized into the carbonyl group even when the carboxyl
group is orthogonal to the ureido ring. Thus, loss of N-CO2
-
1
double bond character in 2 or N -carboxybiotin may not make
the nitrogen appreciably more nucleophilic. This might explain
-
why carboxyl exchange of 2 is several orders of magnitude
-
1
slower than for the simple aliphatic carbamates such as lithium
diethylcarbamate, which has a CO2 exchange rate constant of
-
1
)
1
7
-
1
21
at least 50 min under the same conditions, Figure S1. The
The CO2 exchange experiments and the acid-dependent
decarboxylation together suggest two pathways for elimination
1
-
rate for CO2 elimination from N -carboxybiotin and 2 is slowed
since the nitrogen lone pair is not as available for protonation
as in a simple alkylcarbamate.
-
1
of CO2 from 2 and, by inference, from N -carboxybiotin. An
acid-independent pathway is operative in the uncatalyzed
carboxyl exchange experiments and a more rapid acid-dependent
pathway is operative in acid-dependent decarboxylation. The
rate for elimination of CO2 from H2 is almost 3 orders of
Inasmuch as rapid carboxyl exchange would facilitate car-
boxyl transfer in the carboxylase enzymes, it is counterintuitive
that the biochemical cofactor eliminates CO so much more
2
-
magnitude more rapid than from 2 . This is consistent with the
slowly than simple aliphatic carbamates, which could be
biosynthesized from much simpler precursors. Indeed, at least
1
general stability of N -carboxybiotin and N-carboxyimidazoli-
done at neutral and higher pH and suggests that CO2 release
could then be controlled by the enzymatic system to occur only
in the carboxyl transferase active site by enzyme-facilitated
three enzymes employ N-carboxylation of the amino side chain
of a lysine residue to generate a carbamate group for the purpose
22-24
of coordinating a metal ion.
These carbamates are prepared
1
catalysis or proton transfer to N -carboxybiotin.
from a readily available endogenous protein residue and would
Comparing Scheme 3b and 3c, we see that the proposed
mechanisms for the magnesium-catalyzed carboxyl exchange
and the acid-dependent decarboxylation are chemically similar.
In both cases, a Lewis acid coordinates to the carbonyl oxygen.
The predicted effect is to polarize the C-N double bond away
from the carboxyl group and more toward the carbonyl group.
1
appear to be kinetically more reactive than N -carboxybiotin.
However, the organization of the BDC enzymes into two
1
independent active sites coupled only by N -carboxybiotin led
to the hypothesis that the two subunits may have originally
2,8
evolved as separate enzymes. In this case, a carboxyl transfer
reagent would be required that was sufficiently stable to
decarboxylation to freely diffuse between two isolated proteins.
There would then be a chemical advantage to having a carboxyl
carrier that was more stable to dissociation but whose CO2
elimination could be triggered, for example, by interaction with
a Lewis acid or by proton transfer. This stability can also serve
-
This would weaken the N-CO2 bond and thereby facilitate
dissociation of CO2. Thus, Scheme 3b and 3c are similar up
through the point of CO2 dissociation. However, in Scheme 3c
CO2 dissociation is irreversible because of the very large forward
driving force resulting from the tautomerization equilibrium that
strongly favors the formation of 1. While 1† should be reactive
toward CO2,18 tautomerization (i.e., migration of the proton from
O to N) competes with carboxylation, thereby preventing reentry
1
to minimize abortive decarboxylation prior to N -carboxybiotin
translocation and substrate binding. Our chemical studies lend
some support to this idea by showing that unimolecular
elimination of CO2 from the N-carboxyimidazolidone functional
group is slow but can be catalyzed by noncovalent interactions
with the ureido headgroup. Indeed, a kinetic analysis of the
carboxyltransferase subunit of acetyl-CoA carboxylase has
identified one or more ionizable residues in the active site that
+
of CO2. The more ionic interactions in Mg3 mean that even if
migration of Mg2 from O to N occurs, Mg3 remains reactive
+
+
toward CO2. As a result, magnesium ion catalyzes reversible
CO2 dissociation instead of decarboxylation. The Mg2+ and acid-
dependent experiments therefore show that the rate for unimo-
lecular CO2 exchange can be tuned by influencing specific
interactions with the ureido ring, which is an important step
toward understanding how CO2 elimination might be triggered
in biotin enzymes.
It has also been suggested that decarboxylation of N1-
carboxybiotin involves rotation of the carbamyl group out of
planarity,19 which might result if the N-CO2- double bond
(
20) Johnson, S. L.; Morrison, D. L. J. Am. Chem. Soc. 1972, 94, 1323-1334.
Ewing, S. P.; Lockshon, D.; Jencks, W. P. J. Am. Chem. Soc. 1980, 102,
3072.
(21) Deposited in Supporting Information
(
22) Rubisco: Schreuder, H. A.; Knight, S.; Curmi, P. M. G.; Andersson, I.;
Cascio, D.; Sweet, R. M.; Branden, C. I.; Eisenberg, D. Protein Sci. 1993,
2, 1136-1146. Belknap, W. R.; Portis, A. R., Jr. Biochemistry 1986, 25,
1864-1869. Schreuder, H. A.; Knight, S.; Curmi, P. M. G.; Andersson, I.;
character is decreased by interactions between Lewis acidic sites
Cascio, D.; Br a¨ nden, C.-I.; Eisenberg, D. Proc. Natl. Acad. Sci. U.S.A.
1993, 90, 9968-9972. Cleland, W. W.; Andrews, T. J.; Gutteridge, S.;
Hartman, F. C.; Lorimer, G. H. Chem. ReV. 1998, 98, 549-561.
1
on the protein and the carbonyl group of N -carboxybiotin. In
(
23) Urease: Park, I.-L.; Carr, M. B.; Hausinger, R. P. Proc. Natl. Acad. Sci.
U.S.A. 1994, 91, 3233-3237. Park, I.-L.; Hausinger, R. P. Science 1995,
267, 1156-1158. Pearson, M. A.; Schaller, R. A.; Michel, L. O.; Karplus,
P. A.; Hausinger, R. P. Biochemistry 1998, 37, 6214-6220.
(
(
(
17) Izutsu, K. Acid-base dissociation constants in dipolar aprotic solVents;
Blackwell Scientific Publications: Oxford, U.K., 1990.
18) Hegarty, A. F.; Bruice, T. C.; Benkovic, S. J. J. Chem. Soc., Chem.
Commun. 1969, 1173-1174.
(24) Zinc-dependent phosphotriesterase: Shim, H.; Raushel, F. M. Biochemistry
2000, 39, 7357-7364. Hong, S.-B.; Kuo, J. M.; Mullins, L. S.; Raushel,
F. M. J. Am. Chem. Soc. 1995, 117, 7580-7581.
19) Thatcher, G. R. J.; Poirier, R.; Kluger, R. J. Am. Chem. Soc. 1986, 108,
2
699-2704.
11340 J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002

Kinetics and Mechanism for CO
2
Scrambling
A R T I C L E S
are critical for carboxyl transfer,25 suggesting that these may
be part of a triggering mechanism for N -carboxybiotin dis-
sociation.
that the reaction occurs via an acid-dependent proton-transfer
preequilibrium. The Lewis acid catalyzed carboxyl exchange
and the acid-dependent decarboxylation of N-carboxyimida-
zolidone were explained within a common mechanistic frame-
work in which C-N bond dissociation is facilitated by
1
Summary
We have presented data showing that Li2 can undergo
carboxyl exchange directly with CO2 in DMSO. This chemical
exchange reaction is suggested to occur via a unimolecular C-N
bond dissociation pathway based on CO2 exchange kinetics. The
model studies suggest that unimolecular CO2 dissociation is a
2+
+
interaction of a cation (Mg or H ) with the carbonyl group
of the ureido ring. This implied that carboxyl transfer enzymes
1
might trigger release of CO2 from N -carboxybiotin through
noncovalent interactions with the ureido ring.
1
viable reaction for N -carboxybiotin and that it should be
Acknowledgment. Support from the National Science Foun-
dation (CHE-9985266) is gratefully acknowledged.
considered in developing a chemical mechanism for carboxyl
2+
transfer in biotin enzymes. We observed that Mg catalyzed
-
the exchange of 2 with free CO2, which relates to the ability
Supporting Information Available: Figure S1 showing
-
of the metal ion to coordinate to 2 . The acid-dependent
kinetic traces for CO exchange by lithium diethylcarbamate.
2
-
decarboxylation rate of 2 in aprotic media is dependent on
This material is available free of charge via the Internet at
the concentration of the acid and on its pKa, which suggests
http://pubs.acs.org.
(25) Blanchard, C. Z.; Waldrop, G. L. J. Biol. Chem. 1998, 273, 19140-19145.
JA026753Y
J. AM. CHEM. SOC.
9
VOL. 124, NO. 38, 2002 11341