Synthesis of fluorinated NAD as a soluble coenzyme for enzymatic chemistry in fluorous solvents and carbon dioxide

Article abstract of DOI:10.1016/S0040-4020(02)00259-4

The synthesis of the coenzyme nicatinamide adenine dinucleotide (NAD) with an covalently attached fluorinated polymer is reported. The fluorinated NAD (FNAD) was rendered soluble in both fluorous solvents and liquid carbon dioxide due to the attachment of a perfluoropolyether. The enzyme horse liver alcohol dehydrogenase (HLADH) was active in catalyzing oxidation/reduction reactions using FNAD as a soluble coenzyme in a fluorous solvent, methoxynonafluorobutane (HFE), and liquid carbon dioxide. In both solvents, the activity of HLADH using FNAD was greater than the same molar amount of unmodified (insoluble) NAD, indicating that a soluble coenzyme results in more efficient reactions.

Full text of DOI:10.1016/S0040-4020(02)00259-4

TETRAHEDRON
Pergamon
Tetrahedron 58 *2002) 4091±4104
Synthesis of ¯uorinated NAD as a soluble coenzyme for enzymatic
chemistry in ¯uorous solvents and carbon dioxide
Janice L. Panza,a,b Alan J. Russella,b,c and Eric J. Beckmana,b,p
aDepartment of Chemical and Petroleum Engineering, University of Pittsburgh, 1249 Benedum Engineering Hall,
Pittsburgh, PA 15261, USA
bCenter for Biotechnology and Bioengineering, University of Pittsburgh, Pittsburgh, PA 15219, USA
cMcGowan Institute for Regenerative Medicine, Pittsburgh, PA 15261, USA
Received 13 November 2001; revised 7February 2002; accepted 9 February 2002
AbstractÐThe synthesis of the coenzyme nicatinamide adenine dinucleotide *NAD) with an covalently attached ¯uorinated polymer is
reported. The ¯uorinated NAD *FNAD) was rendered soluble in both ¯uorous solvents and liquid carbon dioxide due to the attachment of a
per¯uoropolyether. The enzyme horse liver alcohol dehydrogenase *HLADH) was active in catalyzing oxidation/reduction reactions using
FNAD as a soluble coenzyme in a ¯uorous solvent, methoxynona¯uorobutane *HFE), and liquid carbon dioxide. In both solvents, the activity
of HLADH using FNAD was greater than the same molar amount of unmodi®ed *insoluble) NAD, indicating that a soluble coenzyme results
in more ef®cient reactions. q 2002 Elsevier Science Ltd. All rights reserved.
1. Introduction
catalyst in the ¯uorous phase, which then can be recycled.
FBS allow easy product separation and recycle of catalysts.
Other promising technologies using ¯uorous solvents include
¯uorous separations, ¯uorous synthesis, and ¯uorous mixture
synthesis.8 In these applications, the catalyst, reagent, or
substrate is rendered soluble in a ¯uorous phase by attaching
a ¯uorinated `ponytail' to the compound.
Fluorous solvents, known as per¯uorinated solvents,
per¯uorocarbons *PFC), or just ¯uorocarbons, are charac-
terized by high density, high stability, and low solubilities in
both water and most organic solvents.1,2 Fluorous solvents
as reaction media are of growing interest in that they are
considered environmentally benign due to their unique
properties. For example, ¯uorous solvents exhibit low
toxicity and are used in many biomedical and clinical appli-
cations.3 They are easy to separate from both aqueous and
organic solvents due to their immiscibility in both;4
however, many organic/¯uorous mixtures may become
miscible at increased temperatures.5 Finally, they do not
deplete the ozone layer;6 although one drawback is that
they are potent greenhouse gases.7
It would be interesting to combine the `green' bene®ts of
¯uorous solvents with that of green catalysts *enzymes) to
expand non-aqueous biocatalysis into environmentally
benign solvents. Enzymes can be considered environ-
Synthesis in ¯uorous biphasic mixtures *FBS) is an emerg-
ing ®eld that takes advantage of the bene®ts of ¯uorous
solvents. This process involves catalysis in two immiscible
phases, a ¯uorous and a non-¯uorous phase. The catalyst is
totally soluble in the ¯uorous phase while the substrates and
products prefer the non-¯uorous phase. After reaction, the
product is recovered in the non-¯uorous phase and the
Keywords: nicatinamide adenine dinucleotide; reduced nicatinamide
adenine dinucleotide; alcohol dehydrogenase; horse liver alcohol dehydro-
genase; oxidation/reduction; ¯uorous solvents; carbon dioxide; solubility;
high pressure.
p
Corresponding author. Address: Department of Chemical and Petroleum
Engineering, University of Pittsburgh, 1249 Benedum Engineering Hall,
Pittsburgh, PA 15261, USA. Tel.: 11-412-624-9641; fax: 11-412-624-
9639; e-mail: beckman@engrng.pitt.edu
Figure 1. *a) Structure of NAD, *b) structure of NADH.
0040±4020/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S0040-4020*02)00259-4

4092
J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
both reactions occurred when the two enzymes and NAD
were lyophilized together.11 They suggested that the essen-
tial water bound to the lyophilized solid containing both
enzymes offered a layer of hydration so that the cofactor
could be released from the active site of E1 to migrate to the
active site of E2 through the hydration layer. When the
enzymes were lyophilized separately, the two enzymes did
not share a hydration layer; therefore, NAD*H) could not
travel from E1 to E2 through the organic solvent. We
propose to overcome the limitation of NAD*H) transport
in organic solvent by derivatizing NAD*H) to affect solu-
bility. Even if a one-enzyme recycling system is used *or if
two enzymes are lyophilized together), solubilization of the
coenzyme should increase reaction ef®ciency.
Figure 2. In situ enzymatic recycling of NAD/NADH.
mentally benign catalysts in that they are derived from
biological systems, have very high activity, and are
selective in the reactions that they catalyze.
Covalent immobilization of NAD onto polymers has been
accomplished previously, speci®cally to facilitate reuse and
regeneration of the coenzyme. Applications include NAD
attachment to water-soluble polymers for use in synthetic
and analytical reactors,12 and NAD immobilization with
water-insoluble polymers for use as the sorbent material
for af®nity chromatography.13 Typically, the covalently
modi®ed NAD was used in an aqueous environment,
whether or not the NAD was modi®ed with a water soluble
or insoluble polymer. Although it was covalently modi®ed,
the solubility of modi®ed NAD was not investigated in
organic solvents. We hypothesized that NAD covalently
modi®ed with a polymer will demonstrate solubility in an
organic phase, in this case a ¯uorous solvent.
Oxidoreductases, commonly known as dehydrogenases,
catalyze oxidations and reductions. Dehydrogenases require
the participation of nicatinamide adenine dinucleotide
*NAD) in the catalysis of these reactions. NAD *Fig. 1*a))
accepts a hydride ion from the substrate, being reduced to
NADH *Fig. 1*b)) with the concurrent oxidation of the
substrate to product:
AH2 1 NAD O A 1 NADH 1 H1
where AH2 is the reduced substrate and A is the oxidized
product. NAD is required in stoichiometric amounts, but it
can be regenerated after it participates in a reaction. For
example, in situ enzymatic recycling of NAD is common.
Once NAD has been reduced to NADH, it can be regener-
ated back to NAD by providing a second reaction *Fig. 2).
One enzyme *E1) oxidizes NADH to NAD while a second
enzyme *E2) reduces NAD back to its original state. At the
same time, substrate *S1) is reduced to a product *P1), while
a second substrate *S2) is oxidized to a second product *P2).
Two separate enzymes can be employed to catalyze the
reactions, but sometimes only one enzyme is necessary, as
long as the enzyme has broad substrate speci®city.
We have examined the covalent modi®cation of NAD with a
¯uorinated polymer to enhance its solubility in a ¯uorous
phase. Interestingly, molecules designed for solubility in
¯uorous solvents also demonstrate appreciable solubility
in CO2. Although CO2 is a poor solvent for most
compounds,14 it has been shown to dissolve a number of
¯uorinated polymers at relatively ambient temperatures
and moderate pressures.15,16 In addition, attachment of
¯uorinated polymers have rendered other compounds
soluble in CO2.17,18
Herein, we report the synthesis of NAD with a covalently
attached ¯uorinated polymer tail *FNAD). The solubility of
FNAD was examined in both ¯uorous solvents and liquid
CO2. The ability for FNAD to act as a soluble coenzyme in
enzyme-catalyzed oxidation/reduction reactions was
explored in a ¯uorous solvent. To brie¯y summarize, issues
such as how FNAD fared as a coenzyme when compared
with native NAD, how the ¯uorinated polymer affected
enzyme activity, and equilibrium of the recycling reaction
system were explored. MALDI-TOF mass spectroscopy was
used to substantiate that FNAD actually participated in a
reaction catalyzed by a dehydrogenase. Finally, FNAD as a
soluble coenzyme in liquid CO2 was investigated.
When NAD is used with an enzyme in an organic solvent,
both NAD and the enzyme are insoluble. The enzyme and
NAD are therefore typically lyophilized together; neverthe-
less, they are surrounded by what is termed `essential
water', or the bound water that is required for the enzyme
to function in the organic solvent.9 Once NAD participates
in a reaction, it is reduced to NADH *or vice versa). It has
been hypothesized that NAD*H) must be released from the
active site in order for it to be regenerated,10 but since
NAD*H) is not soluble in organic solvents, NAD*H) will
not likely leave the active site as required. If NAD*H) is
hindered from leaving the active site, the enzyme molecule
will become inactive, given that the active site is not free to
bind another NAD*H) molecule to continue the reaction.
This has been shown to be true for a two-enzyme regenera-
tion reaction.11 NAD*H) would not traverse the organic
phase to participate in both reactions *Fig. 2) if two enzymes
*E1 and E2) were lyophilized separately; the insolubility of
the coenzyme in the organic solvent prevented the regenera-
tion of the coenzyme. However, Deetz et al. showed that
2. Results and discussion
2.1. Synthesis of PFPE-modi®ed NAD
FNAD *Fig. 3*b)) was synthesized in a four step process
*Fig. 3*a)). NAD was brominated then reduced. The
bromine was then displaced by hexanediamine *HDA) to

J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
4093
Figure 3. *a) Synthesis of FNAD, *b) structure of FNAD.
form NADH-HDA. Finally, a per¯uoropolyether *PFPE)
was appended to NADH-HDA through an amide linkage.
further con®rm that a PFPE chain was attached to NAD,
and *2) to demonstrate that FNAD acted as a coenzyme in
the enzyme-catalyzed reaction. MALDI was effectively
used to characterize FNAD, which is notable since
MALDI has not been routinely applied to ¯uorinated
polymers.19 Most likely, the attached NAD head group
facilitated analysis by this method. The matrix used for
the analysis was 2,5-dihydroxybenzoic acid *DHB);
however, trans-3-indoleacrylic acid *IAA) also yielded
clean spectra with little background or noise. Different
matrix to analyte molar ratios *,3£104:1±3£101:1) were
investigated; however, there were no apparent changes in
the spectra as a function of matrix/analyte ratio. Therefore,
samples containing FNAD in HFE at a concentration of
15 mM were not further diluted before mixing with the
matrix solution. No cationization agent was necessary,
most likely due to the sites on the NAD molecule capable
of ionization.
To NAD, which has a molecular weight of 664.4 g/mol, we
attached a ¯uorinated polymer, PFPE, which has previously
been shown to have high solubility in CO2.15 FNAD is hence
an amphiphilic molecule with a hydrophilic head group and
a ¯uorophilic/CO2-philic tail. An important question was
how large the molecular weight of the polymer must be in
order to bring the large hydrophilic NAD into solution in a
¯uorous solvent or CO2. For our purposes, a per¯uoro-
polyether *PFPE) with a molecular weight of 2500 *,14
repeat units) was used to begin investigations.
Results showed that the 2500 MW PFPE was suf®ciently
large to solubilize FNAD in many ¯uorinated solvents.
FNAD was soluble in methoxynona¯uorobutane *HFE),
1,1,2-tri¯uorotrichloroethane,
per¯uorodimethylcyclo-
hexane, per¯uoromethylcyclohexane, per¯uoroheptane,
and per¯uorohexane at levels of 200 mg/ml and above.
It is important to point out that while PFPE exhibits an
average molecular weight of 2500 g/mol, it is
polydisperse material. With attached polymer, FNAD
exhibits a molecular weight distribution of 923.41166n,
a
2.1.1. MALDI-TOF MS analysis. MALDI-TOF MS was
used for two purposes in the analysis of FNAD: *1) to

4094
J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
Figure 4. MALDI-TOF mass spectrum of FNAD.
where the 923.4 amu includes the NAD, the attached
HDA, the amide linkage, and the end group of the polymer,
and n is the number of repeat units in the polymer
*Fig. 3*b)).
number of laser shots did not lead to higher molecular
weight species being more visible. Higher molecular weight
hydrophobic polymers are more dif®cult to analyze for
several reasons. The polydispersity of the sample means
that there are a mixture of oligomers in the sample; there-
fore, fewer molecules of each length of oligomer will be
contained in the sample leading to lower sensitivity.20,21
Furthermore, the matrix may not be ef®cient in solvating
the longer chain polymers as it is with the low molecular
weight species.20,21 Lastly, the extraction and post-accelera-
tion energies of the instrument may not be suf®cient to
maintain detector sensitivity for the large molecules.20
A MALDI-TOF mass spectrum clearly displays peaks of
FNAD with attached polymers for nˆ1±7repeat units
*Fig. 4). Larger molecular weight species are not seen at
the relative intensity of the lower molecular weight species;
nonetheless, low intensity peaks representing molecular
weights of 8±20 repeat units were also observed. Matrix
concentration, sample concentration, laser intensity, and
Figure 5. Overlapping mass spectra of the theoretical isotopic abundance of FNAD *nˆ5), the theoretical isotopic abundance of FNADH *nˆ5), and
experimental FNAD *nˆ5).

J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
4095
Figure 6. HLADH catalyzed reactions using FNAD as a soluble coenzyme in HFE.
Nevertheless, the spectrum clearly demonstrated that NAD
was modi®ed with per¯uoropolyether molecules.
unnatural environment. Either one of these issues alone
could dramatically alter the ability of the coenzyme to
function properly.
In the second reaction of the synthesis of FNAD, NAD was
reduced to NADH; however, MALDI analysis showed that
the sample after synthesis was actually a mixture of FNAD
and FNADH. When FNAD accepts a hydride ion, its
molecular weight increases by 1 amu; MALDI is indeed
sensitive enough to distinguish a 1 amu difference. Further,
the isotopic forms of the analyzed molecule can be detected
by MALDI. The MALDI software can predict the theoreti-
cal isotopic abundance of a molecule based on the natural
abundance ratios of the isotopes. Using FNAD with ®ve
repeat units attached *C45H40N9O20P2F35, MWˆ1753.4) as
an example, the theoretical isotopic abundance spectrum for
this molecule is shown in Fig. 5. Overlaid on that is the
spectrum for the theoretical isotopic abundance of the
reduced form, FNADH *C45H41N9O20P2F35, MWˆ1754.4).
The actual spectrum for the sample FNAD is also displayed
in Fig. 5. Isotopic abundance calculations for both FNAD
and FNADH predict four peaks with decreasing intensity,
*FNAD: 1753.4.1754.4.1755.4.1756.4 and FNADH:
1754.4.1755.4.1756.4.1757.4). The actual spectrum
for FNAD does not have the same ratio of isotopic peaks
as either theoretical spectrum; it is a combination of the two
spectra as evidenced by the presence of ®ve isotopic peaks.
This qualitatively demonstrates that NADH-HDA is
partially oxidized to NAD-HDA during the wash step with
TFE. Unfortunately, quantitative amounts of each species
cannot be ascertained from the MALDI data because it is
unknown if one form is more readily desorbed by the
matrix/laser combination.
Activity studies in a ¯uorous solvent were performed in
methoxy-nona¯uorobutane *HFE) a solvent that is intended
to replace ozone-depleting CFCs. HFE's bene®ts include
zero ozone depletion potential, very low toxicity, and non-
¯ammability. Horse liver alcohol dehydrogenase
*HLADH), was used to catalyze both reactions as shown
in Fig. 2. HLADH demonstrates broad substrate speci®city,
so it can be ef®ciently used to catalyze both the oxidation
and reduction reactions. The forward reaction is the reduc-
tion of butyraldehyde *S1) to butanol *P1) using FNADH.
The regeneration reaction is the oxidation of ethanol *S2) to
acetaldehyde *P2) using FNAD. Ethanol is a common
regenerative substrate due to its availability and low cost.
In this study, the functionality of FNAD was investigated by
adding increasing amounts of coenzyme to reaction
mixtures. Each sample contained increasing amounts of
FNAD, from approximately 0 to 0.1 g, which corresponds
to 0±15 mM, respectively. As can be seen from the graph
*Fig. 6), increasing the amount of FNAD increased the rate
of reaction until the butanol concentration reached a steady
state.
Two control reactions were also performed. The sample that
contained FNAD but no added HLADH was not active, as
expected. However, the sample that did contain HLADH but
no added FNAD showed substantial activity; therefore, the
enzyme must contain residual bound NAD. A UV scan of
the enzyme was performed to determine how much NAD
was present; the scan showed that 0.1 mg of HLADH in
H2O had approximately a 46-fold molar excess of NAD
associated with it. NAD is added during the puri®cation of
HLADH in order to stabilize the enzyme during the
puri®cation process22Ðif measures were used to remove
the NAD, the HLADH would probably lose its activity.
2.2. FNAD as a soluble coenzyme in a ¯uorous solvent
2.2.1. HLADH activity using FNAD. Two major
alterations to NAD have occurred that could in¯uence
coenzyme utility: ®rst, the molecule was covalently
modi®ed, and second, the coenzyme was used in an

4096
J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
Figure 7. PFPE as a cosolvent.
Therefore, HLADH without added NAD is expected to have
some native activity due to residual bound NAD.
of substrates and products between the solvent and enzyme,
shifting of chemical equilibria, and mass transfer limitations
are all known solvent effects on enzyme activity.23 Polar
cosolvents have been added to hydrophobic solvents to aid
in solubilization of polar substrates.24 Even water can act as
a cosolvent when biocatalysis is performed in organic
solvents.25 DMSO and formamide have been used to acti-
vate several hydrolytic enzymes in anhydrous media, the
activation most likely due to enhanced conformal ¯exibility
of the enzymes in the anhydrous media.26
2.2.2. Effect of PFPE on HLADH. HLADH without added
FNAD demonstrated activity in HFE *Fig. 6), indicating that
HLADH must have remaining NAD associated with it. This
native activity leads one to question whether FNAD actually
participated in the reaction. It is possible that the NAD head
group did not actually participate in the oxidation/reduction
reactions, but that the entire molecule itself or the PFPE tail
acted as a cosolvent, and hence its presence in the reaction
simply increased the enzyme activity in HFE. Solvent can
have a tremendous effect on the activity of an enzyme.
Binding of solvent molecules to the enzyme, partitioning
To test whether PFPE stimulated HLADH activity by acting
as a cosolvent, the native activity of HLADH was compared
with HLADH activity containing two different PFPE
Figure 8. Comparison of FNAD and NAD.

J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
4097
Figure 9. Overlapping MALDI-TOF mass spectra of unreacted and reacted FNAD *nˆ5).
polymers; one with a single terminal hydroxyl *PFPE-OH)
and the other with a single terminal acid *PFPE-COOH).
Addition of 0.12 g *0.036 mmol) PFPE-OH to the reaction
mixture did not have any effect on the HLADH activity
*Fig. 7). Conversely, supplementing the reaction mixture
with 0.10 g *0.031 mmol) PFPE-COOH completely
inactivated the enzyme. Most likely the acid group had an
adverse effect on the enzyme, whereas the hydroxyl group
was benign with respect to HLADH activity. In either case,
though, it is obvious that PFPE itself did not aid the activity
of HLADH by acting as a cosolvent in HFE.
bound NAD, but the addition of more NAD did not have
any effect.
When NAD was directly lyophilized with HLADH, they
shared a hydration layer. NAD was free to move about
this layer to participate in the reactions; however, the
activity was much less that when the NAD was separate
from the HLADH. A possible explanation was that too
much NAD was present leading to mass transfer limitations.
A lyophilized enzyme particle is a porous structure and a
substrate molecule has to negotiate the pores to reach the
active site.27 By lyophilizing HLADH with NAD, increased
internal diffusional limitations within the lyophilized
enzyme/NAD complex most likely hindered the substrate
from reaching the active site of the enzyme, thereby
decreasing activity. Although, Yang demonstrated that
yeast alcohol dehydrogenase *YADH) activity increased
with an increase in NAD/YADH ratio,10 the maximum
molar ratio of NAD/YADH was 4.25. The NAD/HLADH
herein was about 210 *not including the native, residually
bound NAD). NAD was in great molar excess, which most
likely just increased the solid mass, but did not lead to
ef®cient enzyme activity.
2.2.3. FNAD versus NAD. Since the HLADH alone showed
activity in HFE, one question is whether unmodi®ed NAD
*insoluble NAD) shows the same activity as the soluble
FNAD. Consequently, the activity of NAD was compared
with the same molar amount of FNAD using the same
reaction scheme. Initially, HLADH *10 mg) was added to
250 mM of both butyraldehyde and ethanol in HFE *a total
of 4 ml). The reaction proceeded for approximately 7h, at
which point the reaction was split into two samples. In one
sample, 8.7mg *0.013 mmol) of insoluble NAD were
added. To the other sample, 40.4 mg of soluble FNAD
were added at the same molar amount *0.013 mmol). The
third sample contained 8.7mg *0.013 mmol) of NAD that
were directly lyophilized with the enzyme.
2.2.4. MALDI-TOF analysis of FNAD. MALDI-TOF
mass spectroscopy was used to con®rm the structure of
FNAD and that FNAD was actually a mixture of both the
oxidized and reduced forms of the coenzyme. Moreover,
MALDI was used to verify that the FNAD molecule, not
just native bound NAD, participated in the reactions.
The reactions were allowed to proceed, monitoring the
production of butanol. Evidently *Fig. 8), soluble FNAD
promotes a faster reaction rate than either the insoluble
NAD or the NAD directly lyophilized with the enzyme,
indicating that solubilization of the coenzyme increases
reaction rates. When NAD was added to the reaction as a
solid it is possible that HLADH and NAD maintained their
own hydration layers, but were not able to interact
because they were both insoluble in the solvent. The
HLADH demonstrated native activity due to residually
FNAD was used in the following reaction:
CH3CH2OH 1 FNAD HLOADH CH3CHO 1 FNADH 1 H1
This reaction did not include a regeneration reaction.
Different MALDI samples were prepared by spiking

4098
J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
Figure 10. Equilibrium of HLADH catalyzed reactions using FNAD as a soluble coenzyme.
`unreacted FNAD' with increasing amount of `reacted
FNAD'. It would be expected that peaks on the MALDI
spectra would show an increase in the molecular weight
by 1 amu between the unreacted FNAD and the reacted
FNAD.
occurred *Fig. 10). In a 10 ml glass vial, 250 mM butyral-
dehyde and ethanol, 3 mM FNAD, and 10 mg HLADH in a
total of 4 ml HFE were added. The reaction was shaken at
250 rpm and 308C, and the production of butanol was
followed with time. When it appeared that no further buta-
nol was being produced, the reaction mixture was split into
two, with 2 ml in each vial. To the ®rst vial, 4 mg of fresh
HLADH was added. To the second vial, 250 mM of both
butyraldehyde and ethanol were added. The vials were
further shaken at 250 rpm and 308C and the production of
butanol was followed with time.
Indeed, the MALDI spectra show that peaks of the reacted
FNAD have a greater molecular weight than the peaks of the
unreacted FNAD. Fig. 9 shows the peaks of FNAD *nˆ5) of
the spiked samples superimposed on each other. The peak
molecular weight shifts to a higher molecular weight with
increasing amount of the reacted FNAD in the sample. The
same result was seen with six repeat units. Therefore, more
FNADH was produced during the reaction indicating that
FNAD did participate in the reactions.
The reaction that received the additional 4 mg of HLADH
did not see a great increase in butanol concentration, reach-
ing a ®nal butanol concentration of 80 mM *Fig. 10).
However, the reaction that received 250 mM of both butyr-
aldehyde and ethanol showed an increase in butanol concen-
tration, reaching 120 mM butanol. Apparently the system
was at equilibrium. Adding more substrate shifted the equi-
librium position of the reaction resulting in the production
of more butanol. The equilibrium position for the reaction
starting with 250 mM butyraldehyde and ethanol contains
80 mM butanol and therefore 170 mM butyraldehyde,
which agrees with the data in Fig. 6.
2.2.5. Equilibrium. As shown in Fig. 6, the production of
butanol slowed until there was no further formation beyond
a certain time. Two explanations could account for this
phenomenon: *1) either the enzyme was no longer active
*enzyme death) or *2) the system reached equilibrium.
There are two methods to distinguish whether the system
reached a steady state due to enzyme death or equilibrium:
1. At the apparent equilibrium point, more enzyme can be
added to the reaction mixture. Because catalysts do not
affect equilibrium, if the system were at equilibrium,
adding more enzyme should not change the composition
of the reaction mixture.
2. At the apparent equilibrium point, more substrate can be
added. Adding or removing a reactant or product will
disturb the equilibrium. If the system were at equi-
librium, adding more substrate will shift the equilibrium
to produce more butanol.
Acetaldehyde *P2) is volatile *bpˆ218C), hence it would be
expected to evaporate during the reaction. However, acet-
aldehyde apparently remained in the reaction mixture, as
seen on the chromatograms generated from GC and because
the system was able to reach equilibrium. Several possible
reasons exist to explain why acetaldehyde did not simply
evaporate: *1) NAD has been shown to form adducts with
substrates such as butyraldehyde and acetaldehyde;28 *2)
acetaldehyde condenses with NADH to give 1,2-dihydro-
2-ethylidenenicatinamide adenine dinucleotide;29 and *3)
acetaldehyde interacts with proteins.30
The two options were investigated to determine if the
system had reached equilibrium or if enzyme death had
To drive the reactions forward, acetaldehyde must be

J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
4099
Table 1. TON of FNAD HFE *177 h)
When too much FNAD was added, the turnover number
was low because the enzyme was saturated with the
coenzyme. FNAD thus acted more like a cosubstrate than
a recycling coenzyme at high concentrations. Nonetheless,
the reactions at higher FNAD concentrations *14 mm and
1.29 mM) had reached equilibrium so it is likely that the
TON for these two samples could be much higher. On the
other hand, the sample with the lowest concentration of
FNAD *0.012 mM) had the highest TON but the lowest
production of butanol in the time period. The implication
is that TON and activity must be optimized to achieve the
most ef®cient reactions.
FNAD *mM)
TON
Butanol produced *mM)
[FNAD]/[E]
HFE *177 h)
0.012
0.12
1.3
14
CO2 *57h)
4
1199
192
44
14.4
23.1
56.9
63.7454.4
0.4
3.8
41.6
5
9
34.5
253.2
removed from the reaction mixture, which may be accom-
plished by two simple methods. One technique is to bubble
nitrogen in the reaction mixture and trap it in a vessel
containing semicarbazide to form acetaldehyde semi-
carbazone.31 Another method is to oxidize acetaldehyde to
acetate using the enzyme aldehyde dehydrogenase
*AldDH), which also generates a second equivalent of
reduced coenzyme.32,33 Acetate does not harm enzymes or
coenzymes nor does it complicate product isolation.
2.2.7. Side reaction. n-Butyraldehyde is known to undergo
base-catalyzed aldol reactions.37 Both enzymes38,39 and
NAD have basic moieties that are able to catalyze this
reaction, resulting in a competing reaction for butyralde-
hyde. The side reaction is accelerated by the presence of
contaminating HDA in FNAD and by degradation products
in butyraldehyde; therefore, care must be taken to insure the
purity of FNAD and to use fresh and properly stored
butyraldehyde. Nevertheless, this side reaction was not
completely eliminated.
With the removal of acetaldehyde, the reaction would be
pushed to completion, thereby eliminating the need of
separation of the other substrates from the product of
interest. Furthermore, enzyme may be recovered by simple
®ltration and FNAD by a ¯uorous separation technique.8
2.3. FNAD as a soluble coenzyme in CO2
2.3.1. Solubility studies in CO2. A cloud point curve was
generated to determine the solubility of FNAD in CO2
*Fig. 11). The range of concentrations was approximately
3±5 mM, which corresponds to 1.2±2 wt%, respectively,
and typically, the concentration one would expect to use
in a reaction. As can be seen from the graph, the cloud
point pressures are relatively constant over the weight
percent range of interest, the cloud points being between
1200 and 1400 psi. The goal is to perform enzyme catalyzed
oxidation/reduction reactions in CO2 with a CO2-soluble
coenzyme; therefore, the operating pressure should be set
where the FNAD exists as one phase with the CO2, or at a
pressure above the cloud point curve.
2.2.6. Turnover number. Conventional TON's for aqueous
enzymatic reactions do not exceed 2000,34,35 whereas turn-
overs have reached 2,000,000 in organic solvents.36 The
TONs of FNAD were determined from the reactions
shown in Fig. 6, which took place over the time course of
177 h. The data was corrected by subtracting the native
activity of HLADH in order to distinguish the activity due
to FNAD. Table 1 reports the TON, the concentration of
butanol produced in each reaction, and the ratio of FNAD to
enzyme in the sample. Although 14 mM FNAD produced
the most butanol in the shortest time, the TON was very low,
indicating that the FNAD was not being used ef®ciently.
Figure 11. Cloud point curve of FNAD in CO2 *Tˆ218C).

4100
J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
Figure 12. High-pressure CO2 reactor.
Many different compounds have been covalently modi®ed
with CO2±philic functionalities to enhance their solubility
in CO2. Examples include surfactants,15 a biological
molecule,17 and metal chelating agents.18 In each case, the
length of the CO2±philic tail and the degree of modi®cation
of the molecule all affected the solubility of the molecule in
CO2.15,17,18 By increasing the size of the tail, the number of
CO2±philic interactions increases, thereby increasing solu-
bility. But soon there is a point where increasing the tail
further just leads to an increase in the entropy of mixing,
which decreases the solubility. However, one tail with a
molecular weight of 2500 was suf®cient to solubilize
NAD in CO2.
was also necessary to determine if all of the substrates and
products were soluble in liquid CO2 at the required concen-
trations *250 mM). Solubility of the product was not as
important, as an insoluble product would drive the reaction
forward. The cloud points of butanol and butyraldehyde
were generated over the concentration range of 302.5±
195.8 and 298±193 mM, respectively, which was the
substrate concentration range used in the reactions in
HFE. Both butanol and butyraldehyde were soluble in
CO2 at pressures greater than 1800 psi.40 Ethanol and
acetaldehyde were also miscible with CO2 under those
conditions.
2.3.2. Activity of HLADH using FNAD in high pressure
CO2. Fig. 12 shows the schematic of the high pressure
In addition to determining the solubility of FNAD in CO2, it
Figure 13. HLADH catalyzed reactions using FNAD as a soluble coenzyme in CO2 *Tˆ218C, Pˆ2600 psi).

J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
4101
reactor used to study the activity in CO2. HLADH activity in
liquid CO2 at 2600 psi and room temperature was investi-
gated using FNAD as a soluble coenzyme *Fig. 13). Four
different reactions were run: *1) HLADH alone; *2)
HLADH and 0.4 g *0.125 mmol) FNAD; *3) HLADH and
83 mg *0.125 mM) NAD added separately; and *4) HLADH
lyophilized with 83 mg *0.125 mM) NAD. The FNAD and
NAD were provided in equimolar amounts. The concentra-
tion of FNAD was 4 mM, which at 2600 psi is a single phase
mixture *Fig. 11).
supercritical or near-supercritical ¯uid as it is in a liquid
solvent due to the inherent properties of supercritical and
near-supercritical ¯uids. Yet despite the high activity of
HLADH in CO2, the addition of soluble coenzyme still
does lead to higher production of butanol.
The sample with FNAD was run an extra 10 h longer than
the other reactions. It appeared that the production of buta-
nol may have leveled off at that point, indicating that the
reaction may have reached equilibrium. The concentration
was 89 mM, slightly higher than the equilibrium concentra-
tion in HFE, which was 80 mM; however, the difference
most likely is due to the error in sampling from the high-
pressure reactor. To con®rm that the system actually does
reach equilibrium, the reaction would have to be run for
longer times.
HLADH in each case was active in liquid CO2. To date, no
other studies on puri®ed alcohol dehydrogenase activity in
CO2 have been reported. The only investigation of alcohol
dehydrogenase activity in CO2 has been performed using
whole cells, rather than isolated enzyme.41 The resting
cells of a fungus Geotrichum candidum IFO 5767 were
effectively employed for the reduction of o-¯uoroaceto-
phenone to *S)-1-*o-¯uorophenyl)ethanol in supercritical
CO2 *358C and 10 MPa).41 Other substrates such as aceto-
phenone, acetophenone derivatives, benzyl acetone, and
cyclohexanone were all reduced by the cells. Since whole
cells were used, neither additional NAD nor a regenerating
substrate was provided. However, the cells were suspended
in water containing propan-2-ol before immobilization on a
water-absorbing polymer.
The TON of FNAD in CO2 was calculated in the same
method as the samples in HFE. The data was corrected by
subtracting the native activity of HLADH in order to distin-
guish the activity due to FNAD. The TON of FNAD in CO2
at approximately 57h is 9. The results are reported in
Table 1.
3. Conclusions
As shown in Fig. 13, the activity of each reaction in CO2
followed the same trend as the reactions performed in the
¯uorous solvent. The sample with the soluble coenzyme
demonstrated the highest activity. The sample with enzyme
alone and the sample with added NAD showed similar rates
of butanol production, until 45 h, when the rates deviated.
The enzyme alone continued to produce butanol, whereas
the activity of the enzyme with added NAD leveled off. The
reason for this is unknown. Finally, the HLADH that was
lyophilized with NAD produced very little butanol
*1.2 mM). Once again, the reasons for the trends in activity
seen in HFE also apply here. The high activity for HLADH
with FNAD was due to the solubility of the coenzyme; the
low activity for HLADH lyophilized with NAD was likely
due to mass transfer limitations; and NAD added separately
from the enzyme did not change native HLADH activity
because the enzyme and NAD remained separate during
the reaction.
A ¯uorofunctional NAD *FNAD) was synthesized and
found to be soluble in many ¯uorous solvents. FNAD
showed great utility as a coenzyme. Solubilization of the
coenzyme in HFE resulted in faster reactions when HLADH
was used to catalyze oxidation/reduction reactions, much
faster that the same molar amount of NAD. MALDI-TOF
MS data con®rmed that FNAD participated in the reduction
reaction. FNAD at concentrations between 3 and 5 mM
were also soluble in CO2 at pressures between 1200 and
1400 psi. An oxidation/reduction reaction was catalyzed
by HLADH in CO2 at 2600 psi and room temperature
*well above the cloud points of the substrates and FNAD)
using FNAD as a soluble coenzyme. Like in HFE,
solubilization of the coenzyme led to faster reaction rates
than reactions performed with the same molar amount of
insoluble NAD. In addition, relatively the same amount of
butanol was produced in CO2 over the same time as in HFE.
HLADH without any added coenzyme or modi®ed
coenzyme exhibits high activity, more so than the same
sample in HFE. Although the reactions in CO2 and HFE
were not performed under exactly the same conditions, a
comparison can be made between them to illustrate the high
activity of HLADH in CO2. As shown in Fig. 6, the sample
of HLADH with no added coenzyme or derivative of the
coenzyme produced about 8.6 mM butanol in 56 h.
However, the sample in CO2 resulted in approximately
45.9 mM butanol being produced in 58 h *Fig. 13). Five
times more butanol is produced in CO2 than in HFE within
the same time period. The reason for the difference may be
that the diffusional limitations between solid enzyme and
the native-bound NAD that exist in liquid solvents are
alleviated somewhat in near-critical CO2 due to the bene®ts
of gas-like diffusivities and viscosities. This may indicate
that a soluble version of the coenzyme is not as useful in a
4. Experimental
4.1. Materials
Nicatinamide adenine dinucleotide from Baker's Yeast and
alcohol dehydrogenase from horse liver *HLADH) were
purchased from Sigma Chemical Company. Krytoxw 157
FSL *Dupont), a per¯uoropolyether *PFPE) with a terminal
COOH functionality *MW 2500), was procured from Miller
Stephenson. Methoxynona¯uorobutane *HFE) was obtained
from 3M. Carbon dioxide at 99.9% purity was obtained
from Praxair *3.0 grade). All other chemicals were
purchased from Aldrich Chemical Company and used as
received unless otherwise stated. Spectra cellulose ester
dialysis tubing and Whatman ®lters *size 4) were obtained
from Fisher Scienti®c.

4102
J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
4.2. Bromination of NAD
solution by a glass syringe. The reaction mixture was stirred
for 2 h at 808C, after which, the reaction was cooled to room
temperature and stirred for an additional 6 h.
NAD was brominated at the C-8 position of adenine by a
modi®cation of the method of Lee and Kaplan.42 Typically,
2.5 g of NAD were dissolved in 30 ml of 0.5 M sodium
acetate buffer, pH 4.5. To this solution was added 1 ml
bromine and the mixture stirred at room temperature for
1 h. The unreacted bromine was removed by extraction
with an equal volume of carbon tetrachloride until neither
phase changed color *approximately 7£). The solution was
dialyzed against water in 500 MW cut-off cellulose ester
dialysis tubing *Spectra) for 8 h *or overnight). The result-
ing solution was lyophilized *Labconco Freezedrier) for
The reaction was con®rmed using UV spectroscopy
monitoring a peak maximum shift from 264 to 277 nm.42
The product *NADH-HDA) was precipitated from solution
using a 10£ volume of acetone. The product was recovered
by centrifuging *Sorvall Superspeed RC2-B Automatic
Refrigerated Centrifuge) at 5000 rpm for 5 min to pellet
the precipitate. The supernatant was removed and the
product was washed in 2,2,2-tri¯uoroethanol, then ®ltered
*Whatman ®lters). The ®ltrate was passed through a gel
®ltration column *Sephadex G-10) to remove unreacted
HDA. The solvent was removed under vacuum.
1
48 h. Bromination of NAD was con®rmed by H NMR
*Bruker) using deuterated water as the solvent.
The bromination of NAD occurred on the C-8 position on
the adenine group of NAD. This reaction was con®rmed
using 1H NMR. Before bromination, NAD had ®ve protons
attached to aromatic groups, four on nicatinamide *dˆ9.45,
9.31, 9.28, and 8.42 ppm) and one on adenine
*dˆ8.31 ppm). The position of the resonance frequency at
position C-8 also occurs in this range *dˆ8.62 ppm),
4.5. Reaction of PFPE acid chloride with NADH-HAD
PFPE acid chloride was prepared as in Yazdi and
Beckman.18 The product was con®rmed by a shift of the
IR carbonyl peak from 1776 to 1806 cm21 *Genesis-II IR
spectrophotometer, Mattson).
1
resulting in six peaks on the H NMR spectrum between
8 and 10 ppm. After bromination, only ®ve peaks are
present in the region *nicatinamide protons: dˆ9.35, 9.16,
8.86, and 8.45 ppm, adenine proton: dˆ8.18 ppm) because
the proton at position C-8 was replaced with bromine.
Reaction was complete upon disappearance of the peak
representing the proton at the C-8 position on adenine
*dˆ8.62 ppm).
In the reaction of NADH-HDA with PFPE acid chloride, the
molar ratio of NADH-HDA to PFPE acid chloride was
1.2:1. NADH-HDA and an equimolar amount of triethyl-
amine *TEA) in 50 ml TFE were added to a three neck
round bottom ¯ask. In an addition funnel, PFPE acid
chloride was dissolved in 20 ml 1,1,2- tri¯uorotrichloro-
ethane. The system was ¯ushed and blanketed with argon
or nitrogen. The PFPE acid chloride solution was added
dropwise to the NADH-HDA solution. The mixture was
stirred 12±15 h at 308C. The solvent was removed under
reduced pressure. The product was washed with 1,1,2-
tri¯uorotrichloroethane, ®ltered, and the solvent removed
under reduced pressure. The product was resuspended in
20 ml per¯uorodimethylcyclohexane and placed in a
separatory funnel overnight. The bottom layer was
removed and the solvent removed under reduced
pressure. IR spectroscopy was used to substantiate the
formation of the amide linkage between the free amine on
hexanediamine and the PFPE acid chloride. After the
reaction, the IR peak at 1806 cm21 disappeared and a
peak at 1714 cm21 indicative of a ¯uorinated amide linkage
was present.
4.3. Reduction of NAD-Br
NADH-Br has been found to be more thermally stable than
NAD43 and hence NAD-Br was chemically reduced to
NADH-Br using sodium dithionite. NAD-Br *0.5 g) was
dissolved in 1.3% sodium bicarbonate buffer, which had
been deaerated ®rst by bubbling with argon or nitrogen
for 15 min. A two-fold molar excess of sodium dithionite
*0.25 g) was added, and the solution shaken for 1 h at 258C.
The sodium dithionite remaining in the reaction mixture
was oxidized by bubbling oxygen through the solution for
15 min. The NADH-Br was recovered by precipitating
with 10£ the volume of acetone, the supernatant was
removed, and the product was air dried. The product was
resuspended in 10 mM sodium phosphate buffer, pH 7.8,
followed by lyophilization for 48 h. Reduction was veri®ed
using UV spectroscopy *Perkin±Elmer Lambda 2) by the
increase in absorbance at 340 nm. The molar adsorption
coef®cients of NAD at 260 and 340 nm are 18,600 and
6650 mol/l cm, respectively. Once the reduction was
complete, the ratio of the absorbance of 260±340 nm was
approximately 3.42
The attachment of the spacer molecule to NAD requires an
excess of the diamine. If NAD-HDA is not free of HDA, the
diamine will react with the PFPE-acid chloride instead of
the modi®ed NAD. When HDA reacts with PFPE acid
chloride, an IR peak for the amide linkage between the
diamine and PFPE-acid chloride appears very close to, but
at a somewhat lower wavenumber *1704±1708 cm21) than
that of FNAD *1714 cm21).
4.4. Reaction of hexanediamine and NADH-Br
Another important peak on the IR spectrum that con®rms
the presence of FNAD versus HDA-PFPE is around
1668 cm21, which correspond to the amide linkage on the
nicatamide group. This peak is sometimes not separate from
the peak at 1714 cm21, but a large shoulder will be apparent.
This peak is absent when the excess diamine reacts with
PFPE-acid chloride instead of the NAD.
Hexanediamine was used to displace the bromine at the C-8
position of adenine. Typically, NADH-Br *2 g) was
dissolved in 20 ml DMSO. The solution was stirred and
heated to 808C, and the system was ¯ushed and blanketed
with argon or nitrogen. A 20£ molar excess of hexane-
diamine *HDA) in 20 ml DMSO was injected into the

J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
4103
4.6. MALDI-TOF MS analysis
variable volume high-pressure view cell. A known amount
of sample and CO2 were added to the view cell in order to
get a speci®c weight percent of FNAD in CO2. The pressure
of the CO2 in the system was increased using a gas booster
and syringe pump until the FNAD in the CO2 exhibited one
phase *FNAD was soluble in CO2). The volume of the cell
was then expanded using a second syringe pump to decrease
the pressure until the solution inside the cell became turbid.
At that pressure, the FNAD was no longer soluble in the
CO2 and it was labeled the cloud point pressure. More CO2
was then added to the system to dilute the sample and the
cloud point pressure was determined for the new weight
percent of FNAD in CO2. A cloud point curve was generated
by determining the cloud point pressures at different weight
percents of FNAD in CO2.
MALDI-TOF MS analysis was performed on both an
unreacted FNAD sample and a reacted FNAD sample. To
prepare the unreacted sample, FNAD at a concentration of
15 mM was prepared in a solution of 250 mM ethanol and
250 mM butyraldehyde in HFE. Samples were prepared for
MALDI analysis by mixing the matrix 2,5-dihydroxy-
benzoic acid *DHB, 30 mg/ml in tetrahydrofuran) and
15 mM FNAD solution in a 5:1 ratio, respectively. Two
microliters were spotted on the sample plate and the spot
was allowed to air-dry in order to evaporate the solvent.
To prepare the reacted sample, 5 mg of HLADH were added
to a solution of 15 mM FNAD and 250 mM ethanol in HFE
in a 4 ml glass vial. The reaction mixture was shaken at
250 rpm for 48 h at 308C, after which the enzyme was
removed from the solution by ®ltration *Acrodisc CR
PTFE 0.45 mm syringe ®lters). This sample was labeled
reacted FNAD. A series of different samples were prepared
by spiking the unreacted FNAD with increasing amounts of
reacted FNAD. Spiked samples were prepared for MALDI
analysis by mixing with 30 mg/ml DHB in tetrahydrofuran
in a 5:1 matrix/FNAD ratio. Two microliters were spotted
on the sample plate, followed by air-drying.
4.10. Activity studies in CO2
The high pressure reactor used to study the activity in CO2
had a volume of 30 ml *Fig. 12). Samples were added
through the top of the reactor and the lid screwed in place.
CO2 was added and the pressure in the system increased
using the gas booster. The mixture within the reactor was
continuously stirred *Parr) at room temperature. Samples
were taken by opening a 200 ml sample port and bubbling
the reaction mixture through a solvent, toluene. The samples
were analyzed by GC. Due to the error associated with
sampling, a calibration curve of the ratio of the area of the
butanol peak to the area of the butyraldehyde peak versus
the known butanol concentration inside of the reactor was
created.
MALDI analysis was performed using a Perseptive Bio-
systems Voyager DE STR Biospectrometry Workstation
4087. The spectrometer was operated in linear mode,
20 kV accelerating voltage, 1997laser intensity, 50 shots/
spectrum. Data Explorer Software version 3.5 was used to
analyze the results. Data was saved as ASCII ®les, and
converted in Excel documents for graphical presentation.
To study activity, 250 mM butyraldehyde, 250 mM ethanol,
and 38 mg HLADH were added to the reactor with either *1)
no added coenzyme, *2) NAD, or *3) FNAD. CO2 was added
at room temperature and 2600 psi *well above the cloud
points of the substrates and FNAD). The reaction mixture
was sampled at regular intervals in order to monitor the
production of butanol.
4.7. Enzyme preparation
HLADH from Sigma was obtained as a dry, lyophilized
powder. The enzyme was resuspended in 10 mM sodium
phosphate buffer, pH 7.8, to a ®nal concentration of 5 mg/
ml. The enzyme solution was aliquotted into 1 ml portions
in 15 ml conical tubes *Falcon). The enzyme solutions were
snap frozen in liquid nitrogen and lyophilized for 48 h. The
enzyme was stored at 2208C until use.
References
4.8. Activity studies in HFE
1. Zhu, D. W. Synthesis 1993, 953±954.
2. Betzemeier, B.; Knochel, P. Top. Cur. Chem. 1999, 206, 61±
78.
A typical reaction contained FNAD and 5 mg HLADH in a
solution of 250 mM butyraldehyde and ethanol in HFE in a
4 ml screw cap glass vial *Wheaton). The total reaction
volume was 2 ml. The vials were shaken at 250 rpm and
308C. The reaction mixtures were analyzed for the
product, butanol, with time using a FID-GC *Perkin±
Elmer). The solid phase in the capillary column was cross-
linked-methyl siloxane and the column dimensions were
30 m£0.55 mm£0.88 mm ®lm thickness. A calibration
curve was prepared for butanol in HFE at concentrations
from 0 to 50 mM. The calibration curve was used to convert
the peak areas from the GC to the corresponding concentra-
tions of butanol.
3. Lowe, K. C. Blood Rev. 1999, 13, 171±184.
4. Hudlicky, M. Chemistry of Organic Fluorine Compounds;
Ellis Horwood: Chichester, 1992.
Â
Â
5. Horvath, I. T.; Rabai, J. Science 1994, 266, 72±75.
6. Wuebbles, D. J.; Calm, J. M. Science 1997, 278, 1090±1091.
7. Ravishankara, A. R.; Solomon, S.; Turnipseed, A. A.; Warren,
R. F. Science 1993, 259, 194±199.
8. Curran, D.; Lee, Z. Green Chem. 2001, 3, G3±G7.
9. Schulze, B.; Klibanov, A. M. Biotechnol. Bioengng 1991, 38,
1001±1006.
10. Yang, F. X.; Russell, A. J. Biotechnol. Prog. 1993, 9, 234±
241.
11. Deetz, J. S.; Rozzell, J. D. Ann. N.Y. Acad. Sci. 1988, 542,
230±234.
4.9. Solubility studies in CO2
12. BuÈckmann, A. F.; Kula, M. R.; Wichmann, R.; Wandrey, C.
J. Appl. Biochem. 1981, 3, 301±315.
Solubility studies were performed at room temperature in a

4104
J. L. Panza et al. / Tetrahedron 58 22002) 4091±4104
13. Zappelli, P.; Rossodivita, A.; Re, L. Eur. J. Biochem. 1975,
54, 475±482.
29. Chenault, H. K.; Whitesides, G. M. Appl. Biochem.
Biotechnol. 1987, 14, 147±195.
14. Hyatt, J. A. J. Org. Chem. 1984, 49, 5097±5101.
15. Hoe¯ing, T. A.; Stofesky, D.; Reid, M.; Beckman, E. J.;
Enick, R. M. J. Supercrit. Fluids 1992, 5, 237±241.
16. DeSimone, J. M.; Maury, E. E.; Menceloglu, Y. Z.; McClain,
J. B.; Romack, T. J.; Combes, J. R. Science 1994, 265, 356±
359.
30. Brecher, A. S.; Keterba, A. P.; Basista, M. H. Alcohol 1996,
13, 539±545.
31. Schulman, M. P.; Gupta, N. K.; Omachi, A.; Hoffman, G.;
Marshall, W. E. Anal. Biochem. 1974, 60, 302±311.
32. Wong, C. H.; Whitesides, G. M. J. Org. Chem. 1982, 47,
2816±2818.
33. Wong, C. H.; Whitesides, G. M. J. Am. Chem. Soc. 1983, 105,
5012±5014.
17. Ghenciu, E. G.; Beckman, E. J. Ind. Engng Chem. Res. 1997,
36, 5366±5370.
34. Wong, C. H.; Whitesides, G. M. J. Am. Chem. Soc. 1981, 103,
4890±4899.
18. Yazdi, A. V.; Beckman, E. J. J. Mater. Res. 1995, 3, 530±537.
19. Bahr, U.; Deppe, A.; Karas, M.; Hillenkamp, F. Anal. Chem.
1992, 64, 2866±2869.
35. Dordick, J. S. Enzyme Microb. Technol. 1989, 11, 194±211.
36. Grunwald, J.; Wirz, B.; Scollar, M. P.; Klibanov, A. M. J. Am.
Chem. Soc. 1986, 108, 6732±6734.
20. Belu, A. M.; DeSimone, J. M.; Linton, R. W.; Lange, G. W.;
Friedman, R. M. J. Am. Soc. Mass Spectrom. 1996, 7, 11±24.
21. Wu, K. J.; Odom, R. W. Anal. Chem. 1998, 70, 456A±461A.
22. Glendening, T. Sigma Chemical Company, personal
communication.
37. Tsuji, H.; Yagi, F.; Hattori, H.; Kita, H. J. Catal. 1994, 148,
759±770.
38. Martin, S. F.; Hergenrother, P. J. Biochemistry 1998, 37, 57 5±
5760.
23. Zaks, A.; Klibanov, A. M. Science 1984, 224, 1249±1251.
24. Dordick, J. S. Biotechnol. Prog. 1992, 8, 259±267.
25. Zaks, A.; Klibanov, A. M. J. Biol. Chem. 1988, 263, 8017±
8021.
39. Drohat, A. C.; Jagadeesh, J.; Ferguson, E.; Stivers, J. T.
Biochemistry 1999, 38, 11866±11875.
40. Panza, J. L.; Russell, A. J.; Beckman, E. J. In Clean Solvents,
Abraham, M., Moens, L., Eds.; 2002 in press.
41. Matsuda, T.; Harada, T.; Nakamura, K. Chem. Commun. 2000,
1367±1368.
26. Almarsson, O.; Klibanov, A. M. Biotechnol. Bioengng 1996,
49, 87±93.
27. Kamat, S.; Beckman, E. J.; Russell, A. J. Enzyme Microb.
Technol. 1992, 14, 265±271.
42. Lee, C. Y.; Kaplan, N. O. Arch. Biochem. Biophys. 1975, 168,
665±676.
43. Lehninger, A. L. Methods Enzymol. 1957, 3, 885±887.
28. Everse, J.; Zoli, E. C.; Kahan, L.; Kaplan, N. O. Bioorg. Chem.
1971, 1, 207±233.