Stereoselective reductions with macrocyclic NADH models

Article abstract of DOI:10.1016/S0957-4166(00)00248-2

Macrocyclic NADH models with two (C₂ symmetry) or four (D₂ symmetry) nicotinamide units comprised in a ring have been prepared and found to reduce activated carbonyl compounds in good yields and high enantiomeric excess. The roles of magnesium ions as a cocatalyst and the temperature have also been investigated. The smaller, C₂-symmetric macrocycle gave 96% ee upon reduction of ethyl benzoylformate whereas the best result with the larger D₂-symmetric model was 81% ee for the reduction of methyl benzoylformate. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0957-4166(00)00248-2

Tetrahedron: Asymmetry 11 (2000) 3027±3040
Stereoselective reductions with macrocyclic NADH models
Ulrik Gran,* Olof Wennerstrom and Gunnar Westman
Department of Organic Chemistry, Chalmers University of Technology, S-412 96 Goteborg, Sweden
È
Received 19 June 2000; accepted 27 June 2000
Abstract
Macrocyclic NADH models with two (C2 symmetry) or four (D2 symmetry) nicotinamide units
comprised in a ring have been prepared and found to reduce activated carbonyl compounds in good yields
and high enantiomeric excess. The roles of magnesium ions as a cocatalyst and the temperature have also
been investigated. The smaller, C2-symmetric macrocycle gave 96% ee upon reduction of ethyl benzoyl-
formate whereas the best result with the larger D2-symmetric model was 81% ee for the reduction of
methyl benzoylformate. # 2000 Elsevier Science Ltd. All rights reserved.
1. Introduction
Nicotinamide adenine dinucleotide (NAD+/NADH) is a coenzyme which takes part in many
biological oxidation±reduction reactions,1 such as the conversion of ketones and aldehydes to
alcohols and vice versa. These reactions are catalyzed by enzymes commonly named
dehydrogenases. The reduction of ketones has been of major interest when mimicking the
chemistry of NAD+/NADH. The enzymatic reduction of a prochiral ketone carried out by
NADH proceeds via a selective transfer of one of the two diastereotopic hydrogens in the
dihydropyridine ring of NADH. The hydrogen is transferred stereoselectively to the ketone, thus
generating a chiral alcohol. Since the ®rst asymmetric reduction using an NADH model reported
by Ohno et al. in 1975,2 there has been a large number of di€erent approaches to NADH
mimicking.3 The general concept is to start with nicotinamide and then modify it, for example, by
introducing various chiral auxiliaries in the amide or methyl groups at C-2 and C-4 in the
dihydropyridine ring.
The NADH models presented in this report are designed by a supramolecular approach, where
the substrate to be reduced is bound into a hydrophobic pocket of the model. The models
comprise a chiral cavity as a substrate binding site to keep the substrate in a ®xed orientation
during the reaction, which should give good control of the stereoselectivity. Several nicotinamide
units are incorporated in the macrocyclic framework which creates this hydrophobic pocket.
* Corresponding author. E-mail: gran@oc.chalmers.se
0957-4166/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S0957-4166(00)00248-2

3028
U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
Using nicotinamide as part of a macrocycle gives a versatile host molecule that is soluble in polar
solvents when the nicotinamide is oxidized (cationic form) and soluble in apolar solvents when
reduced. The nicotinamide units can also function as reducing reagents and possibly as parts of a
bioorganic catalyst.
Four nicotinamide units are incorporated in model compound 1H4 and two units in compound
2H2 (Fig. 1), thus creating multicenter redox systems. Metal multicenter systems are often
important as redox catalysts, while organic multicenter systems have attracted less attention. In
fact, to our knowledge there are only two other groups that have used a multicenter system for
NADH mimicking, but their models were not designed for asymmetric reduction of carbonyl
substrates.4,5 Symmetry is important in models of this kind. The symmetry is needed to ensure a
uniform environment around the redox centers, which is crucial for the prediction and analysis of
the stereochemical outcome of a reduction. In addition, the symmetry and the cyclic structure
restrict the number of conformations, which should also simplify the analyses. We have
6
previously reported the synthesis of the C2-symmetric model 2H2 and a preliminary result of an
asymmetric reduction of methyl benzoylformate using the same compound7 and also a
preliminary report on the synthesis of the D2 model 1H4.8 We now present a more thorough study
of their syntheses and their ability to reduce prochiral carbonyl substrates stereoselectively.
Figure 1. The two NADH model compounds, one with D2 symmetry (1H4) and one with C2 symmetry (2H2)
2. Results and discussion
2.1. Synthesis
The building block technique has been used for the synthesis of the NADH models, i.e. the
models were constructed from di€erent molecular subunits, namely a chiral subunit, a nicotinamide
unit and a spacer. A C2-symmetric diamine seemed to be a good choice as a chiral subunit and
trans-1,2-diaminocyclohexane, whose eciency in asymmetric syntheses and molecular recognition
is well-documented,9 turned out to be best.6 The relative rigidity of the cyclic diamine reduces the
¯exibility of the macrocyclic product. The nicotinamide unit is the reactive part as it functions as
a hydride acceptor or donor, but it also functions as a wall of the cavity. The spacers make up the
center of the cavity and should have hydrophobic properties. The spacers used in our models
originated from di€erent bis(bromomethyl)arene building blocks (Fig. 2). The C2-symmetric
NADH models were synthesized in three steps according to Scheme 1 (exempli®ed by the
synthesis of 2H2).

U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
3029
Figure 2. Di€erent building blocks that were incorporated as spacers in macrocyclic NADH models using the same
synthetic route
Scheme 1. Reagents and conditions: (i) nicotinoyl chloride hydrochloride, pyridine, rt; (ii) 1,3-bis(bromomethyl)-
benzene, DMF, 85ꢀC; (iii) Na2S2O4/Na2CO3, H2O, rt
R,R-1,2-Diaminocyclohexane was used in the ®rst step as the original building block and chiral
source. The resulting cyclic bispyridinium compounds from the second step can be reduced as
crude products. The reduced NADH models were then puri®ed by size-exclusion chromatography
(Sephadex LH-20) with methanol as an eluent. An alternative way is to purify the bispyridinium
compounds prior to reduction. Sodium dithionite is a mild and e€ective reducing agent with very
good regioselectivity. It reduces pyridinium rings exclusively at the para-position. This means that
in practice no further puri®cation is needed after the reduction. The latter method is more time-
consuming but on the other hand it leads to easier analyses and handling of the compounds.
Generally, the reduced forms are much more sensitive and labile than are the oxidized forms.
The above-described synthetic route was also used for other spacers such as 1,4-bis-(bromo-
methyl)benzene, 1,4-bis(chloromethyl)-2,3,5,6-tetramethylbenzene and 1,5-dibromopentane. The
®rst two spacers were linked at the para-position of the aromatic ring and resulted in lower yields.
The products were also more dicult to purify than 2H2. The meta-phenylene spacer (used in
2H2) seemed to be best suited for the incorporation into the macrocyclic framework. The
synthesis using 1,5-dibromopentane failed to give a good yield of a de®ned product, probably
because the pentyl chain is too ¯exible.

3030
U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
Compound 2H2 seemed to be the most promising of the C2-symmetric NADH models and its
host±guest properties and reduction properties were then further studied.
The D2-symmetric NADH models were synthesized in a similar way. The ®rst step was
identical, generating the same building block, 3, with the same con®guration (R,R). The rest of
the synthetic route is shown in Scheme 2, illustrated by the synthesis of 1H4.
Scheme 2. Reagents and conditions: (i) 1,4-bis(bromomethyl)benzene (5 equiv.), DMF, 85ꢀC; (ii) 3, DMF, 85ꢀC; (iii)
Na2S2O4/Na2CO3, H2O, rt
The second step was carried out with an excess of the bis(bromomethyl)arene building block
and at high concentration, so that undesired side reactions such as polymerization and ring
closure were minimized. The crude product was used in the next step because of diculties in
purifying 4. In the third step, 4 was ring-closed with another unit of 3. The puri®cation of the
resulting 14+ is worth some comments. Size-exclusion chromatography did not work satisfactorily,
due to the complex mixture of products. However, Geuder et al. have used a mixture of methanol
and aqueous ammonium bromide to purify macrocyclic polypyridinium ions on silica gel.10 This
system worked well for the puri®cation of 14+ too, after a slight modi®cation. Size-exclusion
chromatography was then used to separate 14+ from ammonium bromide. However, one of the
byproducts formed in the third step was not completely eliminated by this method. We found that
dimethylamine present in the DMF reacted with 4 to form a di€erent macrocycley with similar
chromatographic properties as 14+. The rather simple solution to this problem was to use amine-
free DMF in steps 2 and 3.{ Finally, 14+ was reduced to 1H4 using sodium dithionite.
1
y The macrocycle, with 4 cyclized with dimethylamine instead of 3, was characterized by H NMR spectroscopy and
mass spectrometry (electrospray ionization).
{ It is important to keep the DMF under argon.

U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
3031
Three other compounds were also synthesized according to Scheme 2: two D2-symmetric
models with a meta-phenylene spacer (74+) and a 2,6-naphtylene spacer (64+), respectively; the
third (compound 54+) was an achiral stereoisomer of 14+ that had (R,R)-con®guration at one end
and (S,S)-con®guration of the diaminocyclohexane at the other. Host±guest properties of all
these four compounds were studied,11 but only 1H4 was used as a reducing agent. Since the host±
guest properties were studied in water with the models in their oxidized states, the ®nal reduction
has not been performed for 54+, 64+ and 74+.
The oxidized tetrapyridinium macrocycles are very stable both in the solid state and in solution,
even in acidic, aqueous solutions. However, they decompose in basic solutions. Compound 1H4,
the reduced form, is much more unstable and must be kept under an inert atmosphere and stored
in the dark.
The NMR spectra of 1H4 and 2H2, which show relatively few signals due to the high sym-
metry, are easy to interpret, but the chemical shifts are concentration- and solvent-dependent.
The NMR spectra of the two models are similar. To determine the sizes of the macrocycles we
used mass spectrometry, where both FAB and electrospray (ES) can be used to analyze the
polypyridinium salts.
2.2. Reductions
The two macrocycles 1H4 and 2H2 were studied as reducing agents. The reductions described in
this section were carried out in organic solvents, which means that all of the nicotinamide
subunits contained in the models must be in the reduced form to ensure solubility. Generally,
reductions using NADH models are carried out in acetonitrile, but since 2H2 and 1H4 are not
soluble in acetonitrile the reductions were carried out in methylene chloride and chloroform.
Magnesium is the most commonly used cocatalyst.3 Alcohol dehydrogenases have a zinc ion in
their active sites.12 Magnesium and zinc ions are about equally e€ective in biomimetic systems,
and also in the system presented here. Other metal ions such as nickel, cobalt and cadmium have
been used, but they are not as e€ective.13 Methyl benzoylformate is often used as a substrate in
NADH mimicking reductions, because of the resemblance to pyruvate. a-Keto acids in general
are common substrates in enzymatic reductions,14 where the reduction of pyruvate to lactate
catalyzed by lactate dehydrogenase is one of the most well-known reactions. Other activated
carbonyl compounds as well as activated imines have also been tested in NADH modeling systems.
The emphasis on this reduction study has been on compound 1H4. A few experiments have
been made with 2H2 for comparison. Di€erent substrates have been reduced (Fig. 3) and the
in¯uences of magnesium ion concentration and temperature on the stereoselectivity and reaction
rate have been studied.
Figure 3. Examples of substrates used in reductions: alkyl benzoylformate, a,a,a-tri¯uoroacetophenone and ethyl
acetoacetate

3032
U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
2.2.1. Reductions with 1H4
The in¯uence of magnesium ion concentration. A series of reductions of methyl benzoylformate
and a,a,a-tri¯uoroacetophenone using 1H4 were carried out in methylene chloride with varying
concentrations of magnesium. The results are summarized in Table 1.
Table 1
Reductions of methyl benzoylformate and a,a,a-tri¯uoroacetophenone using 1H4 with varying magnesium
concentrationsa
In the absence of magnesium ions, 1H4 was able to reduce a,a,a-tri¯uoroacetophenone, a
rather reactive ketone, but not methyl benzoylformate. With 0.5 equivalents of magnesium ions
(relative to 1H4), 22% of the formate was converted within 3 days, whereas all of the formate was
converted to the resulting mandelate within 4.5 hours when 4 equivalents were used. On addition to
magnesium ions the rate increases signi®cantly more than linearly with the magnesium concentration.
In contrast, the presence of metal ions has little e€ect on the reduction of a,a,a-tri-
¯uoroacetophenone. Only a slight increase of the reaction rate was observed. The stereoselectivity
of this reduction was poor, only a 12% enantiomeric excess of the R-isomer. The reduction
without magnesium ions present gave the same enantiomeric excess, in contrast to Ohno's
model,2 where the reduction of the same substrate gave 16% enantiomeric excess in the presence
of magnesium ions and a racemic product without magnesium ions. On the other hand, the ste-
reoselectivity of the reduction of methyl benzoylformate was strongly dependent on the amount
of metal ions present. The reduction gave R-mandelate and the enantiomeric excess dropped from
62% with 0.5 equivalents to 30% with 4 equivalents of magnesium ions.
The comparatively large di€erence in reactivity between the two substrates when magnesium
ions are present needs an explanation. It seems as if the role of the metal ion is not only to act as a
Lewis acid, but also to form a ternary complex keeping the reagent and the substrate together.
Methyl benzoylformate has three oxygen atoms that can coordinate to a magnesium ion (Fig. 3). The
oxygen of the a-keto group and one of the other two oxygens can simultaneously coordinate to a
magnesium ion, thus forming a ®ve-membered cyclic complex. In a ternary complex between methyl
benzoylformate and 1H4, where two atoms of the substrate and at least one atom of the reducing
agent 1H4 (probably the carboxamide oxygen15) are coordinated to magnesium, the conditions for a
high enantioselectivity should be present. On the other hand, a,a,a-tri¯uoroacetophenone has only

U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
3033
one weak donor atom that can coordinate to a metal ion and should not be able to form a similar
complex with the same stability and well-de®ned structure. However, the large catalytic e€ect of
magnesium ions on the reduction of methyl benzoylformate shown in Table 1 cannot be
rationalized by the formation of one complex only. The drop in enantiomeric excess from 54 to
33% and the signi®cant increase in reaction rate when the amount of magnesium is changed from
1.25 equivalents to 2 equivalents indicate that there is a fast, less enantioselective, process
competing with a slower and more selective reaction at higher magnesium ion concentrations.
The fast process could involve magnesium ions bound both to the substrate and the reagent
which react in a bimolecular process. The rate of such a reaction involving two magnesium ions
should depend on [Mg2+]2. Alternatively, the fast process could be caused by coordination of
several metal ions to the model compound 1H4. The carbonyl oxygen of the nicotinamide units
coordinates strongly to magnesium ions.15 Since 1H4 has four nicotinamide units there is also the
possibility that two or more metal ions can be coordinated at the same time, which could cause a
change of conformation due to electrostatic repulsion. This new conformation would then be
more reactive but less enantioselective.
Temperature dependence. Reductions were also performed where the substrates were added at
^30ꢀC to a solution of 1H4 and magnesium perchlorate. The reaction mixture was then allowed to
slowly reach room temperature (about 15±20 hours). The results are summarized in Table 2.
Table 2
Reductions carried out at low temperature
The results from the limited number of experiments are consistent with the assumption that
more than one process might be involved. The dominating mechanism is governed by the metal
ion concentration and temperature. A slow enantioselective reaction within a ternary complex
between the substrate, a magnesium ion and the reagent dominates at low temperature and low
magnesium ion concentrations, whereas a faster, less selective process dominates at higher
temperatures and magnesium ion concentrations. This is valid for the reaction with methyl
benzoylformate. A noteworthy detail is that the enantiomeric excess is higher at low temperatures
when 1.25 equivalents of magnesium is used, whereas the ee is lower at low temperatures when 4
equivalents are used compared to reductions performed at room temperature.
The reaction with a,a,a-tri¯uoroacetophenone is assumed to be more straightforward,
occurring between the substrate and the reagent in a bimolecular process. The coordination of
magnesium to the reagent slightly increases the reaction rate for this not very enantioselective
reaction.
Other reductions with 1H4 as the reducing agent. The results presented in Tables 1 and 2 are
from reductions of methyl benzoylformate and a,a,a-tri¯uoroacetophenone that were carried out

3034
U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
in methylene chloride. Compound 1H4 was also tested for other substrates and the reductions of
methyl benzoylformate and a,a,a-tri¯uoroacetophenone were tested in chloroform.
Compound 1H4 was not completely soluble in methylene chloride and magnesium perchlorate
was hardly soluble at all. However, when 1H4 and the metal salt were added together in methylene
chloride, a heterogeneous mixture was formed that was reactive. Compound 1H4 seemed to be
completely soluble in chloroform but no reaction took place (Table 3, entries 1 and 2).
Table 3
Other reductions using 1H4
a
Since 1H4 has four nicotinamide units capable of transferring a hydride ion, it was of interest
to investigate how many of these four hydride ions could be transferred. With 4 equivalents of
both magnesium ions and methyl benzoylformate, compound 1H4 converted about 2 equivalents
of the substrate (Table 3, entry 3). This means that an average of two out of the four hydrogens
were transferred.
The reactivity of 1H4 towards ethyl benzoylformate was also tested (Table 3, entry 4). The
e€ect on the reactivity and selectivity as compared to that towards methyl benzoylformate was
only marginal (see entry 5, Table 1). Compound 1H4 was not able to reduce acetophenone (Table 3,
entry 6). However, the reduction of ethyl acetoacetate proceeded at a detectable rate (Table 3,
entry 5).
2.2.2. Reductions with 2H2
Some selected reductions were carried out for the smaller NADH model 2H2. The results are
summarized in Table 4.
Generally, reductions using 2H2 were more stereoselective than reductions using 1H4. The
same enantiomer (R-form) of the resulting mandelate was obtained in as high as 92% enantio-
meric excess upon reduction of methyl benzoylformate. The reduction of ethyl benzoylformate
gave an even higher ee (96%). As in the case of 1H4, the reduction of a,a,a-tri¯uoroacetophenone
was not as selective as the reduction of esters of benzoylformic acid. The reason is probably the
same, namely that the reaction occurred in a bimolecular process in contrast to the reduction of
the esters of benzoylformic acid, which was assumed to proceed via a ternary complex between
the substrate, a metal ion and the reagent 2H2. In contrast to 1H4, the stereoselectivity was not
a€ected by an increase of magnesium ion concentration (entries 2 and 3). A lowering of the
temperature did not considerably a€ect the stereochemical outcome of a reduction either.7 The
reason for these di€erences could be that a fast, less selective, bimolecular reaction does not take

U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
3035
Table 4
Reductions of di€erent substrates by 2H2
a
place in the case of 2H2. Another explanation could be that the macrocyclic framework of 2H2 is
more rigid than that of 1H4, which entails that coordination of more than one metal ion to 2H2
does not cause a conformational change leading to a less selective reaction.
2.3. A comparison of the two NADH models
It now seems appropriate to compare the two NADH models 1H4 and 2H2 in more detail. The
1H NMR spectra of the two compounds are very similar. However, the signals for the two
diastereotopic hydrogens on the dihydropyridine ring, H-4 and H-40, in 2H2 are well separated
and show di€erent coupling patterns with one hydrogen in an axial position and the other in an
equatorial position. From an analysis of the coupling constants shown in Fig. 4, H-40 was
assigned as the axial hydrogen and H-4 as the equatorial one. From deuterium marking
experiments7 it was found that H-40 is the reactive hydrogen, consistent with theoretical calculations.16
However, the spectrum of 1H4 shows that the two diastereotopic hydrogens have almost exactly
the same shift. An attempt to `freeze out' a conformation wherein the signals for the two
hydrogens are well separated failed.
Figure 4. The coupling patterns for the hydrogens at position 4 on the dihydropyridine ring of 2H2
In order to determine the orientation of the carboxamide group (trans or cis to C-2 of the
dihydropyridine ring) zinc tetraphenylporphyrin (ZnTPP) was used as a shift reagent. ZnTPP is
particularly attractive as a shift reagent in this case since zinc is a cocatalyst for NADH-mediated
reductions. It should therefore be possible to determine the coordination site for the metal ion in
the NADH models. The aromatic ring current of the porphyrin ring is large and well-de®ned and

3036
U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
causes a considerable shift change for protons that are close to the porphyrin ring in a complex.
Fukuzumi et al. formed a complex between 1-benzyl-1,4-dihydronicotine amide (BNAH), a
simple NADH model, and ZnTPP.17 Based on the observed shift changes it was suggested that
the zinc was coordinated to the carboxamide oxygen, which is consistent with the results of other
groups.
1
When ZnTPP was added to 2H2, dramatic H NMR up®eld shift changes were observed. The
e€ects were not as large upon addition to 1H4. The shift changes induced by ZnTPP are
summarized in Fig. 5, where the shift changes of the di€erent protons are shown as a percentage
of the largest shift change.
Figure 5. A comparison of the relative induced shift changes for the two NADH models upon addition of ZnTPP
(percentage of maximum induced shift)
The relative induced shifts are very similar for the two macrocycles and indicate that the zinc
ion is coordinated to the carboxamide oxygen and that the carboxamide group is oriented cis to
C-2 of the dihydropyridine ring in both cases. All NMR data show that the two compounds have
similar conformations. They give the same dominant enantiomers with the same prochiral
substrates. However, the smaller macrocycle 2H2 was more enantioselective upon reduction of
esters of benzoylformic acid due to the absence of a competing, less selective process. E€orts have
been made in order to elucidate the reason for the di€erence in mechanism between the two
model compounds in high magnesium ion concentrations, but so far they have not led to any
conclusive evidence for one or the other of the two explanations. The reduction properties of 1H4
have been investigated more thoroughly and the reason is the potential use of 1H4 as a water-
soluble reducing agent. The cavity of 1H4 is large enough to encapsulate a substrate whereas the
cavity of 2H2 is too small. However, in an apolar solvent such as methylene chloride attractive
forces that a hydrophobic cavity can o€er will not be strong enough to drive a substrate into the
cavity. On the contrary, it is more likely that the substrates react on the outside of the model in
apolar solvents, which is probably the case for the reactions presented here. The next step in our
research is to use the same or similar macrocyclic structure to design a water-soluble model
compound able to stereoselectively reduce substrates encapsulated in a hydrophobic cavity. The
results of 1H4 are promising enough to use this compound as the basis of the design for a
water-soluble NADH model.

U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
3037
3. Experimental
Materials. The reactions were carried out with oven-dried equipment. Methylene chloride was
dried by distillation from calcium hydride under nitrogen. Chloroform was dried by passage
through a column of basic alumina (Merck, aluminum oxide 60 active, grade I; 10 g per 14 mL of
solvent). Pyridine was distilled prior to use. DMF was purchased from Fluka (absolute, ꢁ99.5%)
and was stored under argon. Commercially available chemicals were used without further
puri®cation, if nothing else is mentioned.
Methods. Thin-layer chromatography was performed on silica gel plates (Merck, silica gel 60
F254). Column chromatography was performed using silica gel (Matrex, LC 60 A/35±70 mm).
Size-exclusion chromatography was performed using Sephadex LH-20 in methanol. H NMR
1
spectra were recorded at 293 K on a Varian UNITY-400 NMR spectrometer at 400 MHz with
DMSO-d6 or CDCl3 with tetramethylsilane as internal standard as solvents. IR spectra were
recorded on an FT-IR instrument, Perkin±Elmer 1600. Mass spectra were recorded on a VG
ZabSpec instrument. Positive FAB-MS (fast atom bombardment-mass spectrometry) with
thioglycerol/glycerol as the liquid matrix and positive ESI-MS (electrospray ionization-mass
spectrometry) were the methods used. Capillary gas chromatography was performed using a
Varian 3300 gas chromatograph equipped with a methyl silicone fused silica column (column i.d.
0.32 mm, column length 30 m). Optical rotations were measured on a Perkin±Elmer 241
polarimeter using a 1 dm cell with a total volume of 1 mL. The enantiomeric excess was measured
using a Hewlett±Packard 5890 gas chromatograph equipped with a b-cyclodextrin based column
(Chrompack, CP-Chirasil-dex CB, column i.d. 0.25 mm, column length 25 m).
3.1. Synthesis of bispyridine building block 3
Nicotinoyl chloride hydrochloride (18.3 g, 103 mmol) was added at 0ꢀC to a solution of
(1R,2R)-(^)-1,2-diaminocyclohexane (5.0 g, 44 mmol) in pyridine (125 mL). The solution was
stirred under argon at room temperature for 2 days. The pyridine was evaporated and 75 mL of
water was added to the light yellow residue. The pH value was adjusted to pH 9 by NaOH (10%
aq.) and the solution was cooled to 0ꢀC. The white crystals formed were ®ltered o€ by suction
and washed with cold water. Yield 12.5 g (88%). Recrystallization from methanol gave white
needles, mp: 269.5±270.5ꢀC; ‰ꢀŠD20=^131þ1 (c 3.4 mg/mL; pyridine). 1H NMR (DMSO-d6): ꢁ 1.31
(m, broad, 2H), 1.53 (m, broad, 2H), 1.75 (m, broad, 2H), 1.88 (d, broad, 2H), 3.95 (m, broad,
2H), 7.42 (dd, J=5 and 8 Hz, 2H), 7.99 (dtr, J=2 and 8 Hz, 1H), 8.52 (d, J=8 Hz, 2H), 8.61 (dd,
J=1.6 and 5 Hz, 2H), 8.82 (s, 2H).
3.2. Synthesis of macrocyclic bispyridinium compound 22+
A sample of 3 (648 mg, 2 mmol) and 1,3-bis(bromomethyl)benzene (528 mg, 2 mmol) were
dissolved in 100 mL of DMF under argon. The solution was heated to 85ꢀC and stirred under
argon overnight. The mixture was cooled and poured into 400 mL of diethyl ether. After stirring
for 20 min the precipitate formed was ®ltered o€ by suction and washed with diethyl ether (3Â50
mL). The crude product was puri®ed by ¯ash chromatography (silica gel, methanol:NH4Br (25%
aq.), 4:1) followed by size-exclusion chromatography, giving 940 mg (80% yield) of a white
powder. ‰ꢀŠ2D0=^90þ2 (c 3.2 mg/mL; MeOH). FAB-MS: [M^Br]+ 507.140 (calculated 507.140). 1H
NMR (DMSO-d6): ꢁ 1.35 (m, broad, 2H), 1.67 (m, broad, 2H), 1.80 (m, broad, 2H), 1.95 (d,

3038
U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
broad, 2H), 3.92 (m, broad, 2H), 5.94 (s, 4H), 7.01 (s, 1H), 7.56 (tr, J=8 Hz, 1H), 7.69 (d, J=8
Hz, 2H), 8.40 (dd, J=6 and 8 Hz, 2H), 8.90 (d, J=8 Hz, 2H), 9.04 (d, J=8 Hz, 2H), 9.16 (s, 2H),
9.48 (d, J=6 Hz, 2H). IR (KBr): 1662 (CˆO), 3425 (N±H, amide).
3.3. Reduction of 22+ to 2H2
Sodium dithionite (2.78 g, 16 mmol) and sodium carbonate (1.69 g, 16 mmol) were dissolved in
20 mL of water. An aqueous solution of 22+ (940 mg, 1.6 mmol) was added dropwise and the
resulting solution, which turned light red and became viscous within minutes, was stirred at room
temperature under argon for 3 hours. Cold water (20 mL) was added and the yellow-red
precipitate was ®ltered o€ by suction and washed with cold water (3Â20 mL). Yield 650 mg
1
(95%) of 2H2. FAB-MS: [M]+ 430.237 (calculated 430.237). H NMR (CDCl3): ꢁ 1.34 (s, broad,
4H), 1.78 (s, broad, 2H), 1.92 (m, broad, 2H), 2.99 (dd, J=3.6 and 17 Hz, 2H), 3.22 (dm, J=17
Hz, 2H), 3.84 (m, broad, 2H), 4.17 (d, J=16 Hz, 2H), 4.41 (d, J=16 Hz, 2H), 4.70 (m, 2H), 5.24
(d, J=10 Hz, 2H), 5.59 (dd, J=1.2 and 8 Hz, 2H), 7.04 (d, J=8 Hz, 2H) 7.18 (s, 1H) 7.26 (tr,
J=8 Hz, 1H).
3.4. Synthesis of 4
A sample of 3 (648 mg, 2 mmol) and 1,4-bis(bromomethyl)benzene (2.64 g, 10 mmol) were
dissolved in 4 mL of DMF under argon. The solution was heated to 85ꢀC and stirred under argon
for 3 hours. The mixture was cooled and poured into 100 mL of chloroform, whereupon 500 mL
of diethyl ether was added. The mixture was stirred for 20 min and a white precipitate was ®ltered
o€ by suction and washed with diethyl ether (3Â50 mL) to give the crude product 4, which was
used without puri®cation in the next step.
3.5. Synthesis of macrocyclic tetrapyridinium compound 14+
A sample of 3 (639 mg, 1.97 mmol) and 4 (1.68 mg, 197 mmol) were dissolved in 100 mL of
DMF under argon. The solution was heated to 85ꢀC and stirred under argon for 3 hours. The
mixture was cooled and poured into 500 mL of diethyl ether. After stirring for 20 min the
precipitate was ®ltered o€ by suction and washed with diethyl ether (3Â50 mL) to give the crude
product. Flash chromatography (silica gel, methanol:NH4Br (25% aq.), 4:1) followed by size-
exclusion chromatography gave 587 mg (25% yield) of white crystals. ‰ꢀŠ2D0=^73þ2 (c 3.06 mg/
1
mL; MeOH). FAB-MS: [M^Br^]+ 1093.197 (calculated 1093.198). H NMR (DMSO-d6): ꢁ 1.33
(m, broad, 4H), 1.59 (m, broad, 4H), 1.80 (m, broad, 4H), 1.92 (d, broad, 4H), 3.91 (m, broad,
4H), 5.81 (d, J=14 Hz, 4H), 5.90 (d, J=14 Hz, 4H), 7.55 (s, 8H), 8.21 (dd, J=6 and 8 Hz, 4H),
8.58 (d, J=8 Hz, 4H), 9.00 (d, J=6 Hz, 4H), 9.26 (d, J=6 Hz, 4H), 9.33 (s, 4H). IR (KBr): 1650
(CˆO), 3425 (N±H, amide).
3.6. Reduction of 14+ to 1H4
Sodium dithionite (1.74 g, 10 mmol) and sodium carbonate (1.06 g, 10 mmol) were dissolved in
10 mL of water. An aqueous solution of 14+ (585 mg, 0.5 mmol) was added dropwise and the
solution, which turned yellow and became turbid within minutes, was stirred at room
temperature under argon for 6 hours. Methanol (40±50 mL) was added and the solvents were

U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
3039
removed under reduced pressure. The product was dissolved in chloroform, ®ltered to remove
inorganic salts and concentrated. Yield 420 mg (98%) of 1H4 as a red solid. 1H NMR (CDCl3): ꢁ
1.25 (m, broad, 4H), 1.35 (m, broad, 4H), 1.74 (m, broad, 4H), 2.10 (d, broad, J=13 Hz, 4H),
3.05 (m, broad, 8H), 3.70 (m, 4H), 4.15 (d, J=16 Hz, 4H), 4.23 (d, J=16 Hz, 4H), 4.65 (m, 4H),
5.65 (dd, J=1.2 and 8 Hz, 4H), 5.90 (d, J=7 Hz, 4H), 7.05 (s, 4H), 7.20 (s, 8H).
3.7. Synthesis of macrocyclic tetrapyridinium compound 54+
Prepared as described for 14+. In the last step, the S,S-form of 3 was used. Yield 14% of a white
1
powder. FAB-MS: [M^3Br^]+ 935.359 (calculated 935.361). H NMR (DMSO-d6): ꢁ 1.35 (m,
broad, 4H), 1.59 (m, broad, 4H), 1.80 (m, broad, 4H), 1.94 (d, broad, 4H), 3.96 (m, 4H), 5.84 (d,
J=14 Hz, 4H), 5.91 (d, J=14 Hz, 4H), 7.51 (s, 8H), 8.25 (dd, J=6 and 8 Hz, 4H), 8.76 (d, J=8
Hz, 4H), 9.07 (d, J=6 Hz, 4H), 9.31 (d, J=6 Hz, 4H), 9.44 (s, 4H). IR (KBr): 1650 (CˆO), 3412
(N±H, amide).
3.8. Synthesis of macrocyclic tetrapyridinium compound 64+
Prepared from 2,6-bis(bromomethyl)naphthalene as described for 14+. Yield 12% of a white
1
powder. ESI-MS: [M^2Br^]2+ 557.3 (calculated 557.2). H NMR (DMSO-d6): ꢁ 1.32 (m, broad,
4H), 1.58 (m, broad, 4H), 1.78 (m, broad, 4H), 1.93 (m, broad, 4H), 3.95 (m, broad, 4H), 5.97 (d,
J=14 Hz, 4H), 6.04 (d, J=14 Hz, 4H), 7.47 (d, J=8 Hz, 4H), 7.77 (d, J=8 Hz, 4H), 7.92 (s, 4H),
8.20 (dd, J=6 and 8 Hz, 4H) 8.76 (d, J=8 Hz, 4H), 9.06 (d, J=7 Hz, 4H), 9.31 (d, J=6 Hz, 4H),
9.45 (s, 4H). IR (KBr): 1668 (CˆO), 3418 (N±H, amide).
3.9. Synthesis of macrocyclic tetrapyridinium compound 74+
Prepared from 1,3-bis(bromomethyl)benzene as described for 14+. Yield 14% of a white pow-
20
1
der. ‰ꢀŠD =^40þ2 (c 2.75 mg/mL; MeOH) MS: [M^Br^]+ 1093.219 (calculated 1093.198). H
NMR (DMSO-d6): ꢁ 1.36 (m, broad, 4H), 1.64 (m, broad, 4H), 1.81 (m, broad, 4H), 1.95 (d,
J=12 Hz, broad, 4H), 3.99 (m, 4H), 5.87 (d, J=14 Hz, 4H), 5.92 (d, J=14 Hz, 4H), 7.38 (tr,
J=8 Hz, 2H), 7.49 (d, J=8 Hz, 4H), 7.89 (s, 2H), 8.27 (tr, J=6 and 8 Hz, 4H), 8.79 (d, J=8
Hz, 4H), 9.02 (d, J=7 Hz, 4H), 9.27 (d, J=6 Hz, 4H), 9.47 (s, 4H). IR (KBr): 1650 (CˆO),
3425 (N±H, amide).
3.10. General procedure for reductions using models 2H2 and 1H4
A 10 mL round-bottomed ¯ask was charged with magnesium perchlorate and the model
compound (80±100 mmol), sealed with a rubber septum, and ¯ushed with argon. Freshly
distilled methylene chloride (2.5±3 mL) was added via syringe and the mixture was stirred
for 1 to 2 hours, whereupon the substrate was added via syringe. The ¯ask was kept in the
dark and the reaction mixture was stirred under argon. The reduction was monitored by
capillary gas chromatography. The reaction was quenched by adding 7±8 mL of water. The
aqueous phase was extracted with methylene chloride (3Â10 mL) and the combined organic
phases were dried over Na2SO4, ®ltered and concentrated. The optical rotation was measured
to determine the enantiomer in excess and the enantiomeric excess was measured using
capillary GC.

3040
U. Gran et al. / Tetrahedron: Asymmetry 11 (2000) 3027±3040
References
1. Sund, H. Pyridine Nucleotide-dependent Dehydrogenases; W. de Gruyter & Co.: Berlin, Germany, 1977.
2. Ohnishi, Y.; Kagami, M.; Ohno, A. J. Am. Chem. Soc. 1975, 97, 4766±4768.
3. (a) Inouye, Y.; Oda, J.; Baba, N. Asymmetric Synthesis; Academic Press: New York, 1983; Vol. 2. (b) Yasui, S.;
Ohno, A. Bioorg. Chem. 1986, 14, 70±96. (c) Burgess, V. A.; Davies, S. G.; Skerlj, R. T. Tetrahedron: Asymmetry
1991, 2, 299±328. (d) Murakami, Y.; Kikuchi, J.; Hisaeda, Y.; Hayashida, O. Chem. Rev. 1996, 96, 721±758.
4. Murakami, Y.; Aoyama, Y.; Kikuchi, J.-i.; Nishida, K. J. Am. Chem. Soc. 1982, 104, 5189±5197.
5. Kuroda, Y.; Seshimo, H.; Kondo, T.; Shiba, M.; Ogoshi, H. Tetrahedron Lett. 1997, 38, 3939±3942.
6. Skog, K.; Wennerstrom, O. Tetrahedron 1994, 50, 8227±8236.
7. Skog, K.; Wennerstrom, O. Tetrahedron Lett. 1992, 33, 1751±1754.
8. Skog, K.; Wennerstrom, O. Tetrahedron Lett. 1995, 36, 4629±4632.
9. Bennani, Y. L.; Hanessian, S.; Chem. Rev. 1997, 97, 3161±3195, and references cited therein.
10. Geuder, W.; Hunig, S.; Suchy, A. Tetrahedron 1986, 42, 1665±1677.
11. Manuscript in preparation.
12. Stryer, L. Biochemistry, 3rd ed.;, W. H. Freeman and Co.: New York, 1988; p. 363.
13. (a) Park, K. K.; Lee, J. H.; Park, J. W. Bioorg. Chem. 1991, 19, 433±444. (b) Engbersen, J. F. J.; Koudijs, A.; van
der Plas, H. C. J. Org. Chem. 1990, 55, 3647±3654.
14. Wong, C.-H.; Whitesides, G. M. Enzymes in Synthetic Organic Chemistry; Elsevier Science, 1994; pp. 154±158.
15. This is generally accepted and has been proven by, for example, 13C NMR spectroscopy by several groupsÐsee
Ref. 3c, and references cited therein.
16. (a) Wu, Y.-D.; Houk, K. N. J. Am. Chem. Soc. 1991, 113, 2353±2358. (b) Nambiar, K. P.; Stau€er, D. M.;
Kolodziej, P. A.; Benner, S. A. J. Am. Chem. Soc. 1983, 105, 5886±5890.
17. Fukuzumi, S.; Kondo, Y.; Mochizuki, S.; Tanaka, T. J. Chem. Soc., Perkin Trans. 2 1989, 11, 1753±1761.