Synthesis of enantiomerically-pure [13C]aristeromycylcobalamin and its reactivity in dioldehydratase, glyceroldehydratase, ethanolamine ammonia-lyase and methylmalonyl-CoA mutase reactions

Article abstract of DOI:10.1002/chem.200390073

We describe a novel enantio-selective synthesis of aristeromycin, the carbocyclic analogue of adenosine. The seven-step synthesis is also suitable for the preparation of specifically-labelled [6′-¹³C]aristeromycin. Both the unlabelled and

Full text of DOI:10.1002/chem.200390073

FULL PAPER
Synthesis of Enantiomerically-Pure [13C]Aristeromycylcobalamin and Its
Reactivity in Dioldehydratase,Glyceroldehydratase,Ethanolamine
Ammonia-Lyase and Methylmalonyl-CoA Mutase Reactions
Ulrich Weigl,[a] Martin Heimberger,[a] Antonio J. Pierik,[b] and Ja¬nos Re¬tey*[a]
Abstract: We describe a novel enantio-
selective synthesis of aristeromycin, the
carbocyclic analogue of adenosine. The
seven-step synthesis is also suitable for
the preparation of specifically-labelled
[6'-13C]aristeromycin. Both the unlabel-
led and 13C-labelled product was cou-
pled to vitamin B12 to form aristeromy-
cylcobalamin. This carbocyclic analogue
of coenzyme B12 was examined for its
coenzymic activity with several adeno-
sylcobalamin-dependent enzymes. For
glyceroldehydratase and dioldehydra-
tase, the reaction rate (kcat) was 38 and
44% of that measured with adenosylcob-
alamin as coenzyme. In contrast, aris-
teromycylcobalamin showed no detect-
able activity with methylmalonyl-CoA
mutase and ethanolamine ammonia-
lyase. Instead, it was a weak inhibitor
of the former and a strong inhibitor of
the latter enzyme. The slower turnover
rate with glyceroldehydratase raised the
hope of detecting the 6'-deoxyaristero-
mycyl radical intermediate. Comparison
of the EPR spectra of the intermediates
in the glyceroldehydratase reaction,
which used adenosyl- and aristeromy-
cylcobalamines, respectively, as coen-
zyme, revealed a significant shift and
this suggests a different geometric posi-
tion of these cofactors at the binding site
during the cleavage of the carbon-cobalt
bond. However, we found no evidence
for the existence of a 6'-deoxyaristero-
mycyl radical during the reaction with
[6'-13C]aristeromycylcobalamin.
We
Keywords: cobalamines ¥ cofactors
¥ enzyme catalysis ¥ EPR spectros-
copy ¥ kinetics ¥ synthetic methods
conclude that the lifetime of this radical
is still too short to be observed.
partners so that it can rearrange cleanly to the product radical.
Hydrogen atom transfer to the product radical from the 5'-
methyl group of the intermediate 5'-deoxyadenosine is
followed by regeneration of AdoCbl to complete the catalytic
cycle.
Introduction
Adenosylcobalamin (AdoCbl) is an essential cofactor of
various enzymes that catalyse rearrangement reactions.[1 4] A
unique property of AdoCbl, which is also known as coenzy-
me B12, is the covalent cobalt-carbon bond. This bond is stable
in aqueous solution in the dark but undergoes homolysis when
exposed to light.[5] In the enzyme-bound state this homolysis is
reversible, strictly controlled, and is the starting step in the
The corresponding enzymes are able to accelerate the
À
homolysis of the cobalt carbon bond by a factor of up to
10]
1012,[6, 7] but only on substrate binding.[8
Obviously, a
portion of the binding energy is used to induce a conforma-
À
À
rearrangement reactions. The cleavage of the cobalt carbon
tional change which weakens the cobalt carbon bond. The
bond generates the highly reactive 5'-deoxyadenosyl radical,
which is able to abstract a hydrogen atom from a non-
activated position of the substrate. The substrate radical is
equally unstable but is protected from possible reaction
occurrence of radical intermediates can be directly shown by
EPR spectroscopy. When more than one such species are
present, only the more stable species, that is those that have
the longest lifetime, can be detected.
Upon homolysis of the cobalt carbon bond, the relatively
stable paramagnetic cob(ii)alamin and the extremely unstable
deoxyadenosyl radical are formed. Thusfar the deoxyadenosyl
radical has not been observed spectroscopically, but its
intermediacy has been documented by numerous experiments
with deuterium and tritium-labels.[11, 12] In adenosylcobala-
min-dependent enzymes, rather stable organic radicals, which
are generated by hydrogen abstraction by the deoxyadenosyl
radical from the substrate or product, have been detected by
EPR spectroscopy.[8, 13] The localisation of the carbon-centred
radicals in the substrate/product has been experimentally
¬
[a] Prof. Dr. J. Retey, Dr. U. Weigl, M. Heimberger
Institut f¸r Organische Chemie, Lehrstuhl f¸r Biochemie
Universit‰t Karlsruhe
Kaiserstrasse 12, 76128 Karlsruhe (Germany)
Fax (‡49)721-6084-823
E-mail: biochem@ochhades.chemie.uni-karlsruhe.de
[b] Dr. A. J. Pierik
Department of Microbiology
Philipps University Marburg
Karl Von Frischstrasse, 35032 Marburg (Germany)
Fax (‡49)64-212-828-979
E-mail: pierik@mailer.uni-marburg.de
652
¹ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/03/0903-0652 $ 20.00+.50/0
Chem. Eur. J. 2003, 9, No. 3

652 660
determined for ethanolamine
ammonia-lyase,[14]
glutamate
Cl
N
N
1.) Cl-purine, PPh3 - polymer, [Pd2(dba)3]
NEt3,THF, 57°C, 15 h
2.) Bzl-Cl, RT, 17 h
N
N
N
O
O
mutase,[15] and 2-methyleneglu-
tarate mutase (A. J. Pierik, un-
published results) by EPR spec-
troscopy of the enzyme in the
presence of isotopically-label-
led substrates. These studies
and the rapid-freeze experi-
ments performed by Babior
et al.[14] and Leutbecher et al.[16]
exclude substantial concentra-
tions of amino acid-centred and
deoxyadenosyl radical species
as intermediates.
O
N
HO
O
60%
1
2
Cl
N
Cl
13C-PhSO2CH2NO2, NEt3
[Pd2(dba)3], PPh3-polymer
THF, RT, 15 h
N
N
N
O
O
N
Na2CO3, TBAOX
MeOH, RT, 15 h
NO2
S
O
*
N
N
*
O
88%
72%
5
4
Cl
N
H2N
N
N
N
N
conc. NH3, THF
RT, 3 d
LiHB(Et)3, THF, -78°C, 3 h
72%
*
*
HO
N
HO
N
87%
Attempts to detect the 5'-
deoxyadenosyl radical have in-
volved its stabilisation by the
introduction of a double bond
into the ribose ring. The use of
3',4'-anhydroadenosylcobala-
min as a cofactor gave rise to
the more stable allylic 3',4'-
anhydroadenosyl radical. This
analogue lacks the 3'-OH group
and was only slightly active with
B12-dependent enzymes. Its
oxygen sensitivity is very high
and this led to its decomposi-
tion in a few minutes.[17] Abeles
and co-workers have prepared
an analogue of coenzyme B12,
in which instead of the adenosyl
group, racemic aristeromycin
was covalently bound to the
central cobalt.[18] The coenzy-
mic activity of this carbocyclic
analogue in the dioldehydra-
7
6
H2N
NH2
N
7
6
1
5
N
N
N
N
8
NMO, OsO4
*
2
THF/H2O, 0°C, 24 h
5'
HO
N
1'
2'
*
4
N
HO
9
6'
4'
3'
N
3
49% / 38%
+
HO
OH
HO
OH
8
9
HMPTA,
SOCl2
RT, 15 h
HMPTA,
SOCl2
RT, 15 h
84%
84%
H2N
N
H2N
N
N
N
N
N
*
*
Cl
N
N
Cl
HO
OH
HO
OH
10
11
Scheme 1. Synthesis of 6'-13C-labelled aristeromycin (8) and its 6'-chlorinated deoxy-analogue (10).
tase reaction was about 30% of that measured with AdoCbl.
It was concluded that replacement of the ribosyl O-atom by
methylene does not severely affect the reaction or, in other
words, that this oxygen atom does not play an essential role in
the catalytic process. We succeeded in achieving a novel
enantioselective synthesis of aristeromycin and its [6'-13C]
isotopomer for the investigation of the coenzymic activity of
aristeromycylcobalamin with several B12-dependent enzymes.
After the stereoisomerically-pure carbanucleoside was cou-
pled to cobalamin, the coenzyme analogue was used for
kinetic and EPR measurements.
the presence of trisdibenzylidene acetone dipalladium(0)
which was converted in situ into the catalytically-active but
oxygen-sensitive tetrakistriphenylphosphine palladium(0).
We used a commercially available polymer-bound triphenyl-
phosphine instead of free triphenylphosphine. Although this
required a longer reaction time at 578C, this avoided a tedious
removal of the side product, triphenylphosphine oxide. After
the polymer-bound material was removed by filtration, the
intermediate allylic alcohol was reacted with benzoylchloride
to give (1'R,4'S)-4'-benzyloxy-1'-(6-chloro-9H-purin-9-yl)cy-
clopent-2'-ene (2) in 60% yield. Since an excess of triethyl-
amine was present in solution from the previous step, HCl,
which had been simultaneously generated, was immediately
neutralized. In the next step, the benzoate group was
substituted by the hydroxymethyl moiety. Here we could
introduce the 13C-label. The anion formed from phenyl-
sulphonylnitromethane (3) was most suitable for this purpose.
Compound 3 was prepared by the method of Wade et al.[20]
but with major modifications (Scheme 2). As nitromethane is
commercially available with a [13C]enrichment ꢀ99%, it was
treated with sodium phenylsulphinate in the presence of
Results and Discussion
Synthesis of (2'S,3'R,4'R)-(À)-aristeromycin (8), its 2',3'-bis-
epi isomer (9) and its [6'-13C]-isotopomer (Scheme 1): The
starting material of this synthesis was the commercially-
available and enantiomerically-pure (1S,4R)-cis-acetoxycy-
clopent-2-ene-1-ol (1) which was coupled to chloropurine by
the method of Trost et al.[19] The reaction was carried out in
Chem. Eur. J. 2003, 9, No. 3
¹ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/03/0903-0653 $ 20.00+.50/0
653

¬
J. Retey et al.
FULL PAPER
bis-epi analogue (13) in 60 and 55% yield, respectively
(Scheme 3). Due to the high light sensitivity of the alkylcob-
alamins, both of the last reaction steps and the subsequent
purification by HPLC was carried out in the dark. Compounds
12 and 13 and their [6'-13C]isotopomers were characterized by
NMR and mass spectrometry.
O
S
NO2
CH2
O
DBU, iodine, RT, 1 h
40%
*
H3C NO2
+
S
Na
*
O
O
3
Scheme 2. Synthesis of phenylsulphonyl[13C]nitromethane (3) from so-
dium phenyl-sulphinate and [13C]-labelled nitromethane.
HO
OH
A12
A11
A13
iodine and the base 1,8-diazobicyclo[5.4.0]undec-7-ene
(DBU) to give phenylsulphonylnitromethane (3) in 40%
yield. The substitution of the benzoate group of 2 was again
accomplished by the method of Trost et al.[19] using the same
reagents as previously described for the step 1 ! 2; however,
this reaction took place at room temperature. Triethylamine
served both to start the catalytic cycle and to deprotonate 3 to
form the corresponding carbanion. Nitrosulphonate 4 was
obtained in 88% yield as a nearly 1:1 mixture of diaster-
eomers (with the introduction of a new chiral centre!).
Separation of the diastereomers was unnecessary as oxidative
cleavage of nitrosulphonate 4 with tetrabutylammonium
oxone (TBAOX) in MeOH/CH2Cl2 gave the methylester 5
with the concomitant abolition of the asymmetric centre that
was created in the previous step.[19] The last two steps also
offer the possibility of specifically introducing an isotope label
into other carbocyclic nucleosides such as carbovir or
NH2
A6
A14
N
A5
A8
13CH2
A16
A15
N A4
N
N
A2
CH2CH2CONH2
H2NCOCH2
CH2CONH2
N
N
CH2CH2CONH2
Co+
C10
N
N
H2NCOCH2
O
CH2CH2CONH2
HN
L3
L1
O-
L2
O
B4
N
B9
B10
B5
HO
P
O
B2
O
R3
R4
N
R2
O
B8
B6
B7
R1
B11
HOCH2
R5
Scheme 3. [6'-13C]aristeromycylcobalamin (12) and its 2',3'-bis-epi isomer
(13). The NMR signals were annotated with the following letters: A:
aristeromycyl, B: benzimidazole, C: corrin, L: loop, R: ribosyl. Position of
the 2'- and 3'-hydroxy groups of the aristeromycyl moiety: 12: anti, 13: syn.
26]
abacavir, which are important antiviral agents.[21
Reduction of methylester 5 by ™superhydride∫ (lithium
triethylborohydride) at À788C afforded the alcohol 6 in 72%
yield. The use of stronger reducing agents could attack the
chloropurine moiety. In fact, even with ™superhydride∫, the
reaction mixture had to be quenched with ethylacetate before
it was allowed to warm to room temperature. The chloro-
substituent of the purine ring was then replaced with an amine
group by stirring 6 in aqueous ammonia at room temperature.
The resulting carbocyclic nucleoside 7 was easily purified and
gave colourless crystals in 87% yield. Finally cis-hydroxyla-
tion of 7 using osmium tetroxide in catalytic amounts afforded
aristeromycin (8) and its 2',3'-bis-epimer 9. The reaction was
carried out at 08C in the presence of excess methylmorpho-
line-N-oxide and gave 8 as the major product (49%) and its
2',3'-bis-epimer 9 (38%). Compounds 8 and 9 were separated
by HPLC.
Kinetic investigations: Aristeromycylcobalamin served as
coenzyme for diol- and glyceroldehydratase. KM values were
identical within experimental error to those of adenosylcob-
alamin for both these enzymes. However, their kcat values
were reduced to 44 and 38% of the values for dioldehydratase
and glyceroldehydratase, respectively. Considering the high
amino-acid sequence identity of the two enzymes, their similar
acceptance of aristeromycylcobalamin was not surprising.
Despite their high sequence identity, glyceroldehydratase has
a much higher affinity for both coenzymes (Table 1). It seems
that the ether oxygen of the ribosyl moiety in AdoCbl does
not play a role in the mechanism of the reaction. The
somewhat reduced kcat values may arise from the slightly
different geometry of the cyclopentane moiety with respect to
the furanose ring.[28]
The synthesis of [6'-13C]-(À)-aristeromycin ([6'-13C](8)) was
achieved from the monoacetate 1 in seven steps with an
overall yield of 12%.
Table 1. Kinetic properties of coenzyme B12 and aristeromycylcobalamin with
glycerol-dehydratase (C. freundii) and dioldehydratase (S. typhimurium).
Synthesis of aristeromycylcobalamin (12),and 2 ',3'-bis-epi
aristeromycylcobalamin (13): Aristeromycin (8) and its 2',3'-
bis-epi isomer (9) were converted into their 6'-chloro-6'-
deoxy-analogues 10 and 11 as previously described.[27] Thion-
ylchloride in hexamethylphosphoramide was used as chlori-
nation agent. In this way, only the primary hydroxyl group was
substituted. The products were purified by ion exchange
chromatography.
In the last step of the synthesis, hydroxycobalamin was
reduced with sodium borohydride in presence of cobalt(ii)-
chloride to cob(i)alamin (vitamin B12s).[12] Cob(i)alamin was
then alkylated with the 6'-chloro-6'-deoxyaristeromycins (10)
and (11) to afford aristeromycylcobalamin (12), and its 2',3'-
Compound
Glyceroldehydratase
Dioldehydratase
KM [mm]
KM [nm] kcat [10À3 sÀ1
]
kcat [10À3sÀ1
]
adenosylcobalamin
aristeromycylcobalamin 11 Æ 2
10 Æ 1
173 Æ 3
65 Æ 1
0.54 Æ 0.06 233 Æ 8
0.57 Æ 0.03 102 Æ 1
Surprisingly, methylmalonyl-CoA mutase and ethanol-
amine ammonia-lyase seemed unable to use aristeromycylcob-
alamin as coenzyme. While the mutase was only weakly
inhibited, ethanolamine ammonia-lyase was strongly inhib-
ited by the coenzyme analogue (Table 2). In our kinetic
investigations we used the ™parallel∫ method, that is the
simultaneous addition of AdoCbl and aristeromycylcobala-
654
¹ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/03/0903-0654 $ 20.00+.50/0
Chem. Eur. J. 2003, 9, No. 3

Aristeromycin Analogues
652 660
radical 7 ä away in the mutases[14] and >10 ä away in the
dehydratases.[33, 34] These predictions were corroborated by
the X-ray crystallographic studies of glutamate mutase and
dioldehydratase.
Table 2. Kinetic properties of coenzyme B12 and aristeromycylcobalamin
with methylmalonyl-CoA mutase (P. shermanii) and ethanolamine ammo-
nia-lyase (S. typhimurium).
Compound
Methylmalonyl-CoA
mutase
Ethanolamine
ammonia-lyase
On incubation for less than 30 s of glyceroldehydratase
from Citrobacter freundii (overexpressed in E. coli) with
racemic 1,2-propanediol (100 mm) and the coenzymes
AdoCbl, aristeromycylcobalamin (12) and [6'-13C]aristeromy-
cylcobalamin ([6'-13C](12)), the ™doublet∫ EPR feature that is
characteristic of the organic radical partner of the cob(ii)ala-
min-radical pair was observed (Figure 2). The features, two
Km [nm]
Ki [nm]
Km [nm]
Ki [nm]
adenosylcobalamin
aristeromycylcobalamin
57 Æ 4
65 Æ 5
830 Æ 45
16 Æ 2
min.[29] The experimental data were analyzed by Dixon
linearization. For ethanolamine ammonia-lyase details for
determination of the apparent Ki value are shown in Figure 1.
The Ki value for the lyase is four times lower than the value of
KM for AdoCbl and is comparable to that for hydroxycobal-
amin (Ki ˆ 12.7 nm[30]).
Figure 2. EPR spectroscopy of glyceroldehydratase (C. freundii) incubat-
ed with glycerol and coenzyme B12, [13C]-labelled and unlabelled arister-
omycylcobalamin. Tˆ 77 K; microwave power, 1.2 mW; modulation am-
plitude, 1.0 mT; receiver gain, 5 Â 104; frequency, 9.431 GHz.
sharp derivative-shaped lines with a P1/2 of ꢁ10 mW at 77 K,
compare well with those of the dehydratase type of coupled
spectra, in particular with those of the glyceroldehydratase
from Klebsiella pneumoniae ATCC 25955 (formerly Aero-
bacter aerogenes, PZH 572)[31]. In addition, broad signals with
g values between 2.3 and 2 were seen, these signals are
assigned to the low spin CoII of the B12 present in the enzyme-
substrate complex. The exact positions of the derivative-
shaped features of the doublet system are g ˆ 2.041 Æ 0.001
and g ˆ 1.927 Æ 0.003 for AdoCbl and at g ˆ 2.032 Æ 0.001 and
g ˆ 1.955 Æ 0.003 for aristeromycylcobalamin (12) and [6'-
13C]aristeromycylcobalamin ([6'-13C](12)). The linewidths for
aristeromycylcobalamin (12) and [6'-13C]aristeromycylcoba-
lamin ([6'-13C](12)) are identical within experimental error
but are significantly larger than those obtained with AdoCbl.
Also the ratio of intensity between the low and high field
feature is lower (2.6:1) than in the spectrum with AdoCbl
(4.1:1). This ratio is sensitive to both angular constants and the
distance between CoII and the radical as reflected by the
coupling constant.[33] One can conclude that the geometry of
the complex with aristeromycylcobalamin is similar but has
subtle differences with respect to that with adenosylcobala-
min. It seems that this distance becomes shorter when
glyceroldehydratase is activated by aristeromycylcobalamin.
This again suggests that the small geometrical difference
between the furanose and cyclopentane ring has a large
influence on the transition state of the B12-dependent enzymic
reactions. No other signals from a (coupled) intermediate
adenosyl radical have been observed.
Figure 1. Dixon linearization obtained by the ™parallel∫ addition method
for determination of the Ki value of aristeromycylcobalamin with ethanol-
amine ammonia-lyase (S. typhimurium). Rate measurements were started
with synchronous addition of coenzyme B12 and the inhibitor to the assay
mixture. Ki of aristeromycylcobalamin calculated by this method is 16 Æ
2 nm.
In the methylmalonyl-CoA mutase reaction the value
of Ki for aristeromycylcobalamin is ꢁ14 times higher
than the value of KM for AdoCbl and this reveals a weaker
binding.
The analogue 2',3'-bis-epi-aristeromycylcobalamin (13) was
neither a coenzyme nor an inhibitor for any of the four
enzymes tested. This indicated that the correct steric position
of the ribose hydroxyl groups is required for the coenzyme to
bind.
EPR spectroscopy: Substrates, products and their analogues
convert the EPR silent adenosylcobalamin in adenosylcob-
alamin-dependent dehydratases and mutases into EPR-active
10, 16, 31, 32]
species at a catalytically competent rate.[8
Simula-
tion of the EPR spectra has provided evidence for the
presence of cob(ii)alamin that is coupled to a carbon-centred
Chem. Eur. J. 2003, 9, No. 3
¹ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/03/0903-0655 $ 20.00+.50/0
655

¬
J. Retey et al.
FULL PAPER
chromatography. The final enzyme concentration of 0.18 mgmLÀ1 (20 mL
in 50% glycerol) was calculated by the method of Warburg & Christian.
Conclusion
The specific activity of the enzyme was 24 UmgÀ1
.
Enantiomerically pure aristeromycin (8) and its [6'-13C]iso-
topomer have been synthesized in seven steps from the
commercially-available (1S,4R)-cis-acetoxycyclopent-2-ene-
1-ol (1). The stereoselectivity of key substitution steps was
increased with Pd0 catalysts and polymer-bound triphenyl-
phosphine to facilitate the purification procedures. To the best
of our knowledge, this is the shortest synthesis of aristeromy-
cin that introduces an isotopic label at the 6'-position.
Aristeromycin (8) and its 2',3'-bis-epimer (9) as well as their
[6'-13C]isotopomers have been coupled to cobalamin via their
6'-chloroderivatives 10 and 11. Both coenzyme B12 analogues
were assayed for their activity as cofactors of four coenzyme
B12-dependent enzymes. Aristeromycylcobalamin (12) had
44% and 38% of the activity of AdoCbl with dioldehydratase
and glyceroldehydratase, respectively, although its KM values
were similar to those of AdoCbl.
In the cases of the enzymes methylmalonyl-CoA mutase
and ethanolamine ammonia-lyase, aristeromycylcobalamin
(12) did not serve as a coenzyme, but was a weak inhibitor for
the former and a very strong inhibitor for the latter. EPR
measurements with glyceroldehydratase and racemic pro-
pane-1,2-diol as substrate revealed small but significant
changes of the doublet signal in comparison with those
obtained using AdoCbl. However no difference in the EPR
spectra was observed when unlabelled (12) and [6'-13C]aris-
teromycylcobalamin ([6'-13C](12)) was used as coenzyme and
this indicates that the putative intermediate 6'-aristeromycyl
radical has a too short lifetime to be detectable or that the 6'-
13C is too far away to broaden the signal.
Glyceroldehydratase: Host cells of E. coli HMS174 cells with a pQE30
vector from Qiagen which contained the gene for the glyceroldehydratase
from Citrobacter freundii. The enzyme had a His6-Tag introduced at the
N-terminus of the enzyme.[36] The cells were cultured and harvested
following as described above for those containing the ethanolamine
ammonia-lyase vector, but expression was induced with only 10 mm IPTG.
The enzyme was isolated and purified from
a cellfree extract by
immobilized metal affinity chromatography according to standard proto-
cols.[37] The enzyme solution was concentrated to 800 mL and the final
protein concentration was calculated from (1.55 À 0.76, A280 nm À A260 nm) to
be 21 mgmLÀ1. The preparation had a specific activity of 129 UmgÀ1 with
racemic 1,2-propanediol as substrate.
Dioldehydratase: The dioldehydratase gene from Salmonella typhimurium
was inserted into a pT7-7 vector and overexpressed in E.coli BL21. The
cells were cultured and harvested as described above for ethanolamine
ammonia-lyase. Dioldehydratase was isolated and purified from to the
protocol described in the literature.[38] After disruption of the cells by
sonication, the sediment was dissolved by the addition of the sodium salt of
cholic acid and the resulting supernatant was applied to a HPLC gel
filtration column (Superdex 200). The eluted enzyme was concentrated to
1 mL and had a concentration of 11 mgmLÀ1 The specific activity was
95 UmgÀ1 using racemic 1,2-propanediol as substrate.
Methylmalonyl-CoA mutase: Methylmalonyl-CoA mutase (EC 5.4.99.2),
methylmalonyl-CoA epimerase (EC 5.1.99.1) and methylmalonyl-CoA
transcarboxylase (EC 2.1.3.1) were isolated from P. shermanii according
to previously published methods.[39, 40] Succinyl-CoA was freshly pre-
pared.[29]
Kinetic measurements: Kinetic constants (KM, kcat) for adenosylcobalamin
and aristeromycylcobalamin (12) were determined by means of standard
linearization methods. 1 10 mL of 0.001 1 mm solutions of the cofactor
were used to obtain 7 10 data points. Linearization was by the method of
Lineweaver Burk, Hanes and Eadie Hofstee. From these three methods
the average values of kinetic constants were calculated.
The inhibition constants (Ki) for the reaction of aristeromycylcobalamin
with ethanolamine ammonia-lyase and methylmalonyl-CoA mutase were
determined by the ™parallel∫ method.[29] In this assay adenosylcobalamin
and aristeromycylcobalamin were added simultaneously to the mixture. At
three constant concentrations of adenosylcobalamin (0.5, 1 and 2.5 mm), the
inhibitor concentration was varied (5 7 data points). Linear regression was
performed by a Dixon plot, in which the aristeromycylcobalamin concen-
tration is rated to 1/v. This results in three lines whose point of intersection
when projected to the [aristeromycylcobalamin]-axis corresponds directly
to Ki.
Experimental Section
Materials and methods: Yeast alcohol dehydrogenase (ADH), l-malate/
NAD oxidoreductase (MDH) coenzyme A and b-NADH Li3 (NADH)
were obtained from Roche Diagnostics Mannheim. Coenzyme B12
,
hydroxycobalamin (vitamin B12a), racemic 1,2 propanediol, succinic anhy-
dride and ethanolamine hydrochloride were products of Fluka Chemie AG.
Compounds (1S, 4R)-cis-acetoxycyclopent-2-ene-1-ol (diastereomeric pu-
rity ꢀ98%) and [13C]nitromethane ([13C]enrichment ꢀ99%) were pur-
chased from Sigma-Aldrich. All other chemicals used are commercially
available and were of highest purity. All solvents were freshly distilled and
dried prior to use. For kinetic measurements coenzyme B12 and the
aristeromycylcobalamin analogues were dissolved in bidistilled water and
kept at 0 48C in the darkness. Under these conditions the solutions were
stable for months (>98% pure from analytical HPLC).
All following enzymatic assays were performed in the darkness and with
degassed buffers.
Ethanolamine ammonia-lyase: A coupled assay with yeast alcohol dehy-
drogenase (ADH) was employed from the method of Blackwell and
Turner[41] with some modifications. In a plastic cuvette, ethanolamine
hydrochloride (10 mm) in 50 mm K2HPO4/KH2PO4 buffer, pH 7.5, was
mixed with NADH (30 mm, 10 mL), ADH (5 U, 5 mL) and the appropriate
amount of ethanolamine ammonia-lyase (usually 2 3 mL of the prepared
solution). The mixture was incubated at 378C for 5 min and the enzymic
reaction was started by addition of coenzyme B12. The decrease of
absorbance at 340 nm was recorded for 5 min at 378C. The rate of
ethanolamine transformation to acetaldehyde was calculated from the
change of absorption at 340 nm, which is due to the consumption of NADH
in the coupled enzyme assay. Inhibition kinetics were performed with the
same assay using the ™parallel∫ method.[29]
Bacterial strains,cell cultivation and enzyme isolation
Ethanolamine ammonia-lyase: E. coli cells BL21 (DE3) were used as host.
The ethanolamine ammonia-lyase gene from Salmonella typhimurium was
inserted into a gene coding for a pT7 7 vector. The cells were cultivated at
308C for 24 h on Luria Bertani agar plates which contained ampicillin and
kanamycin. One colony was transferred into 50 mL Luria Bertani media
and incubated at 308C for 16 h. From this culture, a 10 mL portion was
added to 1 L Luria Bertani media and shaken at 378C for 2 4 h until the
A600 reached 1.0. The expression of the enzyme was induced by the addition
of IPTG (1 mm) and the cells were incubated at 378C for another 4 h. The
cells were harvested by centrifugation at 5000 Â g for 10 min and stored at
À728C until use.
Glyceroldehydratase and dioldehydratase: The ADH/b-NADH coupled
assay used was based on the method of Bachovchin et al.[42] with some
minor modifications. In a plastic cuvette, 0.1% racemic 1,2-propanediol in
20 mm K2HPO4/KH2PO4 buffer, pH 8.0, was mixed with NADH (30 mm,
10 mL), ADH (5 U, 5 mL) and the appropriate amount of glycerol- or diol-
dehydratase (usually 2 3 mL of a 1:100 dilution of the concentrated protein
solution). The mixture was incubated at 378C for 5 min and the enzymic
reaction was started by addition of coenzyme B12. The decrease of
absorbance at 340 nm was recorded for 5 min at 378C.
The enzyme was isolated from a cellfree extract as described by Rˆther[35]
and further purified by ammonium sulphate precipitation (with no loss of
enzyme activity), gel filtration chromatography and anion-exchange
656
¹ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/03/0903-0656 $ 20.00+.50/0
Chem. Eur. J. 2003, 9, No. 3

Aristeromycin Analogues
652 660
Methylmalonyl-CoA mutase: The enzyme assay was based on the method
of Zagalak et al.[40] with some minor changes. In a microcuvette, sodium
pyruvate (0.1 m, 70 mL), NADH (30 mm, 10 mL), MDH (30 UmLÀ1, 10 mL)
and succinyl-CoA (20 mL, 20 mmol) in Tris-HCl-buffer, pH 7.5 (0.1 m,
765 mL,) were mixed with epimerase (3 UmLÀ1, 20 mL), transcarboxylase
was done with the indicated solvent systems. HPLC-separations were done
with a Merck Hitachi L-6210 pump, an L-4000 UV detector, a D-2500
chromato-integrator, an analytical Nucleosil-10-C18 column (250 Â 4 mm)
and a preparative Nucleosil-7-C18 column (250 mm  1'') from Macherey &
Nagel.
(5 UmLÀ1 5 mL,) and mutase (4.5 UmLÀ1
, , 5 mL). The mixture was
Synthesis of [6'-13C]aristeromycin (8)
incubated at 308C for 5 min and the enzymic reaction was started by
addition of coenzyme B12. The decrease of absorbance at 340 nm was
recorded for 5 min at 308C. Inhibition kinetics were determined by the
™parallel∫ method.[29]
(1'R,4'S)-4'-Benzoyloxy-1'-(6-chloro-9H-purin-9-yl)cyclopent-2'-ene (2):
Polymer-bound triphenylphosphine (1.17 g, 3 mmol/g) and Pd2(dba)3 ¥
CHCl3 (364 mg, 0.35 mmol) was added to a deoxygenated mixture of
(1S,4R)-cis-acetoxycyclopent-2-ene-1-ol (1) (1 g, 7.03 mmol) and chloro-
purine (2.17 g, 14.06 mmol) in THF (30 mL). The mixture was stirred for
10 min until the colour changed from deep red to yellow. Triethylamine
(5 mL) was added and the mixture was heated under reflux at 578C for
15 h. The solution was cooled to room temperature and the polymer-bound
triphenylphosphine was filtered off. Benzoylchloride (2.5 mL, 21.5 mmol)
was added and the mixture was stirred at room temperature for 17 h. It was
then concentrated and diluted with diethyl ether. The organic layer was
washed with water, dried (with MgSO4) and concentrated. The residue was
purified by flash chromatography (ethyl acetate/hexane 1:1) to give 2 as a
white foam (1.44 g, 60%). Rf ˆ 0.26; 1H NMR (500 MHz, CDCl3, 258C):
d ˆ 8.77 (s, 1H, (2)-CH), 8.27 (s, 1H, (8)-CH), 8.00 7.98 (m, 2H, phenyl-
H), 7.57 7.44 (m, 3H, phenyl-H), 6.55 6.53 (m, 1H, (3')-CH), 6.28 (dd,
1H, J1 ˆ 6 Hz, J2 ˆ 2 Hz, (2')-CH), 6.03 (dd, 1H, J1 ˆ 5 Hz, J2 ˆ 2 Hz, (4')-
CH), 5.86 5.84 (m, 1H, (1')-CH), 3.14 (dt, 1H, J1 ˆ 16 Hz, J2 ˆ 8 Hz, (5')-
CH2), 2.13 (dt, J1 ˆ 15 Hz, J2 ˆ 3 Hz, (5')-CH2); 13C NMR (125 MHz,
CDCl3, 258C): d ˆ 165.90 (phenylester-C), 152.06 ((2)-CH), 151.34 ((6)-C),
151.14 ((4)-C), 143.41 ((5)-C), 136.68 ((8)-CH), 133.60 ((3')-CH), 133.44
((2')-CH), 130.04 (phenyl-CH), 129.55 (2 Â phenyl-CH), 129.43 (phenyl-
CH), 128.60 (2 Â phenyl-CH), 77.34 ((4')-CH), 57.53 ((1')-CH), 38.67 ((5')-
EPR spectroscopy: The kinetically-competent EPR-detectable ™doublet∫
intermediate[31, 32] was generated by the incubation of Citrobacter freundii
glyceroldehydratase with racemic 1,2-propanediol. Trapping of this inter-
mediate was performed by manually freezing reaction mixtures in ethanol
which was cooled close to its melting point with liquid nitrogen. This
required skill but produced similar quantities of the EPR-active species[32]
and avoided the problems of trapped, solid oxygen, complicated storage
and loss of sensitivity due to interstitials in the snow produced by rapid-
freeze.
EPR tubes that were preloaded with buffer, glyceroldehydratase and
adenosylcobalamin, [6'-13C]-labelled or unlabelled aristeromycylcobalamin
were cooled on ice. Generation of the doublet species was initiated by the
addition of an ice-cold solution of racemic 1,2-propanediol. Transfer and
mixing was done with a Gilson pipette with a 15 cm piece of Tefzel tubing
(Æ 0.75 mm) linked to a yellow tip with silicon tubing (Æ 1 mm), which
were all precooled in ice-cold buffer. The aerobic reaction mixtures had a
final volume of 250 mL, and had final concentrations of 0.4 m racemic 1,2-
propanediol, 5 mgmLÀ1 glyceroldehydratase, 4 mm adenosylcobalamin,
and [6'-13C]-labelled or unlabelled aristeromycylcobalamin in 20 mm
K2HPO4/KH2PO4 buffer, pH 8.0. To avoid depletion of 1,2-propanediol
and the generation of detrimental amounts of propionaldehyde, the mixing
was performed as rapidly as possible: fastest mixing, transfer into the liquid
nitrogen-cooled viscous ethanol and freezing took between 10 15 s in
total. Control experiments with freezing after 25 45 s showed >3 and >10
fold reduction in the intensity of the doublet signal. The experiment was
carried out under dim red light to avoid formation of hydroxycobalamin by
photolysis.
‡
CH2); MS (EI, 70 eV, 1408C): m/z (%): 341 (6) [M ], 219 (22), 218 (7), 194
(9), 192 (38), 105 (100), 77 (22), 65 (9); HRMS (EI): m/z: calcd for
‡
C17H13Cl1N4O2: 340.7636 [M ]; found: 340.7625; IR (CDCl3): nÄ ˆ 3411,
3068, 2987, 1717, 1592, 1565, 1487, 1454, 1436, 1402, 1369, 1334, 1318, 1296,
1271, 1215, 1170, 1148, 1132, 1100, 1070, 1045, 1026, 1001, 952, 926, 883, 863,
855, 836, 800, 790, 749, 720, 686, 670, 651, 637 cmÀ1
.
Phenylsulphonyl[13C]nitromethane (3): [13C]Nitromethane (250 mg,
4.1 mmol) in DMF (5 mL) was cooled to 08C. Then 1,8-diazabicy-
clo[5.4.0]undec-7-ene (620 mL, 4.10 mmol) was added and the reaction
mixture was stirred for 10 min. After addition of sodium phenylsulphinate
(560 mg, 3.42 mmol) and iodine (786 mg, 3.11 mmol), the solution was
allowed to warm to room temperature and stirred for 1 h. A saturated
solution of sodium sulphite was added to the solution until the colour of the
mixture turned to bright yellow to remove the iodine. The pH was adjusted
to 1 with concentrated HCl. Then the aqueous layer was extracted three
times with diethyl ether and the organic layer was dried (MgSO4),
concentrated and purified by flash chromatography (ethyl acetate/hexane
Samples were stored in the dark in liquid nitrogen. No significant loss of
signal intensity occurred upon storage for several months. The EPR
measurements were performed with a Bruker EMX-6/1 X-band EPR
Spectrometer with a built-in ER-041-1161 microwave frequency counter,
EMX-1101 magnet and power supply composed of an ER-070 6 inch
magnet with 60 mm air gap, an EMX-080 1 kW magnet power supply, an
EMX-032T Hall Field probe with electronics to interface with the magnet
and an ER-4102 standard Universal TE102 rectangular cavity with a
typical resonance of ꢁ9.431 GHz. Initial experiments were performed
using a liquid nitrogen finger Dewar. The spectra presented in this paper
were recorded at 77 K using the ESR-900 Oxford Instruments variable
temperature Helium flow cryostat. Data acquisition was with the software
supplied by Bruker (WINEPR Acquisition program, May 1, 1997, version
2.3.1.), data manipulation (determination of g values, subtraction, baseline
correction and conversion to ASCII files for use with Microsoft EXCEL)
was done with the WINEPR program version 2.11. EPR tubes were 4.7 Æ
0.2 mm outer diameter (0.45 Æ 0.05 mm wall thickness) and 13 cm length
Ilmasil-PN high purity quartz tubes obtained from Quarzschmelze Ilmenau
GmbH (Langewiesen, Germany).
1
1:3) to give white crystals (330 mg, 40%). M.p. 788C; Rf ˆ 0.18; H NMR
(500 MHz, CDCl3, 258C): d ˆ 7.93 7.87 (m, 2H, phenyl-H), 7.74 7.70 (m,
1H, phenyl-H), 7.62 7.55 (m, 2H, phenyl-H), 5.53 (d, 2H, J ˆ 156 Hz,
PhSO2-13CH2-NO2); 13C NMR (125 MHz, CDCl3, 258C): d ˆ 135.71 (d, J ˆ
9 Hz, quart. C), 135.54 (phenyl-CH), 129.70 (2 Â phenyl-CH), 129.29 (2 Â
phenyl-CH), 90.29 (13C label); MS (EI, 70 eV, 458C): m/z (%): 202 (20)
‡
[M ], 158 (8), 141 (56), 94 (6), 78 (5), 77 (100), 51 (16); HRMS (EI): m/z:
‡
calcd for 13C1C6H7N1O4S1: 202.2008 [M ]; found: 202.2012; IR (CDCl3):
nÄ ˆ 3064, 3019, 2951, 2732, 2581, 2472, 2240, 1972, 1910, 1821, 1551, 1449,
1350, 1325, 1210, 1154, 1085, 1024, 998, 933, 918, 742, 698, 682 cmÀ1
.
(1'R,4'S)-1'-(6-Chloro-9H-purin-9-yl)-4'-[nitro(phenylsulphonyl)-[13C]-
methyl]cyclopent-2'-ene (4): Polymer-bound triphenylphosphine (227 mg,
3 mmol/g) and Pd2(dba)3 ¥ CHCl3 (70 mg, 0.068 mmol) were added to a
deoxygenated solution of the benzoate 2 (464 mg, 1.36 mmol) and phenyl-
sulphonyl[13C]nitromethane (3) (330 mg, 1.63 mmol) in THF (20 mL). The
mixture was stirred for 10 min until the colour changed from deep-red to
yellow. Then triethylamine (1.5 mL) was added and the mixture was stirred
at room temperature for 15 h. The polymer-bound triphenylphosphine was
filtered off and the solution was concentrated. The residue was diluted in
saturated aqueous ammonium chloride. The aqueous layer was extracted
three times with chloroform, then the organic layer was dried (MgSO4) and
concentrated. The residue was purified by flash chromatography (ethyl
acetate/hexane 3:1) to give 4 as a white foam (503 mg, 88%). Rf ˆ 0.34;
1H NMR (500 MHz, CDCl3, 258C): d ˆ 8.68 (s, 2H, (2)-CH), 8.12 (s, 2H,
(8)-CH), 7.93 7.60 (m, 10H, phenyl-H), 6.68 6.26 (dm, 2H, J ˆ 145 Hz,
Synthetic part
General methods: Melting points were recorded on a B¸chi 535 melting
point apparatus and are uncorrected. H and 13C NMR spectra were both
1
recorded at 500 MHz and 125 MHz, respectively, with a Bruker DRX 500.
Spin multiplicities are indicated by the symbols s (singlet), d (doublet), dd
(doublet of doublet), t (triplet), dt (doublet of triplet), m (multiplet) and
dm (doublet of multiplet). IR spectra were recorded with a Bruker IFS 88
FT-IR spectrometer using the DRIFT-method (Diffuse Reflectance Infra-
red Fourier Transform). EI-MS (electron impact) was performed with a
Finnigan-MAT-90 high-resolution instrument at 70 eV. Reactions were
monitored by TLC (thin layer chromatography) using silica gel precoated
aluminium plates G254 from Macherey & Nagel. All substances could be
visualized by irradiation with a UV lamp. Column flash chromatography
was performed on silica gel 60 (40 63 mm grade) by Merck, and elution
Chem. Eur. J. 2003, 9, No. 3
¹ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/03/0903-0657 $ 20.00+.50/0
657

¬
J. Retey et al.
FULL PAPER
(6')-13CH), 6.54 6.52 (m, 1H, (2')-CH), 6.03 5.99 (m, 2H, (3')-CH, (2')-
CH), 5.90 5.88 (m, 1H, (3')-CH), 5.80 5.75 (m, 1H, (1')-CH), 5.65 5.62
(m, 1H, (1')-CH), 4.01 3.82 (m, 2H, (4')-CH), 3.22 3.13 (m, 1H, (5')-
CH2), 3.05 2.95 (m, 1H, (5')-CH2), 2.47 2.40 (m, 1H, (5')-CH2), 2.15
2.09 (m, 1H, (5')-CH2); 13C NMR (125 MHz, CDCl3, 258C): d ˆ 151.72
(2 Â (6)-C), 151.12 (2 Â (4)-C), 143.83 (2 Â (2)-C), 135.82 (2 Â (8)-C), 134.37
((2')-CH), 134.32 ((2')-CH), 134.17 (2 Â (5)-C), 132.34 ((3')-CH), 132.21
((3')-CH), 132.09 (2 Â phenyl-C), 129.78 (2 Â phenyl-CH), 129.76 (phenyl-
CH, 4 x), 129.74 (4 Â phenyl-CH), 102.80 (13C label), 61.13 ((1')-CH), 59.95
((1')-CH), 44.17 (d, J ˆ 34 Hz, (4')-CH), 43.84 (d, J ˆ 34 Hz, (4')-CH), 33.93
((5')-CH2), 33.64 ((5')-CH2); MS: (FAB, DMSO/glycerol): m/z (%): 421 (5)
white solid (64 mg, 87%). Rf ˆ 0.38; 1H NMR (500 MHz, [D6]DMSO,
258C): d ˆ 8.21 (s, 1H, (2)-CH), 8.12 (s, 1H, (8)-CH), 7.29 (s, 2H, -NH2),
6.21 6.19 (m, 1H, (1')-CH), 5.98 5.94 (m, 1H, (2')-CH), 5.71 5.65 (m,
1H, (3')-CH), 4.95 (s, 1H, -OH), 3.55 3.82 (dm, 2H, J ˆ 134 Hz, (6')-
13CH2), 3.03 2.96 (m, 1H, (5')-CH2), 2.79 2.70 (m, 1H, (4')-CH), 1.81
1.75 (m, 1H, (5')-CH2); 13C NMR (125 MHz, [D6]DMSO, 258C): d ˆ 150.87
((4)-C), 150.12 ((6)-C), 148.24 ((2)-CH), 142.98 ((8)-CH), 139.45 ((2'-CH),
131.35 ((3'-CH), 130.09 ((5)-C), 63.95 (13CH2-label), 60.95 ((1')-CH), 47.55
((4')-CH), 34.76 ((5')-CH2; MS (EI, 70 eV, 2108C): m/z (%): 233 (11)
‡
[M ‡H], 140 (9), 135 (51), 136 (100), 99 (22), 43 (52); HRMS (EI): m/z:
‡
calcd for 13C1C10H13N5O1: 232.2540 [M ]; found: 232.2534; IR (DMSO):
‡
[M ], 405 (3), 386 (10), 298 (15), 185 (100), 93 (90); HRMS (EI): m/z: calcd
nÄ ˆ 3320, 2905, 1650, 1601, 1478, 1414, 1298, 1252, 1221, 1109, 915, 796,
‡
for 13C1C16H14Cl1N5O4S1: 420.8431 [M ]; found: 420.8439; IR (CDCl3): nÄ ˆ
678 cmÀ1
.
(1'R,2'S,3'R,4'S)-1'-(6-Amino-9H-purin-9-yl)-4'-hydroxy[13C]methylcyclo-
pentane-2',3'-diol; 6'-[13C]aristeromycin (8) and (1'R,2'R,3'S,4'S)-1'-(6-
amino-9H-purin-9-yl)-4'-hydroxy[13C]methylcyclopentane-2',3'-diol; 2',3'-
bis-epi-6'-[13C]aristeromycin (9): Methylmorpholine-N-oxide (51 mg,
0.44 mmol) was added to a solution of the unsaturated alcohol 7 (50 mg,
0.22 mmol) in water (2 mL) and THF (4 mL). After the mixture was cooled
to 08C, a catalytic amount of solid osmium tetroxide (10 mg, 0.04 mmol)
was added and the solution was stirred for 24 h at 08C. The reaction
mixture was quenched by the dropwise addition of saturated aqueous
sodium sulphite. After warming to room temperature and stirring for an
additional 30 min, the brown solution was directly applied to a silica gel
column and purified by flash chromatography (ethyl acetate/methanol 3:1)
to give 72 mg of a mixture of 8 and 9 as a white solid (Rf ˆ 0.30). The crude
product was still contaminated with NMO and it was nearly impossible to
separate the two diastereomers by simple flash chromatography. So the two
products were further purified by preparative HPLC (l 280 nm; n ˆ
4 mLminÀ1; eluents: H2O (A) and MeOH (B), using a linear gradient of
3412, 3271, 3062, 3032, 2926, 1969, 1902, 1819, 1618, 1579, 1558, 1495, 1478,
1448, 1438, 1406, 1335, 1300, 1247, 1221, 1179, 1157, 1119, 1082, 1028, 998,
971, 911, 874, 842, 796, 754, 721, 696, 686, 648 cmÀ1
.
(1'R,4'S)-1'-(6-Chloro-9H-purin-9-yl)-4'-methoxy[13C]carbonylcyclopent-
2'-ene (5): Nitrosulphonate 4 (400 mg, 0.95 mmol) was diluted in deoxy-
genated methanol (15 mL). Anhydrous sodium carbonate (1 g, 9.5 mmol)
was added and the mixture was stirred vigorously for 15 min at room
temperature. A solution of tetrabutylammonium oxide (4.64 g, 2.85 mmol)
in dichloromethane (10 mL) was added slowly to the reaction mixture and
then stirred at room temperature for another 15 h. The solution was
concentrated and the residue was taken up in saturated aqueous
ammonium chloride. The aqueous layer was extracted five times with
diethyl ether and the combined organic layers were dried (MgSO4) and
concentrated. The residue was purified by flash chromatography (ethyl
acetate/hexane 9:1) to give 5 as white crystals (191 mg, 72%). Rf ˆ 0.41;
1H NMR (500 MHz, CDCl3, 258C): d ˆ 8.72 (s, 1H, (2)-CH), 8.24 (s, 1H,
(8)-CH), 6.32 6.30 (m, 1H, (2')-CH), 6.05 6.01 (m, 1H, (3')-CH), 5.91
5.85 (m, 1H, (1')-CH), 3.71 (s, 3H, O-CH3), 3.80 3.69 (m, 1H, (4')-CH),
2.98 2.85 (m, 1H, (5')-CH2), 2.35 2.26 (m, 1H, (5')-CH2). 13C NMR
(125 MHz, CDCl3, 258C): d ˆ 173.08 (13C label), 151.79 ((6)-C), 151.43 ((4)-
C), 150.93 ((2)-CH), 143.92 ((8)-CH), 136.06 ((2')-CH), 131.75 ((3')-CH),
130.44 ((5)-C), 59.33 ((1')-CH), 52.49 (O-CH3), 49.43 (d, J ˆ 58 Hz, (4')-
5
a
20% Bwithin 120 min). [6 '-13C]Aristeromycin (8) eluted after 70 min as
white solid (28 mg, 0.11 mmol, 49%). M.p. 2378C. 2',3'-Bis-epi-6'-
[13C]aristeromycin (9) eluted after 82 min as
0.08 mmol, 38%). M.p. 2268C.
a white solid (22 mg,
6'-[13C]Aristeromycin (8): 1H NMR (500 MHz, [D6]DMSO, 258C): d ˆ 8.17
(s, 1H, (2)-CH), 8.10 (s, 1H, (8)-CH), 7.13 (s, 2H, NH2), 4.89 (d, 1H, J ˆ
7 Hz, -OH), 4.71 4.64 (m, 3H, (1')-CH, 2 Â -OH), 4.34 4.30 (m, 1H, (2')-
CH), 3.82 (dd, 1H, J1 ˆ 8 Hz, J2 ˆ 4 Hz, (3')-CH), 3.64 3.27 (dm, 2H, J ˆ
140 Hz, (6')-13CH2), 2.24 2.18 (m, 1H, (5')-CH2), 2.07 2.02 (m, 1H, (4')-
CH2), 1.75 1.68 (m, 1H, (5')-CH2). 13C NMR (125 MHz, [D6]DMSO,
258C): d ˆ 155.90 ((6)-C), 152.00 ((2)-CH), 149.69 ((4)-C), 140.00 ((8)-CH),
119.30 ((5)-C), 74.53 ((2')-CH), 71.64 ((3')-CH), 63.00 ((6')-13CH2), 59.26
((1')-CH), 45.31 (d, J ˆ 38 Hz, (4')-CH), 29.25 ((5')-CH2); MS (EI, 70 eV,
‡
CH), 34.24 ((5')-CH2); MS (EI, 70 eV, 1058C): m/z (%): 280 (12) [M ], 279
(44), 247 (10), 219 (33), 194 (10), 192 (37), 157 (28), 155 (100), 125 (20), 124
(29), 93 (28), 66 (45), 65 (48); HRMS (EI): m/z: calcd for
‡
13C1C11H11Cl1N4O2: 279.6942 [M ]; found: 279.6951; IR (CDCl3): nÄ ˆ
3433, 3113, 3057, 2992, 2950, 1852, 1727, 1591, 1559, 1491, 1436, 1424,
1402, 1337, 1316, 1253, 1200, 1149, 1134, 1119, 1066, 1003, 958, 928, 911, 858,
791, 769, 741, 686, 652, 637, 627 cmÀ1
.
(1'R,4'S)-1'-(6-Chloro-9H-purin-9-yl)-4'-hydroxy[13C]methylcyclopent-2'-
ene (6): The methylester 5 (150 mg, 0.54 mmol) in deoxygenated THF
(15 mL) was cooled in dry ice/acetone bath to À788C. Superhydride
(LiHB(Et)3) (1.62 mL of a 1m solution) was added dropwise and very
slowly over 30 min as the reaction mixture was stirred vigorously. After
stirring for 3 h at À788C, the reaction was quenched with ethyl acetate
(10 mL) and saturated aqueous ammonium chloride (15 mL) and warmed
to room temperature. The aqueous phase was extracted three times with
ethyl acetate. The combined organic layers were washed with water, dried
(MgSO4) and concentrated. The residue was purified by flash chromatog-
raphy (ethyl acetate/methanol 24:1) to give 6 as a yellow gum (97 mg,
‡
1508C): m/z (%): 266 (4) [M ], 162 (25), 137 (17), 136 (100), 135 (47), 108
‡
(7), 43 (7); HRMS (EI): m/z: calcd for 13C1C10H15N5O3: 266.2687 [M ];
found: 266.2690; IR ([D6]DMSO): nÄ ˆ 3327, 2871, 1650, 1599, 1478, 1417,
1331, 1252, 1212, 1109, 913, 795, 723, 645 cmÀ1
.
2',3'-Bis-epi-[13C]aristeromycin (9):1H NMR (500 MHz, [D6]DMSO, 258C):
d ˆ 8.15 (s, 1H, (2)-CH), 8.10 (s, 1H, (8)-CH), 7.12 (s, 2H, NH2), 5.07 (d,
1H, J ˆ 6 Hz, -OH), 4.99 4.89 (m, 2H, (1')-CH, -OH), 4.50 (dd, 1H, J1 ˆ
8 Hz, J2 ˆ 4 Hz, (3')-CH), 4.10 4.02 (m, 2H, (2')-CH, -OH), 3.78 3.30
(dm, 2H, J ˆ 180 Hz, (6')-13CH2), 2.26 2.18 (m, 1H, (5')-CH2), 2.11 2.02
(m, 1H, (4')-CH), 1.90 1.80 (m, 1H, (5')-CH2); 13C NMR (125 MHz,
[D6]DMSO, 258C): d ˆ 156.30 ((6)-C), 152.35 ((2)-CH), 150.26 ((4)-C),
141.82 ((8)-CH), 118.75 ((5)-C), 74.24 ((2')-CH), 72.51 ((3')-CH), 60.99
((6')-13CH2), 53.59 ((1')-CH), 40.44 (d, J ˆ 35 Hz, (4')-CH), 33.25 ((5')-
1
72%). Rf ˆ 0.38; H NMR (500 MHz, MeOD, 258C): d ˆ 8.70 (s, 1H, (2)-
CH), 8.44 (s, 1H, (8)-CH), 6.23 6.17 (m, 1H, (1')-CH), 5.89 5.82 (m, 1H,
(2')-CH), 5.80 5.71 (m, 1H, (3')-CH), 3.65 3.87 (dm, 2H, J ˆ 122 Hz, (6')-
13CH2), 3.18 3.02 (m, 1H, (5')-CH2), 2.90 2.78 (m, 1H, (4')-CH), 1.98
1.85 (m, 1H, (5')-CH2). 13C NMR (125 MHz, MeOD, 258C): d ˆ 151.62 ((4)-
C), 151.59 ((6)-C), 150.66 ((2)-CH), 144.85 ((8)-CH), 139.91 ((2'-CH),
131.61 ((3'-CH), 129.09 ((5)-C), 64.18 (13CH2 label), 60.66 ((1')-CH), 47.59
((4')-CH), 33.86 ((5')-CH2); MS (EI, 70 eV, 2208C): m/z (%): 252 (10)
‡
CH2); MS (EI, 70 eV, 1508C): m/z (%): 266 (25) [M ], 234 (6), 190 (8), 178
(9), 162 (55), 148 (6), 136 (100), 135 (57), 108 (14), 58 (14), 43 (66); HRMS
‡
(EI): m/z: calcd for 13C1C10H15N5O3: 266.2687 [M ]; found: 266.2689; IR
([D6]DMSO): nÄ ˆ 3328, 3215, 1648, 1600, 1485, 1402, 1322, 1252, 1212, 1101,
901, 788, 723, 640 cmÀ1
.
‡
[M ], 151 (23), 98 (100), 71 (19), 70 (15), 66 (34), 45 (12); HRMS (EI): m/z:
(1'R,2'S,3'R,4'S)-1'-(6-Amino-9H-purin-9-yl)-4'-chloro[13C]methylcyclo-
pentane-2',3'-diol,6 '-chloro-6'-deoxy-6'-[13C]aristeromycin (10) and
(1'R,2'R,3'S,4'S)-1'-(6-amino-9H-purin-9-yl)-4'-chloro[13C]methylcyclo-
‡
calcd for 13C1C10H11Cl1N4O1: 251.6841 [M ]; found: 251.6838; IR (MeOD):
nÄ ˆ 3265, 3059, 2960, 1900, 1692, 1607, 1591, 1531, 1495, 1476, 1450, 1404,
1346, 1328, 1300, 1226, 1196, 1160, 1119, 1064, 1028, 998, 972, 911, 858, 794,
pentane-2',3'-diol,2
',3'-bis-epi-6'-chloro-6'-deoxy-[6'-13C]aristeromycin
721, 696, 632 cmÀ1
.
(11): [6'-13C]Aristeromycin (8) (15 mg, 0.056 mmol) was dissolved in
hexamethylphosphoramide (5 mL) and thionyl chloride (55 mL,
0.28 mmol) was added to the solution. The mixture was stirred at room
temperature for 15 h and then the reaction mixture was quenched with
water (15 mL). The product was purified by ion-exchange chromatography.
The aqueous solution was directly applied to a column of Dowex 50 WX8
(1'R,4'S)-1'-(6-Amino-9H-purin-9-yl)-4'-hydroxy[13C]methylcyclopent-2'-
ene (7): The alcohol 6 (80 mg, 0.32 mmol) was dissolved in a mixture of
THF (3 mL) and concentrated aqueous ammonia (6 mL). The flask was
sealed and stirred for 72 h. The mixture was concentrated and directly
purified by flash chromatography (ethyl acetate/methanol 3:1) to give 7 as a
658
¹ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/03/0903-0658 $ 20.00+.50/0
Chem. Eur. J. 2003, 9, No. 3

Aristeromycin Analogues
652 660
‡
‡
(H form) (10 mL). The column was washed with water (100 mL) and the
[M À aristeromycyl], 1183 (16), 1069 (12), 1053 (21), 971 (25), 843 (26),
‡
product was eluted with 1n aqueous ammonia (100 mL). After evaporating
most of the solvent, the compound crystallized directly from the
concentrated solution. The precipitation was separated and purified by
repeated steps of centrifugation and washing with small amounts of water.
6'-Chloro-6'-deoxy-[6'-13C]aristeromycin (10) was obtained as a white solid
(13.4 mg, 0.047 mmol, 84%).
207 (10), 94 (100); calcd for 13C1C72H102Co1N18O16P1: 1578.6097 [M ].
Acknowledgement
We thank Drs. B. M. Babior[30] and D. Rˆther[35] for the gift of the plasmid
which overexpresses ethanolamine ammonia-lyase from Salmonella typhi-
murium, Prof. G. Gotschalk and Dr. R. Daniel (University of Gˆttingen) as
well as Drs. T. Graf[36] and R. Nitsche[38] for plasmids, which overexpress
glyceroldehydratase from Citrobacter freundii and Dr. T. Bobik (University
of Florida) for a plasmid, which overexpresses dioldehydratase from
Salmonella typhimurium. We would also like to thank Mr. Christian Weber
for isolating methylmalonyl-CoA mutase from Propionibacterium sherma-
nii. Financial support from the Deutsche Forschungsgemeinschaft, the
European Union and the Fonds der Chemischen Industrie is gratefully
acknowledged.
The whole procedure was repeated with 2',3'-bis-epi-[6'-13C]aristeromycin
(9) to obtain 2',3'-bis-epi-6'-chloro-6'-deoxy-[6'-13C]aristeromycin (11) in
the same yield. The crude white products were used in the next reaction
without further purification
[6'-13C]Aristeromycylcobalamin (12) and its bis-epi isomer (13): Vitamin
B12a (hydroxycobalamin hydrochloride, 54 mg, 0.04 mmol) and a catalyic
amount of cobalt(ii) chloride (2 mg) were dissolved in deoxygenated water
(4 mL). A solution of sodium borohydride (15 mg, 0.40 mmol) in deoxy-
genated water (1 mL) was added under argon and the resulting solution
was stirred for 30 min. The deep-red colour of the mixture changed to
brown and finally to greenish-grey. 6'-Chloro-6'-deoxy-6'-[13C]aristeromy-
cin (10; 11 mg, 0.04 mmol) in deoxygenated methanol (3 mL) was added
through a syringe with exclusion of oxygen. The flask was kept dark and the
mixture stirred vigorously at room temperature for 2 h. The solid
precipitate (cobalt oxide) was separated from the mixture in the dark by
centrifugation. The remaining deep-red solution was then directly purified
by HPLC (l 280 nm; n ˆ 4 mLminÀ1; eluents: CF3COOH/H2O (A) and
MeOH (B), with a linear gradient of 20 80% B within 60 min). [6'-
13C]aristeromycylcobalamin (12) eluted after 30 min as a yellow solution
(the yellow colour was from the product in acidified solvent). Evaporation
in the dark and drying in high vacuum yielded bright-red crystals (38 mg,
0.024 mmol, 60%).
[1] B. Zagalak, W. Friedrich,Vitamin B12, Walter de Gruyter, 1979, Berlin,
New York.
[2] B. T. Golding, Mechanism of action of the B12 coenzyme: theory and
models in B12, Vol. 1 (Ed.: D. Dolphin), 1982, Wiley, New York,
pp. 543 582.
[3] B. M. Babior, Biofactors 1988, 1, 21 26.
[4] J. Retey in Chemical aspects of enzyme biotechnology (Ed.: T. O.
Baldwin), Plenum Press, New York, 1990, pp. 223 234.
[5] P. G. Lehnert, D. C. Hodgkin, Nature 1962, 192, 937 940.
[6] B. P. Hay, R. G. Finke, J. Am. Chem. Soc. 1987, 109, 8012 8018.
[7] S. Chowdhury, R. Banerjee, Biochemistry 2000, 39, 7998 8006.
¬
This procedure was repeated with 2',3'-bis-epi-6'-chloro-6'-deoxy-[6'-
13C]aristeromycin (11) to obtain 2',3'-bis-epi-[6'-13C]aristeromycylcobala-
min (13). This compound eluted after 35 min also as a yellow solution.
Evaporation in the dark and drying in high vacuum yielded a bright-red
solid ((34 mg, 0.022 mmol, 55%).
¬
[8] Y. Zhao, J. Such, J. Retey, Angew. Chem. 1992, 31, 215 221; Angew.
Chem. Int. Ed. Engl. 1992, 31, 215 216.
¬
[9] Y. Zhao, A. Abend, M. Kunz, P. Such, J. Retey, Eur. J. Biochem. 1994,
225, 891 896.
¬
[10] A. Abend, V. Illich, J. Retey, Eur. J. Biochem. 1997, 249, 180 186.
[6'-13C]Aristeromycylcobalamin (12): 1H NMR (500 MHz, D2O, 25 8C, only
annotated data from Scheme 3 is presented: d ˆ 8.19 (s, 1H, A2), 8.00 (s,
1H, A8), 7.19 (s, 1H, B7), 6.98 (s, 1H, B2), 6.26 (s, 1H, R1), 6.22 (s, 1H,
B4), 5.95 (s, 1H, C10), 4.82 (m, 1H, A11), 4.78 (m, 1H, R3), 4.50 (m, 1H,
A12), 4.35 (m, 1H, L2), 4.25 (m, 1H, R2), 4.12 (m, 1H, R4), 3.90 (m, 1H,
R5), 3.75 (m, 1H, R5), 3.58 (m, 1H, L1), 3.45 (m, 1H, A13), 3.20 (m, 1H,
L1), 2.22 (s, 3H, B10), 2.21 (s, 3H, B11), 2.12 (m, 1H, A16), 1.85 (m, 1H,
A14), 1.82 (m, 1H, A15), 1.80 (m, 1H, A16), 1.21 (m, 3H, L3), 0.80 (m,
1H, A15); 13C NMR (125 MHz, D2O, 25 8C, only annotated data from
Scheme 3 is presented: d ˆ 155.4 (A6), 152.1 (A2), 148.8 (A4), 141.4 (B2),
140.5 (A8), 137.9 (B9), 133.6 (B6), 131.3 (B5), 130.2 (B8), 118.9 (A5),
118.4 (B4), 110.3 (B7), 94.7 (C10), 86.2 (R1), 76.2 (A11), 73.4 (A13), 72.8
(R3), 71.5 (L2), 60.0 (A12), 46.3 (A16), 43.9 (L1), 31.4 (13C-A15), 30.2 (d,
J ˆ 31 Hz, A14), 18.7 (B10), 18.5 (B11), 18.1 (L3); characteristic data from
the corrin moiety and its side chains is comparable with those of
adenosylcobalamin as previously assigned;[43] MS (FAB, glycerol): m/z
[11] P. A. Frey, R. H. Abeles, J. Biol. Chem. 1966, 241, 2732 2733.
[12] A. Gaudemer, J. Zylber, N. Zylber, M. Baran-Marszac, W. E. Hull, M.
¬
Fountoulakis, A. Kˆnig, K. Wˆlfle, J. Retey, Eur. J. Biochem. 1981,
119, 279 285.
[13] B. M. Babior, T. H. Moss, D. C. Gould,J. Biol. Chem. 1972, 247, 4389
4392.
[14] B. M. Babior, T. H. Moss, W. H. Orme-Johnson, H. Beinert, J. Biol.
Chem. 1974, 249, 4537 4544.
[15] H. Bothe, D. J. Darley, S. P. Albracht, G. J. Gerfen, B. T. Golding, W.
Buckel, Biochemistry 1998, 37, 4105 4113.
[16] U. Leutbecher, S. P. Albracht, W. Buckel, FEBS Letters 1992, 307,
144 146.
[17] O. T. Magnusson, P. A. Frey, J. Am. Chem. Soc. 2000, 122, 8807 8813.
[18] S. S. Kerwar, T. A. Smith, R. H. Abeles, J. Biol. Chem. 1970, 245,
1169 1174.
[19] B. M. Trost, R. Madsen, S. D. Guile, B. Brown,J. Am. Chem. Soc. 2000,
122, 5947 5956.
[20] P. A. Wade, H. R. Hinney, N. V. Amin, P. D. Vail, S. D. Morrow, S. A.
Hardinger, M. S. Saft, J. Org. Chem. 1981, 46, 765 770.
[21] T. Kusaka, H. Yamamoto, M. Shibata, M. Muroi, T. Kishi, K. Mizuno,
J. Antiobiot. 1968, 21, 255 266.
[22] S. Yaginuma, N. Muto, M. Tsujino, Y. Sudate, M. Hayashi, M. Otani, J.
Antibiot. 1981, 34, 359 366.
‡
(%): 1578 (9) [M ], 1561 (11), 1512 (10), 1478 (12), 1421 (15), 1330 (80)
‡
[M À aristeromycyl-part], 1069 (14), 971 (22), 916 (17), 829 (15), 94 (100);
‡
calcd for 13C1C72H102Co1N18O16P1: 1578.6097 [M ].
2',3'-Bis-epi-[6'-13C]aristeromycylcobalamin (13): 1H NMR (500 MHz,
D2O, 25 8C, only annotated data from Scheme 3 is presented): d ˆ 8.19 (s,
1H, A2), 8.00 (s, 1H, A8), 7.19 (s, 1H, B7), 7.00 (s, 1H, B2), 6.29 (s, 1H,
R1), 6.27 (s, 1H, B4), 6.02 (s, 1H, C10), 4.82 (m, 1H, A11), 4.75 (m, 1H,
R3), 4.37 (m, 1H, A12), 4.35 (m, 1H, L2), 4.27 (m, 1H, R2), 4.15 (m, 1H,
R4), 3.92 (m, 1H, R5), 3.78 (m, 1H, R5), 3.58 (m, 1H, L1), 3.29 (m, 1H,
A13), 3.20 (m, 1H, L1), 2.25 (s, 3H, B10), 2.24 (s, 3H, B11), 2.15 (m, 1H,
A16), 1.92 (m, 1H, A14), 1.87 (m, 1H, A15), 1.82 (m, 1H, A16), 1.25 (m,
3H, L3), 0.85 (m, 1H, A15); 13C NMR (125 MHz, D2O, 25 8C, only
annotated data from Scheme 3 is presented): d ˆ 155.8 (A6), 152.7 (A2),
149.2 (A4), 141.9 (B2), 141.1 (A8), 138.0 (B9), 133.8 (B6), 131.7 (B5),
130.3 (B8), 119.2 (A5), 118.5 (B4), 110.3 (B7), 94.6 (C10), 86.1 (R1), 78.6
(A11), 75.7 (A13), 72.5 (R3), 71.5 (L2), 63.2 (A12), 48.7 (A16), 43.7 (L1),
32.1 (13C A15), 32.4 (d, J ˆ 36 Hz, A14), 18.8 (B10), 18.5 (B11), 17.8 (L3).
Remaining data from the corrin moiety and its side chains is comparable
with those of adenosylcobalamin as previously assigned.[43] MS (FAB,
[23] R. I. Glazer, M. C. Knode, J. Biol. Chem. 1984, 259, 12964 12969.
[24] V. E. Marquez, M. I. Lim, Med. Res. Rev. 1986, 6, 1 40.
[25] H. R¸bsamen-Waigmann, Spektrum der Wissenschaft, Dossier 3/97.
[26] R. Storer, T. S. Mansour, Curr. Pharm. Des. 1997, 3, 227 264.
[27] H. P. C. Hogenkamp, Biochemistry 1974, 13, 2736 2740.
[28] K. A. Brameld, W. A. Goddard III, J. Am. Chem. Soc. 1999, 121, 985
993.
¬
[29] L. Poppe, J. Retey, Arch. Biochem. Biophys. 1995, 316, 541 546.
[30] L. P. Faust, B. M. Babior, Arch. Biochem. Biophys. 1992, 294, 50 54.
[31] S. A. Cockle, H. A. O. Hill, R. J. P. Williams, S. P. Davies, M. A. Foster,
J. Am. Chem. Soc. 1972, 94, 275 277.
[32] J. E. Valinsky, R. H. Abeles, J. A. Fee, J. Am. Chem. Soc. 1974, 96,
4709 4710.
‡
glycerol): m/z (%): 1578 (6) [M ], 1554 (10), 1463 (11), 1330 (49)
Chem. Eur. J. 2003, 9, No. 3
¹ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/03/0903-0659 $ 20.00+.50/0
659

¬
J. Retey et al.
FULL PAPER
[33] K. L. Schepler, W. R. Dunham, R. H. Sands, J. A. Fee, R. H. Abeles,
Biochim. Biophys. Acta 1975, 397, 510 518.
[39] R. W. Kellermeyer, H. G. Wood, Methods Enzymol. 1969, 13, 207
215.
¬
[34] J. F. Boas, P. R. Hicks, J. R. Pilbrow, T. D. Smith, J. Chem. Soc. Faraday
II 1978, 74, 417 431.
[40] B. Zagalak, J. Retey, H. Sund, Eur. J. Biochem. 1974, 44, 529 535.
[41] C. M. Blackwell, J. M. Turner, Biochem. J. 1978, 175, 555 563.
[42] W. W. Bachovchin, R. G. Eagar, K. W. Moore, J. H. Richards,
Biochemistry 1977, 16, 1082 1092.
[43] M. F. Summers, L. G. Marzilli, A. Bax, J. Am. Chem. Soc. 1986, 108,
4285 4294.
[35] D. Rˆther, PhD Thesis, University of Karlsruhe (Germany), 2000.
[36] T. Graf, PhD Thesis, University of Karlsruhe (Germany), 2001.
[37] Qiagen, The QIA Expressionist–A handbook for high-level expres-
sion and purification of 6 Â His-tagged proteins, 1997.
[38] R. Nitsche, PhD Thesis, University of Karlsruhe (Germany), 1999.
Received: July 29, 2002 [F4295]
660
¹ 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
0947-6539/03/0903-0660 $ 20.00+.50/0
Chem. Eur. J. 2003, 9, No. 3