Stereochemistry of a bifunctional dihydroceramide δ4- desaturase/hydroxylase from Candida albicans; a key enzyme of sphingolipid metabolism

Article abstract of DOI:10.1039/b303939k

The stereochemical course of the dihydroceramide Δ ⁴-(E)-desaturase from Candida albicans, cloned and expressed in the yeast Saccharomyces cerevisiae strain sur2Δ, was determined using stereospecifically labelled (2R,3S)-[2,3,4,4-²H ₄]-palmitic acid as a metabolic probe. Mass spectrometric analysis of the dinitrophenyl-derivatives of the labelled long-chain bases revealed elimination of a single deuterium atom from C(4) (corresponding to the C(4)-H_R) along with a hydrogen atom from C(5) (corresponding to the C(5)-H_S). This finding is consistent with an overall syn-elimination of the two vicinal hydrogen atoms. Besides the desaturation product sphingosine (93%) minor amounts of a 4-hydroxylated product (phytosphinganine, 7%) were identified that classify the Candida enzyme as a bifunctional desaturase/hydroxylase. Both processes, desaturation and hydroxylation proceed with loss of C(4)-H_R, from the chiral precursor. This finding is in agreement with a two-step process involving activation of the substrate by removal of the C(4)-H_R to give a C-centred radical or radicaloid followed by either disproportionation into an olefin, water and a reduced diiron complex, or to recombination of the primary reactive intermediate with an active site-bound oxygen to yield a secondary alcohol. This result demonstrates the close mechanistic relationship between desaturation and hydroxylation as two different reaction pathways of a single enzyme and strengthens the mechanistic relationship of desaturases from fatty acid metabolism and sphingolipids.

Full text of DOI:10.1039/b303939k

View Article Online / Journal Homepage / Table of Contents for this issue
Stereochemistry of a bifunctional dihydroceramide ꢀ4-desaturase/
hydroxylase from Candida albicans; a key enzyme of sphingolipid
metabolism
Christoph Beckmann,a Janine Rattke,a Petra Sperling,b Ernst Heinzb and Wilhelm Boland*a
a Max-Planck-Institut für Chemische Ökologie, Bioorganische Chemie, Winzerlaer Str. 10,
D-07745 Jena, Germany. E-mail: Boland@ice.mpg.de; Fax: ϩ49 3641 571202;
Tel: ϩ49 3641 571200
b Institut für Allgemeine Botanik, Universität Hamburg, Ohnhorststraße 18, D-22609 Hamburg,
Germany. E-mail: eheinz@botanik.uni-hamburg.de; Fax: ϩ49 4042816254;
Tel: ϩ49 4082816369
Received 11th April 2003, Accepted 23rd May 2003
First published as an Advance Article on the web 10th June 2003
The stereochemical course of the dihydroceramide ∆4-(E)-desaturase from Candida albicans, cloned and expressed in
the yeast Saccharomyces cerevisiae strain sur2∆, was determined using stereospecifically labelled (2R,3S)-[2,3,4,4-2H4]-
palmitic acid as a metabolic probe. Mass spectrometric analysis of the dinitrophenyl-derivatives of the labelled long-
chain bases revealed elimination of a single deuterium atom from C(4) (corresponding to the C(4)-HR) along with a
hydrogen atom from C(5) (corresponding to the C(5)-HS). This finding is consistent with an overall syn-elimination
of the two vicinal hydrogen atoms. Besides the desaturation product sphingosine (93%) minor amounts of a
4-hydroxylated product (phytosphinganine, 7%) were identified that classify the Candida enzyme as a bifunctional
desaturase/hydroxylase. Both processes, desaturation and hydroxylation proceed with loss of C(4)-HR from the chiral
precursor. This finding is in agreement with a two-step process involving activation of the substrate by removal of the
C(4)-HR to give a C-centred radical or radicaloid followed by either disproportionation into an olefin, water and a
reduced diiron complex, or to recombination of the primary reactive intermediate with an active site-bound oxygen
to yield a secondary alcohol. This result demonstrates the close mechanistic relationship between desaturation and
hydroxylation as two different reaction pathways of a single enzyme and strengthens the mechanistic relationship of
desaturases from fatty acid metabolism and sphingolipids.
species abstracting a single and specific hydrogen atom from the
substrate. This energetically demanding homolytic cleavage of a
Introduction
Sphingolipids comprise a large class of highly bioactive com-
pounds with diverse biological functions. A central compound
within sphingolipid metabolism is ceramide, which serves as a
precursor of all other major sphingolipids.1,2 Ceramide is
known to be involved in many important biological processes
such as control of metabolism via ceramide-activated protein-
kinases and -phosphatases, cell cycle arrest, cell differentiation
and apoptosis. Ceramide can be either biosynthesised de novo
or upon breakdown of more complex sphingolipids.3,4 The
de novo biosynthesis of ceramide starts with a serine palmitoyl-
transferase which catalyses the condensation of palmitoyl-CoA
and serine to form 3-ketosphinganine which is further
processed to sphinganine.
Subsequently, sphinganine is N-acylated by transfer of an
acyl-residue to give dihydroceramides. The final step of this
de novo ceramide biosynthesis involves a dihydroceramide
∆4-(E)-desaturase which converts the dihydroceramides into
ceramides (Scheme 1). This reaction is of great importance,
since it transforms the less active dihydroceramides into highly
active ceramides.5 From a mechanistic point of view, sphingo-
lipid desaturases seem to be closely related to the much better
investigated fatty acid desaturases. Two major groups of fatty
acid desaturases have been characterised; a large family of
membrane bound desaturases and a smaller group of soluble
desaturases, the latter only found in plants. The membrane-
bound enzymes have a multi-histidine diiron coordination site
and catalyse in a stereo- and regioselective manner an oxygen-
dependent syn-dehydrogenation of non-activated aliphatic/
olefinic substrates linked to CoA- or glycerolipids. The process
of desaturation is initiated by an enzyme-bound iron-oxo
non-activated C–H bond is associated with a large kinetic iso-
tope effect (KIE; kH/kD approx. 5–8) and is assumed to generate
a carbon centred radical (Scheme 2) for all membrane-bound
enzymes.6,7
Then, a second hydrogen atom from a neighbouring methyl-
ene group is transferred to the diiron-centre deactivating the
highly reactive intermediate(s) by disproportionation into an
olefin, water and a reduced iron-complex. The loss of the
second hydrogen atom is not associated with a noticeable KIE.
In addition, some desaturases exhibit an intrinsic residual
hydroxylase activity directed against the same methylene group
where desaturation starts, most likely due to a common reactive
intermediate that is channelled either into desaturation or
hydroxylation.8
First data on the stereochemical course of a dihydroceramide
∆4-(E)-desaturase became available in the early seventies. In
vivo studies with preparations from rat brain and tritium
labelled precursors gave rise to a kinetic isotope effect (KIE) for
the abstraction of a pro-R hydrogen atom from C(4) of a
sphinganine precursor along with a second hydrogen atom from
C(5) consistent with an overall syn-elimination.9 Other authors
arrived at an overall anti-elimination.10 Since the previous
results were based on experiments with tissue preparations lack-
ing the strength of approaches with cloned genes and hetero-
logously expressed enzymes and considering the contradictory
reports on the stereochemistry of the desaturation, we were
encouraged to re-address this question using transformed yeast
cultures overexpressing a dihydroceramide ∆4-(E)-desaturase
which is lacking in wild type cells. A recent bioinformatics
approach afforded a family of protein sequences from animals,
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 2 4 4 8 – 2 4 5 4
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 3
2448

View Article Online
that this enzyme has a dual catalytic function, namely desatura-
tion and hydroxylation of dihydroceramide.
Results and discussion
Synthesis of (2R,3S )-[2,3,4,4–2H4]palmitic acid
As a model for the stereochemical analysis of a dihydro-
ceramide ∆4-(E)-desaturase, the corresponding enzyme from
Candida albicans was cloned and expressed in the Saccharo-
myces cerevisiae strain sur2∆.11 The stereochemistry of the
desaturation process was investigated with stereospecifically
labelled precursors. After inhibition of the fatty acid bio-
synthesis using cerulenin,12 externally added palmitic acid is
taken up by the yeast cells, converted into acyl-CoA and chan-
nelled into the sphingolipid metabolism along the pathway out-
lined in Scheme 1. Owing to the efficiency of this biocatalytic
conversion of palmitic acid into dihydroceramide, the synthetic
approach can focus on the synthesis of chiral [2,3,4,4-2H4]-
(2R,3S)-palmitic acid 7. After incorporation, this acid is con-
verted into the labelled dihydroceramide 8 with two deuterium
atoms residing at the carbon atoms that are directly involved in
the desaturation process and two others at C(6) facilitating an
unambiguous identification of the metabolite in case of a sim-
ultaneous removal of the deuterium atoms from C(4) and C(5)
(Fig. 1). The isotope labels are easily introduced by the synthetic
protocol outlined in Scheme 3.
Scheme 1 De novo biosynthesis of ceramide.
Scheme 3 Synthesis of (2R,3S)-[2,3,4,4-2H4]-palmitic acid. a) I2,
᎐
imidazole, PPh3; b) 1. Li–C᎐C–CH2–O–THP, 2. p-toluenesulfonic acid–
᎐
MeOH; c) LiAl2H4, 2HCl, 2H2O; d) MnO2; e) NaClO2; f ) anaerobic
biocatalytic H2-reduction with broken cells of Clostridium tyro-
butyricum (strain C. La1; for the stereochemical analysis see
Experimental section).
Scheme 2 General mechanism for desaturation and hydroxylation of
aliphatic substrates (Ref. 6). The oxidising species is assumed to be an
electrophilic 1,2-bis-µ-oxo complex of Fe() or an diiron-(,)-oxo
species after bridge-to-terminal oxo migration (not shown,29). Both
processes involve removal of the same hydrogen atom from the
substrate as the initial step.
Reduction of methyl tridecanoate with lithium aluminium
2
deuteride (98% H) afforded [1,1-2H2]-tridecan-1-ol 1. Alcohol
1 was converted into the corresponding iodide 2 and alkylated
with the lithium salt of 2-prop-2-ynyloxytetrahydropyran in
liquid ammonia. After removal of the protecting group, the
alkynol 3 was reduced with lithium aluminium deuteride fol-
lowed by hydrolysis of the organoaluminium intermediates
with H2O and HCl. Oxidation of the alkenol 4 to the corre-
sponding acid 6 was achieved in two steps by treatment of 4
with manganese dioxide and subsequent oxidation of the result-
ing aldehyde 5 with sodium chlorite in a buffered aqueous solu-
tion. Configurationally pure [2,3,4,4-2H4]-hexadec-2-enoic acid
plants and fungi which were subsequently cloned and expressed
in a Saccharomyces cerevisiae mutant strain (sur2∆) lack-
ing long-chain base C4-hydroxylation.11 Biochemical character-
isation of the transgenic yeast revealed that these proteins
belong to the family of sphingolipid ∆4-(E)-desaturases
(dihydroceramide desaturases). Here, we report on a novel
stereochemical analysis of the desaturation step using the
heterologously expressed dihydroceramide ∆4-(E)-desaturase
from Candida albicans as a model system. We also demonstrate
2
2
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 2 4 4 8 – 2 4 5 4
2449

View Article Online
Fig. 1 Strategy of feeding experiments and analysis of long-chain bases. Metabolites were released by hydrolysis, extracted and converted into their
DNP-derivatives. After TLC purification, separation and identification of the DNP-long-chain bases was achieved by HPLC and mass spectrometry:
11, C18-4-hydroxysphinganine; 12, C18-sphingosine; 13, C18-sphinganine; 14, C20-sphingosine and 15, C20-4-hydroxysphinganine.
6 was thus obtained in 70% overall yield. The key step of the
asymmetric synthesis is the biocatalytic reduction of the
labelled acid 6 with broken cells of Clostridium tyrobutyricum
(strain C. La1) and hydrogen gas.13,14 The reduction of the
α,β-unsaturated acid proceeds stereospecifically with an anti-
2Si,3Si-transfer of two hydrogen atoms across the double bond
of the precursor and afforded (2R,3S)-[2,3,4,4-2H4]-palmitic
acid 7 in 81% yield and very high enantiomeric excess (>97%
per centre) as shown by 1H NMR after conversion of 7 into the
corresponding mandelate diester.15
(RP18) and mass spectrometry using electrospray ionisation
(ESI-MS) in the negative ion mode (Fig. 1). Besides labelled
DNP-sphinganine 13 (85%) and its C20 homolog 15 (6%), two
unsaturated DNP-sphingenines 12 (7%) and 14 (1%), resulting
from desaturation of 8 and its C20 homologue, and a DNP-tri-
hydroxysphinganine 11 (<1%) were identified.
All DNP-derivatives of sphingoid metabolites displayed
intense [M Ϫ H]Ϫ ions. The efficiency of the conversion of 7
into labelled dihydroceramide 8 is demonstrated by Fig. 2B
showing the molecular species of labelled DNP-sphinganine
[2H4]-13 (m/z = 470, 100%) and of native unlabelled [1H]-13 (m/z
= 466, 52%) corresponding to approx. 60% labelled material.
Owing to a large kinetic isotope effect during desaturation20 the
ratio of labelled DNP-sphingosine [2H3]-12 (m/z = 467) to native
[1H]-12 (m/z = 464) drops to approx. 1 : 0.95, Fig. 2C. The
molecular species of [2H3]-12 is consistent with a clear and uni-
form loss of a single deuterium atom together with a single
hydrogen atom. Since the desaturation process involves only
hydrogen atoms from C(4) and C(5) of dihydroceramide [2H4]-
8, two of the remaining deuterium atoms have to reside on
C(6), the other one either on C(4) or on C(5) of the labelled
DNP-sphingosine 12.
The C20 homologue is also desaturated and yields tetra-
deuterated DNP-dihomosphingosine 14 (m/z = 496) with
complete retention of all deuterium atoms originally present in
the precursor. This is in agreement with a two-carbon chain
elongation of the administered [2H4]-palmitic acid 7.
Interestingly, the cluster of the pseudomolecular ions of
the early eluting DNP-derivative of labelled C18-trihydroxy-
sphinganine 11 (see Fig. 1), at m/z = 485/486, demands for a
predominant, but not exclusive loss of a deuterium atom ([2H3]-
11: [2H4]-11, approx. 1 : 0.6) from C(4) of dihydroceramide
8 (corresponding to the C(4)-HR, Fig. 2A).9,10,21 Control
Stereochemical analysis of the dihydroceramide
ꢀ4-(E )-desaturase
Sur2∆ cells of Saccharomyces cerevisiae, lacking the genuine
C(4)-hydroxylase gene (due to disruption of the SUR2 gene)16
but overexpressing the dihydroceramide desaturase from
Candida albicans,11 were grown in the presence of deuterium-
labelled palmitic acid 7 and cerulenin. This fungal metabolite
strongly inhibits the yeast’s intrinsic ability to synthesise fatty
acids12 and, hence, minimises the competition between exter-
nally added precursors and genuine fatty acids in the serine
palmitoyl transferase reaction. To increase the internal sub-
strate level of dihydroceramide, the transgenic cells were sub-
jected to heat shock.17,18 Following conversion of the labelled
palmitic acid 7 into dihydroceramide 8, the sphingolipid was
desaturated to labelled ceramide 9 (R = acyl) by the hetero-
logously expressed Candida enzyme. For analysis of the whole
spectrum of products, the sphingolipids of whole cells were
hydrolysed, and the liberated sphingobases converted with
Sanger’s reagent into their dinitrophenyl (DNP) derivatives as
described before.19 The DNP derivatives were pre-purified by
TLC (SiO2 60) and finally separated and identified by HPLC
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 2 4 4 8 – 2 4 5 4
2450

View Article Online
protocol22 involving dihydroxylation of the (4E)-double bond23
to give 16 followed by cleavage of the carbon backbone with
sodium periodate under phase transfer conditions.24 Since
aliphatic aldehydes generally display very weak molecular
ions under EI-MS conditions, the resulting tetradecanal was
analysed using chemical ionisation with isobutane. [1,2,2-2H3]-
Tetradecanal 17 from oxidative degradation of 12 displayed
the presence of three deuterium atoms accompanied by
minor amounts of di- and monodeuterated isotopomers and
unlabelled aldehyde (2H3–2H2–2H1 approx. 100 : 35 : 10). Pro-
1
longed reaction times (24 h) in H2O further decreased the
2
deuterium content, whereas running the reaction in H2O had
the opposite effect. Accordingly, a slow exchange of the acidic
α-hydrogens of the aldehyde is responsible for the variable
isotopomer composition and not the desaturation process.
In summary, the stereochemical course of the desaturation
by the Candida enzyme can be unambiguously described as a
syn-elinination of the C(4)-HR along with the C(5)-HS hydrogen
atom from the dihydroceramide precursor 8 as outlined in
Scheme 2. Most importantly, with respect to the C(4)-HR both
catalytic activities of the enzyme, namely hydroxylation and
desaturation, start by removal of the same hydrogen atom from
the carbon backbone of the ceramide precursor.
Conclusions
The total number of known and suspected non-heme diiron
enzymes is steadily increasing because of the wealth of
sequence data becoming available from genome-sequencing and
bioinformatics efforts.25 Recently, this resulted in the identifi-
cation of a family of dihydroceramide ∆4-(E)-desaturases from
Homo sapiens, Mus musculus, Drosophila melanogaster and
Candida albicans along with a bifunctional ∆4-desaturase/
hydroxylase from M. musculus.11 From these examples the
heterologously expressed enzyme from C. albicans was chosen
as a model system for a detailed mechanistic and stereo-
chemical analysis of the desaturation process. In agreement
with all other previously examined cases, the enzyme operates
stereo- and regiospecifically with simultaneous syn-elimination
of two vicinal hydrogen atoms, namely the C(4)-HR and the
C(5)-HS. This stereochemical course is fully consistent with the
results obtained by Stoffel et al.9 Buist et al. studied the inter-
molecular kinetic isotope effects of a dihydroceramide ∆4-(E)-
desaturase from rat liver microsomes and found that the intro-
duction of the double bond occurs in two discrete steps with a
significant KIE (kH/kD = 8.0 0.8) for the abstraction of the
C(4)-H atom and almost no KIE for the loss of the hydrogen
atom from C(5) (kH/kD = 1.02 0.07).20 In agreement with other
fatty acid desaturases, the KIE is generally observed for the
carbon atom proximal to the polar head of the substrate mole-
cule. A notable exception for the class of membrane-bound
enzymes is the recently investigated non-specific ∆8-(E,Z)-
sphingolipid desaturase which obeys the rule in case of its trans-
product (attack onto C(8)-HR), but switches to the distal carbon
(attack onto the C(9)-HR) in the case of its cis-product.22 In
contrast, the soluble stearoyl-ACP ∆9-desaturase from castor
seed showed no significant KIE for either the C(9) or C(10) (kH/
kD approx. 1) probably due to masking by another, kinetically
more important step in the catalytic cycle.26
Fig. 2 Mass spectrometric analysis of deuterium-labelled metabolites.
Shown are the clusters of pseudomolecular ions from DNP-4-
hydroxysphinganine 11 (Fig. 2A), DNP-sphinganine 13 (Fig. 2B), and
DNP-sphingosine 12 (Fig. 2C). The group of pseudomolecular ions at
higher masses (e.g. m/z = 485, 486, and 487, Fig. 2A) corresponds to the
labelled metabolites, the cluster at lower masses (e.g. m/z = 482, 483, and
484, Fig. 2A) is due to unlabelled, native compounds.
experiments with yeast cells expressing the empty vector of
pYES2 do not produce any desaturation products or hydroxyl-
ated sphinganines and, hence, the labelled DNP-derivative of
[2H3]-4-hydroxysphinganine 11 has to be likewise considered as
a product of the expressed Candida desaturase. Moreover,
treatment of the extract (see Fig. 1) with excess of NaIO4
resulted in a quantitative cleavage of 11 and, thus, independ-
ently confirmed the structure of this metabolite as a vicinal
3,4-dihydroxy compound. All other compounds shown in Fig. 1
were not affected by the reagent.
In order to determine the position of the remaining deuter-
ium atom at the double bond of deuterium-labelled ceramide 9,
the corresponding DNP fraction was collected by HPLC and
subjected to an oxidative degradation yielding tetradecanal 17
from the aliphatic terminus of the metabolite 12 (Scheme 4).
The cleavage was achieved following a recently published
Apparently, the ∆4-desaturase from C. albicans is also able to
act as a C(4) hydroxylase as minor amounts of a correspond-
ing C18 trihydroxysphinganine 10 were identified among the
reaction products (Fig. 1). Since the hydroxylation, like the
desaturation process, removes the same C(4)-HR, both processes
apparently generate the same reactive intermediate, namely a
C(4)-centred radical or radicaloid that is channelled predomin-
antly (approx. 93%) into desaturation or recombines with an
active-site bound oxygen atom to give an alcohol (approx. 7%,
formal insertion of oxygen into the C(4)-HR bond). The factors
controlling this ratio remain to be established in detail, but
Scheme 4 Oxidative degradation of DNP-sphingosine 12.
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 2 4 4 8 – 2 4 5 4
2451

View Article Online
might be largely due to a proper positioning of the substrate at
the active site. First information on this aspect came from the
exchange of four amino acids in core positions of an oleate
desaturase of Arabidopsis thaliana which converted the enzyme
into a bifunctional desaturase/hydroxylase generating different
product ratios.8 The importance of substrate presentation at the
active site is also obvious from studies with the mechanistically
related methane monooxygenase which converts less tightly
bound artificial substrates such as ethyl benzene into an alcohol
and an olefin.27 Both findings, namely modification of the
active site as well as the conversion of non-natural substrates
resulting in different ratios of desaturation versus hydroxyl-
ation, are consistent with a uniform mechanistic view for both
pathways concerning the early steps. In case of desaturation,
activation of the substrate generates a reactive intermediate
(C-centred radical) whose orientation and conformation at the
active site only allows a disproportionation into an olefin, water
and a reduced diiron complex. The hydroxylation probably
requires a slightly different orientation of the substrate relative
to the diiron-centre. After formation of the same primary radi-
cal as before, the divergent substrate orientation allows a
recombination with the iron-bound hydroxy group yielding an
alcohol along with the reduced diiron-centre (Scheme 1).
Whether the two conformationally different enzyme–substrate
complexes represent the entry points into the respective
catalytical cycles or emerge during the progress of the reaction
is currently unknown and remains to be established. Investi-
gations into this direction are under way with the bifunctional
∆4-(E)-desaturase/hydroxylase from mouse and will be reported
in due course.
chromatography (silica gel 60, CHCl3–MeOH 9 : 1, v/v) as
described before.19
[1,1-2H2]-Tridecan-1-ol (1)
A solution of methyl tridecanoate (10 g, 44 mmol) in abs. THF
(20 ml) was added slowly within 10 min at rt to a suspension of
2
lithium aluminium deuteride (1.0 g, 24 mmol, 98% H) in the
same solvent (50 ml). After 1 h at reflux, the mixture was cooled
to 0 ЊC and hydrolysed with water (10 ml) and dil. HCl (6 M).
Extractive work-up with ether, drying and removal of solvent
under reduced pressure gave alcohol 1 as a colourless solid.
Yield: 8.4 g (95%). Mp: 32.5–33.5 ЊC. δH (400 MHz, CDCl3)
0.87 (3 H, t, J 7.1, CH3), 1.22–1.36 (20 H, m, 10 × CH2), 1.38
(1 H, br s, OH), 1.54 (2 H, t, J 7.2, CH2C2H2); δC(100.6 MHz,
CDCl3) 14.24, 22.83, 25.84, 29.49, 29.59, 29.74, 29.76, 29.79,
29.80, 29.83, 32.06, 32.76, 62.46 (qui., J 21.5, C2H2); m/z (EI)
184.21588 (Mϩ Ϫ H20, 32%. C13H242H2 requires 184.21601)
ؒ
156 (14), 112 (16), 97 (44), 84 (74), 70 (93), 57 (100); νmax(neat)/
cmϪ1 3305br, 2959, 2923, 2852, 1463, 1375, 1166, 1130, 1104,
1073, 962, 713.
[1,1-2H2]-1-Iodotridecane (2)
A chilled solution of triphenylphosphine (13.11 g, 50 mmol)
and imidazole (3.40 g, 50 mmol) in diethyl ether–acetonitrile
(200 ml, v/v, 3 : 1) was treated with iodine (12.69 g, 50 mmol) in
four portions within 20 min. The suspension was allowed to
come to rt, chilled and slowly treated with alcohol 1 (7.4 g,
37 mmol). Stirring was continued for 1 h at rt, followed by
addition of light petroleum (200 ml). The upper layer was
decanted, the residual phase treated with aq. NaHCO3 (5%,
150 ml) and extracted with light petroleum (3 × 50 ml). The
organic layers were concentrated under reduced pressure and
the residue dissolved in pentane (2 × 50 ml). After drying,
evaporation of solvent in vacuo (i.v.), the iodide was purified by
flash chromatography on silica gel using light petroleum for
elution. Colourless oil. Yield: 10.5 g (92%). δH(400 MHz,
CDCl3) 0.90 (3 H, t, J 6.8, CH3), 1.21–1.33 (18 H, m, 9 × CH2),
1.38 (2 H, qui., J 7.1, C(3)H), 1.83 (2 H, t, J = 7.4, CH2C2H2);
δC(100.6 MHz, CDCl3) 7.08 (qui., J 22.9, C2H2), 14.25, 22.83,
Experimental
General methods
Reactions were performed under Ar. Solvents were dried
according to standard methods or commercially available dry
solvents (Fluka) were used instead. IR: Bruker Equinox 55
1
FTIR Spectrophotometer. H and 13C NMR: Bruker AV 400
(Bruker, D-76287 Rheinstetten/Karlsruhe, Germany). Chemi-
1
cal shifts of H and 13C NMR are given in ppm (δ) based on
solvent signals: CDCl3 7.26 ppm (1H NMR) and 77.16 ppm (13C
NMR). For ESI-LC-MS measurements a Micromass Quattro
II (Micromass, Manchester, UK) connected to an Agilent
HP1100 HPLC, equipped with a RP18-column (125 mm ×
2 mm, 3 µm, Grom D-71083 Herrenberg, Germany) was used.
Gas chromatography/mass spectrometry (70 eV) was performed
on a Micromass MasSpec (Micromass, Manchester, UK)
double-focusing magnetic sector field mass spectrometer
(geometry EBE) connected to a Hewlett Packard HP6890 II gas
chromatograph, equipped with a DB-5 (J&W Scientific) non-
polar capillary column (30 m × 0.25 mm, 0.25 µm). Silica gel: Si
60 (0.200–0.063 mm, E. Merck, Darmstadt, Germany) was
used for chromatography.
28.71, 29.50, 29.57, 29.70, 29.77, 29.79, 29.81, 30.61, 32.07,
ϩ
32,51; m/z (EI) 312.12946 (M , 14%. C13H252H2I requires
ؒ
312.12831), 185 (45), 87 (28), 85 (32), 73 (34), 71 (56), 59 (40),
57 (100); νmax(film)/cmϪ1 2955, 2930, 2853, 1466, 1377, 1124,
1107, 978, 887, 721.
[4,4–2H2]-Hexadec-2-yn-1-ol (3)
Finely cut lithium wire (0.35 g, 50 mmol) was added slowly to a
well stirred solution of liquid ammonia (250 ml) and iron()
nitrate nonahydrate (10 mg, 0.025 mmol) at Ϫ78 ЊC. After
complete conversion into the amide the solution turned gray
and 2-prop-2-ynyloxytetrahydropyran (7.00 g, 50 mmol) was
added over a period of 10 min. Stirring was continued for 1 h at
Ϫ60 ЊC, and pre-cooled dry DMSO (100 ml) was added fol-
lowed by slow addition of the iodide 2 (10 g, 32 mmol) within
15 min. The mixture was allowed to warm to rt overnight and
the ammonia evaporated. The residue was dissolved in water
(250 ml) and extracted with diethyl ether (3 × 200 ml). The
organic extracts were washed with water (50 ml), dried and
concentrated under reduced pressure. Chromatography on
silica gel using light petroleum–diethyl ether (v/v, 9 : 1) for elu-
tion gave the THP ether of 3 as a colourless oil. Yield: 7.0 g
(67%). A solution of the THP ether (6.5 g, 20 mmol) and
p-toluenesulfonic acid hydrate (200 mg, 1 mmol) in methanol
(50 ml) was stirred for 1 h at rt. Pyridine (1 ml) was added prior
to evaporation of solvents under reduced pressure. The residue
was treated with water (25 ml) and extracted with diethyl ether
(3 × 50 ml). The organic extracts were dried and concentrated
under reduced pressure. Column chromatography on silica gel
Expression system and incubation experiments
The plasmid pYES2 (Invitrogen) containing the open read-
ing frame of the ∆4-(E)-sphingolipid desaturase from Candida
albicans was used to transform the strain sur2∆ of Saccharo-
myces cerevisiae.11 Transgenic sur2∆ cells were grown in com-
plete minimal medium (110 ml) dropout uracil28 containing 2%
(w/v) raffinose as a carbon source, 2% (w/v) galactose to induce
expression, 25 µM cerulenin (Sigma) and 250 µM labelled palm-
itic acid 12. Cultures were grown aerobically at 25 ЊC for 6 d
(final OD600 ∼0.8), subjected to a heat shock at 37 ЊC for 90 min,
and harvested by centrifugation. Cells (∼ 300 mg fresh weight)
were washed and directly hydrolysed (10% Ba(OH)2 (w/v) in
H2O–dioxane, 1 : 1, 24 h, 110 ЊC). The released long chain
sphingobases (LCB) were converted into the dinitrophenyl
derivatives (DNP-derivatives) and pre-purified by thin layer
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 2 4 4 8 – 2 4 5 4
2452

View Article Online
using light petroleum–diethyl ether (v/v 4 : 1) for elution
afforded alcohol 3 as a colourless solid. Yield: 4.5 g (94%). Mp:
54–55 ЊC. δH(400 MHz, CDCl3) 0.87 (3 H, t, J 6.9, CH3), 1.21–
1.40 (20 H, m, 10 × CH2), 1.45–1.51 (2 H, m, CH2C2H2), 1.63
(2 H, br s, OH), 4.24 (2 H, s, CH2OH); δC(100.6 MHz, CDCl3)
14.24, 18.30 (qui., J 18.4, C2H2), 22.83, 28.57, 28.97, 29.30,
29.50, 29.66, 29.77, 29.79, 29.80, 29.83, 32.06, 51.56, 78.44,
colourless solid. Yield: 1.73 g (89%). Mp: 46.5–48 ЊC. δH(400
MHz, CDCl3) 0.88 (3 H, t, J 6.9, CH3), 1.19–1.36 (20 H, m,
10 × CH2), 1.45 (2 H, m, CH2), 11.84 (1 H, br s, COOH);
δC(100.6 MHz, CDCl3) 14.24, 22.84, 27.86, 29.24, 29.51
29.53, 29.67, 29.77, 29.80, 29.81, 29.83, 31.58 (qui., J 18.5,
C2H2), 32.08, 120.48 (t, J 24.6, C(2)), 152.14 (t, J 23.1, C(3)),
172,48 (COOH); m/z (EI) 258.24952 (M , 13%. C16H262H4O2
ϩ
ؒ
86.77; m/z (EI) 222.23164 (Mϩ Ϫ H20, 4%. C16H262H2 requires
requires 258.24969), 240 (39), 197 (12), 130 (18), 116 (27), 102
(41), 98 (49), 84 (49), 70 (58), 57 (100); νmax(KBr)/cmϪ1 2963,
2923, 2848, 2662, 2555, 1681, 1468, 1419, 1290, 1059, 931, 789,
718.
ؒ
222.23166), 137 (33), 123 (45), 113 (67), 95 (77), 83 (100), 69
(91); νmax(neat)/cmϪ1 3250br, 2954, 2914, 2852, 1472, 1259,
1091, 1042, 1024, 1006, 798.
[2,3,4,4-2H4]-(2E )-Hexadec-2-en-1-ol (4)
(2R,3S )-[2,3,4,4-2H4]-Hexadecanoic acid (7)
A solution of alcohol 3 (4.5 g, 19 mmol) in dry THF (20 ml)
was added in 5 min to a suspension of lithium aluminium
deuteride (0.95 g, 22.5 mmol) in dry THF (50 ml). The mixture
was refluxed for 2 h and cooled to 0 ЊC followed by hydrolysis
with 2H2O (2.0 ml) and cautious addition of 2HCl (7.5 ml, 36%
The sodium salt of 6 (1 g, 3,9 mmol) was dissolved in potas-
sium phosphate buffer (50 ml, 0.1 M, pH = 7) and placed
in a reaction vessel with a thermostat. Methyl viologene
(15 mg, 0.06 mmol) and tetracycline (3.0 mg) were added and
the solvent was degassed by passing a gentle stream of helium
for 1 h through the solution. The system was flushed with argon
and approx. 5 g broken cells of Clostridium tyrobutyricum
(strain C. La1, DSM number 1460) were added with strict
exclusion of oxygen. Then, argon was replaced by hydrogen
(connected to gas burette to monitor the progress of the
reaction) while the vessel was gently shaken at 35 ЊC. After a
brief induction period the suspension turned deep blue. After
48 h no further consumption of hydrogen was observed and the
reaction mixture was acidified with dil. H2SO4 (10%) to pH 2–3.
The mixture was continuously extracted with diethyl ether
for 12 h. The ether layer was dried and concentrated under
reduced pressure. Chromatography on silica gel with light
petroleum–diethyl ether (v/v, 1 : 2) yielded the chiral palmitic
acid 7 as a colourless solid. Yield: 0.75 g (81%). Mp: 61.5–62.5
ЊC. For incubation experiments the acid was additionally
purified by semi-preparative HPLC (RP 18) using a linear
gradient from methanol–water (90 : 10) to 100% methanol
within 20 min. δH(400 MHz, CDCl3) 0.88 (3 H, t, J 7.1, CH3),
1.21–1.37 (22 H, m, 12 × CH2), 1.59 (1 H, d, J 8.4, CH2), 2.32
(1 H, d, J 8.4, CH2), 11.25 (1H, br s, COOH); δC(100.6 MHz,
CDCl3) 14.27, 22.85, 24.21 (t, J 19.4, C(3)), 28.26 (qui., J 19.3,
C2H2), 29.17, 29.52, 29.54, 29.76, 29.80, 29.81, 29.83, 29.84,
29.85, 32.09, 33.82 (t, J 19.4, C(2)), 180.57 (COOH); m/z
2
in H2O). The resulting semi-solid mixture was then acidified
by dilute hydrochloric acid (1 M) until a milky suspension
was obtained. The aq. layer was extracted with diethyl ether
(3 × 50 ml), and the organic extracts washed with sat. NaHCO3
solution (20 ml) and water (20 ml). After drying and evapor-
ation of solvents the alkenol was obtained as a colourless solid.
Yield: 4.1 g (88%). Mp: 36.5–37.5 ЊC. δH(400 MHz, CDCl3) 0.88
(3 H, t, J 7.0, CH3), 1.21–1.38 (22 H, m, 11 × CH2), 4.08 (2 H, s,
CH2OH); δC(100.6 MHz, CDCl3) 14.2, 22.84, 29.11, 29.29,
29.51, 29.66, 29.76, 29.81, 29.82, 29.83, 29.84, 31.44 (qui.,
J 19.4, C2H2), 32.08, 63.91, 128.59 (t, J 23.8, C2H), 133.25
2
ϩ
(t, J 23.1, C H); m/z (EI) 244.27045 (M , 4%. C16H282H4O
requires 244.27042), 226 (14), 141 (8), 127 (13), 99 (45), 85 (63),
71 (55), 59 (100); νmax(KBr)/cmϪ1 3298br, 2917, 2849, 1473,
1043, 1009, 713.
ؒ
[2,3,4,4-2H4]-(2E )-Hexadec-2-en-1-al (5)
A solution of the alkenol 4 (3.67 g, 15 mmol) in CH2Cl2 (50 ml)
was treated while stirring at rt with activated manganese()
oxide (10 g, 115 mmol). After vigorous shaking for 1 h another
portion of manganese dioxide (5 g, 58 mmol) was added and
shaking continued for 1 h. The oxidant was allowed to settle,
and the supernatant was decanted. The decantation was
repeated (3 × 50 ml dichloromethane), and the combined
extracts were filtered over a pad of Celite. After drying, and
removal of solvent in vacuo, chromatography on silica gel using
light petroleum–diethyl ether (v/v, 4 : 1) for elution gave alde-
hyde 5 as a colourless liquid. Yield: 2.87 g (79%). δH(400 MHz,
CDCl3) 0.90 (3 H, t, J 7.1, CH3), 1.21–1.40 (20 H, m, 10 × CH2),
1.51 (2 H, br t, J 7.0, C(5)H), 9.50 (1 H, s, CHO); δC(100.6
MHz, CDCl3) 14.22, 22.81, 27.81, 29.21, 29.47, 29.48, 29.63,
29.73, 29.77 (2 × CH2), 29.79, 31.96 (qui., J 19.4, C2H2), 32.04,
ϩ
(EI) 260.26551 (M , 100%. C16H282H4O2 requires 260.26534),
214 (14), 133 (30), 119 (10), 74 (41), 57 (42); νmax(KBr)/cmϪ1
2964, 2954, 2916, 2848, 1698, 1471, 1353, 1312, 1296, 1257,
1221, 940, 675.
ؒ
Analysis of the enantiomeric purity of (2R,3S )-[2,3,4,4–2H4]-
hexadecanoic acid (7)
(2R,3S)-[2,3,4,4-2H4]-Palmitic 7 acid was converted with methyl
(S)-(ϩ)-mandelate into the corresponding mandelate
1
132.81 (t, J 24.6, C(2)), 158.49 (t, J 22.7, C(3)), 194.17 (CHO);
diester.15,30 The H NMR spectrum was recorded with simul-
ϩ
m/z (EI) 242.25461 (M , 25%. C16H262H4O requires
taneous irradiation at 1.6 ppm to suppress the vicinal coupling
of the protons at C(3) and C(2). The signal of the proton C(2)-
HR (>98%) of the labeled acid appears as a singlet at 2.20 ppm
while the minor C(2)-HS is observed at 2.24 ppm (<2%). Within
the error limits of the NMR method the enantiomeric excess of
the labeled palmitic acid 7 is ≥96%.
ؒ
242.25477), 224 (16), 139 (22), 125 (36), 97 (51), 83 (67), 74
(100), 57 (92); νmax(KBr)/cmϪ1 2955, 2913, 2845, 1692, 1665,
1601, 1468, 1395, 1171, 723.
[2,3,4,4-2H4]-(2E )-Hexadec-2-enoic acid (6)
A solution of sodium chlorite (6.2 g, 55 mmol) and sodium
dihydrogenphosphate (8.0 g, 67 mmol) in water (50 ml)
was added over 20 min to aldehyde 5 (1.81 g, 7.5 mmol) in
tert-butanol (70 ml) and pent-1-ene (15 ml). The yellow opal-
escent solution was stirred for 2 h followed by evapor-
ation of solvent in vacuo. Water (50 ml) was added and the
aqueous phase extracted with diethyl ether (3 × 75 ml). The
combined extracts were washed with aq. NaOH (3 × 75 ml, 1 M
solution), acidified (pH 2–3) with 6 M hydrochloric acid
and extracted with diethyl ether (3 × 75 ml). After drying
and removal of solvents the acid 6 was obtained as a
Isolation and analysis of long-chain bases
The DNP-spingoid bases were separated by HPLC under
reversed phase conditions (GROM-SIL 120 ODS-5 ST, 3 µm,
125 × 2 mm, Grom, Herrenberg) with a flow rate of 0.2 ml
minϪ1. A linear gradient from 60% MeOH–CH3CN–propan-
2-ol (v/v/v 10 : 3 : 1) and 40% water to 20% water within 10 min
and finally to 0% water within further 40 min was used. The
ESI-MS spectra were recorded in the negative ion mode (source
temperature: 100 ЊC, desolvation temperature: 250 ЊC, cone
voltage: 35 Volt).
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 2 4 4 8 – 2 4 5 4
2453

View Article Online
8 J. A. Broadwater, E. Whittle and J. Shanklin, J. Biol. Chem., 2002,
277, 15613–15620.
Oxidative cleavage of DNP-sphingosine
The DNP-derivative of [5,6,6-2H3]-sphingosine 12 was collected
by HPLC and concentrated under reduced pressure to
give approx. 1 µg of product. AD-mix-α (14.0 mg, Aldrich),
t-BuOH–water (100 µl, v/v, 1 : 1) and a catalytic amount of
methane sulfonamide was added. The resulting two phase sys-
tem was vigorously shaken for 12 h at rt. A sat. aq. solution of
sodium sulfite (100 µl) was added and shaking was continued
for 5 min. After extraction with ethyl acetate (3 × 200 µl) the
combined solvents were removed with a gentle stream of argon.
The residue was dissolved in 100 µl dichloromethane, and a
catalytic amount of benzyltrimethylammonium chloride and
sodium periodate (20 µl, sat. aq. solution) was added. The mix-
ture was shaken for 5 min and aliquots were directly analysed
by GC/MS using chemical ionisation with isobutane.
9 W. Stoffel, G. Assmann and K. Bister, Hoppe-Seyler’s Z. Physiol.
Chem., 1971, 352, 1531–1544.
10 A. J. Polito and C. C. Sweeley, J. Biol. Chem., 1971, 246, 4178–
4187.
11 P. Ternes, S. Franke, U. Zähringer, P. Sperling and E. Heinz, J. Biol.
Chem., 2002, 277, 25512–25518.
12 F. Schneider, R. Lessire, J. J. Bessoule, H. Juguelin and C. Cassagne,
Biochim. Biophys. Acta, 1993, 1152, 243–252.
13 H. Simon, J. Bader, H. Gunther, S. Neumann and J. Thanos,
Angew. Chem. Int. Ed. Engl., 1985, 24, 539–553.
14 O. Thum, C. Hertweck, H. Simon and W. Boland, Synthesis, 1999,
2145–2150.
15 G. Görgen, W. Boland, U. Preiss and H. Simon, Helv. Chim. Acta,
1989, 72, 917–928.
16 D. Haak, K. Gable, T. Beeler and T. Dunn, J. Biol. Chem., 1997, 272,
29704–29710.
17 G. M. Jenkins, A. Richards, T. Wahl, C. G. Mao, L. Obeid and
Y. Hannun, J. Biol. Chem., 1997, 272, 32566–32572.
18 R. C. Dickson, E. E. Nagiec, M. Skrzypek, P. Tillman, G. B. Wells
and R. L. Lester, J. Biol. Chem., 1997, 272, 30196–30200.
19 P. Sperling, U. Zähringer and E. Heinz, J. Biol. Chem., 1998, 273,
28590–28596.
20 C. K. Savile, G. Fabrias and P. H. Buist, J. Am. Chem. Soc., 2001,
123, 4382–4385.
21 W. Stoffel and E. Binczek, Hoppe-Seyler’s Z. Physiol. Chem., 1971,
352, 1065–1072.
22 C. Beckmann, J. Rattke, N. J. Oldham, P. Sperling, E. Heinz and
W. Boland, Angew. Chem., Int. Ed., 2002, 41, 2298–2300.
23 H. C. Kolb, M. S. Vannieuwenhze and K. B. Sharpless, Chem. Rev.,
1994, 94, 2483–2547.
Acknowledgements
We thank Dr Teresa Dunn for the kind supply of the
S. cerevisiae mutant sur2∆ and Philipp Ternes for providing us
with the transgenic sur2∆ cells expressing the ∆4-(E)-sphingo-
lipid desaturase from C. albicans. We are indebted to
Dr Dieter Spiteller and Dr Bernd Schneider for assistance with
analytical aspects and Dr Soojin Ryu for helpful discussions.
Financial support by the Fonds der Chemischen Industrie,
Frankfurt a.M., is gratefully acknowledged.
24 T. Takata, R. Tajima and W. Ando, J. Org. Chem., 1983, 48, 4764–
4766.
25 P. Sperling, P. Ternes, T. K. Zank and E. Heinz, Prostaglandins,
Leukotrienes Essent. Fatty Acids, 2003, 68, 73–95.
26 B. Behrouzian, P. H. Buist and J. Shanklin, Chem. Commun., 2001,
401–402.
References
1 Y. A. Hannun and L. M. Obeid, J. Biol. Chem., 2002, 277, 25847–
25850.
2 T. Kolter and K. Sandhoff, Angew. Chem., Int. Ed., 1999, 38, 1532–
1568.
3 Y. A. Hannun, C. Luberto and K. M. Argraves, Biochemistry, 2001,
40, 4893–4903.
4 A. Huwiler, T. Kolter, J. Pfeilschifter and K. Sandhoff, Biochim.
Biophys. Acta Mol. Cell. Biol. Lipids, 2000, 1485, 63–99.
5 A. H. Merrill, Jr., J. Biol. Chem., 2002, 277, 25843–25846.
6 B. Behrouzian and P. H. Buist, Curr. Opin. Chem. Biol., 2002, 6,
577–582.
27 Y. Jin and J. D. Lipscomb, J. Biol. Inorg. Chem., 2001, 6, 717–
725.
28 D. A. Treco and V. Lundblad, in Current Protocols in Molecular
Biology, eds. F. M. Asubel, R. Brent, R. E. Kingston, D. D. Moore,
J. G. Seidman, J. A. Smith, K. Struhl, L. M. Albright, D. M. Coen,
and A. Varki, John Wiley & Sons, 1993.
29 L. Que and W. B. Tolman, Angew. Chem., Int. Ed., 2002, 41, 1114–
7 A. Svatos, B. Kalinova and W. Boland, Insect Biochem. Mol. Biol.,
1999, 29, 225–232.
1137.
30 D. Parker, J. Chem. Soc., Perkin Trans. 2, 1983, 83–88.
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 2 4 4 8 – 2 4 5 4
2454