Fatty acid desaturases from the microalga Thalassiosira
pseudonana
Thierry Tonon
1,
*, Olga Sayanova
2
, Louise V. Michaelson
2
, Renwei Qing
1,
†, David Harvey
1
,
Tony R. Larson
1
,YiLi
1
, Johnathan A. Napier
2
and Ian A. Graham
1
1 CNAP, Department of Biology, University of York, Heslington, York, UK
2 Rothamsted Research Institute, Harpenden, UK
The algae, as a group, represent the third largest aqua-
culture crop (after freshwater fish and molluscs) in the
world today [1,2]. In recent years, considerable atten-
tion has been directed at marine microalgae for the
production of oils and fatty acids, in particular the use
of algal oils containing long chain polyunsaturated
fatty acids (LCPUFAs). The most prominent of these
are the health beneficial omega-3 eicosapentaenoic acid

AY817153 (TpdesA), AY817154 (TpdesB),
AY817155 (TpdesI) and AY817156 (TpdesK)
(Received 13 March 2005, revised 22 April
2005, accepted 9 May 2005)
doi:10.1111/j.1742-4658.2005.04755.x
Analysis of a draft nuclear genome sequence of the diatom Thalassiosira
pseudonana revealed the presence of 11 open reading frames showing
significant similarity to functionally characterized fatty acid front-end
desaturases. The corresponding genes occupy discrete chromosomal loca-
tions as determined by comparison with the recently published genome
sequence. Phylogenetic analysis showed that two of the T. pseudonana
desaturase (Tpdes) sequences grouped with proteobacterial desaturases that
lack a fused cytochrome b5 domain. Among the nine remaining gene
sequences, temporal expression analysis revealed that seven were expressed
in T. pseudonana cells. One of these, TpdesN, was previously characterized
as encoding a D
11
-desaturase active on palmitic acid. From the six remain-
ing putative desaturase genes, we report here that three, TpdesI, TpdesO
and TpdesK, respectively encode D
6
-, D
5
- and D
4
-desaturases involved
in production of the health beneficial polyunsaturated fatty acid DHA
(docosahexaenoic acid). Furthermore, we show that one of the remaining
genes, TpdesB, encodes a D
8

To date, at least one enzyme corresponding to each
of the front-end desaturases and elongases necessary
to convert a-linolenic acid to EPA has been isolated
from diverse origins [7]. ‘Front-end’ desaturation can
be defined as desaturation between a pre-existing dou-
ble bond and the C-terminal end of a fatty acid, as
opposed to the much more prevalent (in plants)
methyl-directed desaturation. Reconstitution of EPA
biosynthesis has been achieved in yeast [8,9] and in
plants [10,11], with encouraging levels of C20-
LCPUFA production. Moreover, the D
4
-desaturase
gene encoding the last step in DHA biosynthesis has
recently been isolated from a number of marine organ-
isms [12–14]. The final elongation step [of C20 polyun-
saturated fatty acids (PUFAs) to C22] catalysed by a
D
5
-elongase was the last outstanding step remaining to
be functionally characterized at the molecular level.
Very recently, characterization of such an activity has
been described in the microalgae Pavlova lutheri [15],
Ostreococcus tauri and Thalassiosira pseudonana [16].
These novel fatty acid elongases were used to success-
fully reconstitute DHA synthesis in yeast. Therefore,
all the activities are now available to engineer plants to
produce this nutritionally important fatty acid. How-
ever, all these enzymes have been isolated from a
diverse array of organisms, for instance several marine

complete genome sequence revealed the presence of a
family of putative front-end desaturases that are obvi-
ous candidates for enzymes involved in the synthesis of
EPA and DHA [18]. However, rather surprisingly, the
first of these genes to be functionally characterized was
found to encode a cytochrome b5 fusion desaturase
exhibiting D
11
-desaturase activity. Here we report the
cloning and characterization of the three desaturases
involved in DHA synthesis, i.e. a D
6
-, a D
5
- and a
D
4
-desaturase. Moreover, heterologous expression of
an additional cytochrome b5 fusion desaturase has
allowed the identification of a new D
8
-sphingolipid
desaturase from Thalassiosira.
Results
Phylogenetic and expression analysis of
T. pseudonana genes with similarity to front-end
desaturases
A recent phylogenetic analysis of the draft genome
sequence of T. pseudonana [18] reported the presence
of 12 sequences showing significant similarity to func-

functional analysis described in this present study.
Based on information contained in the GenBank data-
base we have allocated 10 of the 11 T. pseudonana
Tpdes sequences to specific chromosomes (Fig. 1A).
These 10 Tpdes sequences are distributed among six of
the 24 chromosomes, with three sequences on chromo-
some 5, two on chromosomes 4 and 6, and one each on
chromosomes 3, 7 and 21. Material used to sequence
the T. pseudonana genome was derived from a single
diploid founder and this revealed the presence of two
haplotypes with on average 0.75% polymorphism at
the nucleotide level [17]. However, the Tpdes genes
occupy distinct chromosomal positions and therefore
even the pairings with highest sequence similarity such
AB
Fig. 1. Evolutionary relationship of T. pseudonana putative desaturases (TpDES) with known front end desaturases and expression analysis
of corresponding genes. T. pseudonana sequences were arbitrarily designated TpDESA to TpDESO. (A) The phylogenetic tree of TpDES and
functionally characterized front-end desaturases was established using the
PHYLIP 3.5c software package and based on 148 alignable amino
acid residues [18]. Percentage bootstrap values above 60 are indicated above the nodes. Chromosome location and availability of cDNAs for
each gene is shown after the gene name. (B) For RT-PCR based gene expression analysis, cells were harvested at various time of incubation
and growth stage monitored by measuring the percentage of nitrogen degraded (inset table). PCR analysis was performed with gene speci-
fic primers on undiluted (lane 1) and five-fold serial dilutions (lanes 2–4) of cDNA. Size of the expected cDNA amplified fragment is indicated
in brackets below the gene name. Inset PCR products from genomic DNA are shown in order to validate the TpdesG and TpdesM primer
pairs. M, DNA molecular size ladder.
T. Tonon et al. Microalgal front-end desaturases
FEBS Journal 272 (2005) 3401–3412 ª 2005 FEBS 3403
as TpdesO-TpdesM, TpdesN-TpdesG and TpdesH-
TpdesL are not due to haplotype variation.
The phylogenetic tree of TpDES sequences was

genes only TpdesO is transcribed (Fig. 1B). The
TpdesO ORF is 1425 bp long and encodes a protein of
474 amino acids. Alignment with the corresponding
genomic sequence confirmed the presence of a 99 bp
intron in the TpdesO gene. The amino acid sequence
of PtDEL5 and TpDESO exhibited 63% identity, sug-
gesting that TpDESO could catalyse the D
5
-desatura-
tion step in the PUFA biosynthetic pathway. The
TpDESK gene sequence forms a subgroup with func-
tionally characterized D
4
-desaturases from Thrausto-
chytrium sp. and Euglena gracilis (Fig. 1A). TpDESK
is expressed at a low level relative to TpdesO during
the exponential phase of growth. TpDESN and
TpDESG do not group with functionally characterized
front-end desaturases from other organisms. TpDESN
has already been characterized as a 16:0 specific
D
11
-desaturase [18]. Alignment of the TpdesG genomic
sequence with other desaturases showed that although
it contains a cytochrome b5 domain and three histidine
boxes a start methionine cannot be determined by
in silico analyses alone, indicating that it may represent
a pseudogene. RT-PCR based expression analysis sug-
gested that the gene is not expressed during the growth
phase and we have not investigated it further. In the

genes. TpdesB and TpdesN do not contain introns,
TpdesK, TpdesO and TpdesI each have a single intron
and TpdesA contains three introns. Full-length cDNAs
are not available for TpdesG and TpdesE therefore
definite information on intron position is not available
for these two genes. Intron ⁄ exon junction is not con-
served between any pair of the Tpdes genes analyzed
and our analysis suggests that introns appear to have
evolved independently in each case.
Characterization of PUFAs front-end desaturases
To establish the function of the different putative
front-end desaturases, the full-length cDNAs of
TpdesA, TpdesB, TpdesI, TpdesK and TpdesO were
cloned into the vector pYES2, under the control of an
inducible galactose promoter, to produce the constructs
pYDESA, pYDESB, pYDESI, pYDESK and pYDES-
O. They were first expressed in the Saccharomyces cere-
visiae strain Invsc1 (or W303A1 for pYDESK).
Transformants were incubated in the presence of a
range of potential fatty acid substrates (18:2
D9,12
,
18:3
D9,12,15
, 20:2
D11,14
, 20:3
D11,14,17
, 20:3
D8,11,14

, respect-
ively, the two major fatty acids found in yeast. When
linoleic (18:2
D9,12
) and a-linolenic (18:3
D9,12,15
) acids
were added to the culture medium, 18:3
D6,9,12
and
18:4
D6,9,12,15
were detected (Fig. 2), confirming that
TpdesI encodes a D
6
-desaturase. Analysis of the fatty
acid composition in the pYDESI transformants after
6 days of incubation showed the percentage conversion
of the 16:1
D9
and 18:1
D9
substrates were 29 and 38%,
respectively. For the exogenous substrates 18:2
D9,12
and 18:3
D9,12,15
, TpDESI exhibited a slight preference
for the omega-3 fatty acid, as 68 and 80% of these
substrates were converted to their corresponding

acids, suggesting that pYDESO does not have a
Fig. 2. GC analysis of FAMEs from yeast transformed with the
empty plasmid pYES2 or the plasmid containing TpDESI. Yeast
cells transformed with either pYES2 (bottom chromatogram) or
pYDESI (top chromatogram) were induced for six days in the pres-
ence of 18:2
D9,12
and 18:3
D9,12,15
exogenously fed before sampling
for fatty acid analysis. New fatty acids are underlined. I.S., internal
standard (17:0). The experiment was repeated twice and results of
a representative experiment are shown.
Fig. 3. GC analysis of FAMEs from yeast transformed with the
empty plasmid pYES2 or the plasmid containing TpDESO. Yeast
cells transformed with either pYDESO (top chromatograms) or
pYES2 (bottom chromatograms) were induced for six days in the
presence of 20:3
D8,11,14
and 20:4
D8,11,14,17
(A), 20:3
D11,14,17
(B), and
20:2
D11,14
(C) exogenously fed before sampling for fatty acid analy-
sis. New fatty acids are underlined. The experiment was repeated
twice and results of a representative experiment are shown.
T. Tonon et al. Microalgal front-end desaturases

(juniperonic acid) and 20:3
D5,11,14
(podo-
carpic acid), respectively (Fig. 3B,C). The percentage
conversion of 20:2
D11,14
and 20:3
D11,14,17
to their D
5
-de-
saturated products was 4.7 and 8.4, respectively, which
was significantly less than the percentage conversion
determined for D
5
-desaturation of the D
8
-desaturated
fatty acids 20:3
D8,11,14
and 20:4
D8,11,14,17
.
Heterologous expression of TpDESK in S. cerevisiae
identified this enzyme as a D
4
-desaturase, as feeding of
transformants with 22:5
D7,10,13,16,19
resulted in the

any ability to desaturate sphingolipid LCBs, resulting
in the appearance of one additional (non-native) LCB
(Fig. 5). This activity was more pronounced when the
pYDESB plasmid was expressed in the yeast sur2D
mutant (which lacks the LCB C-4 hydroxylase Sur2p
and hence trihydroxylated LCBs, Fig. 5A), indicating
a preference for dihydroxylated substrates (Fig. 5C).
The molecular ion for the pYDESB-dependent LCB
had an m ⁄ z of 465, consistent with the identification of
this product as a dihydroxylated long chain base of 18
carbons, containing one double bond (data not
shown). The precise regiospecificity of the activity
encoded by TpdesB was further investigated by comi-
gration with authentic standards for desaturated
dihydroxy-LCBS. This indicated that the novel prod-
uct present on expression of pYDESB in sur2D was
not sphingosine (d18:1D
4t
) (Fig. 5D), but instead comi-
grated with the trans-isomer of d18:1D
8
(Fig. 5B) (as
determined by coinjection with LCBs resulting from
expression of the borage sphingolipid D8-desaturase in
sur2D 23,24);. Thus, TpdesB encodes a sphingolipid D
8
-
desaturase with strong preference for dihydroxylated
substrates. In addition, it appears that the TpDESB
desaturase differs from higher plant orthologs, since it

compared with 7% sphinganine-containing sphingoli-
pids. Thus, 93% of T. pseudonana LCBs contain one
or more double bonds.
Discussion
Using a combination of molecular cloning and bio-
informatics analysis of the available T. pseudonana
genome, we have been able to identify 11 putative
front-end desaturases. Two lacked cytochrome b5
fusion domain and grouped with functionally unchar-
acterized putative proteobacterial desaturases. Among
the remaining nine cytochrome b5 fusion domain con-
taining desaturase sequences, seven were shown to be
transcriptionally active in T. pseudonana cells based on
semiquantitative RT-PCR. We were unable to obtain a
full-length ORF for TpdesE due possibly to problems
of secondary structure in the mRNA. We previously
showed that despite having significant sequence similar-
ity, TpDESN actually encodes a D
11
-desaturase specific
for 16:0 and so cannot be considered as a member of
the front-end desaturase functional class [18]. Of the
five remaining sequences we expected that at least some
if not all of these would encode desaturases involved in
the biosynthesis of EPA and DHA from stearidonic
acid (18:4
D6,9,12,15
). According to the phylogenetic ana-
lysis, TpdesI, TpdesO and TpdesK were good candidates
for genes encoding D

T. pseudonana cells [18] suggests the existence of a
D
6
-desaturase that can act on 16:2
D9,12
to produce the
corresponding D
6
fatty acid 16:3
D6,9,12
. TpDESI is a
good candidate for this activity considering its broad
substrate specificity. However, we were unable to test
this hypothesis by direct feeding experiments as, to our
knowledge, 16:2
D9,12
is not commercially available.
TpDESO acts as a D
5
-desaturase on C20 fatty acids
to produce 20:4
D5,8,11,14
and 20:5
D5,8,11,14,17
as predicted
from the clustering of the gene in the phylogenetic tree.
This enzyme is also able to introduce a double bond in
a nonmethylene interrupted pattern at the D
5
-position

formed with TpdesB. Total LCBs were extracted from yeast sur2D
expressing pYDESB, derivatized and separated as described. (A)
LCB profile from yeast mutant sur2D which synthesizes only
dihydroxylated LCBs (e.g. d18:0 ¼ dihydroxylated 18 carbon LCB,
saturated). (B) LCB profile from sur2D yeast expressing the stereo-
unselective borage sphingolipid D
8
-desaturase: note the presence
of cis and trans D
8
-desaturated dihydroxylated LCBs (inset). (C)
LCB profile from sur2D yeast expressing pYDESB: note the pres-
ence of only the trans isomer of the D
8
-desaturated dihydroxylated
LCB. (D) The LCB profile of pYDESB was coinjected with a deriva-
tized authentic standard for sphingosine (d18:1
D4t
): note that
sphingosine does not coelute with the novel D
8t
-LCB which arises
from pYDESB expression.
T. Tonon et al. Microalgal front-end desaturases
FEBS Journal 272 (2005) 3401–3412 ª 2005 FEBS 3407
preferred substrate it is not possible to reach a conclu-
sion on the substrate preference of this enzyme for
omega-3 vs. omega-6 fatty acids with the current data.
TpDESI and TpDESO displayed almost no selectivity
between the omega-3 and omega-6 fatty acid sub-

from Euglena gracilis [6]. In this alternate pathway, a
front end D
8
-desaturase acts on 20:2
D11,14
(eicosadi-
enoic acid) and 20:3
D11,14,17
(eicosatrienoic acid) to
produce 20:3
D8,11,14
and 20:4
D8,11,14,17
. The E. gracilis
D
8
-desaturase is in the same subgroup as TpDESI and
TpDESE in the phylogenetic tree (Fig. 1). We have
functionally characterized TpDESI in the current work
but have been unable to characterize TpDESE. The
fact that we did not detect eicosadienoic acid in
T. pseudonana cells, and only very low level of eicos-
atrienoic acid were measured suggests the D
8
-desatu-
rase alternate pathway is not present in this organism.
Furthermore, acyl-CoA profiling of T. pseudonana
cells, did not detect 20:2
D11,14
or 20:3

the TpDESB sphingolipid desaturases and the predom-
inant form of the enzyme found in higher plants.
Firstly, TpDESB has a strong preference for dihydrox-
ylated LCB substrates (i.e. sphinganine), whereas
almost all higher plant sphingolipid D
8
-desaturases dis-
play greater activity towards trihydroxylated LCBs (i.e.
phytosphingosine) [30]. A recent example of a higher
plant sphingolipid desaturase with activity towards
sphinganine was reported from Aquilegia vulgaris [24].
The introduction of the D
8
-desaturation into dihydroxy-
lated substrates may represent the first step in the syn-
thesis of sphingadienine-containing sphingolipids (i.e.
containing d18:2D
4t,8c ⁄ t
LCBs), by the subsequent
D
4
-desaturation of the D
8
-desaturated LCB. This bio-
synthetic route has been invoked to explain the absence
of sphingosine (i.e. D
4
-desaturated dihydroxysphingo-
sine) in many plant species, even though higher plant
LCB D

first cloned example of stereo-selective sphingolipid D
8
sphinganine desaturase; previous examples of the
higher plant sphingolipid desaturases with this regio-
specificity are stereo-unselective in terms of the double
bond introduced (i.e. a nonequal mixture of cis and
trans configurations), though a stereo-specific D
8t
phy-
tosphingosine desaturase has been reported from the
yeast Kluyveromyces lactis [32] (see Sperling and Heinz,
2003 [30] for an excellent review of the topic of LCB
desaturation). The enzymatic basis for this higher plant
stereo-unselective is currently unclear, but has been
hypothesized to result from a syn-elimination of two
vicinal hydrogen atoms from two different substrate
conformers, making this form of sphingolipid LCB
desaturation distinct from the D
4
-desaturation which
Microalgal front-end desaturases T. Tonon et al.
3408 FEBS Journal 272 (2005) 3401–3412 ª 2005 FEBS
yields sphingosine [30]. In that respect, it is perhaps
surprising to find a more ‘precise’ form of stereo-speci-
fic LCB desaturation in the unicellular diatom T. pseu-
donana, which may be due to an as yet unknown role.
It is also currently unclear as to the role of sphingo-
lipid LCB D
8
-desaturation, though it has been hypo-

-desaturases could prove problematic as it would be
preferable to limit the introduced enzymatic activities
to those essential for EPA and DHA production in
order to avoid the presence of additional fatty acids in
an end product. Nevertheless, this set of desaturases
along with the recently characterized D
5
-elongase from
the same organism represents an attractive biotechno-
logical resource. As highlighted in recent publications
[10], a critical issue for the development of a commer-
cially viable product will be the final yield of EPA and
DHA in the engineered vegetable oil and this will most
likely require the introduction of additional activities
such as acyltransferases and acyl-CoA synthetases [35].
Further mining of the T. pseudonana genome should
lead to identification of genes encoding these addi-
tional enzyme activities.
Experimental procedures
Cultivation of T. pseudonana, RNA extraction and
RT-PCR analysis of gene expression
T. pseudonana was cultivated as described previously [18].
Total RNA was extracted from cells harvested at different
stages of growth using an RNeasy plant mini kit (Qiagen,
Valencia, CA, USA). First-strand cDNA was synthesized
from three lg of DNAse treated RNA using a Prostar
First-strand RT-PCR kit (Stratagene, La Jolla, CA, USA).
PCR with primer pairs specific to each T. pseudonana
desaturase gene (Table 1) were performed using gDNA, or
undiluted and five-fold serial dilutions of cDNAs as fol-

ment for TpdesO, TpdesM, TpdesN and TpdesG was veri-
fied by sequencing after cloning in the pGEM-T EasyVec-
tor (Promega, Madison, WI, USA).
5¢- and 3¢-RACE experiment
The GeneRacer
TM
kit (Invitrogen, Carlsbad, CA, USA)
was used to reverse transcribe T. pseudonana RNA and
cDNA was used to amplify the 5¢-end of TpdesE and the
3¢-end of the TpdesB gene. Fragments generated by nested
PCR were cloned into the pGEM-T EasyVector (Promega)
and sequenced.
Functional characterization of T. pseudonana
putative front-end desaturases in Saccharomyces
cerevisiae
cDNA of the entire desaturase coding region was synthes-
ized from T. pseudonana RNA using the SuperScript
TM
III
RNase H
–
Reverse Transcriptase (Invitrogen) or the
Enhanced Avian Reverse Transcriptase (Sigma) and gene
specific primers pairs (Table 1). Forward primers for
TpdesA, TpdesB, TpdesI and TpdesO gene were designed to
contain an alanine codon (GCT) just downstream of the
start codon not present in the original algal sequences.
Presence of a G at position +4 has been shown to improve
translation initiation in eukaryotic cells [36]. In the case of
TpDesK, activity was detected in S. cerevisiae when a full

USA). Expression of the transgene was induced at D
600
¼
0.2–0.3 by supplementing galactose to 2% (w ⁄ v). At that
time, the appropriate fatty acids were added to a final con-
centration of 50 lm. Incubation was carried out at 25 °C
for 3 days and then 15 °C for another 3 days. For the co-
feeding experiment, the same conditions were applied,
except that both substrates were added to 25 lm final con-
centration. Each feeding experiment was repeated twice,
and FA analysis was carried out on triplicate samples.
For functional characterization of Tpdes genes in the
sur2D background, cultures were grown at 22 °C with sha-
king in the presence of 2% (v ⁄ v) raffinose and induction
was carried out as previously described [37]. All cultures
were then grown for a further 48 h unless indicated. All
analysis was performed on triplicate samples and replicated
three times.
Fatty acid analysis
Microalgae or yeast cells were harvested by centrifugation.
Total fatty acids were extracted and transmethylated as
previously described [14]. Fatty acid methyl esters (FAMEs)
of methyl pentadecanoate (15:0) or methyl heptadecanoate
(17:0) were included as internal standards to enable quanti-
fication. PUFA FAMEs were identified by comparing chro-
matographic traces with transmethylated commercial
Menhaden oil (Supelco, Gillingham, Dorset, UK), and by
identification of picolinyl ester and dimethyl disulphide
adduct structures by GCMS as previously described [18].
Sphingoid base analysis

United Nations, Rome, Italy.
3 Tonon T, Harvey D, Larson TR & Graham IA (2002)
Long chain polyunsaturated fatty acid production and
partitioning to triacylglycerols in four microalgae. Phy-
tochemistry 61, 15–24.
4 Metz JG, Roessler P, Facciotti D, Levering C, Dittrich F,
Lassner M, Valentine R, Lardizabal K, Domergue F,
Yamada A et al. (2001) Production of polyunsaturated
fatty acids by polyketide synthases in both prokaryotes
and eukaryotes. Science 293, 290–293.
5 Qi B, Beaudoin F, Fraser T, Stobart AK, Napier JA &
Lazarus CM (2002) Identification of a cDNA encoding
a novel C18-Delta(9) polyunsaturated fatty acid-specific
elongating activity from the docosahexaenoic acid
(DHA)-producing microalga, Isochrysis galbana. FEBS
Lett 510, 159–165.
6 Wallis JG & Browse J (1999) The Delta8-desaturase of
Euglena gracilis: an alternate pathway for synthesis of
20-carbon polyunsaturated fatty acids. Arch Biochem
Biophys 365, 307–316.
7 Sayanova O & Napier JA (2004) Eicosapentaenoic acid:
biosynthetic routes and the potential for synthesis in
transgenic plants. Phytochemistry 65, 147–158.
8 Beaudoin F, Michaelson LV, Hey SJ, Lewis MJ,
Shewry PR, Sayanova O & Napier JA (2000) Heterolo-
gous reconstitution in yeast of the polyunsaturated fatty
acid biosynthetic pathway. Proc Natl Acad Sci USA 97,
6421–6426.
9 Domergue F, Lerchl J, Za
¨

Identification of a very long chain polyunsaturated fatty
acid D4-desaturase from the microalga Pavlova lutheri.
FEBS Lett 553, 440–444.
15 Pereira SL, Leonard AE, Huang Y-S, Chuang L-T &
Mukerji P (2004) Identification of two novel microalgal
enzymes involved in the conversion of the x3-fatty acid,
eicosapentaenoic acid, into docosahexaenoic acid. Bio-
chem J 384, 357–366.
16 Meyer A, Kirsch H, Domergue F, Abbadi A, Sperling P,
Bauer J, Cirpus P, Zank TK, Moreau H, Roscoe TJ
et al. (2004) Novel fatty acid elongase and their use for
the reconstitution of docosahexaenoic acid biosynthesis.
J Lipid Res 45, 1899–1909.
17 Armbrust EV, Berges JA, Bowler C, Green BR, Marti-
nez D, Putnam NH, Zhou S, Allen AE, Apt KE, Bech-
ner M et al. (2004) The genome of the diatom
Thalassiosira pseudonana: ecology, evolution, and meta-
bolism. Science 306, 79–86.
18 Tonon T, Harvey D, Qing R, Li Y, Larson TR & Gra-
ham IA (2004) Identification of a fatty acid A
¨
11-desa-
turase from the microalga Thalassiosira pseudonana.
FEBS Lett 563, 28–34.
19 Sperling P, Ternes P, Zank TK & Heinz E (2003) The
evolution of desaturases. Prostaglandins Leukot Essent
Fatty Acids 68, 73–95.
20 Watts JL & Browse J (1999) Isolation and characteriza-
tion of a delta5-fatty acid desaturase from Caenorhabdi-
tis elegans. Arch Biochem Biophys 362, 175–182.

hringer U, Reski R & Heinz E
(1998) Identification of a novel D6-acyl-group desaturase
by targeted gene disruption in Physcomitrella patens.
Plant J 15, 39–48.
27 Sayanova O, Davies GM, Smith MA, Griffiths G,
Stobart AK, Shewry PR & Napier JA (1999) Accumula-
tion of D
6
-unsaturated fatty acids in transgenic tobacco
plants expressing a D
6
-desaturase from Borago
officinalis. J Exp Bot 50, 1647–1652.
28 Pereira SL, Leonard AE & Mukerji P (2003) Recent
advance in the study of fatty acid desaturases from ani-
mals and lower eukaryotes. Prostaglandins Leukot
Essent Fatty Acids 68, 97–106.
29 Napier JA, Michaelson LV & Dunn TM (2002) A new
class of lipid desaturase central to sphingolipid bio-
synthesis and signalling. Trends Plant Sci 7, 475–478.
30 Sperling P & Heinz E (2003) Plant sphingolipids: struc-
tural diversity, biosynthesis, first genes and functions.
Biochim Biophys Acta 1632, 1–15.
31 Merrill AH Jr & Sweeley CC (1996) Biochemistry of
Lipids, Lipoproteins and Membranes (Vance DE &
Vance J, eds), pp. 309–339. Elsevier Science B.V.,
Amsterdam, the Netherlands.
32 Takakuwa N, Kinoshita M, Oda Y & Ohnishi M (2002)
Isolation and characterization of the genes encoding
D