In addition to the common fatty acids, plants produce a variety of unusual fatty acids such as hydroxy, epoxy or acetylenic fatty acids . Unsaturated fatty acids are formed by desaturases, which catalyze the introduction of double bonds into preformed acyl chains by removal of a pair of hydrogens, concomitant oxidation of an electron donor and reduction of O2[2,3]. The cloning and sequencing of desaturases from different organisms identified two distinct groups, soluble acyl-ACP desaturases and membrane-bound desaturases, which differ in their consensus motifs . The membrane-bound desaturases contain three characteristic histidine motifs, which are believed to coordinate a di-iron cluster in the active site. The comparison of amino-acid sequences has revealed, that these desaturases may be grouped according to differing regioselectivity, which presently spans from C5–C15 . Genes that encode enzymes closely related to the microsomal Δ12-desaturases, have been cloned and found to catalyse exotic modifications at or around the Δ12 position of the acyl chain, reactions such as hydroxylation, epoxidation and acetylenation. A Δ12-hydroxylase gene, that encodes an enzyme responsible for the production of 12-hydroxy-9-(Z)-octadecenoic acid (ricinoleic acid) from oleic acid, was identified by mass sequencing of cDNAs from Ricinus communis. On the basis of homology, the gene encoding the Lesquerella fendleri oleate Δ12-hydroxylase was subsequently isolated. Unlike the enzyme from R. communis, the L. fendleri had both hydroxylase and desaturase activities . Site directed mutagenesis of the Lesquerella oleate Δ12-desaturase has shown that as few as four amino-acid substitutions can convert a strict desaturase to a bifunctional hydroxylase . In Crepis rubra and Crepis alpina linoleic acid and possibly oleic acid are substrates for the synthesis of 9-(Z)-octadecen-12-ynoic acid (crepenynic acid) [7,8]. In the moss Ceratodon pupureus 9,12-(Z,Z)-octadecadien-6-ynoic and 9,12,15-(Z,Z,Z)-octadecatrien-6-ynoic acid are formed by the generation of an acetylenic bond at the pre-existing Δ6cis-double bond of a C18-fatty acid . Recently, a Δ12-acetylenase from Crepis alpina and a Δ12-epoxidase from Crepis palaestina were cloned and functionally expressed in yeast and Arabidopsis thaliana. Both of these enzymes were similar in biochemical properties and sequence to the microsomal Δ12-desaturases. Therefore, at least four reactions (desaturation, hydroxylation, epoxidation and acetylenation) have been shown to be catalyzed by closely related desaturase-like enzymes . More recently, it has been shown that even the conversion of linoleic acid (9Z,12Z-18:2) to conjugated octadecatrienoic acids such as α-eleostearic acid (9Z,11E,13E-18:3) and calendic acid (8E,10E,12Z-18:3) is catalyzed by Δ12-desaturase-like enzymes [10,11]., Regioselectivity appears to place greater constraints on the sequence of fatty-acid desaturase and desaturase-related enzymes than actual reaction outcome (i.e. formation of cis- or trans-double bonds, triple bonds, epoxy or hydroxy groups) . A comparison of sequences from cloned desaturases has revealed, that Δ5-, Δ6- and Δ8-desaturases all fall into the same regioselectivity group [3,12]. We speculated, that an enzyme that could catalyze acetylenation at C6 is likely to be related to the Δ6-desaturases, of which most are cytochrome b5-fusion proteins . This predicted similarity was used to clone an acetylenase with a regioselectivity different from the Δ12 position. Here we report the PCR-based isolation of a bifunctional Δ6-acetylenase/desaturase and a Δ6-fatty-acid desaturase from the moss C. pupureus. Their expression in Saccharomyces cerevisiae and feeding experiments with different fatty-acid substrates, has led to their functional identification.
Many plant genes have been cloned that encode regioselective desaturases catalyzing the formation of cis-unsaturated fatty acids. However, very few genes have been cloned that encode enzymes catalyzing the formation of the functional groups found in unusual fatty acids (e.g. hydroxy, epoxy or acetylenic fatty acids). Here, we describe the characterization of an acetylenase from the moss Ceratodon purpureus with a regioselectivity differing from the previously described Δ12-acetylenase. The gene encoding this protein, together with a Δ6-desaturase, was cloned by a PCR-based approach with primers derived from conserved regions in Δ5-, Δ6-fatty-acid desaturases and Δ8-sphingolipid desaturases. The proteins that are encoded by the two cloned cDNAs are likely to consist of a N-terminal extension of unknown function, a cytochrome b5-domain, and a C-terminal domain that is similar to acyl lipid desaturases with characteristic histidine boxes. The proteins were highly homologous in sequence to the Δ6-desaturase from the moss Physcomitrella patens. When these two cDNAs were expressed in Saccharomyces cerevisiae, both transgenic yeast cultures desaturated Δ9-unsaturated C16- and C18-fatty acids by inserting an additional Δ6cis-double bond. One of these transgenic yeast clones was also able to introduce a Δ6-triple bond into γ-linolenic and stearidonic acid. This resulted in the formation of 9,12,15-(Z,Z,Z)-octadecatrien-6-ynoic acid, the main fatty acid found in C. pupureus. These results demonstrate that the Δ6-acetylenase from C. pupureus is a bifunctional enzyme, which can introduce a Δ6cis-double bond into 9,12,(15)-C18-polyenoic acids as well as converting a Δ6cis-double bond to a Δ6-triple bond.
distortionless enhancement by polarization transfer
fatty-acid methyl ester
gas-liquid chromatography mass spectrometry
heteronuclear multiple bond correlation
heteronuclear multiple quantum coherence
Materials and methods
Single-stranded cDNA and cDNA library
Poly(A)+ RNA was isolated from about 400 mg of C. purpureus protonemata (a gift of P. Beutelmann, Institut für Allgemeine Botanik, Universität Mainz, Germany) with a Quickprep Micro (Amersham Pharmacia Biotech Inc.) according to the manufacturer's instructions. The mRNA was precipitiated and used to synthesize first-strand cDNA with oligo dT(12−18) and MMLV reverse transcriptase (Gibco BRL). This cDNA was used directly for PCR. The moss λ-ZAP cDNA library was a generous gift from F. Thümmler (Botanisches Institut, Universität München, Germany) and was prepared as follows: total RNA was extracted from light-grown and 7-day dark adapted C. pupureus wt3 protonemata and from a 1:1 mixture of this RNA poly(A)+ RNA was prepared. cDNA, prepared from this poly(A)+ RNA, was used to construct a directional cDNA library with EcoRI/XhoI-cut uni-ZAP XR vector arms (Stratagene) .
PCR amplification with degenerated primers
For a PCR-based cloning, single-stranded cDNA was used as a template. One PCR fragment of 557 bp (Cer3) was amplified with the degenerated forward primer A 5′-TGGTGGAA(A/G) TGGA(A/C)ICA(C/T)AA-3′ and the reverse primer C 5′-AT(A/T/G/C)T(T/G)(A/T/G/C)GG(A/G)AA(A/T/G/C)A(A/G) (A/G)TG(A/G)TG-3′. These primers were derived from the amino-acid sequence WWKW(N/T/K)H(N/K) and (I/M)(H/Q/N) PF(L/F)HH, respectively. A second 560 bp fragment (Cer1) was amplified with another forward primer B 5′-(T/G)GITGGAA(A/G)(T/G)(G/A)I(A/C)AICA(C/T)AA-3′ derived from the amino-acid sequence (G/W)WK(E/D/W)(N/Q/K)H(N/K) and the reverse primer C, mentioned above. The PCR amplification was carried out using the following program: 10 min denaturation at 94 °C, brake for ‘hot start’ with Taq DNA polymerase at 72 °C, followed by 32 cycles of 20 s at 94 °C, 1 min at 45 °C, 1 min at 72 °C and terminated by 10 min extension at 72 °C. The PCR fragments of the expected length (500–600 bp) were ligated into pGEM-T (Promega) and transformed into Escherichia coli Xl1blue MRF′ Kan (Stratagene). Plasmid-DNA minipreparations  p1 and p3 were sequenced by the dideoxy chain termination method using the ABI Prism Big Dye Terminator Cycle Sequencing Ready Reaction Kit (Perkin-Elmer, Weiterstadt).
cDNA library screening for full length clones
The inserts from the plasmids p1 and p3 were excised with NcoI and PstI and gel purified. These were radiolabelled by random priming with [α-32P]dCTP and used together to screen the λ ZAP cDNA library from C. pupureus wt3 protonemata by hybridization. Filters were hybridized at 55 °C and washed at 55 °C with 1 × NaCl/Cit and 0.1% SDS. Individual plaques, with sequences that corresponded to either p1 or p3, were further identified, and distinguished from each other, by PCR analysis with the primer pairs 5′-CGAATGAGTGCGACGAAC-3′ and 5′-AATAACCTGGGCTCTCAC-3′ (for the p1 sequence), and 5′-ATGAGGATATTGATACTCTC-3′ and 5′-GCAATCTGGGCATTCACG-3′ (for the p3 sequence). To identify full length cDNA clones individual plaques were analyzed by PCR with the T7 and T3 primer to identify clones containing the longest inserts. Two λ clones, one representing the 1p sequence and the other the 3p sequence, were investigated further. These had inserts of ≈ 2.2 kb in size. The corresponding bacteriophages were isolated by precipitation with poly(ethylene glycol) and from these DNA was isolated. The inserts were excised with EcoRI and KpnI and cloned into the EcoRI and KpnI sites of pUC19, resulting in plasmids pCer1 and pCer3 and sequenced on both strands, which corresponded to the sequences 1p and 3p, respectively.
Expression in S. cerevisiae
The ORFs of pCer1 and pCer3 were cloned behind the galactose-inducible promotor GAL 1 of the yeast expression vector pYES2 (Invitrogen). In order to achieve this, a new 5′ KpnI (upstream of the deduced translation start) and a 3′ EcoRI site (downstream the stop codon) were introduced into the inserts of pCer1 and pCer3 by PCR. These modified inserts were digested with KpnI and EcoRI and ligated into the corresponding restriction sites of pYES2 to yield pYCer1 and pYCer3. Their sequences were verified by DNA sequencing, as described above. These plasmids and the empty vector pYES2 were transformed into S. cerevisiae strain INVSc1 (Invitrogen) using the polyethylenglycol method .
Transformed cells were grown in complete minimal-dropout-uracil medium (CMdum) supplemented with 2% raffinose as the only carbon source  and 1% Tergitol NP-40 [(w/v); Sigma] for the solubilization of fatty acids (Sigma) . For expression experiments, test cultures in CMdum medium supplemented with 0.003% of the corresponding fatty acid [(w/v); stock solution solubilized in 5% Tergitol] were aerobically grown to logarithmic phase, then induced with 1.8% galactose [(w/v); final concentration] and finally grown to saturation for 24 h at 30 °C.
Gas-liquid chromatography mass spectrometry (GC-MS) analysis of fatty acids
C. pupureus and Dicranum scoparium cultures (generous gifts from E. Hartmann, Institut für Biologie, Freie Universität Berlin, Germany and from P. Beutelmann, Institut für Allgemeine Botanik, Universität Mainz, Germany) as well as cell pellets (≈ 100 mg) from wild-type (pYES2) and transgenic yeast (pYCer1 and pYCer3) were subjected to acid methanolysis and fatty-acid methyl esters (FAME) were obtained, as described before . FAME were analyzed by gas-liquid chromatography on a polar capillary column (Chrompack, WCOT Fused Silica, CP-Wax-52 CB, 25 m, 0.32 mm) with the following temperature program: 1 min at 170 °C, linear temperature gradient (5 °C·min−1) up to 240 °C and 5 min at 240 °C. FAME identities were confirmed by comparison with appropriate reference substances (Sigma).
For the determination of double or triple bond positions, FAME (about 200 µg) were converted into their 4,4-dimethyloxazoline (DMOX) derivatives as described . GC-MS of the FAME and DMOX derivatives of the fatty acids was carried out with a Hewlett-Packard Model 5989 equipped with a nonpolar capillary column (HP-MS) using a temperature gradient from 150 °C to 320 °C with a temperature rise of 5 °C·min−1. EI-mass spectra were recorded at 70 eV and CI-mass spectra were obtained at 105 eV with ammonia as reactant gas. The ion-source temperature in both cases was 250 °C. The EI-mass spectra of the DMOX derivatives are shown in Fig. 3, whereas the spectrum of the 18:3A methylester is listed here with m/z-values and relative intensities in brackets: M+(–), 257 (2), 173 (17), 159 (7), 145 (24), 131 (40), 119 (20), 117 (62), 105 (46), 91 (100), 79 (72), 67 (60), 55 (45).
For the preparation and isolation of pure FAME in mg quantities suitable for NMR analysis C. pupureus and D. scoparium cultures (500 mg and 300 mg fresh weight, respectively) were subjected to acid methanolysis as described . Both mixtures containing significant amounts of 18:3A of FAME (as detected by GC) were further separated by preparative TLC using silicagel 60 plates (20 × 20 cm, thickness 0.25 mm, Merck) in petroleum ether/dietyl ether (9:1, v/v). FAME and acetylenic acid methyl esters (Rf values 0.3 and 0.18, respectively) were detected by UV light after spraying the plates with 0.2% (w/v) methanolic 8-anilinonaphthalene-1-sulfonic acid. From these plates acetylenic acid methyl esters were eluted with petroleum ether resulting in 1.4 mg of 18:3A methylester (purity 98% as detected by GC) from C. pupureus and 1.2 mg 18:3A methylester (98%) from D. scoparium.
Both 1D 1H- and 2D homonuclear and H-detected heteronuclear 1H,13C-correlation spectra [heteronuclear multiple quantum coherence (HMQC), and heteronuclear multiple bond correlation (HMBC)] were recorded on a 600-MHz spectrometer (Bruker Avance DRX-600). 13C-NMR and distortionless enhancement by polarization transfer (DEPT) spectra were run at 90.6 MHz (Bruker DPX-360). The methyl ester of 18:3A (isolated from C. pupureus and D. scoparium, as mentioned above) was dissolved in 500 µL CDCl3 (99.96%, Cambridge Isotope Laboratories, Andover, MA, USA) and NMR spectra were recorded at 300 K with internal TMS (δH = 0.000) or CDCl3 (δC = 77.0) as reference. 1D (1H-, 13C, and DEPT) and 2D homonuclear (1H,1H COSY) and H-detected heteronuclear (1H,13C HMQC, 1H,13C HMBC) experiments were performed using standard Bruker software (xwinnmr, version 2.6) and resulted in the following assignment for 9,12,15-(Z,Z,Z)-octadecatrien-6-ynoic acid (18:3A) methylester: 1H-NMR (600 MHz, CDCl3): δ 0.977 (3H, t, J = 7.5 Hz, H-18), 1.517 (2H, m, H-4), 1.724 (2H, t, J = 7.2, 10.0 Hz, H-3), 2.077 (2H, m, H-17), 2.175 (2H, tt, J = 3.3 Hz, H-5), 2.328 (2H, t, J = 7.5 Hz, H-2), 2.815 (4H, m, H-11 and H-14), 2.932 (2H, m, H-8), 3.669 (3H, s, Me at C-1), 5.27–5.48 (6H, m, H-9, H-10, H-12, H-13, H-15, and H-16); 13C-NMR (90.5 MHz, CDCl3): δ 14.2 (C-18), 17.2 (C-8), 18.5 (C-5), 20.6 (C-17), 24.2 (C-3), 25.5 (C-11 and C-14), 28.4 (C-4), 33.6 (C-2), 51.4 (Me at C-1), 78.5 (C-7), 79.5 (C-6), 125.4 (C-9), 126.9 (C-15), 127.4 and 128.8 (C-12 and C-13), 129.2 (C-10), 132.1 (C-16), and 173.9 (C-1).
PCR based cloning
Degenerated primers matching the three conserved histidine regions  were deduced from the aligned amino-acid sequences of cloned Δ5-, Δ6-acyl lipid desaturases and Δ8-sphingolipid desaturases (EMBL accession numbers: Z81122, U79010, AJ222980, AF031477, X87143, AJ224160, AJ224161) . Single-stranded cDNA from the moss C. pupureus was used as a template for PCR. A number of PCR fragments, that were of the expected length, were cloned and sequenced. Databank searches and alignments with these new sequences revealed that two of the fragments (1p and 3p) had similarities to Δ5- and Δ6-acyl lipid desaturases. The deduced amino-acid sequences of these partial sequences had 64% amino-acid identity to each other, and 74% (3p) and 63% (1p) identity to the Δ6-desaturase from the moss Physcomitrella patens. Because sequence 3p (557 bp) was more similar to the Physcomitrella enzyme than sequence 1p (560 bp), we assumed that the 3p fragment might be derived from a gene encoding a Δ6-desaturase and the 1p fragment from a gene encoding a Δ6-acetylenase of C. pupureus.
Isolation of full-length cDNA
To isolate full-length cDNA clones, the inserts from 1p and 3p were isolated, radiolabelled and used to screen a cDNA library  of light-grown and 7-day dark adapted protonemata. On a plate containing about 25 000 plaque-forming units, approximately 80 hybridizing plaques were observed. The analysis of individual hybridizing plaques with primers, specific to either the 1p or 3p sequence and vector DNA flanking the cDNA inserts were used to identify clones that contained the longest inserts (see Materials and methods). Two λ clones were taken for further analysis. As we were not able to excise phagmids form the cDNA library, inserts from these clones were subcloned into the plasmid pUC19 to give the plasmids pCer1 and pCer3. The inserts from these plasmids were sequenced on both strands. Plasmid pCer1 contained a cDNA of size 2003 bp, excluding its poly(A) tail, and encoded a putative ORF of 483 amino acids between the nucleotide positions 176–1624 (CPAcet6; Fig. 1). Two in-frame stop codons preceeded CPAcet6, which indicated that the sequence upstream of the initiation codon, deduced for CPAcet6, is likely to be the 5′-untranslated region. The pCer3 cDNA had a length of 2142 bp, but no stop codon was detected in its 5′-untranslated region. The nucleotide sequence between positions 159–1718 encoded an ORF of 520 amino acids, which was designated CPDes6 (Fig. 1).
Comparison with other desaturases
Both the predicted proteins, CpDes6 and CpAcet6 (Fig. 1), were slightly shorter than the Δ6-desaturase from Physcomitrellapatens (PpDes6), which has an ORF of 525 amino acids . A comparison of cytochrome b5 fusion desaturase sequences  indicated that PpDes6 consists of a hydrophilic cytochrome b5 domain (from residue 96–208), followed by a C-terminal desaturase domain. The CPDes6 and CPAcet6 sequences were most identical (67% and 52%, respectively) to PpDes6 . They are 33% identical with the Δ6-desaturase from the fungus Mortierella alpina and only 21–24% identical to the Δ6-desaturase from Borago officinalis, Caenorhabditis elegans, mammals (, EMBL accession number AB021980) and cyanobacteria [25,26].
The desaturase domains of the two proteins have the three conserved histidine regions characteristic to all membrane-bound desaturases . In addition, the sequence motif QIEHHLFPXMPRXN of the third histidine region is present in these two C. pupureus proteins as well as in all the Δ6-desaturases with an N-terminal cytochrome b5 fusion. In the Δ6-desaturases from cyanobacteria [25,26], which lack a cytochrome b5 domain, the equivalent motif is reduced to QXXHHLFP. The exchange of histidine to glutamine present in the third histidine cluster seems to be a common feature of all Δ5-, Δ6- and Δ8-acyl lipid desaturases and Δ8-sphingolipid desaturases . Sequence comparisons of the desaturase domain revealed 17 highly conserved amino acids in addition to the seven invariant histidines and the glutamine, which are present in the Δ6-acetylenase and in all Δ6-desaturases, known so far (Fig. 1, underlined). Therefore, the signature for the conserved motif of Δ6-desaturases must be amended to (W/F)X1−2H(D/E)XXHX20−22GXSX3WX3HXX1−2HHX3 NX116−134HX11−20(W/F)X3QX14−15WX2GXLX2QXXHHLFP X17CX6Y.
The cytochrome b5-domain of both proteins contained the eight invariant residues that are characteristic for the cytochrome b5 superfamily . The deviations in sequence, that are described for the Helianthus annuus cytochrome b5-containing fusion protein  were also found in CpDes6 and CpAcet6. A comparison of N-terminal cytochrome b5 domains revealed five additional amino acids apart from those eight invariant residues mentioned above, which are highly conserved in all Δ6-fusion desaturases (Fig. 1, underlined). Both CpDes6 and CpAcet6 had putative N-terminal extensions of about 89 and 52 residues, respectively. The function of this sequence, which is also present in the Δ6-desaturase from the moss P. patens (PpDes6; Fig. 1) is unknown . However, if this N-terminal extension codes for a separate domain, the cytochrome b5 domain of these proteins would represent an internal domain.
The comparison of the amino-acid sequence of the Δ6-acetylenase with the Δ6-desaturase sequences from C. pupureus, P. patens, M. alpina, B. officinalis, C. elegans, mammals (, EMBL accession number AB021980) and cyanobacteria [25,26] revealed no strictly conserved residues common to all of these desaturases, but absent in the acetylenase. When the Δ6-acetylenase/desaturase domain was compared with the two moss Δ6-desaturase domains (starting at position 209; Fig. 1), 56 amino-acid substitutions were identified. Of these substitutions, 43 were changes between amino acids that were significantly different in chemical property. In contrast to this, only seven strictly conserved amino-acid changes could be identified in the sequences of the Δ12-fatty-acid hydroxylases when compared to the Δ12-fatty-acid desaturases . As few as four to six amino acids were adequate to change the function of these proteins from a desaturase to a hydroxylase or vice versa. The Δ12-acetylenase from Crepis alpina deviates from Δ12-desaturases in 29 positions. Only four of these changes are conservative substitutions. Taken together, these data suggest, that a switch from desaturase to acetylenase might involve more extensive changes in sequence than what is required to change a desaturase to a hydroxylase and vice versa. However, it is unclear to what degree these changes are required to facilitate merely substrate recognition (as the acetylenases, in contrast to the desaturases and hydroxylases, presumably recognize substrates with a double bond already present at the Δ6 or Δ12 position) or a differing reaction outcome.
Functional expression in S. cerevisiae
For a functional identification, both CpDes6 and CpAcet6 were expressed in S. cerevisiae. The cDNAs CpAcet6 and CpDes6 were cloned into a yeast expression vector to give pYCer1 and pYCer3, respectively. These, and the empty vector pYES2 were transformed into S. cerevisiae strain INVSc1. Transformed yeast cells were grown to saturation after induction with 1.8% galactose in minimal medium. S. cerevisiae contains only saturated and monoenoic fatty-acid substrates, but no di-, tri- or polyenoic fatty acids which may be required for Δ6-desaturation or Δ6-acetylenation. Therefore, the medium in which the yeast cells were grown was supplemented with linoleic (18:2Δ9,12), α-linolenic (18:3Δ9,12,15), γ-linolenic (18:3Δ6,9,12) and stearidonic acid (18:4Δ6,9,12,15). In subsequent analyses of total fatty acids recovered from pYCer1- or pYCer3-transformed cells, 16:2Δ6,9, 18:2Δ6,9, 18:3Δ6,9,12 and 18:4Δ6,9,12,15 were detected, depending on the fatty-acid precursors offered (Table 1). Yeast with pYCer1 also produced the Δ6-acetylenic fatty acids 18:2AΔ6a,9,12 and 18:3AΔ6a,9,12,15 (see below), whereas control cells harbouring pYES2 produced none of these fatty acids. These results confirmed, that CpDes6 encodes a Δ6-desaturase and CpAcet6 an enzyme able to function as both a Δ6-acetylenase and a Δ6-desaturase.
|% of total fatty acids|
To elucidate the substrate specificity of CpDes6 and CpAcet6, transgenic yeast cultures were fed with different fatty-acid substrates. GC analysis of FAME recovered from whole cells expressing pYCer3 (CpDes6) showed, that linoleic acid (18:2) served as the best substrate for Δ6-desaturation to γ-linolenic acid (γ-18:3) (Fig. 2). Cells harbouring pYCER1 were also able to convert incorporated 18:2 into γ-18:3, but in these cultures the newly introduced Δ6-double bond was further converted into a triple bond, resulting in the production of 9,12-(Z,Z)-octadecadien-6-ynoic acid (18:2A). When cells with CpAcet6 were fed with γ-18:3, they also desaturated the preexisting Δ6-double bond to form a triple bond resulting in the production of 18:2A (Fig. 2). Furthermore, the yeast cells with CpAcet6 converted stearidonic acid (18:4ω3) into 9,12,15-(Z,Z,Z)-octadecatrien-6-ynoic acid (18:3A), which represents the main acetylenic fatty acid found in the moss C. pupureus.
Both CpDes6 and CpAcet6 were able to desaturate α-linolenic acid (α-18:3) to stearidonic acid (18:4ω3) (Table 1). When cells were not supplemented with an exogenous fatty acid, the Δ9-monoenoic fatty acids present in the yeast served as substrates for Δ6-desaturation. Palmitoleic acid (16:1Δ9) was a better substrate than oleic acid (18:1Δ9) (Table 1). The fatty acids 16:3Δ7,10,13, 20:2Δ8,11, 20:3Δ8,11,14 and 20:4Δ5,8,11,14 did not serve as substrates for Δ6-desaturation or Δ6-acetylenation (data not shown). Petroselinic acid (18:1Δ6), which has no Δ9-double bond, was not a substrate for Δ6-acetylenation (data not shown). From these results we concluded, that CpDes6 and CpAcet6 were able to introduce a Δ6-double bond into Δ9-unsaturated fatty-acid substrates, thus extending the characteristic divinylmethane pattern one step further towards the carboxyl end of the acyl chain. The CpDes6 has a strong preference for linoleic acid (18:2Δ9,12), whereas CpAcet6 preferred γ-linolenic acid (18:3Δ6,9,12). Palmitoleic acid (16:1Δ9) was a better substrate for the Δ6-acetylenase than for the Δ6-desaturase. Therefore, the Δ6-acetylenase seemed to be less chain length specific than the Δ6-desaturase. A comparison (Table 1) suggests, that in yeast CpAcet6 is a better Δ6-desaturase with a broader substrate specificity than CpDes6.
GC-MS analysis of acetylenic fatty acids
Feeding experiments with γ-18:3 to yeasts expressing either CpDes6 or CPAcet6 followed by GC-MS analysis of fatty acids were used to confirm the presence of Δ6-olefinic and Δ6-acetylenic bonds in the fatty acids produced by the transgenic yeasts. The methyl ester of 18:3A from the transgenic yeast expressing CpAcet6 was identical with respect to retention time (18.25 min), EI-, and CI-MS fragmentation pattern when compared with18:3A isolated from the moss Ceratodon purpureus and Dicranum scoparium, which in addition was used for NMR analysis (see below). Furthermore, the EI-MS spectrum of the 18:3A methyl ester was found to be identical with that of 9,12,15-octadecatrien-6-ynoic acid described previously .
These data strongly suggested the formation of 18:3A, but the position of the double and triple bonds could not be allocated unequivocally by the EI-MS analysis of 18:3A and 18:2A as their methyl ester derivatives. Therefore, FAME were converted into their DMOX-derivatives. They are considered to be the most suitable derivatives for the localization of unsaturated bonds in fatty acids and are separated by GC with similar resolution as the methyl esters [20,30]. Chemical ionization (CI)-MS analysis of the putative 18:3A (eluting at 15.85 min as compared to 14.31 min of γ-18:3) gave a pseudomolecular ion ([M + H]+ = 330), which is 2 mass units (mu) lower as compared to the substrate γ-18:3 ([M + H]+ = 332). This decrease in molecular mass of the new product can be ascribed either to the introduction of an additional double bond into γ-18:3 or to the desaturation of an olefinic to an acetylenic bond. It should be recalled that CPDes6 produces only γ-18:3 and stearidonic acid 18:4, whereas expression of CPAcet6 in addition leads to the formation of 18:2A and 18:3A. The EI-mass spectra of the two polyolefinic acids, γ-18:3 (Fig. 3A) and stearidonic acid (Fig. 3B) displayed molecular ions as [M]+ with m/z = 331 (γ-18:3) and 329 (18:4), respectively, whereas the derivatives of the two corresponding acetylenic fatty acids produced pseudomolecular ions due to the loss of one hydride [M-H]+ with m/z = 328 for 18:2A (Fig. 3C) and m/z = 326 for 18:3A (Fig. 3D). In addition, the series of characteristic fragment ions in the high m/z region confirms the existence of double bonds at positions 9,12 (and 15) in all four compounds (Fig. 3) including the two acetylenic fatty acids recovered from transgenic yeasts (Fig. 3C,D). On the other hand, even DMOX derivatives do not allow the unambiguous localization of a triple bond located between C2-C8. Diagnostic fragment ions of the expected size are not formed (compare Fig. 3C,D with Fig. 3A,B), and structural assignment by GC-MS has to be based on the spectral identity of the unknown and a reference compound. Therefore, we isolated the putative 18:3A in mg quantities  from two different mosses for independent confirmation of structural details by NMR spectroscopy and as source for obtaining the required reference mass spectrum of its DMOX derivative.
The 1H-NMR spectra of 9,12,15-(Z,Z,Z)-octadecatrien-6-ynoic acid (18:3A) methylester isolated from C. pupureus and D. scoparium were found to be identical (details in Materials and methods). With the exception of the olefinic protons (H-9,10,12,13,15,16) all protons in 18:3A appeared as well resolved signals which allowed the complete assignment of all resonances. The presence of a triple bond between C-6 and C-7 could be deduced by a 1H,1H-COSY experiment showing cross peaks between H-5 (δ 2.175 p.p.m.) and H-8 (δ 2.932 p.p.m.) which originated from a diagnostic long range coupling known to be present between methylene protons separated by an acetylene group . This interpretation was further corroborated by 1D 13C-NMR where characteristic signals for triple bonded carbons were assigned to C-6 (δ 79.5) and C-7 (δ 78.6) which were lacking in the DEPT spectrum as expected for an acetylene group. The other signals could be assigned with the help of 1H,13C-HMQC and 1H,13C-HMBC- experiments, the latter showing unambiguously connectivities over three carbon atoms, thus further allowing assignment especially of C-6 and C-7 as well as other 1H and 13C NMR signals, including olefinic resonances which could not be completely achieved in previous communications . Part of the 18:3A methyl ester isolated from the mosses was converted to the DMOX derivative, which gave an EI-mass spectrum identical with that shown in Fig. 3D. In conclusion, our EI-MS and NMR analysis clearly showed that both (CPDes6 and CPAcet6) can introduce a cis-Δ6 double bond into a suitable precursor, and CPAcet6 can further desaturate the cis-Δ6 double bond to form a triple bond at this position.
The cDNA sequences CPAcet6 and CPDes6 encoding a novel Δ6-acetylenase and a Δ6-desaturase from C. purpureus were cloned from a cDNA library using a PCR-based approach. The deduced Δ6-acetylenase protein shared 57% identity with the Δ6-desaturase from C. purpureus, which is about the same value for the Δ12-acetylenase from C. alpina to the Δ12-desaturase from C. palaestina. On the other hand only four amino-acid substitutions are enough to convert a Δ12-desaturase into a Δ12-hydroxylase . These data suggest, that a similarly sized change in amino-acid composition is necessary for a switch from desaturation to acetylenation independent from regioselectivity (Δ6 or Δ12), but that it involves a more extensive change in sequence at different positions than a switch to hydroxylation. The moss Δ6-desaturases from C. pupureus and P. patens share 67% identity, but only less than 33% identity to Δ6-desaturases from other organisms [21,26], whereas a high value (i.e. more than 64% identity for microsomal Δ12-desaturases) is usually found for all desaturases sharing the same regioselectivity (Δ9, Δ12 or Δ15) and the same subcellular compartment, even among phylogenetically diverse species. In contrast, Δ6-desaturases show a similarly high conservation only among species belonging to the same phylum.
The same arrangement of functional domains (a N-terminal extension, followed by a cytochrome b5 and a C-terminal desaturase domain) present in the Δ6-desaturase from P. patens is verified in the Δ6-acetylenase and in the Δ6-desaturase from C. pupureus, which represents a unique feature of these moss desaturases, so far. Therefore, the Δ6-acetylenase can be assigned as a new member of the growing family of cytochrome b5 fusion proteins . The essential role for enzymatic activity of the cytochrome b5-domain in these fusion desaturases has been demonstrated for the yeast Δ9-acyl-CoA desaturase  and for the Δ6-acyl lipid desaturase from borage . Furthermore, the presence of a cytochrome b5-related domain suggests a microsomal rather than a plastidial localization of these enzymes, because plastidial desaturases use ferredoxin as electron donor . Beside this, CPDes6 and CPAcet6 contain a N-terminal extension also found in the Δ6-desaturase from P. patens (about 52–95 amino acids) , which is absent in other presently known desaturases. The function of this extension is unclear, as it shows no significant homology to any known protein or targeting sequence. The desaturase domain downstream of the cytochrome b5 of both proteins shows the three histidine boxes conserved in all desaturases . Furthermore, in the Δ6-acetylenase and Δ6-desaturase from C. pupureus the first histidine of the third box appearing in all Δ9-, Δ12- and Δ15-desaturases is substituted by a glutamine residue (QXXHH). This substitution seems to be a common feature of all N-terminal cytochrome b5 fusion desaturases with a Δ5-, Δ6- or Δ8-regioselectivity, known so far. As for these fusion desaturases, the Δ6-acetylenase can be assigned as ‘front-end’ desaturase, which recognizes a pre-existing double bond at the Δ9 position (x) and introduces a new double and triple bond, respectively, extending the divinylmethane pattern (x − 3 = Δ6 position) towards the carboxyl end of the fatty acyl chain.
Functional analysis in S. cerevisiae
The expression of CPDes6 and CPAcet6 in S. cerevisiae with or without supplementation of different fatty-acid substrates resulted in newly formed desaturation products neither present in control cells transformed with the empty vector nor in wild-type cells. The structure including the double and the triple bond positions, respectively, of these newly formed fatty-acid products were confirmed by GC-MS of their DMOX-derivatives. Based on the presented data, we conclude that CPDes6 codes for a Δ6-desaturase and that CPAcet6 codes for a Δ6-acetylenase, both of which are able to desaturate Δ9-unsaturated C16- and C18-fatty acids, resulting in the production of 18:3Δ6,9,12, 18:4Δ6,9,12,15, 16:2Δ6,9 and minor amounts of 18:2Δ6,9. A similar substrate specificity for Δ9,12(15)-polyenoic and Δ9-monoenoic acids has also been observed for the Δ6-desaturases from P. patens, M. alpina and C. elegans. Apart from the same broad substrate specificity of a Δ6-desaturase, CPAcet6 can further generate a triple bond by desaturating 18:3Δ6,9,12 to 18:2AΔ6a,9,12 and 18:4Δ6,9,12,15 to 18:3AΔ6a,9,12,15. Based on these data, we assume that CPAcet6 codes for a bifunctional Δ6-acetylenase/desaturase, which can use both Δ9(12)-unsaturated fatty acids for the introduction of a double bond and, independent from this, Δ6,9,12(15)-unsaturated substrates for the formation of a triple bond. At least in yeast, the occurrence of concomitant Δ6-desaturation might be due to a limited supply of the preferred Δ6,9,12(15)-unsaturated fatty-acid substrates for acetylenation. On the other hand, there are more examples for bifunctional enzymes. When expressed in yeast, the C. alpina acetylenase  and the L. fendleri hydroxylase  showed Δ12-desaturase activity, the A. thalianaΔ12-oleate desaturase  lost its chain length specificity by using palmitoleic acid and the Δ8-sphingolipid desaturases from Brassica napus and A. thaliana lacked stereoselectivity by introducing a cis- and trans-double bond, respectively. In yeast expressing CPAcet6 18:2 was desaturated two times at the same positions to yield 18:2A. Furthermore, γ-18:3 was a better substrate for acetylenation than 18:4, whereas α-18:3 only served as a substrate for Δ6-desaturation but not for Δ6-acetylenation. These data are in agreement with the pathway suggested for 18:3A synthesis in the moss C. pupureus: linoleic acid (18:2) is desaturated twice in the Δ6 position, yielding 18:2A, which undergoes a Δ15-desaturation to yield 18:3A . At present, very little is known about the catabolism and function of Δ6-acetylenic fatty acids in plants. Recently, an acetylene hydratase was purified from Pelobacter acetylenicus catalyzing the degradation of acetylene to acetaldehyde .
We thank Dr P. Beutelmann (Institut für Allgemeine Botanik, Universität Mainz) for providing us with the poly(A)+ RNA, Dr F. Thümmler (Botanisches Institut, Universität München) for the λ-ZAP cDNA library from Ceratodon purpureus. The moss cultures were generous gifts from Dr E. Hartmann (Institut für Biologie, Freie Universität Berlin) and from Prof. Dr P. Beutelmann. Sequencing of cDNA library full length clones was kindly performed by Dr O. da Costa e Silva (BASF AG, Ludwigshafen). We thank H. Moll for excellent GC-MS analyses. This work was supported by BASF AG, Ludwigshafen.
*Present address: Scandinavian Biotechnology Research AB., Herman Ehles väg 2–4, S-268 31 Svalöv, Sweden
†Present address: Dow AgroSciences, 5101 Oberlin Drive, San Diego, CA 92121, USA.