Histidine-41 of the Cytochrome b₅ Domain of the Borage Δ⁶ Fatty Acid Desaturase Is Essential for Enzyme Activity

Abstract

Unlike most other plant microsomal desaturases, the Δ⁶-fatty acid desaturase from borage (Borago officinalis) contains an N-terminal extension that shows homology to the small hemoprotein cytochrome (Cyt) b₅. To determine if this domain serves as a functional electron donor for the Δ⁶-fatty acid desaturase, mutagenesis and functional analysis by expression in transgenic Arabidopsis was carried out. Although expression of the wild-type borage Δ⁶-fatty acid desaturase resulted in the synthesis and accumulation of Δ⁶-unsaturated fatty acids, this was not observed in plants transformed with N-terminally deleted forms of the desaturase. Site-directed mutagenesis was used to disrupt one of the axial heme-binding residues (histidine-41) of the Cyt b₅ domain; expression of this mutant form of the Δ⁶-desaturase in transgenic plants failed to produce Δ⁶-unsaturated fatty acids. These data indicate that the Cyt b₅ domain of the borage Δ⁶-fatty acid desaturase is essential for enzymatic activity.

The synthesis of unsaturated fatty acids is catalyzed by acyl-lipid-desaturases that introduce double bonds into preformed acyl chains by oxygen- and electron-donor-dependent desaturation (Heinz, 1993; Shanklin and Cahoon, 1998). The immediate electron donor for many microsomal desaturases is Cyt b₅ (Lederer, 1994), a small hemoprotein that functions in a number of oxidation/reduction reactions in plants, including NADH-dependent acyl-group desaturation (Smith et al., 1990, 1998). Cyt b₅ is also involved in fatty acid hydroxylation in higher plants (Kearns et al., 1991; Smith et al., 1992). Apart from the “free” microsomal form of the protein, Cyt b₅ domains or folds have been found in a number of unrelated proteins, such as nitrate reductase, sulfite oxidase, and l-lactate dehydrogenase (Lederer, 1994; Sperling et al., 1995; Napier et al., 1997). These domains include the characteristic Cyt b₅ motif His-Pro-Gly-Gly (HPGG), which forms the core of the heme-binding domain.

Recently, Cyt b₅ domains have been identified in various positions in desaturases and hydroxylases. Thus, a fused C-terminal Cyt b₅ domain has been detected in the Saccharomyces cerevisiae OLE1 microsomal Δ⁹-desaturase (Mitchell and Martin, 1995) and in OLE1 homologs from fungi (Meesters et al., 1997) and red alga (Itoh et al., 1998). Amino-terminal Cyt b₅ domain have recently been identified in Δ⁶-fatty acid desaturases from plants (Sayanova et al., 1997), the nematode worm Caenorhabditis elegans (Napier et al., 1998), and mammals (Cho et al., 1999), and in Δ⁵-fatty acid desaturases from animals and fungi (Knutzon et al., 1998; Michaelson et al., 1998a, 1998b). These domains have also been detected in Δ⁸-sphingolipid desaturases from plants (Sperling et al., 1995, 1998) and in the FAH1 sphingolipid α-hydroxylase of yeast (Mitchell and Martin, 1997; Napier et al., 1997). The Δ⁶-fatty acid desaturase from the moss Physcomitrella patents also contains an additional N-terminal extension of 100 amino acids preceding its Cyt b₅ domain, thus resembling fusion proteins such as nitrate reductase (Girke et al., 1998).

Although these Cyt b₅-domain sequences have significantly diverged, this has occurred without modification of the HPGG motif, which forms the heme-binding core of the protein (Napier et al., 1997, 1999). The heme-binding nature of these domains has been demonstrated in sunflower sphingolipid desaturase by expression of the first 122 residues (essentially the entire Cyt b₅ domain) in Escherichia coli (Sperling et al., 1995). The resulting recombinant protein exhibited redox absorbance spectra similar to those of plant microsomal Cyt b₅ (Smith et al., 1994).

The essential role of the Cyt b₅-like domain in these fusion-desaturases has been demonstrated for yeast OLE1p Δ⁹-acyl-CoA desaturase (Mitchell and Martin, 1995). Expression of the OLE1 gene rescued yeast double mutants that lacked both OLE1 and microsomal Cyt b₅ genes, whereas rescue of ole1 mutants by a rat microsomal Δ⁹-desaturase required the additional presence of a functional Cyt b₅ gene. Truncation or disruption of the Cyt b₅ domain of the OLE1 desaturase, even in cells with wild-type levels of “free” Cyt b₅, resulted in unsaturated fatty acid auxotrophy (Mitchell and Martin, 1995). Since OLE1 complements ole1 mutants in a Cyt b₅-deletion yeast strain (unlike the rat Δ⁹-desaturase, which lacks the C-terminal Cyt b₅ domain), this implies that the 100-residue extension present in OLE1 functions as the electron donor (Mitchell and Martin, 1995). However, in that study, the disruption of the Cyt b₅ domain was achieved via an internal 100-bp deletion, which may have resulted in a perturbation of the secondary structure of the enzyme and rendered it nonfunctional.

Although it is possible that the presence of a fused Cyt b₅ domain allows for a more efficient reaction mechanism, it is interesting that none of the acyl-desaturases with Δ¹²- and Δ¹⁵-regio-selectivities (Arondel et al., 1992; Okuley et al., 1994) or the related oleate hydroxylase (van de Loo et al., 1995) have a Cyt b₅ domain, even though the Δ¹²- and Δ¹⁵-desaturases are much more prevalent in the plant kingdom. However, all of the enzymes that contain N-terminal Cyt b₅ domains are involved in “proximal” or “front-end” modifications of lipid substrates, and it is possible that this fusion is an obligate requirement for correct function of this distinct class of proteins (Sperling et al., 1998; Napier et al., 1999).

Recently, we isolated a cDNA clone (pBdes6) encoding the microsomal fatty acid Δ⁶-desaturase from developing seeds of borage (Borago officinalis). This cDNA encoded a protein with a N-terminal Cyt b₅ fusion domain and was functionally characterized by expression in transgenic tobacco plants, resulting in the accumulation of γ-linolenic acid (GLA; 18:3^6,9,12) and octadecatetraenoic acid (OTA; 18:4^6,9,12,15) (Sayanova et al., 1997). As part of our continuing studies on this enzyme, we sought to determine the functional role of the Cyt b₅ domain. We used two complementary approaches, N-terminal deletion and site-directed mutagenesis, to define this domain. Thus, two N-terminally truncated forms of the borage Δ⁶-desaturase were generated, as well as a point-mutated form in which the conserved His (His-41) of the heme-binding HPGG motif was replaced with an Ala. These three mutant forms of the desaturase were expressed in transgenic Arabidopsis and functionally compared with lines expressing the full-length wild-type borage Δ⁶-desaturase.

MATERIALS AND METHODS

Plasmid Construction for Plant Expression

PCR products encoding N-terminally deleted forms of the Δ⁶-desaturase were amplified by removing 112 and 146 residues from the N terminus of the borage Δ⁶-fatty acid desaturase. Two oligonucleotides were synthesized based on the pBdes6 coding sequence: primer M2, 5′GCGTCGACATGTTTGCAACTTTGTGC 3′ (annealing to the Met-113, indicated in boldface type) and primer M3, 5′-GCGTCGACATGGGGTTTCTTTGGATTC-3′ (annealing to the Met-147, indicated in boldface type). The SalI restriction site is underlined. Reverse primer D, containing a SmaI site, was as described in Sayanova et al. (1997) and these primers were used for PCR amplification with pBdes6 as described previously. Reactions were run on a DNA thermal cycler (Perkin-Elmer Cetus, Foster City, CA) using a program of 1 min at 94°C, 1 min at 55°C, and 2 min at 72°C for 32 cycles, followed by extension for 10 min at 72°C. PCR amplification products were gel-purified and subcloned into vector pJD330. Digestion of the two resulting plasmids with XbaI released fragments containing ORFs of different lengths (starting either with Met-113 or Met-147 and designated Δ112 and Δ146, respectively), together with the CaMV 35S promoter containing an Ω-translational enhancer and the nopaline synthase (nos) termination sequence. These XbaI-fragments were gel-purified and cloned into pBIN19 to generate plasmids pBdesM2 and pBdesM3, which were transformed into Agrobacterium tumefaciens strain GV 3101 + VirG⁺) by electroporation.

Plant Transformation

The Columbia ecotype of Arabidopsis was transformed by A. tumefaciens-mediated transformation according to the in planta vacuum infiltration method (Bechtold et al., 1993) Transformed seeds were selected on plates containing selective medium: Murashige and Skoog basal salts (4.3 g/L), 10 g/L Suc, and 50 mg/L kanamycin. Kanamycin-resistant seedlings (the T₁ generation) were transferred to soil for production of T₂ seeds.

Fatty Acid Analysis

Fatty acids from seeds, roots, siliques, and leaves were extracted and transmethylated according to the method of Miquel and Browse (1992). Analyses of fatty acid methyl esters (FAMEs) were carried out using a GC (model 5880A, Hewlett-Packard, Wilmington, DE) equipped with a 25- × 0.32-mm bonded capillary column (RSL-500BP, Hewlett-Packard) with a flame ionization detector. Fatty acids were identified by comparison of retention times with FAME standards (Sigma, St. Louis) separated on the same GC.

Site-Directed Mutageneses

Mutagenesis of His-41 to Ala was performed using a site-directed mutagensis kit (QuikChange, Stratagene, La Jolla, CA) according to the manufacturer's protocol. Two mutagenic oligonucleotide primers, each complementary to the opposite strands of the p35Bdes6 construct, were synthesized as shown.

Forward primer A, 5′GATTGGGTGAAAGACGCTCCAGGTGGCAGC 3′, and reverse primer B, 5′GCTGCCACCTGGAGCGTCTTTCACCCAATC 3′. The bold letters indicate altered nucleotides. Mutation of His-41 to Ala was performed on the plant expression construct containing the borage Δ⁶-desaturase coding sequence. The sequence of the mutated fragment was checked for the presence of the correct alteration and was used to ensure that other mutations had not occurred. This plasmid was used for Arabidopsis transformation via A. tumefaciens (see above).

Nothern-Blot Analyses

RNA from expanding Arabidopsis leaves was extracted using a kit (RNAeasy, Qiagen, Valencia, CA). About 10 μg of total RNA was run on 1% (v/v) formaldehyde gel and transferred to a nylon membrane (Hybond N⁺, Amersham, Uppsala) according to the manufacturer's protocol. Probes were made from the truncated Δ146 form of the pBdes6 cDNA by random priming using a kit (Prime-it II, Stratagene) according to the supplier's instructions.

RESULTS AND DISCUSSION

Expression of a Δ⁶ Fatty Acid Desaturase Gene in Arabidopsis

The full-length (wild-type) borage Δ⁶-desaturase cDNA (Sayanova et al., 1997) was initially expressed in Arabidopsis plants and the distributions of fatty acids in total lipid fractions prepared from leaves, roots, siliques, and seeds compared with those in control transformants (Table I). GLA and OTA accumulated to levels of 4.8% and 9.4% of the total fatty acids, respectively (all fatty acid percentages are expressed as a mole percentage of the total). This bias toward the accumulation of OTA may be have been due to the preference of the Δ⁶-desaturase for α-linolenic acid as a substrate or to high levels of endogenous leaf Δ¹⁵-desaturase activity, reducing the levels of linoleic acid available for conversion to GLA. The endogenous levels of both 18:2 and α-18:3 in roots were reduced by expression of the borage Δ⁶-desaturase, with a significant increase in GLA. In mid-stage (S2) developing seeds, the accumulation of GLA and OTA was low (Table I) and declined to undetectable levels as the seeds matured (data not shown). This may have been due to reduced levels of the Δ⁶-desaturase enzyme (as a result of reduced promoter activity) or to the exclusion of Δ⁶-desaturated fatty acids from seed lipids by endogenous acyltransferases.

Table I.

Fatty acid composition (moles percent of total) of seed and non-seed tissues of transgenic Arabidopsis transformed with either full-length borage Δ⁶-desaturase gene (AT6) or a control transformation (WT)

Fatty Acid	Leaves		Roots		S2 Seeds
	WT	AT6	WT	AT6	WT	AT6
16:0	15.5	15.5	25.0	27.0	20.1	14.0
16:1	3.2	4.4	1.6	1.0	–	–
16:3	16.0	22.0	–	–	–	–
18:0	0.5	0.7	2.0	2.5	2.7	2.5
18:1	1.3	2.1	2.7	3.0	12.4	10.4
18:2	11.0	10.1	40.0	30.1	35.1	34.0
18:3	46.0	35.0	26.2	15.2	26.0	28.7
GLA	–	4.90	–	15.1	–	0.5
OTA	–	9.30	–	7.0	–	1.3

Measurements were determined from a number of runs; se for all measurements was <3%. S2 seeds are defined as being at the midstage of seed development.

We recently determined the lipid classes present in the vegetative tissues of homozygous lines of transgenic tobacco plants expressing the borage Δ⁶-desaturase (Sayanova et al., 1997) and also in seed tissue (O. Sayanova, P.R. Shewry, and J.A. Napier, unpublished data). GLA accumulated in the leaves of the transgenic tobacco plants at levels between 12.9% of total fatty acids in young leaves and 20.1% in mature leaves, with OTA levels of 8.5% and 14.9%, respectively. However, there was only a low level of accumulation of GLA (2.6%) in mature seeds with insignificant levels of OTA. Thus, there is a difference in the proportions of the GLA and OTA in Arabidopsis and tobacco leaves, with GLA being the predominant Δ⁶-fatty acid in tobacco and OTA being predominant in Arabidopsis. This difference may reflect the relative levels of desaturation occurring in the chloroplasts of the two species, as some 50% of cellular glycerolipids are synthesized in the plastids of 16:3 plants such as spinach and Arabidopsis, compared with only about 20% in 18:3 plants such as tobacco. Such observations confirm the contribution of the plastid in glycerolipid synthesis (Somerville and Browse, 1991), especially in 16:3 species such as Arabidopsis.

Truncation of the Cyt b₅-Domain of the Borage Δ⁶ Fatty Acid Desaturase

Comparison of the deduced amino acid sequences of the borage Δ⁶-desaturase with other microsomal desaturases indicated that borage protein contained an N-terminal Cyt b₅ domain, with residues 1 to 112 corresponding to the hydrophilic portion of microsomal Cyt b₅ (see Fig. 1.). If this N-terminal Cyt b₅ domain was the result of a gene fusion event, then Met-113 in the borage Δ⁶-desaturase could correspond to the initiating Met of the ancestral desaturase domain. Alternatively, the ancestral desaturase ORF could start at Met-147, which corresponds to the end of plant microsomal Cyt b₅ (Fig. 1). To determine whether the desaturase domain of the borage Δ⁶-enzyme could function with other (independent) electron donors instead of its own (linked) Cyt b₅ domain, two truncations of the N-terminal domain were generated by deleting either the first 112 or 146 residues. These deletions, designated Δ112 and Δ146, respectively, were expressed in transgenic Arabidopsis plants.

Figure 1.

Sequence alignments of the first 167 residues of the borage Δ⁶-desaturase (boofd6) with the Arabidopsis FAD2 (arthfad2) and FAD3 (arthfad3) microsomal desaturases, sunflower Δ⁸-sphingolipid desaturase (heand8), cyanobacteria Δ⁶ desaturase (synd6), and borage Cyt b₅ (boofb5). The conserved HPGG motif is underlined, with the essential His (H⁴¹) indicated by a star. The position of the internal Met residues (Met-113 and Met-147) are boxed and arrowed. The conserved His residues of the first “His box” of the microsomal desaturases are also indicated by black dots.

GC of FAMEs prepared from leaf lipids showed no evidence of GLA or OTA in the control transformants or in plants transformed with Δ112 and Δ146, in contrast to plants transformed with the full-length Cyt b₅ Δ⁶-fatty acid desaturase fusion. Northern analysis was therefore performed to confirm that the truncated forms of the borage Δ⁶-desaturase were expressed in the transgenic Arabidopsis lines (Fig. 2). This analysis showed that the full-length and truncated forms of the borage desaturase had reduced levels of transcript accumulation, although all constructs were driven by the same promoter (CaMV 35S). In particular, transcripts of the more severe N-terminal deletion (Δ146) appeared to be less stable, as judged by the signal from the northern blot.

Figure 2.

Northern-blot analysis of transgenic Arabidopsis plants expressing the full-length borage Δ⁶-desaturase (AT6) compared with lines expressing two N-terminally truncated forms (Δ112 and Δ146). The approximate size of the detected transcript is indicated. A control transformed Arabidopsis (WT) sample was also probed. Ten micrograms of RNA was loaded in each lane.

It is possible that the failure to detect Δ⁶-fatty acids in plants expressing the deleted forms of the desaturase could be due to alterations in mRNA processing or translation of the synthetic transgene, rather than from any loss of enzyme activity. Alternatively, deletion of the Cyt b₅ domain could result in aberrant folding of the modified desaturase, rendering the enzyme nonfunctional or reducing stability. Unfortunately, the lack of a suitable antibody prevented us from comparing the levels of wild-type and N-terminally deleted forms of the Δ⁶-desaturase. The role of the C-terminal Cyt b₅ domain of the yeast OLE1 Δ⁹-desaturase was similarly studied via internal deletion of 34 amino acid residues by restriction digestion and re-ligation (Mitchell and Martin, 1995). In that study, the mutated form of OLE1 was nonfunctional. Moreover, in gene-replacement yeast strains, steady-state transcripts for the mutant OLE1 were at a reduced level compared with that observed for the wild-type OLE1 gene. An equally important observation was that even in the functional OLE1 enzyme, the protein displayed a very short half-life (<2 mins) (Mitchell and Martin, 1995); therefore, altered forms of the protein may be even less stable.

Site-Directed Mutagenesis

In view of our results for the truncated Δ⁶-desaturase and the above-mentioned studies of Michell and Martin (1995) on the OLE1 Δ⁹ desaturase, it was clear that a different approach was required to study the role of the Cyt b₅ domain. Site-directed mutagenesis is a more precise tool for the definition of important amino acid determinants in a protein, and such mutations would not be expected to affect mRNA stability and translational efficiency. We therefore mutated the highly conserved His of the HPGG heme-binding domain in the N-terminal Cyt b₅ extension. This residue is considered essential for the function of Cyt b₅ by acting as a heme ligand (Lederer, 1994). His-41 was converted to Ala by a 3-bp mutation in the desaturase coding sequence. Ala was chosen as the replacement residue because it is uncharged and should therefore not be able to act as a heme-binding residue. The mutated Δ⁶-desaturase (named H41A) was introduced into Arabidopsis, and a number of transgenic lines were regenerated. GC analysis of the leaf lipids extracted from these transgenic plants again showed no accumulation of GLA and OTA (data not shown; compared with approximately 15% Δ⁶ desaturate fatty acids in line AT6 expressing the wild-type Δ⁶-desaturase; compare with Table I), although northern blots showed mRNA levels equivalent to that in plants expressing the wild-type borage enzyme (Fig. 3). This suggests that the conserved residue His-41 of the HPGG-binding domain is likely to be essential for enzymatic function of the Δ⁶-fatty acid desaturase and is most likely to coordinate the axial heme-binding center of that domain.

Figure 3.

Northern-blot analysis of transgenic Arabidopsis plants expressing either the full-length borage Δ⁶-desaturase (AT6) or a mutated form in which His-41 was converted to an Ala (H41A). Approximately 10 μg of RNA was loaded for both samples, as judged by ethidium bromide staining (B), and the approximate size of the detected transcript is indicated (A).

The above results also suggest that the presence of an N-terminal Cyt b₅ domain in the so-called front-end desaturases (i.e. enzymes that desaturate between pre-existing double bonds and the carboxyl-group) has arisen as an integral part of the reaction mechanisms of this class of enzymes. The situation for other desaturase-like enzymes involved in proximal lipid modification is less clear and is summarized in Table II. The presence of an N-terminal Cyt b₅ domain has been observed in higher plant Δ⁸-sphingolipid desaturase (Sperling et al., 1998) and in yeast FAH1 sphingolipid α-hydroxylase (Mitchell and Martin, 1997). However, in FAH1, an Arabidopsis homolog of this enzyme did not contain any domain with homology to Cyt b₅. When this plant FAH1 homolog was expressed in yeast fah1 deletion mutants, it was capable of α-hydroxylation of substrate long-chain fatty acids. However, the plant enzyme did not fully restore hydroxylated fatty acids to wild-type levels even though it was overexpressed via a GAL1 promoter on a multicopy plasmid (Mitchell and Martin, 1997). It may be that the efficiency/activity of the (Cyt b₅ domain-lacking) plant enzyme is reduced, though obviously other factors (such as expression in a heterologous system) could also be influential. The authors of that study also noted the presence of an N-terminal Cyt b₅ domain in a putative FAH1 homolog from C. elegans, but this domain was missing in a Schizosaccharomyces pombe FAH1 homolog. This would indicate that, at least in the α-hydroxylation of the sphingolipid long-chain fatty acids, the Cyt b₅ domain is not an obligate requirement for enzyme function. This is in contrast to our observations with the Δ⁶-fatty acid desaturase and to that reported for the OLE1 Δ⁹-desaturase (Mitchell and Martin, 1995). It will be of interest to determine if further examples of front-end fatty acid-modifying enzymes exist as fusions with their electron donor.

Table II.

Examples of “front-end” desaturases and related proximal lipid modifying enzymes

Enzyme	Organism	Accession No.	b5
Δ5-Fatty acid desaturases
Mortierella alpina	Fungus	AF054824	✓
Caenorhabditis elegans	Nematode	AF078796	✓
Dictyostelium discoideum	Slug	AB022097	✓
Δ6-Fatty acid desaturases
B. officinalis	Plant	U79010	✓
Physcomitrella patens	Moss	AJ980222	✓a
C. elegans	Nematode	AF031477	✓
Homo sapiens	Human	Z44979	✓
Synechocystis	Cyanobacteria	L11421	✗b
Δ8-Sphingolipid desaturases
Arabidopsis	Plant	AJ224161	✓
Helianthus annuus	Plant	X87143	✓
Brassica napus	Plant	AJ224160	✓
Ricinus communis	Plant	AF005096	✓
FAH1 sphingolipid α-hydroxylase
S. cerevisiae	Yeast	Z49260	✓
C. elegans	Nematode	Z81038	✓
Arabidopsis	Plant	AF021804	✗
Schizosaccharomyces pombe	Yeast	Z97209	✗

The presence of a N-terminal Cyt b₅ domain is indicated by a tick. The GenBank accession numbers for all sequences are given.

^aThe Cyt b₅ domain present in this Δ⁶-desaturase is internal. 

^bThe electron donor for cyanobacterial desaturases is ferredoxin. 

CONCLUSIONS

We have shown that His-41 of the Cyt b₅ domain is likely essential for the enzymatic activity of the borage Δ⁶-fatty acid desaturase, as determined by site-directed mutagenesis and expression in transgenic plants. These results support the hypothesis that the presence of a N-terminal Cyt b₅ domain in this front-end desaturase is an obligate requirement for enzymatic activity.