Annotation of Genes Involved in Glycerolipid Biosynthesis in Chlamydomonas reinhardtii: Discovery of the Betaine Lipid Synthase BTA1_Cr

Abstract

Lipid metabolism in flowering plants has been intensely studied, and knowledge regarding the identities of genes encoding components of the major fatty acid and membrane lipid biosynthetic pathways is very extensive. We now present an in silico analysis of fatty acid and glycerolipid metabolism in an algal model, enabled by the recent availability of expressed sequence tag and genomic sequences of Chlamydomonas reinhardtii. Genes encoding proteins involved in membrane biogenesis were predicted on the basis of similarity to proteins with confirmed functions and were organized so as to reconstruct the major pathways of glycerolipid synthesis in Chlamydomonas. This analysis accounts for the majority of genes predicted to encode enzymes involved in anabolic reactions of membrane lipid biosynthesis and compares and contrasts these pathways in Chlamydomonas and flowering plants. As an important result of the bioinformatics analysis, we identified and isolated the C. reinhardtii BTA1 (BTA1_Cr) gene and analyzed the bifunctional protein that it encodes; we predicted this protein to be sufficient for the synthesis of the betaine lipid diacylglyceryl-N,N,N-trimethylhomoserine (DGTS), a major membrane component in Chlamydomonas. Heterologous expression of BTA1_Cr led to DGTS accumulation in Escherichia coli, which normally lacks this lipid, and allowed in vitro analysis of the enzymatic properties of BTA1_Cr. In contrast, in the bacterium Rhodobacter sphaeroides, two separate proteins, BtaA_Rs and BtaB_Rs, are required for the biosynthesis of DGTS. Site-directed mutagenesis of the active sites of the two domains of BTA1_Cr allowed us to study their activities separately, demonstrating directly their functional homology to the bacterial orthologs BtaA_Rs and BtaB_Rs.

Knowledge of genes and proteins involved in lipid metabolism in plants has greatly increased over the past decade. Many genes encoding enzymes of glycerolipid biosynthesis and fatty acid desaturation have been identified by genetic and biochemical means (10, 16, 33), and even lipid trafficking phenomena are beginning to be unraveled by genetic approaches (57). In addition, the availability of the Arabidopsis genome sequence (51) has facilitated the annotation of many uncharacterized gene products which are likely to have activities in lipid biosynthesis, trafficking, and catabolism (4). Although Arabidopsis is undoubtedly the most widely used plant system currently under investigation, the eukaryotic green alga Chlamydomonas reinhardtii is a well-established model for many processes, such as photosynthesis (31), phototaxis and flagellar function (47), posttranscriptional gene silencing (56), and nutrient acquisition (8). Recent large-scale expressed sequence tag (EST) projects and sequencing of the C. reinhardtii genome (21, 46) and the development of insertional mutagenesis (50), RNA interference methods (17, 48), and a molecular map (25) make Chlamydomonas an attractive genetic model. These attributes make Chlamydomonas also an ideal model organism with which to study the biosynthesis and physiological functions of nonphosphorous lipids, such as the plastidic galactolipids and sulfolipids and the betaine lipid diacylglyceryl-N,N,N-trimethylhomoserine (DGTS) (18, 29, 39), as well as phospholipids, such as phosphatidylethanolamine (PtdEtn) (58) and phosphatidylglycerol (PtdGro) (11, 34, 45).

We have begun to explore Chlamydomonas as a model for plant lipid metabolism by using a high-throughput robotic screening method to find mutants with defects in lipid metabolism (35). The initial goal of the study reported here was to identify and annotate genes involved in acyl-lipid metabolism in the Chlamydomonas version 2.0 draft genome sequence available at http://genome.jgi-psf.org/chlamy/. Enzymes of fatty acid biosynthesis, glycerolipid synthesis, and fatty acid desaturation were identified and organized so as to reconstruct lipid biosynthesis pathways and posit their subcellular localizations. As a specific example, this study allowed us to identify and characterize the enzyme responsible for the synthesis of the betaine lipid DGTS, which replaces the phospholipid phosphatidylcholine (PtdCho) in extraplastidic membranes of Chlamydomonas. Due to its structural similarity to PtdCho and the lack of PtdCho in Chlamydomonas, DGTS is presumed to take on the functions of PtdCho as the major component of the extraplastidic membranes of this organism (29, 40, 42).

Previous work elucidated the pathway for DGTS biosynthesis in Rhodobacter sphaeroides (23), and genetic analysis with this bacterium identified two gene products, BtaA and BtaB, which are required for phosphate stress-induced DGTS production in R. sphaeroides (27). The btaA gene was proposed to encode an S-adenosylmethionine (AdoMet):diacylglycerol (DAG) 3-amino-3-carboxypropyltransferase giving rise to the intermediate diacylglycerolhomoserine (DGHS), and the btaB gene was proposed to encode an AdoMet-dependent N-methylase converting DGHS to DGTS. Here we describe the discovery of the C. reinhardtii BTA1 (BTA1_Cr) gene, encoding a Chlamydomonas betaine lipid synthase that was shown to encompass both activities in a single polypeptide.

MATERIALS AND METHODS

Annotation of C. reinhardtii genes involved in lipid metabolism.

Version 2.0 of the C. reinhardtii whole-genome shotgun sequence assembly is available at http://genome.jgi-psf.org/chlamy. Genes encoding enzymes and other components of lipid metabolism (e.g., acyl carrier proteins [ACPs]) were manually annotated based on key word searches executed against the automatically annotated gene loci (21), followed by verification of similarity by BLASTP searches (1) and CLUSTALW multiple-sequence alignments (52) with verified protein sequences. In addition, direct BLASTP and TBLASTN searches of the version 2.0 assembly were carried out with known protein sequences from plants, animals, fungi, and bacterial sources to add additional evidence and to identify genes that were incorrectly annotated by the automated system. For the tentative prediction of subcellular localizations, the programs TargetP and ChloroP were used (12, 13). When the N terminus was not available due to sequence gaps and/or ambiguity in the prediction of the first exon, tentative localizations were predicted on the basis of clustering with isoforms from other species for which subcellular localizations are known.

Materials.

Restriction enzymes, T4 DNA ligase, calf intestinal phosphatase, and the Klenow fragment were obtained from New England Biolabs (Beverly, Mass.). SuperScript II RNase H reverse transcriptase was obtained from Invitrogen (Carlsbad, Calif.). Taq and Pfu DNA polymerases were obtained from Stratagene (La Jolla, Calif.). All other chemicals and solvents were of reagent grade and were obtained from Sigma Chemical Co. (St. Loius, Mo.), EM Science (Gibbstown, N.J.), or J. T. Baker Chemical Co. (Phillipsburg, N.J.).

Cloning and expression of BtaA_Rs, BTA1_Cr, and derivatives of BTA1_Cr in Escherichia coli.

A summary of all strains and plasmids used in this study is shown in Table 1. The coding region of btaA_Rs was PCR amplified from plasmid pRKL323 (27) for expression in pACYC-31 (9) with the following primers (restriction sites are underlined): btaA forward (SphI), 5′-ACATGCATGCAGTGACGCAGTTCGCCCTC-3′, and btaA reverse (KpnI), 5′-CGGGGTACCAGGACGATCCGCTCGAACCG-3′. PCR was carried out with Taq DNA polymerase according to the manufacturer's specifications, except that 10% (vol/vol) dimethyl sulfoxide was added to each reaction mixture to overcome difficulties in PCR due to the high G+C content of R. sphaeroides DNA. The PCR product was cloned into the EcoRI site of pBluescript SK(+), sequenced at the Michigan State University (MSU) Genomics Technology Support Facility, and subcloned into pACYC-31 by using the restriction sites underlined in the PCR primer sequences shown above. The resulting construct was designated pBtaA-LC (btaA in pACYC-31).

TABLE 1.

Strains and plasmids used in this work

Strain or plasmid	Descriptiona	Source
Strains
C. reinhardtii
CC125	Wild type; mt+	Chlamydomonas Genetics Center, Duke University
dw15.1	nit1-305 cw15; mt+	Arthur Grossman
E. coli TOP10F′	Cloning and expression strain; φ80lacZΔM15 F′ [lacIq Tn10 (Tetr)]	Invitrogen
Plasmids
pBluescript SK(+)	Cloning vector; Ampr	Stratagene
pACYC-31	Expression vector; Camr	9
pQE-31	Expression vector; Ampr	Qiagen
pRKL323	btaAB operon in pCHB500	27
pBtaA-LC	SphI-KpnI fragment of btaA amplified from pRKL323 in pACYC-31	This work
pBTA1	BamHI-KpnI fragment of the BTA1Cr coding sequence in pQE-31	This work
pBTA1-μA	V340A-D341A derivative of pBTA1	This work
pBTA1-μB	V121A-D122A derivative of pBTA1	This work

^amt⁺, mating type⁺.

C. reinhardtii strain CC125 was grown to mid-log phase in 50 ml of TAP medium (22) and harvested by centrifugation. RNA was purified with Trizol reagent (Invitrogen) according to the manufacturer's instructions. Total RNA (1.0 μg) was reverse transcribed with SuperScript II RNase H reverse transcriptase. The coding region of BTA1_Cr was amplified from an aliquot of cDNA with Pfu DNA polymerase and the following primers (the native start codon of BTA1_Cr in the forward primer is shown in bold type, and BamHI and KpnI sites are underlined in the forward and in the reverse primers, respectively): CrBTA1 forward, 5′-CAGGATCCAATGGGGTCGGGTCGT-3′, and CrBTA1 reverse, 5′-CAGGTACCGCCGCCAGCTGCTTA-3′. The product was cloned in frame with the N-terminal His₆ tag of pQE-31, giving rise to pBTA1.

Site-directed mutagenesis of pBTA1 was carried out with a Quikchange II XL kit (Stratagene) according to the manufacturer's instructions. For the V340A-D341A mutant version of pBTA1 (pBTA1-μA), the mutagenic primers were μA-fw (5′-GCCAGGTGGTGTCGGCGGCATGCAACCCCGCGCAG-3′) and its reverse complement μA-rev. For the V121A-D122A mutant (pBTA1-μB), the mutagenic primers were μB-fw (5′-CAAGTCCATCTACGTGGCCGCCCTGTGCCACTCGCTG-3′) and its reverse complement μB-rev. Primers were designed such that in addition to the missense mutations for changing codons, a diagnostic restriction site was created or eliminated (e.g., “μB” destroys a SalI site and “μA” introduces an SphI site) in order to easily identify strains carrying mutated plasmids.

pBluescript and pQE-31 derivatives were propagated in Luria-Bertani (LB) medium containing ampicillin at 100 μg/ml, and pACYC-31 derivatives were propagated in LB medium containing chloramphenicol at 25 μg/ml. Strains containing compatible combinations of plasmids were constructed by cotransformation of heat shock-competent cells of TOP10F′ (Invitrogen) with 100 ng of each plasmid followed by selection on LB agar containing both ampicillin and chloramphenicol.

In vivo assays of DGHS and DGTS synthesis in E. coli.

E. coli TOP10F′ cells harboring compatible combinations of plasmids were grown as 2-ml overnight cultures, and 0.1-ml samples of these cultures were used to inoculate 10 ml of LB medium containing appropriate antibiotics. After growth to an optical density at 600 nm of ∼0.6, the cultures were harvested by centrifugation and dispersed into 10 ml of M9 minimal medium (37) containing 0.2 mM isopropyl-β-d-thiogalactoside (IPTG) and 0.5 μCi of [1-¹⁴C]methionine (American Radiolabeled Chemicals, St. Louis, Mo.). Cultures were incubated for 3 h and harvested by centrifugation; these steps were followed by extraction with 2 ml of chloroform-methanol (1:1 [vol/vol]) and phase separation by the addition of 0.5 ml of 1 M KCl-0.2 M H₃PO₄. A portion of the organic phase from each extraction was spotted on activated (120°C, 2 h) thin-layer chromatography (TLC) plates (Si 250; Baker), resolved with chloroform-acetone-methanol-acetic acid-water (10:4:2:2:1 [vol/vol]), and subjected to autoradiography.

In vitro BTA1_Cr activity assays.

E. coli TOP10F′ cells harboring pBTA1 were grown in 250 ml of LB medium-ampicillin at 37°C to an optical density of 0.7 and induced with 0.25 mM IPTG; these steps were followed by an additional 4 h of growth at 28°C. Cells were harvested by centrifugation, and the cell pellet was suspended in 10 ml of cold buffer (50 mM HEPES, 1 mM dithiothreitol, 1 mM EDTA [pH 7.3]). The resuspended cells were sonicated three or four times for 30 s each time with a microprobe tip, and the lysates were centrifuged at 2,000 × g for 10 min to remove unbroken cells and cellular debris. Aliquots (1 ml) of the cell extracts were frozen in liquid N₂ and stored at −80°C prior to use. Activity under these storage conditions did not decrease appreciably for at least 1 month.

Assays for the pH-activity profile were conducted with a 100-μl final volume by combining 48.75 μl of cell extract with 48.75 μl of 100 mM HEPES, Tris-Cl, or morpholineethanesulfonic acid, 1 mM dithiothreitol, and 1 mM EDTA at various initial pHs to give a final pH in the range of 5.5 to 8.6 when mixed with the cell extract (initial pH, 7.3). Reactions were initiated by the addition of 25,000 dpm of [1-¹⁴C]AdoMet (American Radiolabeled Chemicals; 2.5 μl; final concentration of AdoMet, 2.1 μM), incubated for 30 min at 28°C, and terminated by the addition of 400 μl of chloroform-methanol (1:1 [vol/vol]) and 100 μl of 0.9% (wt/vol) NaCl to separate the aqueous and organic phases. Under these conditions, product formation was linear over time up to 1 h at all concentrations of AdoMet tested. The organic phase was transferred to a new tube, dried under a stream of nitrogen, and dissolved in 50 μl of chloroform-methanol (1:1 [vol/vol]). This lipid extract was spotted on silica TLC plates (Baker) and developed with the solvent chloroform-acetone-methanol-acetic acid-water (10:4:2:2:1 [vol/vol]); these steps were followed by quantification of the signal for DGHS and methylated products on a PhosphorImager screen (Molecular Dynamics, Piscataway, N.J.) with the ImageQuant software package. Determination of the K_m was carried out at pH 7.2 with various concentrations of AdoMet by direct fitting of a hyperbola to the experimental data in the Origin software package (OriginLab, Northampton, Mass.).

Structure confirmation for DGTS produced by BTA1_Cr in E. coli.

E. coli cells harboring pBTA1 were cultured as described above for the production of crude extracts for enzyme assays. DGTS was isolated from lipid extracts by preparative TLC on ammonium-sulfate-impregnated silica TLC plates with the solvent chloroform-acetone-methanol-acetic acid-water (10:4:2:2:1 [vol/vol]). The DGTS zone was scraped from the plates, eluted with chloroform-methanol (1:1 [vol/vol]), and concentrated under a stream of nitrogen. Fast-atom bombardment-mass spectrometry was done as previously described (5), except that nitrobenzyl alcohol was used as a matrix. ¹H nuclear magnetic resonance (NMR) spectroscopy was conducted by dissolving approximately 200 μg of lipid in CDCl₃, and data were collected as previously described (5).

RESULTS

Annotation of genes involved in acyl-lipid biosynthesis in the Chlamydomonas genome. (i) Overall results.

A metabolic reconstruction of glycerolipid metabolism in Chlamydomonas from genomic sequence data was performed in silico. For this purpose, we searched all available genomic sequences of Chlamydomonas, essentially covering the entire genome, for deduced protein sequences with significant similarity to lipid biosynthetic enzymes with experimentally verified activities from plants and other organisms. For building of the pathway models, we also considered analytical data published for the lipid compositions of various membranes in Chlamydomonas and plants in general. Figures 1 and 2, in conjunction with Table 2, show schemes for membrane lipid biosynthesis, starting with fatty acid synthesis in the plastid (Fig. 1, steps 1 to 4 and 16), plastidic glycerolipid biosynthesis (Fig. 1, steps 5 to 15), extraplastidic phospholipid and betaine lipid biosynthesis (Fig. 1, steps 16 to 31), and fatty acid desaturation (Fig. 2, steps 32 to 40). Genes are identified by their version 2.0 gene model transcript number, as well as by three-letter abbreviations based on the tentative assignment of function. These assignments were made in accordance with the recommendations of the Commission on Plant Gene Nomenclature and the Mendel database (28).

FIG. 1.

Pathways of lipid biosynthesis which are known or hypothesized to occur in Chlamydomonas and their presumed subcellular localizations. Numbers in bold type correspond to the reactions and steps summarized in Table 2. Synthesis of the betaine lipid DGTS (step 23) is highlighted and was characterized in detail. Abbreviations: ACP, acyl carrier protein; AdoMet, S-adenosylmethionine; ASQD, 2′-O-acyl-sulfoquinovosyldiacylglycerol; CDP, cytidine-5′-diphosphate; CoA, coenzyme A; CTP, cytidine-5′-triphosphate; DAG, diacylglycerol; DGDG, digalactosyldiacylglycerol; DGTS, diacylglyceryl-N,N,N-trimethylhomoserine; Etn, ethanolamine; FA, fatty acid; G-3-P, glycerol-3-phosphate; Glc, glucose; Glc-1-P, glucose-1-phosphate; Ins, inositol; Ins-3-P, inositol-3-phosphate; Met, methionine; MGDG, mono- galactosyldiacylglycerol; P-Etn, phosphoethanolamine; PtdEtn, phosphatidylethanolamine; PtdGro, phosphatidylglycerol; PtdGroP, phosphatidylglycerophosphate; PtdIns, phosphatidylinositol; PtdOH, phosphatidic acid; Ser, serine; SQ, sulfoquinovose; SQDG, sulfoquinovosyldiacylglycerol; UDP, uridine-5′-diphosphate.

FIG. 2.

Pathways of acyl chain desaturation, with numbers in bold type serving to cross reference the specific reactions with the respective enzymes and gene models summarized in Table 2. Fatty acids esterified to the sn1 and sn2 positions of the glycerol moieties of the respective lipids are indicated as “fatty acid at sn1/fatty acid at sn2.” Fatty acids are identified as “carbons:double bonds.”

TABLE 2.

Assignment of candidate genes encoding enzymes catalyzing the reactions diagrammed in Fig. 1 and 2

Reaction or step	Descriptiona	Gene model(s)	Gene(s)	EST(s)b
Plastid and mitochondrial pathways
1	Multimeric acetyl-CoA carboxylase components
1a	Alpha-carboxyltransferase	C_160059	ACX1	23
1b	Beta-carboxyltransferase	C_1150054	BCX1	39
1c	Biotin carboxylase	C_2010018	BCR1	18
1d	Biotin carboxyl carrier protein	C_590042	BXP1	16
		C_250035	BXP2	27
2	Acyl carrier protein	C_650040	ACP1	38
		C_210028	ACP2	51
3	Malonyl-CoA:ACP transacylase	C_240122	MCT1	4
		C_7350001	MCT2	16
4	Type II fatty acid synthase components
4a	3-Ketoacyl-ACP synthase	C_140066	KAS1	17
		C_720014	KAS2	7
		C_1320034	KAS3	2
4b	3-Ketoacyl-ACP reductase	C_460059	KAR1	19
4c	3-Hydroxyacyl-ACP dehydratase	C_120121	HAD1	7
4d	Enoyl-ACP-reductase	C_190023	ENR1	64
5	Glycerol-3-phosphate:acyl-ACP acyltransferase	C_1210022	ATS1	8
6	Lyso-phosphatidate:acyl-ACP acyltransferase	No candidate
7	CDP-DAG synthetase	C_30067	CDS1	12
8	Phosphatidylglycerophosphate synthase	C_140036	PGP1	21
		C_2150006	PGP2	4
9	Phosphatidylglycerophosphate phosphatase	No candidate
10	Phosphatidate phosphatase	No candidate
11	UDP-sulfoquinovose synthase	C_60107	SQD1	46
12	Sulfolipid synthase	C_250126	SQD2a	5
		C_650023	SQD2b	0
13	Sulfolipid 2′-O-acyltransferase	No candidate
14	Monogalactosyldiacylglycerol synthase	C_21260001	MGD1	3
15	Digalactosyldiacylglycerol synthase	C_230128	DGD1	1
16	Acyl-ACP thiolase	C_5710003	FAT1	5
		C_200208	FAT2	7
Extraplastidic pathways
17	Long-chain acyl-CoA synthetase	C_1580041	LCS1	11
		C_7940001	LCS2	0
18	Glycerol-3-phosphate:acyl-CoA acyltransferase	No candidate
19	Lyso-phosphatidate:acyl-CoA acyltransferase	No candidate
20	Phosphatidate phosphatase	C_1170013	PAP1	1
		C_300097	PAP2	4
21	CDP-DAG synthetase	C_630023	CDS2	4
22	AdoMet synthetase	C_750005	SAS1	82
23	Betaine lipid synthase	C_570065	BTA1	42
24	CDP-ethanolamine:DAG-ethanolamine phosphotransferase	C_960032	EPT1	2
25	Serine decarboxylase	C_250130	SDC1	43
26	Ethanolamine kinase	C_290062	EKI1	0
27	CTP:phosphoethanolamine cytidylyltransferase	C_960052	ECT1	15
28	Phosphatidylglycerophosphate synthase	C_1350041	PGP3	0
29	Phosphatidylglycerophosphate phosphatase	No candidate
30	Inositol-3-phosphate synthase	C_860083	INO1	31
31	CDP-DAG:inositol phosphotransferase	C_260084	PIS1	3
Fatty acid desaturase plastidic isoforms
32	Stearoyl-ACP-Δ9-desaturase	C_170008	FAB2	10
33	PtdGro palmitate-Δ3t-desaturase	No candidate
34	MGDG palmitate-Δ7-desaturase	C_330009	FAD5	70
35	ω6-Desaturase	C_790066	FAD6a	264
		C_290028	FAD6b	0
36	ω3-desaturase	C_250125	FAD7	39
37	MGDG 16-carbon Δ4-desaturase	No candidate
Fatty acid desaturase cytosolic isoforms
38	Oleate desaturase	C_590003	FAD2a	20
		C_4630001	FAD2b	0
39	Linoleate desaturase	C_250037	FAD3	4
40	Linoleate or linolenate-Δ5-desaturase	No candidate

^aProteins were identified on the basis of similarity to known proteins, and predicted subcellular targeting was based on an analysis of the N terminus and on similarity to isoforms from other species for which the subcellular localization is known. Candidates for dual targeting to chloroplasts and mitochondria are in bold type; subunits are in italic type. 3t, 3-trans. Abbreviations are as in Fig. 1.

^bEST counts represent data available from the Joint Genome Institute genome browser in May and June 2004.

(ii) Lipid metabolism in plastids.

Genes encoding enzymes of plastidic lipid metabolism are readily identifiable in the genome and appear to be most similar to those of flowering plants. For example, it is clear that all components of the multimeric bacterium-type acetyl coenzyme A (acetyl-CoA) carboxylase (Fig. 1 and Table 2, 1a to 1d) and fatty acid synthase complexes (Table 2, 4a to 4d) are accounted for and that these are most similar to those that have been described for flowering plant plastids. We identified single orthologs for many of the genes encoding enzymes central to fatty acid biosynthesis, e.g., 3-hydroxyacyl-ACP-dehydratase and enoyl-CoA reductase (Table 2, 4c and 4d), components of the type II fatty acid synthase, and the predicted proteins have high and approximately equal probabilities of being targeted to both mitochondria and chloroplasts, as judged by TargetP analysis (12).

The pathway of sulfoquinovosyldiacylglycerol (SQDG) biosynthesis (Fig. 1, steps 11 and 12) has been thoroughly characterized in Arabidopsis and shown to be the function of the products of two genes, namely, SQD1 and SQD2 (14, 59). SQD1 has been characterized in Chlamydomonas by both insertional inactivation (35) and heterologous complementation of a sulfolipid-deficient sqdB cyanobacterial mutant (43), and the SQD1 gene, encoding UDP-sulfoquinovose synthase (Fig. 1, step 11), is highly expressed, as judged by EST abundance. This disproportionately high mRNA abundance might be due to the relatively large amount of SQDG in Chlamydomonas compared to that in Arabidopsis, coupled with the fact that the k_cat for the reaction catalyzed by SQD1 is very low (38), a feature which might necessitate increased abundance of the enzyme. Genes encoding enzymes of galactolipid biosynthesis are also present in the genome, with MGD1 encoding monogalactosyldiacylglycerol synthase (Fig. 1, step 14) and DGD1 encoding digalactosyldiacylglycerol synthase (Fig. 1, step 15).

Several functions in plastidic lipid biosynthesis are still without candidate genes; these include the phosphatidylglycerophosphate phosphatase (Fig. 1, step 9), which is unknown in plants and algae (4), and a putative activity utilizing SQDG and an unidentified acyl donor to form 2′-O-acyl-SQDG (Fig. 1, step 13). 2′-O-Acyl-SQDG has been shown to be a minor component of Chlamydomonas lipids; however, it is apparently absent from flowering plants (35). It should also be noted that, unlike the endoplasmic reticulum (ER) isoforms, the plastidic isoform of phosphatidate phosphatase (Fig. 1, step 10), which is responsible for generating DAG for use in galactolipid and sulfolipid synthesis, is not known yet in flowering plants.

(iii) Extraplastidic lipid metabolism.

Membrane lipid metabolism outside the plastid in Chlamydomonas differs from that in flowering plants by both the presence of unique pathways and the apparent absence of certain central pathways and branches of pathways. Most notable is the tentative absence of a gene for phosphoethanolamine methyltransferase, which provides the pathway for the methylation of phosphoethanolamine to form phosphocholine. This activity is responsible for the generation of phosphocholine for conversion into CDP-choline and then PtdCho, and the enzymes from spinach and Arabidopsis have been functionally identified by their ability to rescue a cho2 mutant of Schizosaccharomyces pombe and an opi3 mutant of Saccharomyces cerevisiae, respectively (6, 32). The apparent lack of both a pathway for PtdCho precursor biosynthesis in Chlamydomonas and the PtdEtn methylation pathway, whereby PtdEtn is N trimethylated to form PtdCho, correlates well with the lack of PtdCho in cellular membranes (19).

Phosphatidylserine (PtdSer) is a major precursor of PtdEtn in yeasts, bacteria, and plants through the action of PtdSer decarboxylase (53); however, Chlamydomonas has been reported to lack PtdSer as a component of its membranes (19). This fact is corroborated by our survey of the genome, in that there is an apparent lack of genes encoding PtdSer synthase or phospholipid base exchange enzymes. These data lead us to propose that the biosynthesis of PtdEtn is the function of a single pathway, consisting of serine decarboxylase (36), to generate ethanolamine (Fig. 1, step 25); this step is followed by phosphorylation of ethanolamine (Fig. 1, step 26), activation with CTP to form CDP-ethanolamine (Fig. 1, step 27), and transfer of the phosphoethanolamine moiety to DAG to form PtdEtn (Fig. 1, step 24). The Chlamydomonas CTP:phosphoethanolamine cytidylyltransferase enzyme was recently characterized by expression in E. coli and biochemical characterization of the recombinant protein (58).

Certain pathways are assumed to be present in multiple compartments. For example, PtdGro biosynthesis is proposed to be present in the plastid, mitochondria, and ER, and genes encoding three isoforms of phosphatidylglycerophosphate synthase are present in the genome. Two of these (Fig. 1, step 8) are expressed, as judged by EST analysis, and both proteins contain a predicted plastid or mitochondrial targeting sequence. The other isoform is predicted to be cytosolic but is not represented in the EST database. This finding may correlate with the fact that PtdGro is predicted to be present at a much lower abundance in the cytosol than in the plastid (19); thus, the gene encoding the cytosolic activity might not be as highly expressed as those directing PtdGro synthesis in the organelles. It is also noteworthy that we were unable to identify a gene for cardiolipin synthase, an enzyme typically present in the mitochondria of eukaryotes.

(iv) Fatty acid desaturases.

In addition to the genes involved in the synthesis of the glycerolipids themselves, fatty acid desaturase genes were identified and annotated. The FAD gene complement of Arabidopsis has been well characterized at the genetic and biochemical levels (33), and orthologs are present in the Chlamydomonas genome, as outlined in Fig. 2. Of note are the putative extraplastidic ω6- and ω3-desaturase genes (Fig. 2, steps 38 and 39), as well as predicted plastidic orthologs of the Arabidopsis FAD5, FAD6, and FAD7 or FAD8 genes (33) (Fig. 2, steps 34 to 36). The plastidic ω6-desaturase gene of Chlamydomonas (Fig. 2, step 35) was identified previously in a mutant screening for high-chlorophyll fluorescence mutants and subsequently was cloned and characterized (41, 44). We were unable to assign candidate genes for the Δ4-desaturase involved in the generation of the unusual fatty acid 16:4^Δ4,7,10,13 (carbons:double bonds^Δpositions) in the plastid (Fig. 2, step 37) or the Δ5-desaturase involved in 18:3^Δ5,9,12 and 18:4^Δ5,9,12,15 synthesis in the ER (Fig. 2) (40).

BTA1_Cr is a bifunctional protein sufficient for DGTS biosynthesis.

The pathway of betaine lipid biosynthesis in the bacterium R. sphaeroides and the proposed involvement of the enzymes BtaA_Rs and BtaB_Rs are shown in Fig. 3. In this bacterium, the betaine lipid DGTS is formed under phosphate-limited growth conditions (5). Chlamydomonas is unusual in that its membranes are devoid of PtdCho and constitutively contain DGTS in its place. Searching of the whole-genome shotgun sequence assembly with the bacterial BtaA_Rs protein sequence (27) revealed a predicted protein (encoded by transcript C_570065) which we have named BTA1_Cr. It shows 27% identity and 40% similarity to BtaA_Rs over about 400 amino acids. In addition, portions of BTA1_Cr were similar to the bacterial BtaB_Rs protein. Indeed, examination of the genomic structure of BTA1_Cr (C_570065) revealed that the protein consists of an N-terminal domain with similarity to BtaB_Rs and a C-terminal portion with similarity to BtaA_Rs (Fig. 4A). The genomic locus is approximately 6 kb long, and the transcript is interrupted 11 times by introns (Fig. 4A). The predicted mRNA also contains a relatively long 3′ untranslated region of ∼1.6 kb which is common to many Chlamydomonas transcripts.

FIG. 3.

Reaction scheme for DGTS biosynthesis as determined in R. sphaeroides and Chlamydomonas. The bacterial reactions are carried out by individual proteins BtaA and BtaB, while Chlamydomonas contains the BTA1 protein, which is a fusion of BtaA-like and BtaB-like domains. Site-directed mutagenesis of the putative AdoMet binding motifs in the individual domains causes lesions in the pathway, as indicated by μA and μB.

FIG. 4.

Identification of a BtaA-BtaB fusion protein encoded by the Chlamydomonas genome. (A) An approximately 6-kb locus (gene model C_570065) encodes a protein with N- and C-terminal domains showing similarity to the bacterial betaine lipid biosynthesis proteins BtaB and BtaA, respectively. 5′ and 3′ untranslated regions are indicated in grey, and exons and introns are indicated as thick and thin line segments, respectively. (B) Multiple-sequence alignment of the BTA1_Cr protein and its bacterial counterparts in R. sphaeroides. The residues marked with arrowheads were targeted for alanine mutagenesis in the μA and μB proteins. The sequences of BtaB_Rs and BtaA_Rs (GenBank accession numbers AAK53561 and AAK53560, respectively) were treated as a single entity here; the starting methionine of BtaA_Rs is indicated by double underlining. The BtaB_Rs protein sequence is indicated by a solid line; the BtaA_Rs protein sequence is indicated by a broken line. Grey shading and black shading indicate conserved and identical residues, respectively; dashes indicate gaps.

Figure 4B shows a sequence alignment of BTA1_Cr and the BtaA_Rs and BtaB_Rs proteins. The residues identified for BtaA_Rs as being potentially important in AdoMet binding are conserved in BTA1_Cr, and two residues in predicted AdoMet binding motif II (24) of the BtaA-like domain (V340 and D341) are indicated in Fig. 4B. To demonstrate functional homology between the two bacterial proteins and the two domains of BTA1_Cr, amino acids V340 and D341 were chosen for the mutagenesis experiments described below, because we expected that a V340A-D341A mutant would likely be incapable of binding AdoMet. This mutation would render the protein inactive in the DGHS synthesis reaction, but since the native conformation of the protein would probably be retained, the BtaB-like portion would be expected to still be active in the methylation steps. Additionally, the analogous positions in AdoMet binding motif II (24) of the BtaB-like portion (V121 and D122) were targeted for alanine mutagenesis (Fig. 4B). In this situation, it was presumed that DGHS synthesis activity would be intact but that the methyltransferase portion would be unable to bind AdoMet and therefore would be inactive. Cloning of the BTA1_Cr cDNA (GenBank AY656806) and expression in E. coli led to the accumulation of relatively large amounts of DGTS, allowing structure elucidation. Figure 5A shows a TLC separation of lipids from E. coli expressing BTA1_Cr, and it is apparent that there is a new species accumulating in the expression strain. This lipid was subsequently identified as DGTS by positive-ion fast-atom bombardment-mass spectrometry (Fig. 5B) and NMR spectroscopy. It consists of molecular species that are typical of E. coli membrane lipids; e.g., the species at m/z 710.7 is consistent with DGTS carrying 32:1 (carbons:double bonds) fatty acids, and that at m/z 736.7 is a 34:2 species of DGTS. The ¹H NMR spectrum (data not shown) exhibited resonances consistent with those previously reported (5), most notably, a strong nine-proton resonance at 3.22 ppm, which corresponds to the methyl protons of the DGTS quaternary ammonium group.

FIG. 5.

BTA1_Cr is sufficient to catalyze the biosynthesis of DGTS. (A) E. coli harboring pBTA1 was grown and expressed as described in Materials and Methods, and lipids were separated by TLC and visualized with iodine vapor. A culture containing empty vector pQE-31 served as a negative control. CL, cardiolipin. (B) Fast-atom bombardment-mass spectrometry in the positive-ion mode was used to confirm the identity of the putative DGTS spot on the TLC plates. Peaks at m/z 710.7 and m/z 736.7 represent the most abundant molecular species, with acyl chain lengths and total double bonds (carbons:double bonds) of 32:1 and 34:2, respectively.

Kinetic analysis of BTA1_Cr.

To characterize the activity of the BTA1_Cr protein, we expressed the gene in E. coli, broke the cells by sonication, and assayed the incorporation of [1-¹⁴C]AdoMet into labeled DGHS, partially methylated DGHS derivatives, and the end product DGTS. In Fig. 6, the pH-activity profile and the determination of the AdoMet K_m for the overall reaction of BTA1_Cr are shown. The relative stoichiometries of the methylation products change with increasing pHs (Fig. 6A), indicating a difference in the pH optima between the BtaA-like and the BtaB-like domains. Figure 6B shows the quantification of the total counts in each lane of Fig. 6A, plotted against the pH; this analysis indicates that the pH optimum of the overall reaction is ∼7.2. Figure 6C shows a plot of the rate of overall product formation against substrate concentration; this analysis was used to determine, by direct fitting of a hyperbola to the experimental data, that the apparent K_m for AdoMet was 16 μM.

FIG. 6.

Biochemical properties and site-directed mutagenesis of BTA1_Cr. (A) TLC of products of BTA1_Cr at various pHs. Cell lysates were prepared and adjusted to the proper pHs as described in Materials and Methods, followed by incubation with [1-¹⁴C]AdoMet (4.2 μM) for 30 min. Products were extracted and separated by TLC, followed by exposure to a PhosphorImager screen. DGM/DS, a mixture of mono- and di-N-methylated DGHS. (B) The radioactivity in each lane of the image in panel A was quantified with the ImageQuant software package and used to determine the pH optimum for DGHS synthase activity. (C) The apparent K_m for AdoMet was determined at pH 7.2; the concentrations of AdoMet were varied. The apparent K_m was determined by direct fitting of a hyperbola to the experimental data in the Origin software package.

Site-directed mutagenesis of AdoMet binding domains of BTA1_Cr.

To corroborate the relatedness of the two domains of BTA1_Cr and the respective bacterial proteins, site-directed mutagenesis was used to change residues predicted to be critical for AdoMet binding in the active sites of the BtaA-like and BtaB-like domains. These were designated BTA1_CrμA and BTA1_CrμB, respectively (Fig. 3). The BTA1_CrμA protein was inactive in the synthesis of DGHS, as shown in Fig. 7, second lane, with only background levels of label in the DGTS or DGHS regions of the chromatogram. BTA1_CrμB, however, produced DGHS (Fig. 7, third lane), as confirmed by cochromatography with the product of a strain carrying btaA_Rs (Fig. 3) on a low-copy-number plasmid, which caused the accumulation of DGHS (Fig. 7, fourth lane). Figure 7, fifth lane, shows DGTS accumulating in the pBTA1-containing E. coli strain, and Fig. 7, sixth lane, shows that the BtaB-like methyltransferase domain in pBTA1-μA was still capable of methylating DGHS produced by BtaA_Rs. These results establish the functions of the two domains of BTA1_Cr as being comparable to those of the individual bacterial orthologs and corroborate the hypothesis that BTA1_Cr is capable of carrying out all reactions necessary for DGTS biosynthesis in Chlamydomonas.

FIG. 7.

Analysis of BtaA-like and BtaB-like domains of BTA1_Cr by site-directed mutagenesis. E. coli harboring the combinations of plasmids indicated above the lanes was labeled with [1-¹⁴C]methionine. The lipids were separated by TLC, followed by exposure to a PhosphorImager screen.

DISCUSSION

With the exception of our case study of BTA1_Cr, all of our conclusions are based on an in silico analysis of genome sequences of Chlamydomonas. We present an annotation of glycerolipid biosynthetic genes and a subsequent metabolic pathway reconstruction, which was guided in part by our general knowledge of lipid biosynthesis in Arabidopsis and other plants. Genome comparisons between less well-studied organisms, such as Chlamydomonas, and more well-studied model organisms, such as Arabidopsis, at least with regard to lipid metabolism, provide great opportunities to rapidly assess the evolution or adaptation of, in this case, Chlamydomonas; they also permit one to focus on the essential differences, e.g., the absence of PtdCho and the biosynthesis of DGTS in Chlamydomonas. However, one should not forget that annotation by an in silico analysis primarily serves to build hypotheses, which subsequently need to be rigorously corroborated to gain generally accepted knowledge. For example, the absence of a certain enzyme or pathway does not mean that alternative pathways do not exist. Likewise, predicting the subcellular localizations of proteins from signal sequences is possible only with limitations. Keeping these caveats in mind, we cautiously draw the conclusions discussed below, which hopefully will stimulate and guide further membrane lipid research with Chlamydomonas.

Chlamydomonas lipid metabolism is less complex than that of flowering plants.

During the course of bioinformatic analysis and annotation, several themes became apparent. First, the core pathways of fatty acid biosynthesis and glycerolipid assembly in Chlamydomonas appear to be simpler than those in Arabidopsis, reflected both in the presence and/or in the absence of certain pathways and in the apparent sizes of the gene families that represent the various activities. A case in point is that of galactolipid synthesis. In the Chlamydomonas genome, we have identified only one MGDG synthase gene (MGD1) and one DGDG synthase gene (DGD1), as opposed to three MGD1 isoform-encoding genes (3) and two DGD1 isoform-encoding genes (26) in the Arabidopsis genome. This phenomenon may be thought of as being parallel to the lack of PtdCho and its presumed replacement by the nonphosphorous analog DGTS in Chlamydomonas. In Arabidopsis, phosphate deprivation has been shown to up-regulate the expression of MGD2, MGD3, and DGD2, concomitant with a loss of PtdCho and an accumulation of DGDG in extraplastidic membranes (10). This scenario has also been observed in phosphate-starved oat, where DGDG, presumably produced via the phosphate starvation-inducible biosynthetic pathway, replaces a large proportion of PtdCho in the plasma membrane (2). This phenomenon is presumed to be an adaptation to phosphate limitation, in that a nonphosphorous lipid is substituting for a phospholipid, thus allowing reallocation of the limiting nutrient into more critical components, such as nucleic acids. Given that Chlamydomonas apparently has no need to replace PtdCho with DGDG due to the presence of DGTS, the phosphate-inducible pathway of DGDG biosynthesis may be expendable and has been lost during the course of evolution, or it may simply have evolved only in flowering plants.

Potential for dual targeting of proteins involved in lipid metabolism.

A second theme is apparent in the biosynthesis of fatty acids, in that some of the core components of this process may be dually targeted to both the plastid and the mitochondrion. Several lines of evidence support this hypothesis. First, it is known from studies of isolated organelles of flowering plants that fatty acid synthesis occurs in both plastids and mitochondria (54), the former being responsible for the majority of fatty acid biosynthesis to supply precursors for membrane lipid synthesis and the latter being a function of the biosynthesis of lipoic acid, a critical prosthetic group for certain mitochondrial enzymes (49). Several genes encoding enzymes that are central to fatty acid synthesis, e.g., components of the multimeric acetyl-CoA carboxylase and the type II fatty acid synthase, as well as ACPs, are potentially represented by only one gene (two genes in the case of ACPs) in the Chlamydomonas genome. Since the activities are present in multiple compartments but may be the products of a single gene, these proteins may be imported by both plastids and mitochondria; this phenomenon has a growing number of precedents, especially for proteins involved in core metabolic processes, such as nucleotide biosynthesis and DNA replication (7, 55).

In agreement with these observations, the Chlamydomonas proteins in question have a relatively high probability of dual targeting to mitochondria and plastids, as judged by TargetP analysis. The recent finding of a potentially dually targeted multifunctional acetyl-CoA carboxylase in grasses (15) also lends credence to this hypothesis. Additionally, Beisson et al. (4) were unable to predict candidates for the mitochondrial 3-ketoacyl-ACP-reductase and 3-hydroxyacyl-CoA-dehydratase activities of Arabidopsis. Given that these activities have been demonstrated implicitly by the fact that mitochondria can synthesize fatty acids, it is reasonable to posit that dual targeting of these proteins could be operative in both Arabidopsis and Chlamydomonas, although direct experimental evidence is currently lacking.

BTA1_Cr is a bifunctional enzyme involved in DGTS biosynthesis.

BTA1_Cr was identified based on similarities to the bacterial betaine lipid synthesis proteins BtaA_Rs and BtaB_Rs, encoding the presumed AdoMet:DAG 3-amino-3-carboxypropyltransferase and the DGHS methylase, respectively (Fig. 3). To experimentally verify the predicted enzymatic activities of BTA1_Cr, we expressed the BTA1_Cr cDNA (GenBank AY656806) in E. coli. This procedure resulted in DGTS biosynthesis from the expression of a single protein, thus confirming that BTA1_Cr catalyzes all of the reactions specific to the pathway. The production of the BtaA-like portion of BTA1_Cr by itself in E. coli was expected to result in DGHS accumulation, as was shown for BtaA_Rs (Fig. 7, fourth lane); however, we were unable to show that the truncated protein has DGHS synthesis activity in E. coli (data not shown).

As an alternative way of revealing the functions of the two domains of BTA1_Cr, we altered residues in the predicted AdoMet binding sites of the BtaA-like and BtaB-like domains. The results of this experiment were informative on several levels. First, the fact that the BTA1_Cr V340A-D341A mutant (BTA1_CrμA) was inactive in the synthesis of DGHS and DGTS lends credence to the model that the BtaA-like domain binds AdoMet in a methyltransferase fold. Given the structure of the cognate binding site of methyltransferases, V340 is expected to make favorable hydrophobic interactions with the adenine moiety of AdoMet, and D341 is proposed to form hydrogen bonds to the 2′ and 3′ hydroxyls of the ribose portion. Mutation of these residues to alanine rendered the DGHS synthase function of BTA1_Cr inactive or at least compromised it to below the limits of detection. The DGHS trimethylase function was left unaffected, given that the BTA1_CrμA protein was still capable of methylating DGHS produced by simultaneous expression of the bacterial btaA_Rs gene (Fig. 7, sixth lane). This finding provided an additional piece of information regarding the sequence of reactions, showing that the BtaB-like domain of BTA1_Cr is able to take DGHS from the membrane as a substrate, rather than that it is transferred directly from the active site of the BtaA-like domain, as would be the case for substrate channeling. These data are in agreement with the findings of in vitro assays conducted with isolated C. reinhardtii membranes (29) as well as the results presented in Fig. 6A, showing an accumulation of a transient pool of DGHS in in vitro assays.

Enzyme mechanisms for DGTS biosynthesis are conserved across kingdoms.

BTA1_Cr activity was measured in a cell-free system, relying on endogenously formed DAG, the presumed lipid substrate, and exogenous radiolabeled AdoMet. The analysis was somewhat complicated by the fact that DGHS synthase and methyltransferase activities were present on the same polypeptide; thus, multiple products were formed. However, measurement of all products on the same TLC plate (Fig. 6A) made this a trivial point, in that the sum of DGHS and its methylated products gave the total flux through the BtaA-like domain, and this quantity was used in the analyses shown in Fig. 6B and C. In fact, Fig. 6A is informative as to the activity of the BtaB-like (N-methyltransferase) portion, showing that the formation of the mono- and dimethylated DGHS intermediates is preferred as the pH increases and that the final methylation (forming DGTS) is favored at an even higher pH (∼8). The pH dependence of these subsequent methylation reactions may reflect a change in the pK_a values of amine species with higher degrees of methylation. The apparent K_m for recombinant BTA1_Cr activity is presumably directly comparable to that for the activity that was measured in C. reinhardtii membranes (29). However, the value for the recombinant enzyme (∼16 μM) was about fivefold lower than that for the native enzyme, indicating that the membrane composition and environment of the enzyme may affect the affinity for AdoMet of the DGHS synthase module. This scenario is not surprising, considering that a thorough kinetic description would require both AdoMet and DAG concentrations to be varied; i.e., at higher DAG concentrations (as might occur in E. coli membranes), the apparent K_m of the BTA1_Cr A domain for AdoMet might be lower. The question of the true K_m remains unanswered given our current inability to reconstitute BTA1_Cr into the defined lipid environment of a synthetic liposome.

Taken together, the results presented here show that the enzymes and mechanisms of DGTS biosynthesis are conserved between eukaryotes and prokaryotes, with the exception that the eukaryotic activities are fused into a single polypeptide. This latter phenomenon is common among metabolic enzymes, such as urease, which consists of three separate structural components in Klebsiella aerogenes that are fused together on a single polypeptide in legumes (20), and acetyl-CoA carboxylase, which is present as a multimeric complex in bacteria but consists of a single multifunctional polypeptide in eukaryotes (30). Apparently, the same regulatory advantages that are in play in those systems are important in DGTS biosynthesis as well; i.e., having one gene to encode the machinery for the entire pathway presumably alleviates the coordination required for the proper expression of multiple individual genes.

Conclusions.

In this study, we identified a set of genes encoding products that are likely to be involved in membrane glycerolipid biosynthesis in Chlamydomonas. However, the assembly of a biological membrane is more than the sum of its biosynthetic reactions, and there are undoubtedly layers of complexity which this study fails to reach. One area in which this gene set is lacking is lipid transport, which is likely to be critical in maintaining the proper lipid composition of a given membrane. For example, the situation of a lipid being synthesized in one membrane and transported to another must be accounted for when one is devising a mechanistic description of membrane biogenesis. Likewise, simply knowing the genes involved in PtdEtn and DGTS biosynthesis (or that of any other set of lipids in a given membrane) implies nothing about how the final stoichiometric ratio between them is regulated so as to form a functional membrane with the correct biophysical properties. Given the apparent lack of redundancy in the pathways described above, the absence of complicating factors, such as tissue specificity, and the potential for functional analysis by gene silencing, Chlamydomonas is poised to be a useful model for studying how plant cells sense and control the flux of precursors into different polar lipid biosynthesis pathways in order to produce a functional membrane.