Functional Overlap of Long-Chain Acyl-CoA Synthetases in Arabidopsis

Abstract

Long-chain acyl-CoA synthetases (LACSs) play diverse and essential roles in lipid metabolism. The genomes of model eukaryotic organisms encode multiple LACS genes, and the substrate specificities of LACS homologs often overlap substantially. Homologous LACSs tend to differ in their expression patterns, localizations, and, by extension, the metabolic pathways to which they contribute. The Arabidopsis genome encodes a family of nine LACS genes, which have been characterized largely by reverse genetic analysis of mutant phenotypes. Because of redundancy, distinguishing the contributions of some Arabidopsis LACS genes has been challenging. Here, we have attempted to clarify the functions of LACSs that functionally overlap by synopsizing the results of previous work, isolating a suite of higher-order mutants that were previously lacking, and analyzing oil, wax, cutin, cuticle permeability, fertility and growth phenotypes. LACS1, LACS2, LACS4, LACS8 and LACS9 all affect cuticular lipid metabolism, but have different precise roles. Seed set, seed weight and storage oil amounts of higher-order lacs1, lacs2, lacs4, lacs8 and lacs9 mutants vary greatly, with these traits subject to different effects of fertility and oil synthesis defects. LACS4, LACS8 and LACS9 have partially redundant roles in development, as lacs4 lacs8 and lacs4 lacs9 double mutants are dwarf. lacs4 lacs8 lacs9 triple mutants were not recovered, and are assumed to be non-viable. Together, these results sketch a complex network of functions and functional interactions within the Arabidopsis LACS gene family.

Sequence data from this article can be obtained from the Arabidopsis Genome Initiative database under the following accession numbers: LACS1 (At2g47240), LACS2 (At1g494300), LACS4 (At4g23850), LACS8 (At2g04350) and LACS9 (At1g77590)

Introduction

Fatty acids are building blocks for a variety of lipids. In plants, they are incorporated into membrane glycerolipids, sphingolipids and storage triacylglycerols, and they serve as precursors of surface waxes, cutin and suberin. The synthesis and breakdown of fatty acids and their downstream products is divided among many organelles in plant cells; for example, fatty acids are synthesized in plastids, elongated on the endoplasmic reticulum (ER) and catabolized by β-oxidation in peroxisomes. Both for transport between organelles and for passage through membranes to enter or exit organelles, fatty acids are generally esterified to coenzyme A (CoA). Fatty acids are also activated with CoA for most metabolic processes they are involved in. Thus, the activities of long-chain acyl-CoA synthetases and very-long-chain acyl-CoA synthetases (LACSs and VLACSs) are of fundamental importance to lipid metabolism.

The spatial distribution of LACS enzymes within the cell is a factor in channeling fatty acids toward a specific metabolic fate (Digel et al. 2009, Bu and Mashek 2010). Consistent with this, in most eukaryotes LACS activity is encoded by gene families that differ in their subcellular locations and expression patterns, but often exhibit significant overlap in their substrate specificities, such as the fatty acid translocases, LACSs and ‘bubblegum’ protein families in human (Digel et al. 2009), and Saccharomyces fatty acid transport proteins and fatty acyl-CoA synthetases (Black and Dirusso 2007). This is also the case in Arabidopsis, which contains one of the largest known LACS families with nine LACS enzymes (Shockey et al. 2002). All Arabidopsis isoforms have been reported to be metabolically active after expression in Escherichia coli, and in vitro enzyme assays demonstrated that all LACSs can effectively activate a broad range of substrates. For most, the highest levels of activity were detected with saturated and monounsaturated 16-carbon fatty acids and mono- and polyunsaturated 18-carbon fatty acids (Shockey et al. 2002, Lü et al. 2009, Pulsifer et al. 2012). Additionally, the peroxisomal isoforms LACS6 and LACS7 exhibited high activity with 20:1 (Fulda et al. 2002, Shockey et al. 2002), and LACS1 and LACS2, which are involved in cuticular wax biosynthesis, also had high activity with C₂₀–C₃₀ very-long-chain fatty acids (Lü et al. 2009).

Mutants of most of the Arabidopsis LACS genes have been isolated. Characterization of these mutants, and analysis of expression patterns and subcellular localizations, has revealed a complex network of redundant or partially redundant LACS activities involved in different aspects of lipid metabolism. This redundancy accounts for the observation that only the cer8/lacs1 mutant, deficient in cuticular wax biosynthesis, has been identified in a forward genetic screen (Lü et al. 2009). Reverse genetic characterization of the LACS family showed that LACS9, uniquely localized in the outer membrane of the plastid envelope (Schnurr et al. 2002, Breuers et al. 2012), and LACS4, which is localized to the ER, are both required for storage lipid biosynthesis as well as lipid transport from the ER to the plastid (Jessen et al. 2015). Additionally, LACS9 and LACS4 functionally overlap with LACS8, another ER-specific isoform, as the disruption of LACS8 in the lacs4 lacs9 double mutant background results in lethality (Jessen et al. 2015). Characterization of the lacs6 lacs7 double mutant demonstrated that LACS6 and LACS7 proteins have overlapping roles in fatty acid beta-oxidation in the peroxisome. Whereas single T-DNA insertional mutants were indistinguishable from the wild type, the lacs6 lacs7 double mutant was defective in seed oil mobilization required for seedling establishment (Fulda et al. 2004). Similarly, several LACS isoforms activate fatty acids for the production of cuticular lipids. These include ER-localized LACS1, LACS2 and LACS4 (Lü et al. 2009, Weng et al. 2010, Jessen et al. 2011), and possibly LACS3, which is strongly expressed in stem epidermis (Suh et al. 2005) but has not been characterized to date. Chain length specificity analyses and phenotypes of lacs1 mutants suggest that LACS1 has a primary role in generating very-long-chain fatty acyl-CoAs that serve as precursors for cuticular wax components. LACS2 appears to act in concert with LACS1 in the activation of VLCFAs for the production of wax components and plays a role in the incorporation of C₁₆ and C₁₈ acyl groups into cutin (Lü et al. 2009, Weng et al. 2010). The ER-resident LACS4 was also reported to be partially redundant with LACS1 in providing substrate for cuticular wax biosynthesis on stems and leaves, and pollen coat (tryphine) lipid formation (Jessen et al. 2011). Finally, five of the nine LACS genes are expressed in developing seeds; LACS1, LACS2, LACS4, LACS8 and LACS9 (Shockey et al. 2002, Zhao et al. 2010), and though no single mutants of these genes displayed seed oil-related phenotypes, lacs1 lacs9 and lacs4 lacs9 double mutants both produced less oil than wild-type seeds (Zhao et al. 2010, Jessen et al. 2015).

Clearly, functional redundancy within the Arabidopsis LACS gene family exists with respect to surface lipid barrier, membrane lipid and seed oil synthesis. We endeavored to better understand the division of functions within the LACS family by analyzing a suite of higher-order mutants including LACS1, LACS2, LACS4, LACS8 and LACS9. LACS3 was excluded from our analysis because no null mutants were available at the outset of our investigation, LACS5 was omitted as it is only expressed in anthers (Schmid et al. 2005, Winter et al. 2007), which are not directly relevant to the phenotypes we are investigating, and LACS6 and LACS7 were left out because they are involved in fatty acid β-oxidation, not the aforementioned processes. This study advances and synthesizes our current understanding of the functions of Arabidopsis LACS enzymes and presents several new questions regarding the function of this gene family as a whole.

Results

The LACS4 gene is expressed throughout the plant

Of the five LACSs of interest, LACS4 has been studied the least. Yet, its expression pattern suggests that it may be involved in the synthesis of membrane, storage and surface lipids; we therefore initiated our investigation by further characterizing this gene. Shockey et al. (2002) previously showed using reverse transcription-polymerase chain reaction (RT-PCR) that the LACS4 gene is expressed in roots, leaves, flowers, developing seeds and germinating seedlings. More recently, transcription of LACS4 in stems was also demonstrated by real-time polymerase chain reaction (qPCR) (Zhao et al. 2010). To determine the tissue-specific expression profiles of the LACS4 gene more comprehensively, we carried out promoter::β-glucuronidase (GUS) analysis. The LACS4 5′ promoter fragment was cloned upstream of the GUS reporter cDNA sequence and was transformed to wild-type Arabidopsis plants. Different tissues from 10 independent T₁ lines were observed to establish GUS expression patterns. Consistent with published qPCR data (Zhao et al. 2010), GUS activity in transgenic plants carrying the LACS4promoter::GUS was detected in a variety of tissues including seedlings, roots (except the elongation zone), stems, leaves, flowers and siliques (Fig. 1A–L). Additionally, GUS staining was strong throughout embryo development (Fig. 1M). LACS4 expression in the embryo was validated by in situ hybridization at 7 days post anthesis (DPA; Fig. 1N). The seed-specific expression of FAE1 shown by in situ hybridization was used as a positive control (Fig. 1O; Rossak et al. 2001).

Fig. 1

Tissue-specific expression patterns of LACS4 in Arabidopsis. (A–M) Stained tissues expressing the reporter gene encoding GUS driven by the LACS4 promoter: seedlings (A); root (B); root tip (C); stem (D); stem cross-section (E); rosette leaf (F); flowers (G, H); anther (I); pollen grains (J); silique (K, L) and developing embryos from 3 to 12 DPA (M). (N) Expression of LACS4 in embryo as detected by in situ hybridization at 7 DPA using an antisense DNA probe. (O) Transcript of FAE1 shown in a 7 DPA embryo by in situ hybridization as a positive control. (P) No detection of LACS4 expression in embryo by in situ hybridization at 7 DPA using a sense probe as a negative control. Scale bars = 1 mm (A, D, F,G, H, K, L), 100 µm (B, C, E, J, M, N, O, P).

The lacs4 mutant is phenotypically indistinguishable from the wild type

We investigated the functions of LACS4 through a reverse genetic approach. Eight T-DNA insertional mutant lines were obtained from the Arabidopsis SALK collection (Alonso et al. 2003). Three of the lines had identical T-DNA insertions; the six independent alleles we collected were designated lacs4–1 to lacs4–6. Homozygous mutants were identified by two sets of PCR amplifications to ensure the insertions were homozygous. Gene structure and exact locations of the T-DNA in each line were established by sequencing the T-DNA-gene junction (Fig. 2A; Supplementary Table S1).

Fig. 2

Mutant alleles of LACS4 gene. (A) The precise sites of T-DNA insertions in the LACS4 gene were determined and are indicated by vertical arrows above the gene. Boxes represent exons, gray boxes represent untranslated regions and solid lines represent introns and intergenic areas. P1 and P2 are the two sets of primers designed for RT-PCR. (B) RT-PCR analyses of LACS4 mRNA in lacs4 mutants compared to the Col 0 wild type (WT). Total RNA was extracted from leaf tissue and the expression level of GAPC was employed as a loading control.

Total RNA isolated from leaves of wild type and homozygous mutants was used for RT-PCR analysis to determine the effect of gene disruption on transcript accumulation (Fig. 2B). The GAPC gene was used as an internal control. A small fragment (P1) from the 3′ end of the transcript, which includes the lacs4–2 insertion site, could not be amplified from cDNA of lacs4–2 or lacs4–3. A larger fragment from the middle of the cDNA sequence (P2), which includes the lacs4–1 insertion site, could not be amplified from lacs4–1. The remaining lacs4 mutants all had either reduced or wild-type levels of both P1 and P2 fragments. Based on the positions of T-DNA insertions and transcript accumulation, lacs4–1 and lacs4–2 were selected for phenotypic analysis. It is notable that fragment P1 was detected in lacs4–1, and trace levels of fragment P2 could be detected in lacs4–2. The presence of these truncated transcripts may have contributed to subtle differences in the phenotypes of these two lacs4 alleles.

The lacs4 mutants were indistinguishable from wild-type plants in appearance, growth and development under our growth conditions. This is not surprising considering that LACS4 shares high amino acid sequence similarity with several other LACS family members including LACS1, LACS2, LACS3 and LACS5 (Shockey et al. 2002). In addition, LACS1, LACS2, LACS8 and LACS9 were also shown to be co-expressed with LACS4 in the stem or in the seed (Zhao et al. 2010). To verify that LACS4 has overlapping functions with other LACSs, we generated double and triple mutants among LACS4, LACS1, LACS2, LACS8 and LACS9.

lacs1 lacs2 lacs4 triple mutant exhibits reduced fertility and severe wax deficiency

The lacs1–3 and lacs2–3 mutant alleles were crossed to the lacs4–1 and lacs4–2 mutants respectively to generate lacs1 lacs4 and lacs2 lacs4 double mutants, which were crossed to each other to produce a lacs1 lacs2 lacs4 triple mutant. As reported previously, lacs1, lacs1 lacs2 and lacs1 lacs4 mutants had reduced seed set, reduced silique size and wax-deficient stems (Lü et al. 2009, Weng et al. 2010, Zhao et al. 2010, Jessen et al. 2011; Fig. 3). The lacs2 and lacs2 lacs4 mutants developed both fertile and sterile siliques (without seeds) on the same stem. The lacs1 lacs2 lacs4 triple mutant, however, showed more pronounced phenotypes than any of the single or double mutants, including dramatically reduced seed set and severely wax-deficient stems (Fig. 3).

Fig. 3

Photograph of inflorescence stems and siliques from 6-week-old plants. The lacs1 single, double and triple mutants all exhibited greener stems, indicating wax deficiency, as well as reduced seed set and silique size. The lacs2 single, double and triple mutants all produced both fertile and sterile siliques.

lacs1 lacs2 lacs4 triple mutant leaves have permeable cuticles and exhibit increased water loss

To assess cuticle functionality, we first investigated cuticle permeability of the single, double and triple mutants using the toluidine blue test (Fig. 4A). Toluidine blue is a metachromatic dye that selectively binds free anionic groups such as carboxylate, phosphate and sulfate radicals (Tanaka et al. 2004). If the cuticle is permeable, the dye will penetrate epidermal cells and stain the leaf. Leaves of mutants with defects in wax or cutin biosynthesis have been demonstrated to stain rapidly, while wild-type leaves do not stain even after prolonged exposure to the dye (Tanaka et al. 2004). When rosette leaves of lacs2–3 were immersed in toluidine blue solution they had a patchy staining pattern. Lacs1 lacs2 and lacs2 lacs4 were more intensely stained, and the most prominent staining was detected in the lacs1 lacs2 lacs4 triple mutants. In contrast, wild-type leaves, as well as lacs1, lacs4, or lacs1 lacs4 leaves were not stained by toluidine blue, indicating that their cuticles excluded the dye. These results demonstrate that lacs1 lacs2 lacs4 triple mutant leaves have the greatest cuticle permeability, followed by the lacs1 lacs2 and lacs2 lacs4 double mutants and the lacs2 single mutant.

Fig. 4

Cuticle permeability assays. (A) Toluidine blue staining of rosette leaves. Leaves were placed in an aqueous solution (0.025%) of toluidine blue for 4 min and rinsed with distilled water three times. Representative rosette leaves of each genotype compared with the Col-0 WT are shown. The lacs2 single, double and triple mutants were all stained in blue with increased intensity in the double and triple mutants. (B) Chlorophyll leaching from rosette leaves. Rosettes were submerged in 80% ethanol and aliquots were taken at the given time points to assess the chlorophyll concentration of the ethanol. Results shown are means from three replicates [±standard deviation (SD)]. To reduce the figure complexity, only one allele of each genotype is included. The experiments were repeated once with similar results. (C) Water loss rate of rosette leaves. Rosette leaves were excised from the whole plant and weighed at the indicated time points. Water loss rate was determined as percentage of water loss over total water loss over 24 h. Numbers are means of three samples ± SD. Only one allele of each genotype is shown to reduce figure complexity. The experiments were repeated once with similar results.

To further evaluate the cuticle defects of the double and triple mutants, we carried out chlorophyll leaching analysis (Fig. 4B). Ethanol can penetrate epidermal surfaces of leaves and stems, and releases chlorophyll through the cuticle. More rapid chlorophyll discharge indicates a more permeable cuticle, which may result from defects in cutin or wax deposition (Yephremov et al. 1999, Schnurr et al. 2004, Kurdyukov et al. 2006). When rosette leaves of wild type, single, double and triple mutants were submerged in 80% ethanol, lacs1 and lacs2 showed similar chlorophyll leaching rates that were about 3-fold higher than the wild type and the lacs4 single mutant. The lacs1 lacs4 double mutant released chlorophyll at a similar rate to the lacs1 or lacs2 single mutants. Lacs1 lacs2 and lacs2 lacs4 double mutants lost chlorophyll 25-fold and 20-fold more quickly than the wild type, respectively. The release of chlorophyll from the leaves of lacs1 lacs2 lacs4 triple mutant was the fastest, an indication that the triple mutant has the most permeable cuticle.

To examine the physiological impact of the cuticular barrier defects we had detected, we measured the water loss rate of the fully expanded rosette leaves of the wild type and mutants. Whole rosette leaves from 5-week-old plants were cut off and weighed over the course of 8 h (Fig. 4C).

The water loss from the leaves of lacs4 and lacs1 lacs4 was only slightly faster than that in wild type and lacs1 plants. Lacs2 and lacs1 lacs2 lost water two times faster than the wild type, whereas the lacs2 lacs4 double and lacs1 lacs2 lacs4 triple mutant leaves exhibited the most rapid loss of water; over 70% and 80% of total water was lost in the first 3 h after leaf excision in these two genotypes, respectively. These data indicate that the ability of the leaf cuticle to limit non-stomatal water loss is most severely impaired in the lacs2 lacs4 and lacs1 lacs2 lacs4 mutants, less impaired in the lacs2 and lacs1 lacs2 mutants, only marginally impaired in the lacs4 and lacs1 lacs4 mutants, and not impaired in the lacs1 mutant.

Stem wax and cutin loads are altered in the lacs1 lacs2 lacs4 triple mutant

To obtain biochemical evidence that the biosynthesis of cuticle components of the mutants is disrupted, we analyzed stem wax and cutin loads. The lacs1, lacs1 lacs2, lacs1 lacs4 and lacs1 lacs2 lacs4 mutants displayed a glossy stem phenotype, which is a hallmark of altered cuticular wax composition and/or load. This was reflected in the total wax load determined by gas chromatography equipped with a flame ionization detector (GC-FID). The total wax load in lacs1 was reduced to approximately 75% of the wild-type level, whereas the lacs1 lacs2 and lacs1 lacs4 double mutants had wax loads that were approximately 45% and 54% of wild-type level, respectively (Fig. 5). Similar quantitative differences were reported previously by other groups (Lü et al. 2009, Weng et al. 2010, Jessen et al. 2011). The most severely decreased stem wax amount, only 27% of that in wild type, was detected in the lacs1 lacs2 lacs4 triple mutants. The loss of LACS1 activity specifically led to decreased amounts of aldehydes, alkanes, ketones and secondary alcohols, which are formed through the alkane pathway, and increased quantities of fatty acids. Combined loss of LACS1 and LACS2 activities resulted in a further reduction in amounts of products from the alkane pathway, as well as reduction of output from the acyl reduction pathway. The lacs1 lacs4 double mutants had a decreased alkane, ketone and secondary alcohol content and a slightly increased accumulation of aldehydes and fatty acids. Overall, all wax components, except esters, were less abundant in the cuticles of the lacs1 lacs2 lacs4 triple mutants, whereas the lacs2, lacs4 mutants and lacs2 lacs4 double mutants had similar amounts of all waxes on their stems as wild-type plants (Supplementary Table S2).

Fig. 5

Total cuticular wax load of inflorescence stems of 6-week-old plants. Total cuticular wax load on inflorescence stems of the mutants and Col-0 WT were investigated. Each value represents the mean of three independent replicates ±[standard deviation (SD)]. Statistically significant differences between samples are labeled with different letters at P < 0.01 using a one-way analysis of variance (ANOVA) and Tukey’s test.

We also examined the stem wax deficiency of the mutants by scanning electron microscopy (SEM; Fig. 6). The wild-type stems were covered with a dense layer of irregularly shaped flat plates. These were largely absent from the lacs1 mutants, and were further diminished in the lacs1 lacs2 and lacs1 lacs4 double mutants. All epicuticular wax structures were completely lacking in the lacs1 lacs2 lacs4 triple mutants, leaving the stem surface smooth.

Fig. 6

Scanning electron micrographs of epicuticular wax crystals on inflorescence stems of dry Arabidopsis plants. Structures and densities of wax crystals were obviously reduced in lacs1 mutants compared with the Col 0 WT. The lacs1 lacs2 and lacs1 lacs4 double mutants displayed a greater reduction of wax crystals, and the lacs1 lacs2 lacs4 triple mutants showed the smoothest stem surfaces. Scale bars = 10 µm.

Total cutin monomer amounts on stems of lacs1, lacs4 and lacs1 lacs4 mutants were similar to that of Col-0 wild-type plants. In contrast, the lacs2 single and lacs1 lacs2 double mutants both showed reduced cutin loads as previously reported by other researchers (Bessire et al. 2007, Lü et al. 2009, Weng et al. 2010). Additionally, the lacs2 lacs4 and lacs1 lacs2 lacs4 mutants also contained decreased amounts of cutin; among these genotypes, the lacs1 lacs2 lacs4 triple mutants had the most severely diminished cutin loads (Fig. 7A).

Fig. 7

Cutin monomer deposition on 6-week-old inflorescence stems of Arabidopsis. Total (A) and each cutin monomer (B, C) loads on inflorescence stems of the mutants and Col-0 WT were analyzed. Each value represents the mean of four independent replicates ± [standard deviation (SD)]. Significant reductions of cutin monomers in the lacs1 lacs2 lacs4 triple mutants were detected comparing with lacs1 lacs2 double mutants. *P < 0.05.

Cutin compositional analysis revealed that lacs1 single mutants accumulated lower levels of saturated dicarboxylic acids (DCAs), 10,16-dihydroxy hexadecanoic acid and 18-hydroxy octadecadienoic acid. Except for ferulate, all cutin monomers were further reduced in lacs1 lacs2 lacs4 triple mutants compared to the amount in the parent single or double mutants (Fig. 7B,C). The mutants can be ordered according to their cutin load of DCAs and hydroxy acids in descending order as lacs1 > lacs2 > lacs1 lacs2 > lacs1 lacs2 lacs4. In sum, the cutin monomer reductions on stems were associated with mutations in LACS1, LACS2 and LACS4 genes.

Seed storage lipids are reduced in the lacs1 lacs2 lacs4 triple mutant

Consistent with previously published results (Lü et al. 2009, Weng et al. 2010, Zhao et al. 2010, Jessen et al. 2011), the lacs1, lacs1 lacs2, lacs1 lacs4 and lacs1 lacs2 lacs4 mutants were all found to be partially sterile and yield fewer seeds, with the most severe phenotypes observed in the lacs1 lacs2 lacs4 triple mutants. The lacs2 and lacs2 lacs4 mutants also had reduced seed set, we assume due to the failure of some flowers to open. Upon closer inspection of the seeds under a dissecting microscope, we noticed that seeds of the lacs1 lacs2, lacs1 lacs4 and lacs1 lacs2 lacs4 mutants were malformed, suggesting compromised seed storage oil accumulation (Fig. 8).

Fig. 8

Photographs of the seeds showed the aberrant appearance of the mutant seeds. Arrows point to the seeds with aberrant appearance. Scale bars = 500 µm.

Reduced oil amount as a percentage of seed weight was confirmed by GC-FID analysis in the lacs1 lacs2, lacs1 lacs4 and lacs1 lacs2 lacs4 seeds, which exhibited a 14%, 17% and 30% reduction of the wild-type oil level, respectively (Fig. 9A). No significant decrease in seed oil amount was detected in the lacs1, lacs2, lacs4 single or lacs2 lacs4 double mutants, which did not show the malformed seed phenotype (Fig. 9A).

Fig. 9

Fatty acid contents and weights of dry seeds. (A) Fatty acid content. Data are expressed as mean percentages ± [standard deviation (SD)] (n = 3). (B) Seed weight of 1000 dry seeds. Values are means ± SDs (n = 3). Statistical analysis using one-way ANOVA and Tukey’s test was performed to show the significant differences between samples, which are indicated by different letters at P < 0.01.

Even though double and triple mutants produced fewer normal seeds, some of these seeds seemed larger. We therefore determined the weight of 1000 seeds from wild type and different mutant genotypes (Fig. 9B). The lacs1, lacs2, lacs4 single and lacs2 lacs4 double mutant seeds had a similar weight to the wild type, whereas seeds of the lacs1 lacs2 and lacs1 lacs4 double mutants reached 140% of the wild-type weight. The most prominent increase of seed weight was detected in the lacs1 lacs2 lacs4 triple mutants that weighed 171% of the wild type.

LACS4 displays overlapping functions with LACS8 and LACS9

In addition to probing the functional overlap of LACS4 with LACS1 and LACS2, we also wanted to explore whether LACS4 exhibits redundancy with LACS8 and LACS9 by creating double and triple mutants. However, consistent with a previous study (Jessen et al. 2015), our attempt to identify the lacs4 lacs8 lacs9 triple mutant failed, suggesting that inactivity of these three genes leads to embryo lethality. Compared with the wild type, both the lacs4 lacs8 and lacs4 lacs9 double mutant seedlings developed slowly, with a more pronounced phenotype exhibited in lacs4 lacs9 mutants (Fig. 10A). At 4 weeks the wild type had fully developed elongated rosette leaves, while leaves of the lacs4 lacs8 and lacs4 lacs9 double mutants were short, round, curly and disorganized. At reproductive stage, the overall size of the lacs4 lacs8 and lacs4 lacs9 double mutant plants was reduced when compared with wild type, and both mutants had spindly stems and small siliques (Fig. 10B). These phenotypes were not observed in lacs4, lacs8 or lacs9 single mutants or the lacs8 lacs9 double mutant (data not shown).

Fig. 10

Morphology of the lacs4 lacs8 and lacs4 lacs9 mutants. (A) 12-day-old seedlings; (B) 4-week-old rosette leaves; (C) 6-week-old whole plants; (D) Photographs of the seeds.

Examination of seeds under the dissecting microscope revealed that the lacs4 lacs8 seeds were slimmer than wild-type seeds and that lacs4 lacs9 seeds were misshapen (Fig. 10D). Seed weights of the lacs4 lacs8 and lacs4 lacs9 double mutants were decreased to 90% and 76% of wild type, respectively (Fig. 11A). Total fatty acid content per seed in the lacs4 lacs8 and lacs4 lacs9 double mutants were also reduced to 88% and 53% of the wild-type level, respectively (Fig. 11B). Seed oil amount as a percentage of seed weight of the lacs4 lacs8 double mutant was not significantly changed, whereas the oil amount, as a percentage of seed weight, in the lacs4 lacs9 double mutants reached only 70% of the wild-type level (Fig. 11C).

Fig. 11

Phenotypes of the lacs4 lacs8 and lacs4 lacs9 mutants. (A) Seed weights of the mutants compared with wild type. (B) The fatty acid amounts per individual seed of mutants and wild type. (C) The fatty acid contents of the mutants and wild type as a percentage of seed weight. (D) Total wax loads on the inflorescence stems of the mutants and wild type. All values are represented as means ± SDs (n = 3). Statistical analysis using one-way ANOVA and Tukey’s test was performed to show the significant differences between samples, which are indicated by different letters at P < 0.01.

Even though lacs4 lacs8 and lacs4 lacs9 did not have bright green and shiny stems indicative of wax deficiency, we carried out cuticular wax analyses. As indicated in Fig. 11D and Supplementary Table S2, stems of the lacs4 lacs9 double mutants contain only about 74–80% of the total wild-type wax load, with decreased quantities of very-long-chain fatty acids, alkanes, ketones, secondary alcohols, primary alcohols and esters, and slightly increased amounts of aldehydes. In contrast, the wax load on the lacs4 lacs8 double mutant stems was similar to wild type (Supplementary Table S2).

Discussion

There has been a continued effort to decipher the functions and functional redundancy of members of the Arabidopsis LACS gene family (Supplementary Table S3). The focus of the present work has been LACS4 and its overlapping functions with LACS1, LACS2, LACS8 and LACS9. All of these homologs were already known to have biological functions that are, in varying degrees, redundant; we have attempted to complete and synopsize these investigations. Although LACS3 also has an overlapping expression pattern and substrate preference with LACS4, a suitable knock-out mutant allele has not been isolated to date. The ease and accessibility of CRISPR/Cas9 systems for generating mutants is likely to quickly resolve this issue. LACS5 was omitted from this study due to its unique expression pattern in anthers; however, it will be interesting to investigate its role in fertility, and possible redundancy with LACS1 and LACS4. LACS6 and LACS7 were omitted due to their known, paired role in peroxisomal β-oxidation of fatty acids.

LACSs and cuticular lipids

Past studies have described the overlapping functions of LACS1, LACS2 and LACS4 in cuticular lipid metabolism, using single and double mutants (Lü et al. 2009, Weng et al. 2010, Jessen et al. 2011). LACS1 is unique in that its mutant has an obvious glossy stem phenotype, caused by a substantial reduction in alkane and alkane-derived waxes. This is only partially compensated for by increased free fatty acid content. The increase we detected in the amount of free fatty acids in lacs1 is exponentially greater than that described previously by Lü et al. (2009) and Jessen et al. (2011), but is in agreement with Jenks et al. (1995). This dramatic increase is only seen in the lacs1 single mutant and lacs1 lacs8 and lacs1 lacs9 double mutants (Zhao et al. 2010). The other double mutants involving lacs1 do not accumulate as much free fatty acid; the combination of the lacs1 mutation with either lacs2 or lacs4 appears to offset this phenotype (Supplementary Table S2).

The requirement of a very-long-chain acyl-CoA synthetase for the production of cuticular wax alkanes is puzzling, since the very-long-chain fatty acid precursors of cuticular waxes are elongated as acyl-CoA thioesters. There are no known acyl-CoA thioesterases that participate in cuticular wax metabolism, and there is no obvious reason why the acyl-CoA products of elongation would be released from CoA only to be re-esterified for subsequent alkane synthesis. Regardless, if alkane production does require LACS1 activity, the lacs1 mutation should indeed result in a buildup of free fatty acids at the chain lengths required for cuticular wax biosynthesis. Why lacs2 or lacs4 mutations compensate for this increased production and deposition of free fatty acids is unclear. Perhaps the activation of fatty acids that LACS2 and LACS4 catalyze provides substrate for the elongation of fatty acid precursors of wax. Thus, in their absence, less VLCFAs are produced.

Only LACS1 and LACS2 were previously reported to contribute to stem cutin biosynthesis. Lü et al. (2009) and the present study both detected compositional differences in the cutin of the lacs1 single mutant (Fig. 7). Lü et al. reported an increase in 18:2 DCA, and either a decrease or no change in the load of all other aliphatic cutin monomers. We observed either no change or a slight decrease in all aliphatic monomers, including 18:2 DCA. However, these changes were subtle, and in many cases were insignificant. Even among our own experimental replicates and analyses of different lacs1 alleles, we observed substantial variation in cutin composition. This could result in the observation of either no change or a subtle decrease in total cutin load, compared to the slight increase reported by Lü et al. We attribute this variation to slight changes in growth conditions, development at the time of tissue harvesting, and allele-specific phenotypes. The phenotype of the lacs2 mutant was more robust; substantial decreases in the load of all aliphatic monomers and the total cutin load were detected by both Lü et al. and in the present work. Similarly, both studies reported a more severe reduction in the cutin load of lacs1 lacs2 double mutants. We found that the addition of lacs4 in the lacs1 lacs2 lacs4 triple mutant worsened the cutin load deficiency, although the lacs4 lesion did not affect cutin in either the single mutant or the lacs1 lacs4 and lacs2 lacs4 double mutants. We conclude that LACS2 plays a major role in cutin biosynthesis in stems, and that LACS1 and LACS4 can also participate in the production of acyl-CoA substrates for cutin biosynthesis. Overall, the phenotypes of the lacs2 mutant and higher-order mutants including a lacs2 lesion were the most conspicuous with respect to fertility, cutin composition and cuticle function.

It is interesting to note that none of the lacs mutants investigated to date have been reported to have suberin defects. This may be because assays such as tetrazolium salt penetration for seed permeability, or Fluorol yellow staining of root endodermis, have not been systematically used to screen lacs mutants.

LACSs, seed development and seed oil

Multiple acyl-lipid metabolic pathways contribute directly and indirectly to the size and oil content of seeds. It is therefore not surprising that we observed a range of seed size and oil content phenotypes among the lacs mutants we analyzed. The development of lacs2 flowers seems to vary according to growth conditions and the allele studied. There are two reports of lacs2 flowers opening normally (Lü et al. 2009, Weng et al. 2010), but we observed fusion of floral organs that interfered with anthesis in lacs2 (Fig. 3). Lacs2, and the double and triple mutants we generated that included the lacs2 lesion, therefore had a reduced seed set due to the presence of both normal siliques and siliques that had failed to develop entirely.

Jessen et al. (2011) reported and quantified a tryphine defect in lacs1 that resulted in a humidity-dependent reduction in male fertility. LACS4 was shown to contribute to male fertility as well, as lacs1 lacs4 double mutants were male-sterile at low humidity and had reduced tryphine content relative to lacs1 (Jessen et al. 2011). Our experiments produced similar results. The contribution of lacs4 to male fertility is not straightforward, however, as the single lacs4 mutant was found to have increased tryphine content, and decreased amounts of internal pollen lipids (Jessen et al. 2011).

The seeds of lacs1 lacs2 and lacs1 lacs4 double mutants were heavier than wild type and single mutants, and had reduced oil amount as a percentage of seed weight (Fig. 9). Increased seed size is commonly observed in mutants with fertility defects, and reflects increased allocation of resources to individual seeds when the seed set is reduced. While the cause of the decreased relative oil amount of these mutants is less obvious, we speculate that it could be due to oil biosynthesis not keeping pace with the enhanced growth of individual seeds. Additionally, analysis of lacs1 lacs9 double mutants (Zhao et al. 2010) revealed that LACS1 contributes to oil metabolism. Perhaps the lacs1 single mutant does not have an oil-associated phenotype under normal conditions, but its contribution becomes apparent when seed growth is intensified in the lacs1 lacs2 and lacs1 lacs4 double mutants.

The importance of LACS1, LACS4, LACS8 and LACS9 for oil metabolism has been investigated in two previous reports (Zhao et al. 2010, Jessen et al. 2015). Reduced fatty acid amount as a percentage of seed mass was previously reported for lacs1 lacs9 (Zhao et al. 2010) and reduced fatty acid content per seed was reported for lacs4 lacs9 (Jessen et al. 2015). Here, we showed that both lacs4 lacs8 and lacs4 lacs9 seeds had reduced mass compared to the wild type, that both double mutants also had reduced oil content (per seed), and that lacs4 lacs9 has reduced fatty acid amount as a percentage of seed mass (Fig. 11). It is not surprising that lesions in LACS9 resulted in defective oil metabolism, as its product is the only LACS protein localized to the chloroplast (Schnurr et al. 2002, Shockey et al. 2002, Jessen et al. 2015). Indeed, it is intriguing that oil production as well as other lipid metabolic pathways, such as cuticular wax biosynthesis, are not affected more substantially in lacs9 mutants.

Conclusions

Through the isolation and characterization of a large set of higher-order mutants, multiple research groups have collectively succeeded in deciphering, in general terms, the roles of most Arabidopsis LACS genes. With respect to the mutants investigated in this work, we can summarize that LACS1 has a major role in cuticular wax biosynthesis, and that LACS2 and LACS4 also contribute to wax metabolism. Analogously, LACS2 has a major role in cutin biosynthesis, and LACS1 and LACS4 also participate in this process. LACS1 and LACS4 both contribute to tryphine lipid metabolism, and therefore, male fertility. LACS1, LACS4 and LACS9 all participate in oil biosynthesis. Finally, LACS4, LACS8 and LACS9 are all required for normal vegetative growth.

These studies have highlighted several questions that require further investigation. First, it must be established which LACSs are used for suberin biosynthesis. More broadly, the in vivo substrate specificities of several LACSs require further investigation. Substrates of LACS1 and LACS2 are apparent from the wax and cutin profiles of their mutants. Jessen et al. (2015) explored the in vivo substrates of LACS4 and LACS9 using the lacs4 lacs9 double mutant, which revealed a complex role for these enzymes in triacylglycerol metabolism as well as lipid import into the plastid. Free fatty acid, acyl-CoA, and in general more extensive lipid profiling of other LACS single and higher-order mutants will certainly reveal more intricacies of their biological functions. Further, the role of LACS4 in both tryphine and internal pollen lipid metabolism requires elucidation. Finally, the contributions of LACS3 and LACS5 to acyl-lipid metabolism have yet to be investigated.

Materials and Methods

Plant material and growth conditions

Seed stocks of Arabidopsis (Arabidopsis thaliana) lacs1–1 (At2g47240; SALK_127191), lacs1–3 (SALK_138782), lacs4–1 (At4g23850; SALK_101543), lacs4–2 (SALK_126610), lacs4–3 (SALK_110108), lacs4–4 (SALK_079329), lacs4–5 (SALK_147465), lacs4–6 (SALK_140210), lacs8–1 (At2g04350; SALK_136060) and lacs9–2 (At1g77590; SALK_111835) were provided by the Arabidopsis Biological Resource Center (Alonso et al. 2003) (http://www.arabidopsis.org/). The seeds of lacs2–3 (At1g49430; GABI_368C02) were as published by Bessire et al. (2007). The seeds of the lacs1–1 lacs2–3 double mutant were obtained from Dr. Owen Rowland (Lü et al. 2009). Single mutants were genotyped using primers generated by the SALK T-DNA primer design program (http://signal.salk.edu/tdnaprimers.2.html). Double and triple mutants were obtained by crossing homozygous mutant lines and selecting the appropriate individual plants from the F2 generation by genotyping. The exact location of the T-DNA insertion in the mutant alleles was determined by sequencing the PCR product amplified using the T-DNA border primer and one gene-specific primer. All seeds were germinated on Arabidopsis medium (Somerville and Ogren 1982) supplemented with agar (7 g·l⁻¹). Seven-day-old seedlings were transplanted into soil (Sunshine Mix 4, SunGro) and grown under continuous light (90–110 µEm⁻²·s⁻¹ of photosynthetically active radiation) at 20°C in an environmental chamber until maturity.

Plasmid construction and plant transformation

To generate ProLACS4:GUS, a 1664-bp long promoter region of LACS4 was amplified from leaf genomic DNA extracted from Col 0 wild-type plant using high fidelity Pfx polymerase (Invitrogen, Carlsbad, CA, USA) and primers (5′-ACGCGTCGACTGTCGCAATTATTCTTCACAAAT-3′ and 5′-CGGGATCCAGACATGCTGCAATAAACAATAA-3′). Both the genomic fragment and the binary vector pBI101.1 (Clontech, Mountain View, CA, USA) were digested with restriction enzymes SalI and BamHI (Invitrogen). Purified vector and insert were ligated using T4 DNA ligase (Invitrogen) at room temperature (RT) for 1 h. The ligation product was transformed into chemically competent E. coli cells (DH5α) and selected based on resistance to kanamycin (50 mg l⁻¹). Insertion of the fragment was verified by colony PCR and enzyme digestion test. Sequencing confirmed that the PCR product corresponded to the region −1072 to +592 bps relative to the LACS4 translational start point in the Arabidopsis database.

The construct described above was introduced into the competent Agrobacterium tumefaciens GV3101 (pMP90) cells (Koncz and Schell 1986) and used for transformation of Arabidopsis inflorescences (Bechtold et al. 1993). Four-week-old Col 0 WT plant flowers were dipped in the suspension of bacteria containing the construct and covered for 24 h. After senescence, T1 seeds were bulk harvested and screened for kanamycin resistance conferred by the pBI101.1 plasmid.

GUS histochemical assays and RNA in situ hybridization

Tissues from transgenic plants containing the LACS4promoter::GUS construct were removed and immersed in GUS staining buffer containing 0.5 mM potassium ferricyanide, 0.5 mM potassium ferrocyanide, 100 mM Na₂HPO₄, 100 mM NaH₂PO₄, 0.2% Triton-X-100 and 1 mM 5-bromo-4-chloro-3-indolyl-β-D-glucuronide (X-gluc) (Jefferson et al. 1987). Air was eliminated from the tissues by vacuum infiltration. Infiltrated tissues were incubated in the staining buffer for 4 h, or overnight at 37°C. Stem and leaf tissues were cleared of chlorophyll by incubation in a 75% ethanol solution. Stained and cleared samples were examined visually either directly or under dissecting microscope or compound light microscope.

In situ hybridization of Arabidopsis developing seeds (tissue fixation, sectioning, hybridization, signal detection and probe synthesis) was carried out as published previously (Hooker et al. 2002, Hepworth et al. 2005). To generate the probes for hybridization, DNA templates were amplified by PCR from the leaf cDNA of Col0 WT using primers that add the T7 RNA polymerase binding site at the 5′ end. For the sense probe, the primers used for LACS4 gene expression were 5′-AATACGACTCACTATAGGGATGTCGCAGCAGAAGAAATACATC-3′ and 5′-CTACCCTCTGGAAGCAAATTTTG-3′. For the antisense probe of LACS4 that was used as a negative control, the primers were designed as 5′-AATACGACTCACTATAGGGCTACCCTCTGGAAGCAAATTTTG-3′ and 5′-ATGTCGCAGCAGAAGAAATACATC-3′.

RNA isolation and RT- PCR

Rosette leaves of Col 0 wild-type Arabidopsis and homozygous SALK T-DNA insertional mutants were collected and frozen immediately using liquid nitrogen. Total RNA was isolated using Trizol Reagent (Invitrogen) according to manufacturer’s protocol. By adding DNase I (Invitrogen) to each sample, possible residual DNA in the RNA sample was removed. For reverse transcription, 1 μg of total RNA, oligo-dT and SuperScript II reverse transcriptase were mixed to synthesize the first-strand cDNA as specified by the manufacturer (Invitrogen). Amplification of a specific region overlapping two exons of LACS4 gene or glyceraldehyde-3-phosphate dehydrogenase C (GAPC) for expression analysis by RT-PCR was carried out using the following primers: LACS4P1 forward: 5′-ATGAAAGGGTTCGAGATCATC-3′, reverse: 5′-GACCGAGAGTGAGGAGGAAT-3′; LACS4P2 forward: 5′-GCATTGTGACTCTAATCGCTGGAG-3′, reverse: 5′-ATCTTCATGCTTCCATCTGG-3′ (Jessen et al. 2015); and GAPC forward: 5′-ACTCGAGAAAGCTGCTAC-3′, reverse: 5′-ATTCGTTGTCGTACCATG-3′.

Chlorophyll extraction and quantification

Four-week-old rosette leaves were collected, and the fresh weight (FW) was measured and recorded for each sample. Then the leaves were carefully submerged in 4 ml of 80% ethanol at RT. At each given time points, 800 μl aliquots of the supernatant were added to cuvettes, and the absorbance was measured at 664 and 647 nm, respectively (Hiscox and Israelstam 1979). The micromolar concentration of total chlorophyll per microgram of FW of tissue was calculated using the formula: total micromoles chlorophyll = 7.93(A₆₆₄) + 19.53(A₆₄₇) (Lolle et al. 1997).

Measurement of water loss

To analyze the water loss rate, 5-week-old rosette leaves were removed from the plants and kept in petri-dishes at RT. At indicated time points, the samples were weighted and recorded. To dry the samples completely, rosettes were put in an 80°C oven overnight. Total water was calculated as the final dry weight (DW) subtract from the FW. The water loss at each time point was expressed as percentage of water loss over total water.

Cuticular wax extraction and analysis

Cuticular waxes were extracted from the top 10 cm primary stems of 6-week-old Arabidopsis. Stem surface area was calculated by photographing stems prior to wax extraction, measuring the number of pixels, converting the values to cm² and multiplying by π. Stems were immersed for 30 s in chloroform containing 10 µg n-tetracosane (C24 alkane), which was used as an internal standard for amount calculation. After extraction, samples were blown down under a stream of nitrogen and re-dissolved in 10 µl of pyridine (Fluka) and 10 µl of N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) with 1% trimethylchlorosilane (Sigma). Samples were then derivatized for 60 min at 80°C. After derivatization, excess BSTFA and pyridine were removed by blowing down under nitrogen, and samples were suspended in 50 µl of chloroform. Wax analyses were performed on an Agilent 7890 GC-FID and an HP1 methyl siloxane column with helium as the carrier gas. GC was carried out with temperature-programmed on-column injection and oven temperature set at 50°C for 2 min, and then raised by 40°C min^–1 to 200°C, held for 1 min, increased by 3°C min^–1 to 320°C, and held for 15 min. One microliter of each sample was injected. Quantification of wax components was carried out by comparing their FID peak areas to that of the internal standard.

Analysis of cutin monomers

Cutin chemical analyses were carried out on stems of 6-week-old wild type and mutants following the method described by Jenkin and Molina (2015) with some modifications. Briefly, dry cell wall-enriched residues remaining after solvent extractions (50 mg per tube) were depolymerized using sodium methoxide-catalyzed transmethylation, adding 50 µg methyl heptadecanoate and 50 µg omega-pentadecanolactone as internal standards. Acetyl derivatives of cutin monomers were analyzed by GC-MS using a Thermo Scientific TRACE 1300 gas chromatograph equipped with a Thermo Scientific ISQ LT Single Quadrupole mass spectrometer. A split injection (5:1 ratio, 310°C) was used with a 30 m TG-5MS column (30 m length, 0.25 i.d., 0.25 mm film thickness). Temperature settings were as follows: ion source 300°C, MS transfer line 300°C, oven temperature program was set to 180°C for 3 min and increased to 280°C at a rate of 3°C·min⁻¹. The helium flow rate was set at 1.0 ml·min⁻¹.

SEM

SEM was used to view epicuticular wax crystallization patterns of stems collected from dry plants. The top 1 cm of stems were mounted onto the SEM stubs and sputter coated with gold particles for 10 min in a SEMPrep2 sputter coater (Nanotech, Vancouver, Canada). A Hitachi S4700 field emission SEM (Toronto, Canada) with an accelerating voltage of 5 kV and a working distance of 12 mm were used to examine the samples.

Seed fatty acid analysis

About 2.5 mg of dry seeds from Col 0 WT and each mutant line were weighed and transferred into 1 cm × 10 cm glass tubes (pre-rinsed with chloroform twice and dried) with Teflon screw caps. One milliliter of freshly prepared 5% (v/v) concentrated sulfuric acid in methanol, 25 µl of BHT solution (0.2% w/v butylated hydroxyl toluene in methanol), and 300 µl of toluene with internal standard (triheptadecanoin, 12.5 µg/300 µl) were added to each tube. All the tubes were then vortexed for 30 s and heated at 90°C for 2 h. After cooling on ice, 1.5 ml of 0.9% NaCl (w/v) were added to each sample. Fatty acid methyl esters (FAMEs) from each tube were twice extracted with 2 ml of hexane each time, evaporated under a stream of nitrogen, dissolved in 50 µl of hexane and transferred to glass vials. FAMEs were separated by gas-liquid chromatography (GC) as described previously (Kunst et al. 1992). For preparing FAMEs from single seeds, one seed of wild type and each mutant was placed into a screw top glass tube. Transmethylation was achieved by adding 0.5 ml of 1 N HCl in methanol (Supelco) and 300 µl of hexane to each tube, with 10 µl of 0.1 mg ml⁻¹ 17:0-methyl ester as internal standard. After 2 h of heat treatment at 80°C, the samples were cooled down and 0.5 ml of NaCl (0.9%) was added. The tubes were vortexed vigorously, and the top phase was carefully drawn and transferred to a clean GC vial. The hexane was blown off under nitrogen stream and 20 µl of the hexane was added to dissolve FAMEs, which were then separated by GC as described above.

Funding

Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grants to L.K. and I.M.

Acknowledgments

We are grateful to the Salk Institute for supplying the T-DNA insertional mutants; Dr. Owen Rowland (Carleton University) for the lacs1–1 lacs2–3 seeds; Dr. Christiane Nawrath (University of Lausanne) for providing lacs2–3 seeds; the University of British Columbia Bio Imaging Facility for the use of microscopes and Dr. Vesna Katavic for the LACS4pro::GUS transgenic plants.

Disclosures

The authors have no conflicts of interest to declare.