The ORFs coding for the GL2 (GRMZM2G098239; Zm00001d002353) and GL2-LIKE (GRMZM2G315767; Zm00001d024317) proteins were codon-optimized for expression in yeast using GeneOptimizer (GeneArt, LifeTechnologies) and in Arabidopsis (OptimumGene (GenScript; www.genscript.com), respectively. These sequences were chemically synthesized (GenScript; GeneArt, LifeTechnologies). These DNA fragments were cloned into high-copy episomal yeast plasmids, pAG426 (URA3) or pAG423 (HIS3) (Invitrogen, Carlsbad, CA) (Alberti et al., 2007) using either the Gateway^® cloning system (Invitrogen, Carlsbad, CA) or In-Fusion^® cloning system (Takara Bio USA, Inc., Mountain View, CA). Depending on the experiment, expression from the episomal plasmids was under the control of a galactose-inducible promoter (GAL1) or the constitutive glyceraldehyde-3-P-dehydrogenase (GPD) promoter. All recombinant yeast shuttle vectors were confirmed by DNA sequencing and were maintained in E. coli TOP10 cells (Invitrogen, Carlsbad, CA), using Luria Bertani (LB) media supplemented with the appropriate antibiotics.

The yeast strains, INVSc1, BY4743, BY4741, and BY4742 were obtained from Open Biosystems (ThermoFisher Scientific, Rockford, IL) and maintained in YPD (yeast peptone dextrose) or synthetic complete (SC) dropout media. Yeast strains carrying mutations in the endogenous FAE component genes and complemented by maize FAE component genes were previously described (Campbell et al., 2019; Stenback et al., 2022). All yeast strains expressing the maize GL2 or GL2-LIKE proteins were selected by their ability to grow on minimal medium (SD) lacking the appropriate amino acid or nucleobase (e.g., uracil). The induction of expression from the GAL1 promoter was accomplished by replacing glucose with 2% (w/v) galactose in SC medium. Yeast cultures were grown according to standard procedures in appropriate media at 30°C (Adams and Kaiser, 1998). For fatty acid analysis experiments, all strains were grown for 72 hours with exception of the INVSc1 strain that expressed maize proteins from the GAL1 promoter; these strains were grown for 48 hours.

Plasmids were transformed into yeast using a standard lithium acetate transformation protocol (Gietz and Woods, 2002). Briefly, 3 µL of salmon sperm DNA, 1 µg of plasmid DNA, and 100 µL transformation mix (200 µL of 2 M lithium acetate; 800 µL of 50% (w/v) PEG; 7.7 µL of β-mercaptoethanol) were added to a 1.5 mL microcentrifuge tube and mixed by vortexing. Yeast cells from a large colony were added to the mix, vortexed, and incubated at 37°C for 30 min. Samples were then centrifuged at 900 g for 5 min, and the cell pellet was resuspended in 200 µL of sterile water and plated on selective solid media. The plates were incubated at 30°C for 2-3 days. The non-recombinant vector was transformed into each strain and used as a control.

Cell pellets were lyophilized, and after the addition of 10 μg of nonadecanoic acid as an internal standard, fatty acids were extracted and converted to methyl esters by the addition of 1 mL 5% sulfuric acid in methanol, followed by incubation of the mixture at 80°C for 1 h. After cooling, 2 mL of 0.9% (w/v) aqueous NaCl was added, and fatty acid methyl esters were extracted thrice with 1 mL of 4:1 hexane: chloroform. After each extraction, phases were separated and the organic phase was removed and pooled, and samples were dried under a stream of nitrogen gas to a final volume of ~250 µL. One microliter of each sample was analyzed by gas chromatography-flame ionization detection (GC-FID) or gas chromatography-mass spectrometry (GC-MS).

GC-FID analysis was conducted with an Agilent 6890 GC, equipped with a DB-1 MS capillary column (15 m x 0.25 mm x 0.25 μm, Agilent 122-0112). Chromatography was conducted with helium gas, at a flow-rate of 1.2 mL/min, and an inlet temperature at 280 °C. The column oven temperature was initially held at 80°C for 1 min, then ramped at 15°C/min to 230°C and held for 2 min, and then ramped at 15°C/min to 340°C and held there for 2 min. For peak identification purposes, chromatograms were compared with fatty acid methyl ester standards (8:0-30:0) and parallel GC-MS analyses were queried against the NIST 14 Mass Spectral Library. Samples that were analyzed by GC-MS were first silylated with 50 µL N,O-Bis(trimethylsilyl)trifluoroacetamide (BSTFA) with 1% trimethylchlorosilane (TMCS) (Sigma-Aldrich), prior to chromatography.

Statistical significance of the differences in fatty acid profiles were determined by analyzing 3-6 replicates of each yeast strain, as defined in the captions of each figure. In all cases, statistical comparisons are made only from cultures that were grown in parallel in the same incubator. Fatty acid abundance data are reported in Supplementary Table 1 , and statistical significance between genotypes was determined by Tukey’s Honest Significant Difference (HSD) test following ANOVA. The p-values for all Tukey’s HSD tests for each experiment are included in Supplementary Table 2 .

Protein extracts were prepared from yeast cell pellets using the YeastBuster Protein Extraction Reagent, (Novagen Sigma-Aldrich, Inc., St. Louis, MO). In brief, weighed cell pellets were suspended in YeastBuster Reagent (5 ml/g cell pellet) and 0.5 M Tris(hydroxypropyl) phosphine (50 µl/g cell pellet). The cell suspensions were agitated for 15-20 min at room temperature and then centrifuged at 16,000 g for 20 min at 4°C. The supernatant was collected and subjected to SDS-PAGE, and proteins were electrophoretically transferred from the polyacrylamide gel to a nitrocellulose membrane using a Bio-Rad Criterion™ blotter (Bio-Rad, Hercules, CA). Membranes were first incubated in a solution of 5% (w/v) non-fat milk powder, dissolved in TBST buffer (0.1% (v/v) Tween 20, 0.15 M NaCl, 2 mM KCl, 20 mM Tris-HCl, pH 7.5) that contained GL2-specific antiserum (1:1000 dilution) (Alexander et al., 2020), and then with a solution containing horseradish peroxidase-linked to goat anti-mouse IgG antibody (Bio-Rad) (1:1000 dilution). After extensive washing with TBST buffer, the antigen-antibody complexes were detected using the Pierce ECL chemi-luminescent detection system (ThermoFischer Scientific) and visualized on the ChemiDoc XRS+ gel documentation system (Bio-Rad).

Yeast strains expressing maize proteins were grown in liquid cultures and cells were pelleted by centrifugation. RNA was extracted from the resultant cell pellets utilizing the RNeasy Mini Kit (Qiagen, Germantown, MD) according to manufacturer’s instructions. Contaminating DNA was removed from the RNA samples by treatment with RQ1 RNase-Free DNase (Promega, Madison, WI), and 1 µg of RNA was used for cDNA synthesis using either the Superscript IV Synthesis Kit (ThermoFisher Scientific) or PrimeScript Synthesis Kit (Takara Bio USA, San Jose, CA) according to manufacturer’s instructions. PCR was conducted using the resultant cDNA as template, GoTaq Green Master Mix (Promega) and gene-specific primer pairs to assess the expression of ZmKCS6 (Primer 1: 5’-GTGAACCTCAAGCACGTCAA-3’ and Primer 2: 5’-CTCTTGTCGTCGTCGCTGAT-3’), GL2 (Primer 1: 5’-ATGGTTTTCGAACAACACGAAG-3’ and Primer 2: 5’-TTAAGCAACATGTAAAGCAGAACCC-3’), and GL2-LIKE (Primer 1: 5’-ATGGTTGTTGAGGCTAACTCTG-3’ and Primer 2: 5’-TCAAGCAACTCTAAGTGCATCC-3’). ZmKCS5 expression was assessed by utilizing Q5 High-Fidelity Polymerase with GC enhancer (NEB, UK) with gene-specific primer pairs (Primer 1: 5’- CAGAAGAACCTGCAGCTGTC-3’ and Primer 2: 5’- GCCGCTGCCGAAGCCGATCTGCCAG-3’). Each cDNA sample was also subjected to PCR analysis by using primers specific for the yeast GLC7 gene (Primer 1: 5’-CCAGATCTATATTCATAAAGCAACCC-3’ and Primer 2: 5’-GATAATTAGATTCTGGCGGGAATC-3’). This test PCR assay established that all cDNA samples lacked the GLC7 intron, indicating that the template samples were not contaminated by genomic DNA. The thermal cycling program for PCR was initiated at 95°C for 3 min, and then 35 cycles of incubations at 95°C for 30 s, 56°C for 45 s, and 72°C for 1.5 min; followed by a final extension step at 72°C for 5 min.

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers: GRMZM2G098239 and Zm00001d002353 (Gl2); GRMZM2G315767 and Zm00001d024317 (Gl2-like); GRMZM2G393897 and Zm00001d009608 (ZmKCS4); AC233893.1_FG003 and Zm00001d048061 (ZmKCS5); GRMZM2G164974 and Zm00001d028241(ZmKCS6); GRMZM2G160417 and Zm00001d039053 (ZmKCS15); AC205703.4_FG006 and Zm00001d017111(Gl8a; ZmKCR1); GRMZM2G087323 and Zm00001d050992 (Gl8b; ZmKCR2); GRMZM2G151087 and Zm00001d039856 (ZmHCD).

Based on the carbon chain lengths of the alkyl derivatives that occur in the cuticular waxes of maize (Bianchi et al., 1985), the maize FAE system should have the ability to produce VLCFAs as long as 34-carbon atoms. Moreover, in planta characterizations of mutations in the maize gl2 gene indicate that it may be involved in the ability of plant cells to produce VLCFAs beyond 26:0 and 28:0, and up to 34:0 (Bianchi et al., 1975; Alexander et al., 2020). Therefore, analogous to the Arabidopsis FAE system, where interactions between the Arabidopsis CER2-LIKE proteins and KCS paralogs stimulate the terminal elongation cycle(s) of the FAE system to generate longer chain VLCFAs (Haslam et al., 2017), yeast expression experiments evaluated whether GL2 or GL2-LIKE could affect the VLCFA profiles produced by co-expression with ZmKCS homologs. Initial experiments were conducted with ZmKCS5 or ZmKCS6, which are the closest homologs to the Arabidopsis CER6-encoded KCS (Campbell et al., 2019) that specifically interact with CER2 in the Arabidopsis FAE system (Haslam et al., 2015, 2017).

Figure 1 shows the results of such experiments using the diploid INVSc1 strain as the expression host. The wild-type INVSc1 strain yields a VLCFA titer of ~7.5 µmol/g dry weight, and these VLCFAs range from 20:0 to 28:0, with the most abundant VLCFA being 26:0. The individual expression of ZmKCS5, ZmKCS6, GL2 or GL2-LIKE does not significantly affect the VLCFA titers, as compared to the wild-type host strain ( Figures 1A, B ). However, the co-expression of ZmKCS5 ( Figures 1A, C ) or ZmKCS6 ( Figures 1B, D ) with either GL2 or GL2-LIKE induced an approximately 2-fold increase in the VLCFA titer ( Figures 1A, B ), and this is particularly associated with increased accumulation of the 24:0, 26:0 and 28:0 VLCFAs ( Figures 1C, D ). The fact that these increased titers occur only upon the co-expression of ZmKCS and GL2 homologs, and not when these maize proteins were individually expressed, is indicative that the GL2 homologs affect the activity of the ZmKCS proteins; possibly via interactions between the co-expressed proteins, analogous to the CER2-KCS interactions in Arabidopsis (Haslam et al., 2017).

However, these results do not recapitulate the in planta expectation based on the gl2 mutant phenotype, which indicates that this gene may be involved in the ability of plant cells to produce VLCFAs of beyond 28:0 and up to 34:0 (Bianchi et al., 1975; Alexander et al., 2020). Therefore, we considered whether this may be associated with either the promoter that was used in the co-expression experiments (i.e., the galactose inducible GAL1 promoter) or by the particular yeast strain that was used, INVSc1, whose provenance is proprietary, and unknown. Hence, both the GL2 and GL2-LIKE proteins were individually co-expressed with either ZmKCS5 or ZmKCS6 under the control of the constitutive, glyceraldehyde-3-phosphate dehydrogenase (GPD) promoter (Bitter et al., 1987) in the diploid strain, BY4743, which has a well-defined pedigree (Brachmann et al., 1998).

In contrast to the results obtained with the GAL1 promoter, these experiments with the GPD promoter in either the INVSc1 or BY4743 diploid strains generated qualitatively different VLCFA profiles ( Figure 2 ) that more accurately recapitulate the in planta chemotypes expected for the GL2 protein (Bianchi et al., 1975). Specifically, in both the INVSc1 ( Figures 2A, B ) and BY4743 ( Figures 2C, D ) strains, the co-expression of GL2 or GL2-LIKE with either ZmKCS5 or ZmKCS6 induced the accumulation of 30:0 VLCFA, a product that was undetectable in the other strains that singularly expressed ZmKCS5, ZmKCS6, GL2 or GL2-LIKE (compare Figures 1 , 2 ). These distinct observations between the use of the GPD and GAL1 promoters in different diploid strains indicate that additional insights are required to dissect the complexity of co-expressing FAE components in heterologous biological systems, such as plant FAE components in yeast strains that also carry an endogenous FAE system.

To overcome the complexity of genetically adding heterologous maize FAE components to an intact FAE system (i.e., the yeast FAE), we co-expressed ZmKCS paralogs with either GL2 or GL2-LIKE in a yeast strain that lacked the analogous endogenous yeast FAE component enzyme. These experiments were conducted in a haploid strain (i.e., BY4742, which is one of the haploid progenitors of the diploid strain, BY4743 (Brachmann et al., 1998)) that carries mutations in genes coding for yeast FAE component genes. The yeast FAE system does not utilize KCS-type condensing enzymes, but rather utilizes ELO-type condensing enzymes, encoded by the ELO1, ELO2 or ELO3 genes (Toke and Martin, 1996; Oh et al., 1997). Because ELO3 has the ability to catalyze the terminal cycles of fatty acid elongation using preexisting 24:0 and 26:0 acyl-CoAs (Oh et al., 1997; Denic and Weissman, 2007), we first recapitulated the above experiments in the BY4742 haploid strain that carried an elo3 null allele (note that this strain carries functional ELO1 and ELO2 genes). In these experiments we individually expressed ZmKCS4, ZmKCS5, ZmKCS6 and ZmKCS15 in the elo3 mutant yeast strain, and compared the VLCFA profiles when these ZmKCS isozymes were co-expressed with either GL2 or GL2-LIKE ( Figure 3 ). The rationale for including ZmKCS4 and ZmKCS15 isozymes is based on the fact that these two plant isozymes have been shown to be functional in yeast (Stenback et al., 2022).

As with the diploid strains, the wild-type BY4742 haploid strain produced VLCFAs that range between 20 and 28 carbon chain lengths, with 26:0 accounting for 80% of the VLCFAs. The effect of the elo3 mutation was to eliminate the accumulation of 26:0 and 28:0, and thus the predominant VLCFA in this mutant yeast strain was 24:0. The expression of ZmKCS5 or ZmKCS6 in the elo3 mutant strain had no effect on the fatty acid profiles ( Figures 3A, B ). However, the expression of ZmKCS4 or ZmKCS15 in the elo3 mutant strain resulted in partially restoring the ability of the strain to produce the longer VLCFAs; specifically these strains produced small but statistically significant quantities of 26:0 ( Figures 3C, D ). Thus, in contrast to ZmKCS4 or ZmKCS15, neither ZmKCS5 or ZmKCS6 were capable of replacing the elo3 function. Moreover, the co-expression of either GL2 or GL2-LIKE with each of these four ZmKCS isozymes was incapable of modifying the abilities of the resulting strains to produce larger quantities of the longer chain VLCFAs (i.e., 26:0, 28:0, or 30:0) ( Figure 3 ).

As demonstrated by the synthetic lethality associated with the elo2 elo3 double mutant, the ability to produce VLCFAs is essential for yeast viability (Toke and Martin, 1996; Oh et al., 1997). Moreover, prior characterizations have established that five ZmKCS isozymes (i.e., ZmKCS2, ZmKCS4, ZmKCS11, ZmKCS15, and ZmKCS20) are capable of genetically rescuing this lethality of the elo2 elo3 double mutant (Stenback et al., 2022). Therefore, we used these rescued strains to evaluate whether the co-expression of GL2 or GL2-LIKE could modify the product profile of the FAE system, indicative of interactions with the ZmKCS isozymes.

As previously reported (Stenback et al., 2022), in the absence of either GL2 or GL2-LIKE, the expression of ZmKCS2 enables the elongation of fatty acids to 22-carbon chain length, whereas ZmKCS4, ZmKCS11, and ZmKCS20 enables the elongation of fatty acids to 24-carbon chain length, and ZmKCS15 can elongate fatty acids up to 26-carbon atoms ( Figure 4 ). These acyl-products include VLCFAs and 2-hydroxy-VLCFAs, which are hydroxylated post-synthesis in the assembly of ceramide-based lipids (Erdbrügger and Fröhlich, 2021). The co-expression of either GL2 or GL2-LIKE with each of these ZmKCS isozymes did not cause any qualitative change in the VLCFA profiles; namely there were no new VLCFA products generated that were absent from the strains that only expressed the ZmKCS isozyme. Rather, there were statistically significant quantitative changes in the accumulation of some of the VLCFA products. For example, GL2-LIKE caused increased accumulation of 2-hydroxy-22:0 or 2-hydroxy-24:0 when it was co-expressed with ZmKCS11 ( Figure 4C ) or ZmKCS15 ( Figure 4D ), respectively; the latter change was accompanied by the decreased accumulation of 26:0 ( Figure 4D ).

The findings observed with the GL2 and GL2-LIKE proteins ( Figure 4 ) are not consistent with the model developed with the Arabidopsis homologs of these proteins (i.e., the CER2 and CER2-LIKE proteins) (Haslam et al., 2017). This difference between the maize and Arabidopsis homologs may be associated with the fact that the CER2 and CER2-LIKE proteins mediate this effect by specifically interacting with the CER6- or CER60-encoded KCS isozymes (Haslam et al., 2015). The difference between the two systems (i.e., maize versus Arabidopsis) may be due to the fact that the ZmKCS isozymes used in the current study are in a phylogenetic clade that is distinct from the Arabidopsis CER6- or CER60-encoding KCS isozymes. In maize, the phylogenetic homologs to the CER6-and CER60-encoding KCSs are ZmKCS5 and ZmKCS6 (Campbell et al., 2019). However, because ZmKCS5 and ZmKCS6 are incapable of complementing the yeast elo2 elo3 double mutant strain (Stenback et al., 2022), it was not possible to co-express them individually with GL2 or GL2-LIKE in the yeast system lacking the native condensing enzymes. Therefore, we co-expressed either ZmKCS5 or ZmKCS6 in the yeast elo2 elo3 double mutant strain that was rescued by genetic complementation by the expression of ZmKCS15. This complemented strain was chosen because it was capable of producing the longest VLCFAs (i.e., up to 26:0) (Stenback et al., 2022). Figure 5 shows that the co-expression of GL2 or GL2-LIKE with either ZmKCS5 or ZmKCS6 in the ZmKCS15-rescued elo2 elo3 double mutant had no effect on the VLCFA profiles that were generated.

Comparing the VLCFA profiles in Figures 4D and 5A , we found that the co-expression of ZmKCS15 with ZmKCS5 in the elo2 elo3 double mutant is no different than solely expressing ZmKCS15 in this strain, and the co-expression of GL2 or GL2-LIKE with both ZmKCS15 and ZmKCS5 had little effect on the profile ( Figure 5A ). Similar results were obtained when ZmKCS15 expression was combined with ZmKCS6 expression; the most pronounced apparent effect, which is not statistically significant, was a 50% reduction in the accumulation of 2-hydroxy-26:0 upon the co-expression with GL2-LIKE ( Figure 5B ).

Prior studies have indicated that protein-protein interactions among the FAE component enzymes are important for VLCFA biosynthesis; in the Arabidopsis system this includes interactions between CER2-LIKE proteins and the KCR, HCD and ECR components (Kim et al., 2022). Therefore, using the yeast strains that we previously developed to functionally characterize these additional components of the maize FAE system (Campbell et al., 2019), we evaluated whether the GL2 or GL2-LIKE proteins can interact with additional FAE components (i.e., ZmKCR1, ZmKCR2 and ZmHCD) and thereby affect the VLCFA product-profile of the FAE system.

We evaluated if either GL2 or GL2-LIKE could affect changes in VLCFA profiles upon co-expression with either ZmKCR1 (gl8a) or ZmKCR2 (gl8b) (Xu et al., 1997, 2002; Dietrich et al., 2005). These co-expression experiments were conducted in yeast strains that lacked the endogenous yeast KCR (i.e., the ybr159Δ mutant strain) (Han et al., 2002). The yeast ybr159Δ mutation is lethal, but can be rescued by the expression of either ZmKCR1 or ZmKCR2 (Campbell et al., 2019). The resulting rescued yeast strains produced VLCFA profiles that are similar to those that occur in the wild-type strain, with 26:0 being the most abundant VLCFA ( Figure 6 ). The co-expression of either GL2 or GL2-LIKE with the two maize KCR isozymes, ZmKCR1 ( Figure 6A ) or ZmKCR2 ( Figure 6B ), produced minor, but statistically significant quantitative changes in the VLCFA profiles, affecting only 22:0, 24:0, and 26:0.

Finally, using the same strategy we investigated if GL2 or GL2-LIKE interacts with ZmHCD to affect VLCFA profiles, by co-expressing the two combinations in the yeast hcd mutant strain (Campbell et al., 2019). Figure 6C compares the VLCFA profiles of the ZmHCD-rescued hcd mutant yeast strains, with and without the co-expression of either GL2 or GL2-LIKE. Although the VLCFA profile of the yeast strain is not altered when ZmHCD is expressed in the wild-type background, the profile is strikingly different when ZmHCD is expressed in the hcd mutant background. Specifically, ZmHCD complementation induces the appearance of 3-hydroxy-VLCFAs of 20 and 22 carbon chain lengths and a 50% reduction in 26:0. We suggest that this is because the maize HCD is not as efficient in replacing the endogenous yeast HCD in converting 3-hydroxyacyl-CoA to the enoyl-CoA, and thus 3-hydroxy-VLCFAs accumulate. Even so, the co-expression of ZmHCD with either GL2 or GL2-LIKE did not further alter the VLCFA profiles of the rescued strains ( Figure 6C ).

The expression of ZmKCS2, ZmKCS4, ZmKCS11, ZmKCS15 and ZmKCS20 proteins in yeast cells was genetically confirmed because their presence rescued the lethality associated with the yeast elo2 elo3 double mutant strain (e.g., Figure 4 ). In addition, changes in VLCFA profiles in a number of yeast strains indicate that the GL2 and GL2-LIKE proteins were successfully expressed (e.g., Figures 1 , 2 ). However, in some strains that were engineered to express these proteins, the resultant VLCFA profiles were very similar to those generated by the host strain (e.g., Figures 3 – 5 ). Therefore, we sought molecular evidence to confirm successful expression of the GL2 and GL2-LIKE proteins, and ZmKCS isozymes. Such confirmation was obtained by either Western blot or RT-PCR analyses.

For example, Figure 3E shows Western blot analyses of protein extracts from a subset of the yeast elo3-mutant strains, which confirms that the GL2 protein was successfully expressed in these experiments, even though its expression did not generate a change in the VLCFA profile. Analogously, RT-PCR analysis confirmed the expression of the Gl2 and Gl2-like mRNA in the yeast elo2 elo3 double mutant strains that were genetically complemented by the expression of the ZmKCS2, ZmKCS4, ZmKCS11, ZmKCS15 or ZmKCS20 proteins ( Figure 4F ), even though the expression of Gl2 and Gl2-like did not qualitatively change the VLCFA profiles ( Figures 4A–E ).

Similarly, we evaluated the expression of mRNAs encoding for GL2 or GL2-LIKE proteins and either ZmKCS5 ( Figure 5B ) or ZmKCS6 ( Figure 5D ), which were co-expressed with ZmKCS15 in the yeast elo2 elo3 double mutant strain. This double mutant strain is only viable because of the expression of the ZmKCS15 protein ( Figure 4D ). Thus, in these strains ZmKCS15 was co-expressed with either ZmKCS5 or ZmKCS6, in the presence or absence of either GL2 or GL2-LIKE. RT-PCR analyses confirmed the expression of Gl2 and Gl2-like mRNAs and either ZmKCS5 ( Figure 5B ) or ZmKCS6 ( Figure 5D ), even though these genetic modifications did not affect changes in the VLCFA profiles ( Figures 5A, C ).