Fatty Aldehyde Dehydrogenase Multigene Family Involved in the Assimilation of n-Alkanes in Yarrowia lipolytica*

Background: Fatty aldehydes are converted to fatty acids in the metabolism of n-alkanes of Yarrowia lipolytica. Results: Four fatty aldehyde dehydrogenases (FALDHs) are required for the assimilation of long-chain n-alkanes. Conclusion: FALDHs oxidize fatty aldehydes to fatty acids in the peroxisome and/or the endoplasmic reticulum in n-alkane metabolism. Significance: FALDH genes may have been multiplicated and functionally diversified in Y. lipolytica. In the n-alkane assimilating yeast Yarrowia lipolytica, n-alkanes are oxidized to fatty acids via fatty alcohols and fatty aldehydes, after which they are utilized as carbon sources. Here, we show that four genes (HFD1–HFD4) encoding fatty aldehyde dehydrogenases (FALDHs) are involved in the metabolism of n-alkanes in Y. lipolytica. A mutant, in which all of four HFD genes are deleted (Δhfd1–4 strain), could not grow on n-alkanes of 12–18 carbons; however, the expression of one of those HFD genes restored its growth on n-alkanes. Production of Hfd2Ap or Hfd2Bp, translation products of transcript variants generated from HFD2 by the absence or presence of splicing, also supported the growth of the Δhfd1–4 strain on n-alkanes. The FALDH activity in the extract of the wild-type strain was increased when cells were incubated in the presence of n-decane, whereas this elevation in FALDH activity by n-decane was not observed in Δhfd1–4 strain extract. Substantial FALDH activities were detected in the extracts of Escherichia coli cells expressing the HFD genes. Fluorescent microscopic observation suggests that Hfd3p and Hfd2Bp are localized predominantly in the peroxisome, whereas Hfd1p and Hfd2Ap are localized in both the endoplasmic reticulum and the peroxisome. These results suggest that the HFD multigene family is responsible for the oxidation of fatty aldehydes to fatty acids in the metabolism of n-alkanes, and raise the possibility that Hfd proteins have diversified by gene multiplication and RNA splicing to efficiently assimilate or detoxify fatty aldehydes in Y. lipolytica.

In the n-alkane assimilating yeast Yarrowia lipolytica, n-alkanes are oxidized to fatty acids via fatty alcohols and fatty aldehydes, after which they are utilized as carbon sources. Here, we show that four genes (HFD1-HFD4) encoding fatty aldehyde dehydrogenases (FALDHs) are involved in the metabolism of n-alkanes in Y. lipolytica. A mutant, in which all of four HFD genes are deleted (⌬hfd1-4 strain), could not grow on n-alkanes of 12-18 carbons; however, the expression of one of those HFD genes restored its growth on n-alkanes. Production of Hfd2Ap or Hfd2Bp, translation products of transcript variants generated from HFD2 by the absence or presence of splicing, also supported the growth of the ⌬hfd1-4 strain on n-alkanes. The FALDH activity in the extract of the wild-type strain was increased when cells were incubated in the presence of n-decane, whereas this elevation in FALDH activity by n-decane was not observed in ⌬hfd1-4 strain extract. Substantial FALDH activities were detected in the extracts of Escherichia coli cells expressing the HFD genes. Fluorescent microscopic observation suggests that Hfd3p and Hfd2Bp are localized predominantly in the peroxisome, whereas Hfd1p and Hfd2Ap are localized in both the endoplasmic reticulum and the peroxisome. These results suggest that the HFD multigene family is responsible for the oxidation of fatty aldehydes to fatty acids in the metabolism of n-alkanes, and raise the possibility that Hfd proteins have diversified by gene multiplication and RNA splicing to efficiently assimilate or detoxify fatty aldehydes in Y. lipolytica.
A variety of yeasts have acquired the ability to assimilate n-alkanes, some of the most hydrophobic compounds, as sole carbon and energy sources. In n-alkane assimilating yeasts, n-alkanes are incorporated into cells and hydroxylated to fatty alcohols by cytochromes P450ALK (P450ALKs), which are classified into the CYP52 family of P450 (1) present in the endoplasmic reticulum (ER) 3 membrane (2)(3)(4)(5)(6)(7). Fatty alcohols are oxidized to fatty aldehydes by fatty alcohol dehydrogenase (FADH) in the ER or fatty alcohol oxidase (FAOD) in the peroxisome. Fatty aldehydes are oxidized to fatty acids by fatty aldehyde dehydrogenase (FALDH) in the ER or in the peroxisome. Fatty acids are finally activated to fatty acyl-CoA and metabolized through ␤-oxidation in the peroxisome, or are used in lipid synthesis.
In marked contrast to P450ALKs, FADH, FAOD, and FALDH that are involved in the metabolism of n-alkanes have remained elusive. FADH that is involved in the oxidation of fatty alcohols of more than 10 carbons has remained unidentified in all n-alkane-assimilating yeasts. FAOD activity has been detected in cell extracts of a subset of n-alkane-assimilating yeasts, including C. maltosa, C. bombicola, C. parapsilosis, C. tropicalis, and D. hansenii, and coding genes have been characterized in some of these yeasts (24,25). An orthologous gene, however, has not been identified in the genome sequence of Y. lipolytica. It was recently shown that alcohol dehydrogenase genes, FADH and ADH1-ADH7, and a fatty alcohol oxidase gene, FAO1, are involved in the oxidation of -hydroxy fatty acids in Y. lipolytica; however, the deletion mutant strain lacking all these genes is able to oxidize dodecanol (26). FALDH activities have been reported in C. tropicalis, Y. lipolytica, and Candida intermedia (27)(28)(29), but the enzymes catalyzing the reactions remain unidentified. It has been reported that C. maltosa P450 52A3 expressed heterologously in Saccharomyces cerevisiae oxidizes n-alkane of 16 carbons to the corresponding fatty alcohol, fatty aldehyde, and fatty acid in vitro (30).
ALDH3A2, one of the mammalian FALDH, belongs to a superfamily of NAD(P) ϩ -dependent aldehyde dehydrogenases (31). Human ALDH3A2 have high activities toward saturated and unsaturated fatty aldehydes of 6 -24 carbons in length (32). ALDH3A2 is involved in various cellular processes: it has been suggested that ALDH3A2 is involved in the breakdown of phytanic acid in the peroxisome (33,34), and ALDH3A2 is further reported to have a protective role against oxidative stress associated with lipid peroxidation (35). Mutations in the ALDH3A2 gene have been shown to result in Sjögren-Larsson syndrome (36). S. cerevisiae has an ortholog of the ALDH3A2 gene, HFD1, and S. cerevisiae Hfd1p and mammalian ALDH3A2 are responsible for the conversion of hexadecenal to hexadecenoic acid in the degradation process of sphingosine 1-phosphate (37).
The focus of this study was four genes predicted to encode FALDHs, whose amino acid sequences exhibit similarities to those of FALDHs of S. cerevisiae (Hfd1p) and mammals (ALDH3A2), in the Y. lipolytica genome. These four genes were named HFD1, HFD2, HFD3, and HFD4 and were analyzed in terms of the roles they play in the n-alkane assimilation. Our data suggest that Hfd proteins are involved in the oxidation of fatty aldehydes in the n-alkane metabolism process in Y. lipolytica.

EXPERIMENTAL PROCEDURES
Yeast Strains and Growth Conditions-Yeast strains used in this study are shown in Table 1. Y. lipolytica strains CXAU1 and CXAU/A1 were used as wild-type strains (15,21).
Deletion of 4 HFD genes was performed using pop-in/popout method as described previously (19). pBPOT1-mCherry-ADE1 was digested with EcoRV and XbaI, and the ADE1-carrying fragment was introduced into the CXAU1 strain to obtain the POC strain. Correct integration was confirmed by PCR and Southern blot analysis.
Plasmids-Plasmids used in this study are shown in Table 2. Sequences of used primers are listed in Table 3.
The deletion cassettes for HFD genes were constructed as follows. The 5Ј-and 3Ј-adjacent regions of the HFD genes were amplified from CXAU1 total DNA by PCR using the primers listed in Table 4 and cloned into pBURA3 (19). The deletion cassettes for HFD genes were obtained by digestion of the plasmids with restriction enzymes shown in Table 4.
Plasmids, pSEGFP-HFD1, pSEGFP-HFD2, pSEGFP-HFD3, and pSEGFP-HFD4, to express Hfd1p, Hfd2p, Hfd3p, and Hfd4p fused with enhanced green fluorescent protein (EGFP) at their N termini, respectively, from their own promoters, were constructed as follows. The 5Ј-adjacent regions with their start codons and the ORFs with the 3Ј-adjacent regions of HFD genes were amplified from CXAU1 total DNA by PCR using primers, HFD1P-F3, HFD1P-R2, HFD1-F, and HFD1T-R2 for HFD1, HFD2P-F3, HFD2P-R2, HFD2-F, and HFD2T-R2 for HFD2, HFD3P-F3, HFD3P-R2, HFD3-F, and HFD3T-R2 for HFD3, and HFD4P-F3, HFD4P-R2, HFD4-F, and HFD4T-R2 for HFD4, and cloned into pSUT5 to obtain pSHFD1pROt, pSHFD2pROt, pSHFD3pROt, and pSHFD4pROt, respectively. The EGFP ORF was amplified from pEGFP by PCR using primers EcoRVStuI-EGFP-F and EGFP-R-EcoRI. The amplified fragment was digested with EcoRV and EcoRI, and cloned into the EcoRV-EcoRI sites of pBluescript II SK(ϩ) to obtain pBS-EGFP. The fragments carrying EGFP ORF obtained by digestion of pBS-EGFP with HindIII and EcoRI, StuI and BamHI, EcoRV and EcoRI, and HindIII and BamHI were cloned into the HindIII-EcoRI sites of pSHFD1pROt, the StuI-BamHI sites of pSHFD2pROt, the EcoRV-EcoRI sites of pSHFD3pROt, and the HindIII-BamHI sites of pSHFD4pROt, to obtain pSEGFP-HFD1, pSEGFP-HFD2, pSEGFP-HFD3, and pSEGFP-HFD4, respectively. The plasmid pBPOT1-mCherry-ADE1 carrying a cassette to express Pot1p fused with mCherry at its C terminus in its chromosomal location was constructed as follows. The coding region of the C-terminal half of Pot1p was amplified from CXAU1 total DNA by PCR using primers EcoRI-POT1-F and POT1-R-EcoT22I and the amplified fragment was digested with EcoRI and EcoT22I. The ORF of mCherry was amplified from pmCherry-N1 by PCR using primers EcoT22I-mCherry-F and mCherry-R-NotI and the amplified fragment was digested with EcoT22I and NotI. The XPR2 terminator region was amplified from pSUT5 by PCR using primers NotI-XPR2T-F and XPR2T-F-BamHI and the amplified fragment was digested with NotI and BamHI. The 3Ј-non-coding region of POT1 was amplified from CXAU1 total DNA by PCR using primers POT1-F and POT1T-R-XbaI and the amplified fragment was digested with BamHI and XbaI. These 4 DNA fragments were cloned into the EcoRI-XbaI sites of pBluescript II SK(ϩ). The resultant plasmid was digested with BamHI and ligated with the ADE1-carrying BamHI fragment of pSAT4 to obtain pBPOT1-mCherry-ADE1.
The plasmids, pSHFD2A and pSHFD2B, were constructed as follows. The DNA fragments containing HFD2 ORF with the 5Ј-adjacent region were amplified from CXAU1 total DNA by PCR using primers HFD2P-F2 and HFD2A-R, and HFD2P-F2 and HFD2B-R, and cloned into the BamHI-ApaI sites of pSUT5, to obtain pSHFD2A and pSHFD2B, respectively.
The plasmids, pSEGFP-HFD2A and pSEGFP-HFD2B, were constructed as follows. The DNA fragments containing HFD2 ORF were amplified from CXAU1 total DNA by PCR using primers HFD2-F and HFD2A-R for pSEGFP-HFD2A and HFD2-F and HFD2B-R for pSEGFP-HFD2B, digested with BamHI and ApaI, and cloned into the BamHI-ApaI sites of pSHFD2pROt. The fragment carrying EGFP ORF obtained by digestion of pBS-EGFP with StuI and BamHI was cloned into the StuI-BamHI sites of these plasmids, resulting in pSEGFP-HFD2A and pSEGFP-HFD2B.
The plasmid pSHFD2A-EGFP to express Hfd2Ap fused with EGFP at its C terminus was constructed as follows. The DNA fragment containing HFD2A with the 5Ј-non-coding region was amplified from CXAU1 total DNA by PCR using primers HFD2P-F3 and HFD2A-R2, digested with NotI and BamHI, and cloned into the NotI-BamHI sites of pSUT5 to obtain pSHFD2AORt. The fragment carrying EGFP ORF obtained by digestion of pBS-EGFP with StuI and BamHI was cloned into the StuI-BamHI sites of pSHFD2AORt, resulting in pSHFD2AORt2. BamTAABspApa-F and BamTAABspApa-R were annealed, phosphorylated, and cloned into BamHI-ApaI sites of pSHFD2AORt2 to obtain pSHFD2A-EGFP.
Southern Blot Analysis-Southern blot analysis was performed as described previously (19).
Rapid Amplification of cDNA Ends (RACE)-Rapid amplification of cDNA ends was performed using the SMARTer TM RACE cDNA Amplification Kit (Clontech), according to the manufacturer's instructions, with primers HFD1_GSP1 and HFD1_NGSP1 for 5Ј-RACE of HFD1, HFD1_GSP2 and HFD1_ NGSP2 for 3Ј-RACE of HFD1, HFD2_GSP1 and HFD2_NGSP1 for 5Ј-RACE of HFD2, HFD2_GSP2 and HFD2_NGSP2 for 3Ј-RACE of HFD2, HFD3_GSP1 and HFD3_NGSP1 for 5Ј-RACE of HFD3, HFD3_GSP2 and HFD3_NGSP2 for 3Ј-RACE of HFD3, HFD4_GSP1 and HFD4_NGSP1 for 5Ј-RACE of HFD4, and HFD4_GSP2 and HFD4_NGSP2 for 3Ј-RACE of HFD4 (Table 5).
Fatty Aldehyde Dehydrogenase Activity Assay-Yeast cells were cultured in the SD medium for 24 h. n-Decane was added into the medium to a final concentration of 1%, and cells were further incubated for 6 h. Cells were collected by centrifugation, washed with phosphate-buffered saline, suspended in homogenization buffer (25 mM HEPES-NaOH (pH 7.3), 100 mM KCl, 10% glycerol, 1 mM dithiothreitol, and 1% Protease inhibitor mixture (Sigma)), and crushed by glass beads of 0.45-0.5-mm diameter for 3 min using Multi-Beads Shocker (YASUI KIKAI). The homogenates were centrifuged twice at 1,000 ϫ g for 10 min at 4°C. Tween 80 was added to the supernatants at a final concentration of 1% (v/v). The mixtures were placed on ice for 20 min and centrifuged at 13,000 ϫ g for 10 min at 4°C. The supernatants were used as cell extracts.
Escherichia coli BL21(DE3) cells harboring pET3a-HFD1, pET3a-HFD2, pET3a-HFD3, or pET3a-HFD4, were cultured to A 600 ϭ 0.5 at 37°C in LB Broth (Lenox) (1% tryptone, 0.5% yeast extract, 0.5% NaCl). Following 4 h of incubation with 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside, cells were harvested and broken using Multi-Beads Shocker with glass beads of 0.3 mm diameter for 5 min in the homogenization buffer. Cell lysate was centrifuged twice at 1,000 ϫ g for 10 min at 4°C. Tween 80 was added to the supernatants at a final concentra-  HFD2P-F, HFD2P-R, HFD2T-F, HFD2T-R EcoT22I HFD3 HFD3P-F, HFD3P-R, HFD3T-F, HFD3T-R BspT104I HFD4 HFD4P-F, HFD4P-R, HFD4T-F, HFD4T-R MunI tion of 1% (v/v), and the mixtures were placed on ice for 20 min and centrifuged at 13,000 ϫ g for 10 min at 4°C. The supernatants were used as cell extracts. Dodecanal or tetradecanal was mixed with M-buffer (25 mM HEPES-NaOH (pH 7.3), 100 mM glycine, 0.25% (v/v) Tween 80) at a final concentration of 1 mM and the mixture was sonicated for 30 s at room temperature. One-hundred forty l of M-buffer containing the fatty aldehyde, 10 l of cell extracts, and 50 l of 4 mM NAD ϩ in M-buffer were mixed and incubated at 30°C, and the absorbance at 340 nm was measured using Microplate Reader SH-8000(Lab) (CORONA ELECTRIC Co., Ltd.). The extinction coefficient of NADH was 6.22 mM Ϫ1 . One unit of activity was defined as the amount of enzyme that catalyzes the production of 1 mol of NADH/min.
Fluorescence Microscopy-Fluorescence microscopic observation was performed as described previously (23).

Four Genes Encoding Fatty Aldehyde Dehydrogenases-We
searched the genome database of Y. lipolytica for orthologs of Hfd1p and ALDH3A2 using the BLAST program provided by the Génolevures project with the amino acid sequences of S. cerevisiae Hfd1p and human ALDH3A2 as queries. The translation products of YALI0F23793g (HFD1) (40), YALI0E15400g, YALI0A17875g, and YALI0F23793g exhibited significant similarities to S. cerevisiae Hfd1p and mammalian ALDH3A2. YALI0E15400g, YALI0A17875g, and YALI0B01298g were named HFD2, HFD3, and HFD4, respectively, and were investigated in terms of their involvement in n-alkane assimilation. The phylogenetic tree of aldehyde dehydrogenases in Y. lipolytica and S. cerevisiae shows that Hfd proteins are in a clade distinct from that of S. cerevisiae Ald proteins, which catalyze the conversion of acetaldehyde to acetate (41), and their orthologs in Y. lipolytica (Fig. 1A).
A RACE analysis of these HFD genes suggested that the open reading frames (ORFs) of HFD1, HFD3, and HFD4 are consistent with those predicted by the Génolevures project and that they encode proteins of 519, 533, and 529 amino acids, respectively. The result of the analysis also suggested that two variants of HFD2 transcripts, named variants A and B, exist (Fig. 1B). As predicted by the Génolevures project, variant A encodes a protein of 521 amino acids, which was named Hfd2Ap. Variant B was generated by splicing at nucleotide 1494 and encodes a 503-amino acid protein with the peroxisomal targeting signal 1 (PTS1)-like sequence at its C terminus. This protein was named Hfd2Bp. The ratio of DNA clones of these two variants obtained in a 3Ј-RACE analysis of cDNA from cells cultured in medium containing glucose was ϳ1:1 (Table 6). Significant differences in this ratio of variants were not observed in cells cultured in the presence of n-alkanes compared with those cultured in the absence of n-alkanes (Table 6). Typical PTS1-like sequences were not identified in the deduced amino acid sequences of HFD1, HFD3, or HFD4. All Hfd proteins were shown to contain amino acid sequences categorized into the  Fatty Aldehyde Dehydrogenases in Y. lipolytica NOVEMBER 28, 2014 • VOLUME 289 • NUMBER 48 aldehyde dehydrogenase family in the Pfam database. A glutamic acid residue that coordinates nicotinamide ribose, and a cysteine residue that provides the catalytic thiol in the active sites of aldehyde dehydrogenases, were found to be conserved in all Hfd proteins (42). Using TMHMM, single transmembrane helices were predicted at the C termini of Hfd1p and Hfd2Ap.
Transcription of the HFD Genes-The transcriptional profiles of HFD1-HFD4 were assessed by Northern blot analysis and qRT-PCR (Fig. 2). To confirm the specificities of the probes used for Northern blot analysis, a deletion mutant strain of each HFD gene was constructed using the pop-in/pop-out method (19,43). The wild-type Y. lipolytica strain and the deletion mutant strain of each HFD gene were cultured to logarithmic phase in glucose-containing medium, after which they were further cultured in medium containing glucose or n-decane for 1 h. Total RNA was extracted from these cells and the mRNA level of each HFD gene was assessed by Northern blot analysis ( Fig. 2A). The transcript levels of HFD1, HFD2, and HFD3 were highly elevated in cells grown in the medium containing n-decane, whereas that of HFD4 was not.
The transcript levels of the HFD genes in wild-type cells cultured with various carbon sources were quantitated by qRT-PCR. The wild-type cells cultured to logarithmic phase in glucose-containing medium were shifted to the medium containing glucose, n-decane, n-dodecane, n-tetradecane, n-hexadecane, or oleic acid, after which they were further incubated for 1 h. In accordance with the results of the Northern blot analysis, the transcripts of HFD1, HFD2, and HFD3 were markedly increased in cells grown in media containing n-alkanes compared with those grown in glucose-containing medium (Fig. 2B). In contrast, the transcript of HFD4 was not found to be significantly higher in cells grown in n-alkane-containing media. The transcripts of HFD1 and HFD2 were shown to be increased in cells grown in n-decane-containing medium compared with cells grown in medium containing longer chain n-alkanes. In contrast, the transcript level of HFD3 was markedly higher in the presence of longer chain n-alkanes. The transcript of HFD1, but not other HFD genes, was found to be increased in cells cultured in the presence of oleic acid.
Involvement of the HFD Genes in the Growth of Y. lipolytica on Long-chain n-Alkanes-Next, double, triple, and quadruple deletion mutants of the HFD genes were constructed by the pop-in/pop-out method. These strains were analyzed in terms of their growth on various carbon sources. The wild-type strain and the deletion mutants of the HFD genes were incubated on glucose for 3 days, on n-alkanes of 10 -18 carbons for 7 days, or on dodecanal for 10 days (Fig. 3A and Table 7). The single deletion mutants exhibited no growth defects on the medium containing glucose or n-alkane. In contrast, the deletion mutant lacking all four HFD genes (⌬hfd1-4 strain) grew normally on glucose, but did not grow on n-alkanes of 12-18 carbons ( Fig.  3A and Table 7). The ⌬hfd1-4 strain was found to grow on shorter chain n-alkanes, n-decane and n-undecane, but showed defective growth compared with the wild-type strain. Of the triple deletion mutants, the ⌬hfd1,2,3 and ⌬hfd1,3,4 strains exhibited slower growth on n-alkanes and formed smaller colonies on n-hexadecane, whereas the ⌬hfd1,2,4 and ⌬hfd2,3,4 strains grew normally on n-alkanes ( Fig. 3B and Table 7). The growth of the ⌬hfd1-4 strains harboring a plasmid to express a HFD gene was comparable with that of the corresponding triple mutant (data not shown). These findings suggest that the HFD genes share an essential function in the assimilation of n-alkanes of 12-18 carbons. The relative growth of the triple deletion mutants also suggests that HFD1 and HFD3 play major roles in n-alkane assimilation. Consistent with this, the double deletion mutant of HFD1 and HFD3 grew slowly on longerchain n-alkanes (Table 7).
On dodecanal, the growth of ⌬hfd1-4 strain was found to be comparable with the growth of the wild-type strain (Fig. 3A). A, Northern blot analysis using total RNAs extracted from the wild-type, ⌬hfd1, ⌬hfd2, ⌬hfd3, and ⌬hfd4 strains. Cells were grown to logarithmic phase in the glucose-containing medium, after which they were cultured in fresh medium containing glucose or n-decane for 1 h. Ribosomal RNAs stained with ethidium bromide are shown as loading controls. To detect HFD mRNA, the probe specific for HFD1, HFD2, HFD3, or HFD4 was used. B, qRT-PCR analysis. The wild-type strain was cultured to logarithmic phase in the glucose-containing medium, after which they were cultured in fresh medium containing glucose (Glc), n-decane (C10), n-dodecane (C12), n-tetradecane (C14), n-hexadecane (C16), or oleic acid (Ole) for 1 h. Total RNAs were extracted from the cells and reverse-transcribed to cDNAs. cDNAs were detected using primers specific for HFD1, HFD2, HFD3, or HFD4. Each result represents an average of three independent experiments Ϯ S.E.

Fatty Aldehyde Dehydrogenases in Y. lipolytica
Because the ⌬hfd1-4 strain exhibited severely defective growth on the n-alkanes, it is possible that fatty aldehydes incorporated from the culture medium were oxidized by enzyme(s) different from those involved in the oxidation of fatty aldehydes produced in the process of n-alkane metabolism. The ⌬hfd1-4 strain also exhibited comparable growth to the wild-type strain on 1-dodecanol (Fig. 3A).
FALDH Activities of Hfd Proteins-Cell extracts were prepared from wild-type and ⌬hfd1-4 strains cultured in medium containing glucose alone or n-decane and glucose. FALDH activity in these cell extracts was assessed using either dodecanal or tetradecanal as a substrate (Fig. 4A). The n-alkane-containing medium was supplemented with glucose to support growth of the ⌬hfd1-4 strain, because glucose does not significantly affect the n-alkane-activated transcription of a subset of genes, including ALK1, involved in the assimilation of n-alkanes (15,16,39). For both substrates (dodecanal or tetradecanal), the FALDH activity in the extract of the wild-type strain cultured in the presence of n-decane was substantially higher than the FALDH activity in extracts of cells cultured with glucose. In contrast, no significant increase in the FALDH activities was observed in the ⌬hfd1-4 strain extract cultured in the presence of n-decane. These findings suggest that the ⌬hfd1-4 strain is defective in n-alkane-induced FALDH production.
To confirm that oxidation of fatty aldehydes to fatty acids is catalyzed by Hfd proteins, we attempted to purify each Hfd protein recombinantly produced in E. coli. We were, however, unable to purify Hfd proteins tagged with His 6 or glutathione S-transferase at their N termini. FALDH activities were thus measured in extracts of E. coli producing each untagged Hfd protein (Fig. 4B). In the extract of cells producing Hfd1p, Hfd2Ap, or Hfd4p, significant FALDH activities to dodecanal and tetradecanal were detected compared with the extract of vector-transformed E. coli. The FALDH activity in the extracts of cells producing Hfd4p in particular was the highest. Weak FALDH activity to dodecanal was detected in the Hfd3p-producing cell extract. These results suggest that Hfd proteins have FALDH activity. HFD1 is predicted to encode a protein of 519 amino acids, whereas Hfd2p produced in E. coli is predicted to be a protein of 521 amino acids. However, the mobility of Hfd1p produced in E. coli is much higher than that of Hfd2p (Fig. 4B). Therefore, it is possible that Hfd1p is degraded in the E. coli lysate and this affects the activity of Hfd1p in vitro.

TABLE 7 Growth of deletion mutants on n-alkanes of various carbon lengths
Deletion mutants were cultivated on indicated n-alkanes for 1 week. Growth equivalent to the wild-type cells was indicated as "ϩϩϩ."

JOURNAL OF BIOLOGICAL CHEMISTRY 33281
Hfd4p) under control of their native promoters. The introduction of these plasmids into the ⌬hfd1-4 strain restored the growth on n-alkanes, indicating that these proteins are functional in Y. lipolytica (data not shown). These plasmids were introduced into the REA1 strain, which produces Sec61p tagged with DsRed as an ER marker (23), and into the POC strain, which produces Pot1p tagged with mCherry as a peroxisome marker. These cells were then cultured to logarithmic phase in medium containing glucose. n-Dodecane was added to the culture medium at a final concentration of 2%, after which the cells were further incubated for 1 h. The fluorescent signals of EGFP-Hfd1p and EGFP-Hfd2p were found to be partially colocalized with those of Sec61p-DsRed or Pot1p-mCherry (Fig. 5), suggesting that Hfd1p and Hfd2p are localized in both the ER and peroxisome. The fluorescent signal of EGFP-Hfd3p was colocalized exclusively with that of Pot1p-mCherry (Fig. 5), indicative of a predominant peroxisomal localization for Hfd3p. The fluorescent signal of EGFP-Hfd4p was weak compared with that of the other Hfd proteins, and they exhibited a punctate pattern distinct from the signals of Sec61p-DsRed and Pot1p-mCherry (Fig. 5). Similar localization patterns were observed for EGFP-Hfd1p, EGFP-Hfd2p, EGFP-Hfd3p, and EGFP-Hfd4p when the cells were cultured in glucose-containing medium (data not shown).

Roles of Two Hfd2p
Variants-To determine which variant of Hfd2p contributes to n-alkane assimilation, plasmids pSHFD2A and pSHFD2B were constructed to express either Hfd2Ap or Hfd2Bp alone, respectively, under the native promoter, and were introduced into the ⌬hfd1-4 strain. These strains were incubated on n-dodecane or n-hexadecane for 7 days, and their growth was observed (Fig. 6A). The growth of both strains was found to be similar to that of the ⌬hfd1-4 strain containing the wild-type HFD2, which produces both variants.
Next, the localization of Hfd2Ap and Hfd2Bp was investigated. Plasmids pSEGFP-HFD2A and pSEGFP-HFD2B were constructed to express Hfd2Ap and Hfd2Bp, respectively, fused with EGFP at their N termini (EGFP-Hfd2Ap and EGFP-Hfd2Bp). These fusion proteins were shown to complement the defective growth of the ⌬hfd1-4 strain on n-alkanes, indicating that they are functional in vivo (data not shown). The pSEGFP-HFD2A and pSEGFP-HFD2B plasmids were then introduced into REA1 and POC strains to allow for subcellular localization of EGFP-Hfd2Ap and EGFP-Hfd2Bp to be assessed. The cells were cultured to logarithmic phase in glucose-containing medium, after which n-dodecane was added to the culture medium at a final concentration of 2%. The cells were further incubated for 1 h. The fluorescent signal of EGFP-Hfd2Ap was colocalized with that of Sec61p-DsRed and Pot1p-mCherry (Fig. 6B). Similar results were obtained when localization of Hfd2Ap tagged with EGFP at its C terminus (Hfd2Ap-EGFP), which was indicated to be functional in Y. lipolytica (data not shown), was observed (Fig. 6B). These results imply that Hfd2Ap is localized in both the ER and the peroxisome. In contrast, the fluorescent signal of EGFP-Hfd2Bp was colocalized with the signal of Pot1p-mCherry, but not with that of Sec61p-DsRed, as expected from the presence of the PTS1-like sequence. These results suggest that Y. lipolytica utilizes FALDH in the peroxisome to assimilate n-alkanes.

DISCUSSION
In this study, the four HFD genes encoding FALDHs in Y. lipolytica were characterized. The ⌬hfd1-4 strain did not grow on n-alkanes of 12-18 carbons and exhibited partial growth defects on n-alkanes of 10 -11 carbons, whereas Y. lipolytica mutant strains expressing one of four HFD genes exhibited substantial growth on n-alkanes. n-Alkane-induced FALDH activity was not detected in the ⌬hfd1-4 strain extract. Significant FALDH activity was, furthermore, detected in extracts of bac- terial cells producing the Hfd proteins. These findings strongly suggest that HFD genes encode FALDHs that oxidize fatty aldehydes to fatty acids in the process of n-alkane metabolism in Y. lipolytica. HFD Genes in n-Alkane Assimilating Yeasts-S. cerevisiae has a single FALDH gene, HFD1, whereas, in contrast, four HFD gene paralogs are present in the Y. lipolytica genome. Interestingly, two paralogs are found in the genomes of a subset of n-alkane-assimilating yeasts, including C. albicans, C. parapsilosis, C. tropicalis, and L. elongisporus (Candida database at Broad Institute). The HFD genes may thus be duplicated or multiplicated to efficiently metabolize or detoxify potentially harmful fatty aldehydes in the process by which these yeasts acquire the ability to assimilate n-alkanes or hydrophobic compounds.
Transcription of HFD Genes-Transcription of HFD1, HFD2, and HFD3 was notably induced by the presence of n-alkanes, whereas that of HFD4 was not significantly induced by n-alkanes (Fig. 2). Transcription of ALK1 and a subset of ALK genes is known to be regulated by transcription activators Yas1p and Yas2p and repressor Yas3p (18,(21)(22)(23). It is thus possible that transcription of HFD1-HFD3 is regulated by the Yas1p-Yas2p-Yas3p system.
qRT-PCR analysis revealed that HFD1 transcription is highly up-regulated by oleic acid (Fig. 2B). In Y. lipolytica, the transcription of genes involved in the metabolism of fatty acids is activated in the presence of fatty acids, and Por1p, a Zn 2 Cys 6 transcription factor, is involved in the transcriptional activation of a subset of these genes by fatty acids (44). Whether or not Por1p is involved in the regulation of HFD1 transcription in response to fatty acids remains to be addressed. MTLY 37, a Y. lipolytica strain, in which POX2, POX3, POX4, and POX5 encoding acyl-CoA oxidases involved in fatty acid ␤-oxidation are deleted, was reported to accumulate hexadecanedioic acid when fed hexadecanoic acid (45). HFD1 may be involved in the oxidation cascade of the -terminal end of fatty acids.
Oxidation of Fatty Aldehydes in n-Alkane Assimilation-According to the model of n-alkane assimilation in yeasts, fatty aldehydes are oxidized to fatty acids in the ER or in the peroxisome (4). The localization of Hfd1p, Hfd2p, and Hfd3p to the ER and/or the peroxisome observed in this study strongly supports this model. Because the cells expressing only Hfd2Bp or Hfd3p, which were suggested to be localized exclusively in the peroxisome (Figs. 5 and 6B), were found to grow on n-alkanes, it can be concluded that fatty aldehydes are oxidized to fatty acids in the peroxisome in the process of n-alkane assimilation. Considering that n-alkanes are hydroxylated to fatty alcohols by P450ALKs in the ER, our findings imply that fatty alcohols and/or fatty aldehydes are transported from the ER to the peroxisome in these strains. It remains to be elucidated how Hfd1p, Hfd2Ap, and Hfd3p lacking typical PTS1 sequences are targeted to the peroxisome. It would be of interest to assess whether the C-terminal sequences of Hfd1p and Hfd2Ap, ALL and IFL, respectively, function as peroxisome targeting signals.
It is not clear whether n-alkanes can be assimilated by the ER-localized FALDH. In mice, variants of FALDH (FALDH-N, -V, -V2, and -V3) are generated by alternative splicing from the ALDH3A2 gene (34,46). Among these variants, FALDH-N, V2, and V3 are localized in the ER and FALDH-V is localized in the peroxisome (34,47). Interestingly, overproduction of FALDH-V or FALDH-N was reported to confer resistance to dodecanal on HEK293 cells, suggesting that FALDH variants are involved in protection against the cytotoxicity of dodecanal in the peroxisome or the ER. Fatty aldehydes derived from n-alkanes may thus be converted to fatty acids by the ER-localized FALDH in Y. lipolytica. EGFP-Hfd4p did not appear to be localized in the peroxisome in this study; however, the expression of HFD4 in the ⌬hfd1-4 strain grown on n-alkanes was found to restore the growth partially. Although subcellular localization of Hfd4p was not determined in this study, the fluorescent signal of EGFP-Hfd4p might be partially colocalized with that of the Sec61p-DsRed signal. Further detailed analysis of the localization of Hfd4p would provide important information on whether or not fatty aldehydes are oxidized to fatty acids in the ER or in other organelle(s) in the process of n-alkane assimilation. S. cerevisiae Hfd1p and mammalian ALDH3A2 are known to catalyze the conversion of 1-hexadecenal to hexadecenoic acid in the metabolism of sphingosine 1-phosphate probably in the ER (37). The ER-localized Hfd proteins of Y. lipolytica may thus be involved in the metabolism of sphingosine 1-phosphate.
The ⌬hfd1-4 strain was found to grow on the medium containing dodecanal as the sole carbon source (Fig. 3A), whereas it could not grow on n-alkanes of more than 12 carbons. One potential reason for this is that the enzymes oxidizing fatty aldehydes incorporated from culture medium could be different from those involved in the oxidation of fatty aldehydes produced in the metabolism of n-alkanes. It is possible that there is channeling of n-alkane metabolites, perhaps in the peroxisome. Fatty aldehyde dehydrogenase activity was still detected in the ⌬hfd1-4 strain extract (Fig. 4A), and the ⌬hfd1-4 strain might grow on dodecanal by oxidizing it with this residual activity. Y. lipolytica has four orthologs of the S. cerevisiae ALD genes encoding cytosolic or mitochondrial aldehyde dehydrogenases (Fig. 1A). However, in our preliminary results, a mutant deleted for all of four HFD genes and four ALD orthologs was able to grow on dodecanal. Thus, other dehydrogenase(s) could be involved in the oxidation of incorporated fatty aldehydes in Y. lipolytica. Cytochrome P450 52A3 of C. maltosa was suggested to catalyze the oxidation of fatty aldehydes (30), and thus P450ALKs represent another candidate group for the oxidation FIGURE 6. Functions and subcellular localization of Hfd2Ap and Hfd2Bp. A, the wild-type strain carrying pSUT5 (vector) and the ⌬hfd1-4 strain carrying pSUT5, pSHFD2, pSHFD2A, or pSHFD2B were cultivated on glucose for 3 days or on n-dodecane or n-hexadecane for 7 days. B, cells co-expressing Hfd2Ap or Hfd2Bp fused with EGFP at its N or C terminus and Sec61p-DsRed or Pot1p-mCherry were grown in glucose-containing medium to logarithmic phase. n-Dodecane was added, after which the cells were further incubated for 1 h. Localization of Hfd fusion proteins and Sec61p-DsRed or Pot1p-mCherry was observed as described under "Experimental Procedures." Bars, 3 m. of short chain fatty aldehydes or incorporated fatty aldehydes in Y. lipolytica.
Alternatively, Hfd proteins may function upstream of fatty aldehyde oxidation in the n-alkane metabolism pathway. The initial hydroxylation of n-alkanes was shown to be catalyzed by P450ALKs, and the mutant deleted for all of 12 ALK genes are unable to grow on n-alkanes (19). Therefore, it is less probable that Hfd proteins play critical roles in the hydroxylation of n-alkanes. The ⌬hfd1-4 strain appears not to be defective in the oxidation of incorporated fatty alcohols, because it grew on 1-dodecanol comparably to the wild-type strain (Fig. 3A). However, it is not clear whether Hfd proteins are involved in the oxidation of fatty alcohols produced by the hydroxylation of n-alkanes. Thus, it remains to be determined whether Hfd proteins play additional roles in the n-alkane metabolism pathway.
The ⌬hfd1-4 strain exhibited slight growth on n-decane or n-undecane (Table 7). Because the mutant deleted for all four HFD genes and four ALD orthologs also could grow on these n-alkanes in our preliminary results, other dehydrogenase(s) or P450ALKs might be involved in the oxidation of shorter chain fatty aldehydes, as discussed above.
Transcript Variants of HFD2-Two HFD2 transcript variants, which encode Hfd2Ap without PTS1 and Hfd2Bp with a PTS1-like sequence, were generated from HFD2. Our results suggest that Hfd2Ap is localized in both the ER and the peroxisome, whereas Hfd2Bp is localized exclusively in the peroxisome. The distinct production of two FALDH isoforms localized in the ER or the peroxisome is achieved by alternative splicing in mice and humans (34). Our findings indicate that the determination of FALDH localization in the ER and/or the peroxisome by RNA splicing is conserved from yeasts to mammals. Overproduction of the peroxisome-localized FALDH-V, but not the ER-localized FALDH-N, was reported to protect HEK293 cells from damage caused by phytanic acid, implying an involvement of FALDH-V in the metabolism of phytanic acid in the peroxisome (34). The generation of the transcript variants by RNA splicing could thus contribute to diversification of FALDH proteins. In Y. lipolytica, significant differences were not observed between the ratios of the two HFD2 transcript variants in the cells cultured in the presence or absence of n-alkanes. In mouse liver, the relative mRNA ratios of the transcript variants of the ALDH3A2 gene were not altered significantly in the presence or absence of Wv14,643, an agonist of PPAR␣, which activates transcription of the ALDH3A2 gene (34). The functional differences between Hfd2Ap and Hfd2Bp remain unclear.
Chain-length Specificity of Hfd Proteins-The transcript levels of HFD1 and HFD2 were found to be higher in the presence of n-decane compared with longer chain n-alkanes. In contrast, the transcript level of HFD3 was higher in cells grown on n-hexadecane compared with cells grown on shorter chain n-alkanes. The correlation between the transcription profile of each HFD gene in response to n-alkanes of various carbon numbers and substrate specificities of its product is unclear, because we were unable to measure FALDH activity using fatty aldehydes of 10 or 16 carbons as a substrate. It is possible that the solubility of hexadecanal is too low to be used as a substrate in our FALDH activity assay system. Pathway of n-Alkane Assimilation-In n-alkane-assimilating yeasts, n-alkanes are converted to fatty acids via a three-step oxidation process. It has been shown that the initial oxidation of n-alkanes to fatty alcohols is catalyzed by P450ALKs (2-7). The results of this study suggest that Hfd proteins are responsible for the oxidation of fatty aldehydes to fatty acids in Y. lipolytica. The missing link in the n-alkane assimilation pathway is the enzyme(s) that catalyze(s) the oxidation of fatty alcohols to fatty aldehydes. It has been reported that a deletion mutant lacking the alcohol dehydrogenase genes, FADH and ADH1-ADH7, and a fatty alcohol oxidase gene, FAO1, exhibited defects in the oxidation of long chain -hydroxy fatty acids, but was still able to metabolize n-dodecane and oxidize 1-dodecanol (26). It is thus possible that multiple genes are involved in the oxidation of fatty alcohols, as is the case for n-alkanes and fatty aldehydes. Alternatively, fatty alcohols may be oxidized by P450ALKs as reported for C. maltosa P450 52A3 (30). Our findings provide a crucial step toward elucidating the n-alkane metabolism in yeasts, and will contribute to our deeper understanding of the metabolism of hydrophobic compounds in lower eukaryotes.