Isozyme function of n-alkane-inducible cytochromes P450 in Candida maltosa revealed by sequential gene disruption.

An n-alkane-assimilating yeast Candida maltosa contains multiple n-alkane-inducible forms of cytochromes P450 (P450alk), which can be assumed to catalyze terminal hydroxylation of n-alkanes in the assimilation pathway. Eight structurally related P450alk genes have been identified. In the present study, the function of four major isoforms of P450alk (encoded by ALK1, ALK2, ALK3, and ALK5 genes) was investigated by sequential gene disruption. Auxotrophic markers used for the selection of disrupted strains were regenerated repeatedly through either mitotic recombination between heterozygous alleles of the diploid genome or directed deletion of the marker gene, to allow sequential gene disruptions within a single strain. The strain depleted of all four isoforms could not utilize n-alkanes for growth, providing direct evidence that P450alk is essential for n-alkane assimilation. Growth properties of a series of intermediate disrupted strains, plasmid-based complementation, and enzyme assays after heterologous expression of single isoforms revealed (i) that each of the four individual isoforms is alone sufficient to allow growth on long chain n-alkane; (ii) that the ALK1-encoding isoform is the most versatile and efficient P450alk form, considering both its enzymatic activity and its ability to confer growth on n-alkanes of different chain length; and (iii) that the ALK5-encoding isoform exhibits a rather narrow substrate specificity and thus cannot support the utilization of short chain n-alkanes.

Cytochromes P450 are heme-containing monooxygenases that are distributed widely among living organisms (1). Higher eukaryotes generally contain multiple forms of cytochrome P450 catalyzing diverse oxidative reactions in the metabolism of a large number of endogenous and xenobiotic compounds. Depending on the induction level and substrate specificity of individual cytochromes P450, cellular metabolic processes are often effected by a specific ensemble of isoforms, making it difficult to distinguish their biological function.
In microorganisms, such a cytochrome P450 multiplicity only rarely can be found. One of the microorganisms harboring a cytochrome P450 multigene family is an n-alkane-assimilating yeast Candida maltosa (2)(3)(4)(5)(6). Resembling the situation in Candida tropicalis (7), this yeast species contains n-alkaneinducible forms of P450 (P450alk), and eight structurally related P450alk genes belonging to the CYP52 family were identified. Coupled with NADPH-cytochrome P450 reductase (8), they are assumed to catalyze the terminal hydroxylation of n-alkanes, which represents the first and rate-limiting step in the n-alkane assimilation pathway. The functions of the C. maltosa cytochromes P450 were first apparent from in vitro reconstitution of n-alkane hydroxylase and NADPH-cytochrome P450 reductase purified from n-alkane-grown C. maltosa cells (9) and could be confirmed later by in vivo CO inhibition studies (10). In addition to these investigations, recent studies revealed the enzymatic characters of some P450alk forms by means of heterologous expression in Saccharomyces cerevisiae (11)(12)(13). However, investigations about the cellular significance of individual P450alk isoforms for the n-alkane assimilation pathway are still fragmental. In particular, it remains to be clarified whether indeed such a P450alk multiplicity is required for the n-alkane assimilation pathway of C. maltosa and to what extent each of the gene products of eight P450alk genes contributes to the respective n-alkaneassimilating phenotype. Recent development of the genetic engineering systems including host-vector systems as well as the gene disruption method, which enabled us to disrupt stepwise both alleles of a certain gene in the diploid genome by using two selectable markers, have greatly facilitated molecular biological analyses of C. maltosa (14 -19). As an initial step to address the cellular function of individual isoforms of P450alk, we previously disrupted the first isolated P450alk gene and found that the disrupted stain retained the ability to utilize n-alkanes as a sole carbon and energy source because of the n-alkane hydroxylation activity of the other P450alk forms. In the present paper we describe the sequential disruption of multiple P450alk genes and show the effect of these disruptions on the n-alkaneassimilating property of C. maltosa. The results in this paper provide evidence that the difference in substrate specificity among multiple isoforms of P450alk in vivo can be detected clearly by using the gene disruption technique even in a nonconventional yeast.

EXPERIMENTAL PROCEDURES
Strains and Media-The C. maltosa strains used in this study are listed in Table I. Media for C. maltosa were YPD (1% yeast extract, 2% Bacto-pepton, and 2% glucose) and YNB (0.67% Yeast Nitrogen Base (Difco) without amino acids and either 2% glucose or 1% n-alkanes), which was supplied with appropriate nutrients. 5-Fluoro-orotic acid * This work was supported in part by a grant-in-aid for science research from the Ministry of Education, Science, and Culture of Japan and by Grant 0310257 from the Bundesministerium fü r Forschung und Technologie. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ Recipient of the Japan Society for the Promotion of Science Fellowships.
Plasmid Construction and Yeast Transformation for Gene Disruptions-A 1.8-kb HindIII-EcoT22I fragment of pCMU6 carrying the C. maltosa URA3 gene (18) was subcloned into the HindIII-PstI sites of pUC19 to construct pURAD. A 1.6-kb SspI-DraIII fragment of the C. maltosa ADE1 gene (16) was cloned into the SalI site of pUC119 through the SalI linker to construct pUADE. A 6.7-kb SalI-BamHI fragment carrying entire ALK2-B and ALK3-B genes and their 5Ј-and 3Ј-noncoding regions (4) was subcloned into pUC19 to construct pUA23. For gene disruption of the ALK2-ALK3 locus, a 1.6-kb EcoRV-BamHI fragment of pURAD and a 1.5-kb DraI-BamHI fragment of pUADE replaced a 4.1-kb BglII-EcoRV fragment of pUA23 to construct pUD23U and pUD23A, respectively. The alk2-alk3::URA3 and the alk2-alk3:: ADE1 cassettes for gene disruption were excised from pDA23U and pDA23A, respectively, with PstI-BamHI double digestions in both cases, and transformed into C. maltosa. The transformation of C. maltosa was performed by a modified lithium acetate method (14). For regeneration of the ADE1 marker, pDA23⌬ade was constructed by ligation of a 1.3-kb PstI-EcoRV fragment from pUD23A and a 1.3-kb EcoRV-BamHI fragment from pUA23 into the PstI-BamHI sites of pUC19, and the alk2-alk3::⌬ade1 cassette was generated by digestions with PstI-BamHI for the C. maltosa transformation. In these three cassettes, the complete ALK2 coding region, NH 2 -terminal one-third of the coding region of ALK3, and the ALK2-ALK3 internal region (1.0 kb) were replaced. A 2.1-kb HindIII-HindIII fragment of ALK5-A (6) was subcloned into the HindIII site of pUC19 to construct pUA5. A 0.6-kb BglII-EcoRV fragment of pUA5 was replaced with either the 1.6-kb EcoRV-BamHI fragment of pURAD or the 1.5-kb DraI-BamHI fragment of pUADE to construct pUD5U and pUD5A, respectively. The alk5::URA3 and the alk5::ADE1 disruption cassettes were excised from pUD5U and pUD5A, respectively, with PstI-ScaI digestions in each case. Both cassettes replaced the NH 2 -terminal 40% of the coding region of ALK5.
A 2.8-kb EcoT22I-EcoT22I fragment of ALK1-A (3) was subcloned into the PstI site of pUC119 to construct pUA1. A 0.5-kb EcoRV-EcoRV fragment corresponding to the central one-third of the ALK1 coding region was replaced by a 1.8-kb HindIII-BamHI fragment of pURAD containing URA3 through blunt end ligation to construct pUD1U. The alk1::URA3 disruption cassette was excised from pUD1U with SphI-BamHI digestions. The construction of the other plasmids for the ALK1 disruption, pUD45A and pUD45H, was described previously (17).
Southern Blot Hybridization-Total DNA of C. maltosa was isolated from a culture of 10 ml of YNB-glucose medium as described (18). Southern blot analysis was performed with an ECL (enhanced chemiluminescence) gene detection system (Amersham) in accordance with the instruction of the supplier. For detection of the appropriate gene replacements of the ALK2-ALK3 locus, total DNA from the transformants was digested with BamHI and ClaI and probed with a 0.6-kb HindIII fragment of ALK3-B (see Fig. 2). For the ALK5 and ALK1 replacements, total DNA from the transformants was digested with HindIII and PstI-BglII, respectively, and probed with a 2.1-kb HindIII fragment of ALK5-A and a 2.3-kb HindIII fragment of ALK1-A, respectively. The appropriate replacements were confirmed as follows. Both of the intact alleles of ALK5 gave a 2.2-kb doublet band. The genomic blot of the first step transformant gave a 3.3-kb band as predicted for the replacement with URA3 in addition to the 2.2-kb band of the undisrupted allele. That of the second step transformant gave new 1.0-and 2.0-kb bands as predicted for the replacement with ADE1 in addition to the 3.3-kb band for the URA3-replacement, and the 2.2-kb band for the intact ALK5 allele disappeared. In ALK1 disruptions the genomic blots of the transformants for the replacements with URA3, ADE1, and HIS5 gave 8.2-, 9.1-, and 9.8-kb bands, respectively. Disruption of both alleles of ALK1 was confirmed by detection of each of the two respective bands for the replacements and by the disappearance of the 7.0-kb band for the intact allele of ALK1. A 0.9-kb SalI fragment of pURAD (18), a 0.7-kb EcoRI-EcoRV fragment of ADE1 (16), and a 2.3-kb EcoRI-Hin-dIII fragment of HIS5 (15) were also used as probes to confirm the replacements of URA3, ADE1, and HIS5, respectively. P450 Induction Experiments-To determine the expression level of spectrally active P450 in the wild-type strain CHAU1 as well as in the disruptant strains DA235-24, DA123-14, and DA1235-12, respective cultures were first grown to a density of about 1 ϫ 10 7 cells/ml in a yeast minimal medium containing 1.34% Yeast Nitrogen Base, 2% glucose and, as required, histidine (50 mg/liter), adenine (100 mg/liter), and uracil (40 mg/liter). Then cells were washed in the fresh medium without glucose and cultivated under the same conditions except for 1% dodecane as a sole carbon source. P450 contents were determined after an induction time of 4 h by means of CO difference spectra (10).
Plasmid Construction for Plasmid-based Complementation-A 1.8-kb PstI-XhoI fragment of the TRA region, which contains an autonomously replicating sequence and a centromeric DNA of C. maltosa (19), was isolated from the vector pTRA11 (14) and cloned into the PstI site of pUC18 through a PstI linker. From the resulting plasmid a 1.8-kb PstI fragment was inserted into the AatII site of pUC19 through blunt end ligation. Then a 1.2-kb DraI-XbaI fragment of URA3 from pURAD was inserted into the XhoI site of the TRA region through an XhoI linker to construct pUTU1. A 2.8-kb BamHI-SphI fragment of ALK1-A isolated from pUA1 was inserted into the BamHI-SphI sites of pUTU1 to construct pUTU-ALK1. A 2.6-kb BglII fragment of ALK2-A and a 3.3-kb EcoT22I-EcoRV fragment of ALK3-A were inserted into the BamHI site and the PstI-SmaI sites of pUTU1 to construct pUTU-ALK2 1 The abbreviations used are: 5FOA, 5-fluoro-orotic acid; kb, kilobase(s).  (17), and CHAU1 (18) were described previously.
and pUTU-ALK3, respectively. A 3.5-kb PstI-EcoRI fragment of ALK5-A was inserted into the PstI-SmaI sites through blunt end ligation to construct pUTU-ALK5. Exchanges of the promoter regions of ALK1 and ALK5 genes were done by polymerase chain reaction as follows. To obtain recombinant products having ALK1 promoter and ALK5 coding region, primers 5Ј-AACATCTGGTCATGATTGATGAAATACTTCCT-3Ј and 5Ј-TCATCA-ATCATGACCAGATGTTATAAATG-3Ј were used. For the product having ALK5 promoter and ALK1 coding region, primers 5Ј-TAAACTA-AAAATATGGCTATAGAACAAATTATTG-3Ј and 5Ј-GTTCTATAGCC-ATATTTTTAGTTTAATCTTATTTAAA-3Ј, respectively, were used. Underlines correspond to the translational initiation codons. As flanking primers, M13 forward and reverse sequencing primers were used. Plasmids pUTU-ALK1 and pUTU-ALK5 were used as templates for the first polymerase chain reaction step to obtain the respective promoter and coding region. Then the second polymerase chain reaction step generated exchanged recombinant products. The products were cloned into the KpnI-SphI sites of pUTU1 to construct pUTU-A1/A5 (ALK1 promoter and ALK5 coding region) and pUTU-A5/A1 (ALK5-promoter and ALK1-coding region), respectively.
Heterologous Expression and Enzyme Assay-Construction of the four coexpression vectors used for the simultaneous production of the individual C. maltosa P450alk forms and NADPH-cytochrome P450 reductase was described previously (13). Heterologous expression in S. cerevisiae was done as described previously (11). The enzyme assay was carried out using the microsomal fraction using [1-14 C]dodecane (Sigma) and [1-14 C]hexadecane (Amersham) as substrates as described previously (13).

RESULTS
Disruption of Single Loci-We have shown that four out of the eight P450alk genes of C. maltosa are significantly induced by n-alkanes and may thus encode P450 enzymes directly involved in n-alkane assimilation in the previous study (6). These four genes have been designated ALK1, ALK2, ALK3, and ALK5. Each of them occurs in two allelic variants in agreement with the diploid nature of C. maltosa genome. Previous gene disruption experiments revealed that a C. maltosa strain defective in ALK1 retained the ability to grow on long chain n-alkanes (17). Therefore, it has been the first objective of the present study to disrupt each of the other three n-alkaneinducible P450alk genes (for the experimental strategy applied, see Fig. 1) and to examine on this basis their contribution to the phenotype of n-alkane assimilation.
The growth phenotype of the strains DA23-81 and DA5-61 as well as the ALK1-disrupted strain DA1-44 (17) are shown in Table II. All of these strains were able to utilize at least medium and long chain n-alkanes (C14 and C16) as sole carbon sources for growth, indicating that none of the three loci (ALK1, ALK2-ALK3, and ALK5) was solely essential for n-alkane-assimilating ability of C. maltosa. However, it was obvious that the growth of the strain DA1-44 showed significant but weak growth on the n-alkane of short chain length (C12) compared with the parental and the other strains.
Regeneration of Selectable Markers-To investigate the cellular function of individual P450alk isoforms, the respective P450alk genes had to be disrupted simultaneously within a single strain. However, because the initial round of disruption resulted in a Ura ϩ and Ade ϩ phenotype and only the HIS5 marker was available, methods to restore the selectable markers to the disrupted strains should have been developed to permit reutilization of the markers for further disruption of the remaining P450alk genes. In one method, the introduced selectable marker ADE1 was substituted for a deletion derivative of it by means of cotransformation with a selectable plasmid to facilitate the selection of the replacement. The ADE1-deleting cassette alk2-alk3::⌬ade1 was transformed into strain DA23-81 along with an HIS5-containing autonomously replicating plasmid, pBTH10B (15). Among the resulting His ϩ transformants, some developed red colored colonies characteristic of the ade1 mutant. They were tested for adenine auxotrophy, and the appropriate replacement with the alk2-alk3::⌬ade1 cassette was confirmed by Southern blot analysis of the genomic DNA of an Ade Ϫ strain, DA23-816 (Fig. 2, lane 4). Then, the plasmid pBTH10B was cured from the strain DA23-816 by growth on nonselective medium.
The other methods used for the regeneration of a selectable marker in a single strain were based on the selectability of an ura3 mutant by their resistance to 5FOA, a toxic analog of an intermediate in the uracil synthetic pathway. Application of the 5FOA selection of a ura3 mutant in C. maltosa has already been shown in a previous report (18). Spontaneous 5FOAresistant strains were selected from the strain DA23-816, and their uracil requirement was tested. Southern blot analysis of a Ura Ϫ strain, DA23-8164, indicated that this strain was homozygous for alk2-alk3::⌬ade1 (Fig. 2, lane 5). As previously shown, a recombination event between both alleles occurred frequently during the selection of 5FOA resistance (18). The strain DA23-8146 (his5, ade1, ura3, alk2-alk3::⌬ade1/alk2-alk3:: ⌬ade1) was devoid of ALK2-ALK3 in both alleles; nevertheless, three selectable markers could be utilized further in this strain.
Development of Multiple P450alk Gene Disruptions and Their Effects on n-Alkane Assimilation-Mutants containing multiple P450alk gene disruption were generated according to the scheme shown in Fig. 1. The complete set of disruption in all the three P450alk loci, ALK1, ALK2-ALK3, and ALK5, was obtained by sequential disruption. The complete genotypes of the strains generated are listed in Table I. In each step, the appropriate genotypes were confirmed by Southern blot anal-ysis of the isolated strains (see "Experimental Procedures"), as shown in the case of the ALK2-ALK3 disruption.
The growth properties of P450alk gene disruptants on nalkanes of different chain lengths as sole carbon sources were examined (see Table II). Strain DA1235-12, in which all four of the genes were disrupted, could not utilize n-alkanes of any chain lengths for growth, although this strain was able to grow on n-alcohols and fatty acids as sole carbon sources. The growth property of this strain indicated that the sequential gene disruption resulted in a complete functional block of the n-alkane assimilation at the first step.
Strains DA123-14, DA235-24, and DA15-23, in which two of the three loci were disrupted, showed good growth on C16, indicating that each of the products of the three loci was alone sufficient for the assimilation of long chain n-alkane. Interestingly, a specificity of the n-alkane-assimilating property was observed depending on the chain length of n-alkanes. The strain DA123-14 could not utilize C12 at all for growth, whereas the other strains grew significantly on this substrate, although some of them showed weak growth. And the strain DA123-14 alone showed weak growth on C14. This growth property of the strain DA123-14 indicated that the ALK5 product that remained functional in this strain could function fully only on the long chain n-alkane. Strains devoid of ALK1 showed merely poor growth on C12, whereas strains expressing ALK1 showed good growth on C12. These results indicated that the ALK1 product played an important role for assimilation of short chain n-alkanes.
To prove the expression of spectrally active P450 in the gene disruptant strains, P450s were induced by C12 as described under "Experimental Procedures." Whereas the wild-type strain CHAU1 expressed a significant amount of P450 (0.11 nmol/10 8 cells), no significant expression of P450 was observed in the P450 disruptant strain DA1235-12. Under the same conditions, the presence of P450 could be detected clearly in strain DA235-24 (0.09 nmol/10 8 cells) and DA123-14 (0.02 nmol/10 8 cells). Interestingly, in the strain DA235-24 in which only the ALK1 product remained functional, almost 80% of the P450 amount of the wild-type strain could be found, indicating that this P450 form was most abundant P450 in C. maltosa. In contrast, the amount of P450 in the strain DA123-14 in which only ALK5 remained functional was less than 20% of the wildtype strain.
Plasmid-based Complementation with Each P450alk Gene-Function of individual P450alk genes toward n-alkane assimilation was analyzed further by plasmid-based complementation. Each of four entire genes having 5Ј-and 3Ј-flanking regions was cloned into pUTU1, which carried both an autonomously replicating sequence and a centromeric DNA sequence of C. maltosa (19), transformed into the strain DA1235-121, which was devoid of all four of the genes and in which the   4). The predicted band for the replacement, 3.5 kb, was detected in addition to the 4.7-kb band for the alk2-alk3::URA3 allele. When probed with ADE1, the 4.6-kb band of alk2-alk3::ADE1 in the strain DA23-81 (lane 3) had disappeared in the strain DA23-816 (data not shown). Hybridization of a 5FOA-resistant and Ura Ϫ strain, DA23-8164, showed that the 4.7-kb band for the alk2-alk3::URA3 allele had disappeared and that the 3.5-kb doublet band of the alk2-alk3::⌬ade1 was detected. The 4.7-kb band for the alk2-alk3::URA3 had also disappeared in strain DA23-8146 when probed with URA3 (data not shown).
URA3-selectable marker could be utilized, and examined for n-alkane assimilating property. Table III summarizes the results of the complementation experiments. Complementation with ALK1 resulted in the complete recovery of n-alkane assimilating property, to utilize n-alkanes of different chain length as sole carbon sources for growth. Although the disruption experiments could not distinguish between the function of the products of ALK2 and ALK3, the results of the complementation indicated that both the ALK2 and the ALK3 products solely could function significantly in assimilation of n-alkanes of different chain length. However, both complemented strains grew only weakly on C12 as in the cases of the ALK1-disrupted strains. Unlike these three genes, ALK5 was able to complement the assimilating property of n-alkanes with only C16 but not with C12 and C14.
To distinguish whether the functional specificity of the ALK5 product depending on the chain length of n-alkanes was responsible for its enzymatic activity or for its expression level, two plasmids in which the promoter regions of ALK1 and ALK5 had been exchanged were examined for their ability to complement the assimilation phenotype of the strain DA1235-121 (see Table III). The strain carrying pUTU-A5/A1, which had the ALK5 promoter region and the ALK1 coding region, grew well on both the long and short chain n-alkanes. The strain carrying pUTU-1A/5A, which had the ALK1 promoter region and the ALK5 coding region, grew only on the long chain n-alkane, showing the same growth property of the strain carrying pUTU-ALK5A. The cellular P450 contents after induction with C12 were found to be nearly the same for both strains (0.03 and 0.02 nmol/10 8 cells, for strains carrying pUTU-A1/A5 and pUTU-A5/A1, respectively). These results indicated that although ALK5 was induced by C12 to the significant amount in the cell, the presence of the ALK5 protein itself did not contribute to the growth on C12, disclosing a restricted substrate specificity of the ALK5 product toward C12.
Enzyme Activities of Individual P450alk Isoforms-Enzymatic activities of individual P450alk isoforms were characterized by means of turnover rates toward n-alkanes. Because expression levels of P450 in the complemented strains were limited, heterologous overexpressions of individual P450alk forms in another yeast, S. cerevisiae, in which the related activity is absent, were applied. A highly active P450 monooxygenase reconstitution system consisting of P450alk and NADPH-cytochrome P450 reductase from C. maltosa has been established already in vivo in S. cerevisiae (11,13). Microsomal fractions were prepared from the respective strains overexpressing P450alk and then assayed for P450alk-encoded hydroxylase activities toward C12 and C16 (see Table IV). The ALK1 product displayed the strongest hydroxylase activity, suggesting that it was the most important P450alk form for the primary hydroxylation of n-alkanes. Products of ALK2 and ALK3 showed n-alkane hydroxylation activities of approximately half of the ALK1 product, respectively. No preference of the activity depending on n-alkane chain lengths was observed for these three P450alk forms. The ALK5 product showed relatively weak but significant activity toward C16 (approximately one-forth of the ALK1 product), whereas it showed very little activity toward C12. The substrate specificity of the ALK5 product toward the long chain n-alkanes was congruent with the results of the sequential gene disruption.

DISCUSSION
To our knowledge, this is the first report about gene disruption of multiple forms of cytochromes P450 and about its direct effect on phenotype. The sequential gene disruption of the P450alk genes resulted in a C. maltosa strain (DA1235-12) that could not utilize n-alkanes as sole carbon sources for growth because of the lack of the functional P450alk. Because only the P450alk genes were manipulated in the genome, this result provides direct evidence that P450alk participates in the nalkane assimilation pathway. In the proposed n-alkane assimilation pathway, n-alkane is first hydroxylated at the terminal position to produce n-alcohol and then oxidized successively to fatty acids. The growth property of the disrupted strain that could not assimilate n-alkanes but could grow on n-alcohols and fatty acids as sole carbon sources clearly correlated with the in vitro n-alkane-hydroxylating activity of the P450alk enzymes shown here and in previous studies (5,9,12,13), confirming that P450alk catalyzes the first step of n-alkane assimilation. This growth property also indicates that none of the four isoforms analyzed is essential in the further downstream steps of the assimilation pathway, although considerable fatty acid -hydroxylation activity has been shown for some of them (5,(11)(12)(13).
The overlapping function of the four P450alk forms toward hydroxylation of at least long chain n-alkanes is supported by the results that disruption of all four of the genes was necessary to generate the n-alkane-nonassimilating strain and that each of the four genes could complement the assimilating ability of the disrupted strain. Furthermore, their n-alkane hydroxylation activities were confirmed directly by an in vitro assay using a heterologous overexpression system.
The ALK1 product is the most important P450alk isoform because when it was present in the cells, the growth was significant on n-alkanes of any chain length. Its functional importance was supported by the strongest in vitro enzyme activity among the P450alks examined as well as by the considerable amount in the cells as shown in the strain DA235-24 in which only the ALK1 product was the functional P450alk. The absence of the functional ALK1 product in the series of disruptants resulted in weak growth on the short chain nalkanes, indicating that the ALK1 product is not essential but necessary for full activity for the assimilation of short chain  n-alkanes. Because the previous analysis of the gene disruption of ALK1 had been done using only n-hexadecane (17), we could not find this importance. The ALK2 and the ALK3 products can also hydroxylate n-alkanes of any chain length, although their function on short chain n-alkanes is restricted. However, the ALK5 product demonstrated a narrow spectrum toward long chain n-alkanes. The in vivo complementation experiment using the promoter-exchanged constructs and the direct in vitro enzyme assay (see Tables III and IV) showed that the functional specificity was affiliated not to the expression level of the enzyme but to the enzyme activity itself. Previous analysis on the induction of individual P450alk genes (6) revealed that ALK1 is induced strongly by short chain n-alkanes but relatively weakly by long chain n-alkanes. This result is congruent with the functional importance of the ALK1 product on short chain n-alkanes. However, ALK5 is induced by both long and short chain n-alkanes, whereas its product shows the substrate preference toward long chain n-alkane. Furthermore, although the four P450alk isoforms are not essential for the following steps of n-alkane assimilation pathway, three genes, ALK1, ALK2, and ALK5, are induced weakly but reproducibly by n-alcohols and fatty acids. The regulation of the expression of individual P450alk forms is not simply correlated with their functions.
The P450alk gene family in C. maltosa consists of at least eight members (6). The four isoforms characterized here are significantly induced when the carbon sources are n-alkanes, whereas the other four genes (ALK4, ALK6, ALK7, and ALK8) are induced only at low levels by n-alkanes as well as n-alcohols and fatty acids. The growth property of the n-alkane-nonassimilating strain indicates that these undisrupted four P450alk genes cannot function fully on the assimilation. However, further analysis such as in vitro assay using heterologous overexpression is necessary to determine whether they have enzymatic activity toward n-alkanes.
Methods to restore the auxotrophic requirements to cells previously transformed to respective prototrophy with the selectable marker genes have been developed here in C. maltosa to allow sequential gene disruption within a single strain. According to the strategy used previously (17), only both alleles of a single gene could be disrupted in a stepwise manner using two selectable markers. The cotransformation with a selectable autonomously replicating plasmid facilitated the selection of the regeneration of the ade1 auxotrophy. Moreover, the strain regenerating the ade1 marker could be distinguished easily because of red color development of the colony. Although the condition for the cotransformation was not optimized, a frequency of one ade1 strain among approximately 300 His ϩ transformants was obtained. Such a cotransformation strategy has also been applied in C. albicans where the gene replacement was selected by colony hybridization (20).
The 5FOA selection method also allowed us to regenerate the ura3 marker by means of the positive selection of a ura3 mutant. The strain having heterozygous disrupted alleles, one replaced with URA3 and the other by another marker or its deletion derivative, was rendered homozygous through mitotic recombination. Such a recombination was observed frequently in our previous study, in which ura3 mutants homozygous for disruption were obtained from the strain heterozygous for disruption by selecting the 5FOA resistance (18). Almost all ura3 mutants isolated by the 5FOA selection in this and in the previous studies were homozygous for disruption with a few exceptions, in which a local mutation in the URA3-carrying allele occurred. Such a strategy for the disruption of both alleles of a single gene has been applied previously to C. albicans, where UV irradiation has been used to facilitate recom-bination (21). The repeated use of URA3 for sequential disruptions has also been reported in C. tropicalis (22). In the case of C. tropicalis, since only URA3 has been available as a selectable marker in the host strain, the regeneration required the nystatin enrichment of ura3 strains. In principle, the combination of the two methods for regenerating the auxotrophic markers would allow us disruption of a nonlimiting number of genes in C. maltosa.
The disrupted strain unable to assimilate n-alkanes may provide a useful host for further investigation of structurefunction relationships of P450alk because mutant-type P450alks, the wild-type of which is nonfunctional on a specific substrate, can be positively selected by its n-alkane assimilating phenotype if they are functional. Because the structures of the four isozymes are significantly similar (56 -67% amino acid identities) to each other (4, 6), P450alk is a suitable model for such a study. Industrially, C. maltosa may also be useful for the production of some hydrophobic metabolic intermediates produced by a selected P450alk form such as dicarboxylic acids. Some P450alk isoforms should play important roles for the production because considerable fatty acid -hydroxylation activity has been reported for them (5,(11)(12)(13). In combination with the functional block of the ␤-oxidation system by gene disruption as described in C. tropicalis (22,23) and C. maltosa (24) and gene disruption of the appropriate P450alk forms as described here, it may be possible to construct an efficient strain for production.