The ascomycetous yeast Yarrowia lipolytica (formerly Candida, Endomycopsis, or Saccharomyces lipolytica) is one of the more intensively studied ‘non-conventional’ yeast species. This yeast is quite different from the well-studied yeasts Saccharomyces cerevisiae and Schizosaccharomyces pombe with respect to its phylogenetic evolution, physiology, genetics, and molecular biology. However, Y. lipolytica is not only of interest for fundamental research, but also for biotechnological applications. It secretes several metabolites in large amounts (i.e. organic acids, extracellular proteins) and the tools are available for overproduction and secretion of foreign proteins. This review presents a comprehensive overview on the available data on physiology, cell biology, molecular biology and genetics of Y. lipolytica.
In the last two decades the group of so called ‘non-conventional’ yeasts received more and more attention in fundamental research and biotechnology. The term ‘non-conventional’ was originally used to differentiate this group of yeasts from the more commonly used — ‘conventional’ — and already well studied yeasts Saccharomyces cerevisiae and, with some restrictions, Schizosaccharomyces pombe. Nowadays about ten species among these non-conventional yeasts have been investigated more intensively with respect to their physiology, genetics, molecular biology, and biotechnical application. A summary of these data and the special tools developed has been published recently .
The ascomycetous yeast Yarrowia lipolytica (originally classified as Candida lipolytica) is one of the more intensively studied ‘non-conventional’ species. With the emergence of single-cell (SCP) protein projects in the mid-sixties, a strong industrial interest arose from the fact that strains of this species were able to use n-paraffins (which were cheap and abundant in this period) as sole carbon source. It was also observed that Y. lipolytica was able to produce high amounts of organic acids (2-ketoglutaric acid and citric acid) when grown on these substrates . Large-scale industrial production of citric acid or of SCP using Y. lipolytica thus permitted the accumulation of extensive data on its behavior in very large fermentors.
Three reviews have been already published which describe the historical development of this research and summarize main data on physiology, genetics, molecular biology, and biotechnological application of this fungus [3–5] .
This review will give a comprehensive overview on the physiology and genetics of this yeast. It will also show recent areas of interest which include both fundamental and applied topics.
Natural habitats, taxonomy, and phylogenetic relationship
Y. lipolytica strains are readily isolated from dairy products, but also from shoyu or from salads containing meat or shrimps. The inability of this yeast to survive under anaerobic conditions permits their easy elimination from dairy products, contrary to Kluyveromyces marxianus strains for example . An extensive description of assimilated substrates is done by Kreger-van Rij  and Barnett et al. . Glucose, alcohols, acetate, and hydrophobic substrates like fatty acids or alkanes, but not sucrose are used as carbon sources.
The species was originally classified as a Candida, since no sexual state had been described. The perfect form of C. lipolytica was identified in the late sixties by Wickerham at the Northern Regional Research Laboratory of the USDA at Peoria. A culture isolated in 1945 from a jar of fiber tailings in a corn processing plant was found to form asci attached to hyphal elements when put on suitable media. One to four spores of various size and shape could be isolated from these asci, but spore viability was found to be very low. Two mating types, called A and B, were identified among the progeny. Nearly all other wild type isolates from the species would mate to one of these two types, albeit at very low frequency, suggesting that most natural isolates are haploid (or near haploid). The perfect form was reclassified first as Endomycopsis lipolytica, then as Saccharomycopsis lipolytica, and finally as Yarrowia lipolytica.
Y. lipolytica is not considered as pathogenic .
Several observations suggested that Y. lipolytica may have diverged considerably from other ascomycetous yeasts: high GC content, unusual structure of rDNA genes coupled with a lack of RNA polymerase I consensus sequences found in other yeasts , higher eukaryotic-like size of snRNA , and of 7S RNA . Homologous genes tend to display a low level of similarity (typically in the 50–60% range at amino acid level) with their counterparts in S. cerevisiae, K. lactis or C. albicans, thus hindering in most cases attempts to clone homologous genes by means of Southern hybridization. Naumova et al.  showed that Y. lipolytica genes did not hybridize detectably with DNA from species of the genera Saccharomycopsis, Endomyces or Endomycopsella, some of which were formerly classified in the same genus as Y. lipolytica. Similarly, only few genes of Y. lipolytica seem to be directly expressed in S. cerevisiae, suggesting that RNA polymerase II promoters and/or associated transcriptional factors have diverged considerably. Recent data on 7S RNA genes, on intron structure, or on ARS function confirmed that this species was quite peculiar when compared to most other yeasts. Evolutionary trees based on sequence comparison of genes encoding well conserved functions (glycolytic genes, ribosomal RNA genes) locate Y. lipolytica on an isolated branch, clearly separated from S. pombe on one hand, and from the bulk of other ascomycetous yeasts on the other [16, 17] . A truly positive aspect of this divergence is that the observation of structural conservation between Y. lipolytica and other yeast genes is much more likely to reflect functional constraints than when it happens between closely related yeasts.
Assimilation of carbon sources
Utilization of hydrocarbons as carbon source
Y. lipolytica can use n-alkanes and 1-alkenes as carbon sources [18–21] . Polymethylated and chlorinated alkanes are also assimilated by this fungus [22, 23] . However, only few enzymes are characterized and no gene involved in the early steps of degradation of these compounds has so far been cloned.
The first step in the assimilation is likely to be an emulsification at the cell surface to form small droplets which can be internalized. A 27 kDa extracellular emulsifier, called liposan, is induced in cells growing on n-alkanes [24, 25] . Liposan contains 88% carbohydrate and 12% protein. After entry into the cell, n-alkanes are hydroxylated by a cytochrome P-450 monooxygenase system. Cytochrome P-450 was detected in several strains of Y. lipolytica [26–28] . The synthesis of this enzyme is induced during growth on n-alkanes, but not on glucose, ethanol or acetate. Differences in sensitivity to glucose repression during growth on hexadecane and decane indicate the presence of different regulated cytochrome P-450 genes in Y. lipolytica. Cytochrome P-450 is localized in the membrane of the endoplasmic reticulum as shown by subcellular fractionation of alkane-grown cells of Y. lipolytica. 1-alkanol formed after the first step is further oxidized by a membrane-bound fatty alcohol oxidase [30, 31], not by NAD(P)-dependent alcohol dehydrogenases, which are also present in Y. lipolytica.
Several alkane-non-utilizing mutants have been isolated and characterized [33–36] . Uptake of n-alkanes is inducible and due to active transport. Mutations in 16 loci caused a significant reduction in n-alkane uptake . Mauersberger  characterized three subtypes of alkA mutants, which were unable to utilize either all types of n-alkanes (alkAa), or only short ones in the C8 to C12 range (alkAb), or only long alkanes in the C16 range (alkAc). These data indicate that Y. lipolytica contains either several length specific alkane-uptake systems or specific cytochrome P-450 monooxygenases.
Fatty acids biosynthesis and degradation
Fatty acid biosynthesis and degradation were initially investigated by Kamiryo et al.  and a few reports were issued on the use of Y. lipolytica for the stereospecific conversion of hydroxylated long chain fatty acids into γ-lactones .
Kohlwein and Paltauf  detected at least two fatty acid carrier systems in Y. lipolytica, one being specific for C12 or C14 fatty acids, the other for C16 and C18 saturated or unsaturated fatty acids. Octanoic acid and decanoic acid are not taken up by this yeast. Contrary to S. cerevisiae, an intracellular 100 kDa fatty acid binding protein is readily induced in Y. lipolytica when grown on palmitate . A low molecular mass fatty acid binding protein (15 kDa) has been isolated from the cytosol of this yeast recently .
An anabolic acyl-CoA synthetase I activity was found in peroxisomes, mitochondria and cytoplasm and was shown to be required for the incorporation of exogenous fatty acids into cellular lipids. Mutants deficient in this activity have been isolated .
A catabolic acyl-CoA synthetase II was found exclusively in peroxisomes, and was required for the β-oxydation of fatty acids. A mutant initially reported as affecting this activity was later shown to affect another step (Kamiryo, personal communication).
Five genes encoding acyl-CoA oxidases have been identified and cloned (Nicaud, personal communication).
Use of Y. lipolytica cells for the production of γ-decalactone from alkyl ricinoleate (a derivative of castor oil) has been patented by BASF (DE4126997, 1993), [44, 38] . Other possible application include production of wax esters from fatty alcohols .
The POT1 gene encoding a catabolic 3-oxoacyl-CoA-thiolase, which catalyzes the thiolytic cleavage of 3-oxoacyl-CoA thioesthers during β-oxidation of fatty acids, was cloned and characterized by Berninger et al. . Disruption of POT1 inhibited the utilization of oleate but not the elongation of externally added tridecanoic acid to higher-chain-length homologues.
Assimilation of alcohols
Y. lipolytica does not produce ethanol, but uses ethanol as carbon source at concentrations up to 3%. Higher concentrations of ethanol are toxic. Several NAD+- and NADP+-dependent alcohol dehydrogenases were observed in Y. lipolytica. There probably exist two NAD+-dependent alcohol dehydrogenases which differ in substrate specificity. Synthesis of both enzymes seems not to be repressible by glucose or inducible by ethanol . Three NADP+-dependent alcohol dehydrogenases were detected which exhibited different substrate specificities. The occurrence of these enzymes varies depending on growth phase and carbon source of the media .
Assimilation of acetate
Most strains of Y. lipolytica grow very efficiently on acetate as sole carbon source. Concentrations up to 0.4% sodium acetate are well tolerated, higher concentrations reduce the growth rate and concentrations above 1.0% inhibit the growth.
Several mutants have been isolated and characterized, which are blocked in the utilization of acetate [33, 36, 47–50] . Nothing is known about the uptake of acetate by this yeast. Mutants blocked in the activity of acetyl-coenzyme A synthetase were characterized by Kujau et al. . Acetyl-coenzyme A is needed for the induction of the glyoxylate cycle, which is not induced in acetyl Coenzyme A deficient mutants [36, 47].
The induction of the glyoxylate pathway is necessary for the utilization of alkanes, fatty acids, alcohols and acetate. Isocitrate lyase, the key enzyme of this cycle, and its encoding gene is well studied [49–51, 36, 47, 48] . The structural gene has been cloned and sequenced .
Dominant mutations in the gene GPR1 (glyoxylate pathway regulator)  make the cells sensitive to low concentrations of acetate or ethanol even in the presence of glucose.
Secretion of metabolites
Wild type strains of Y. lipolytica secrete a mixture of citric and isocitric acid when grown on n-paraffins as carbon source: a total yield of 130% compared with the substrate and a ratio of 60:40 (citric to isocitric) was reported for strain ATCC 20114 by Akyiama et al. . When the cells were supplied with monofluoroacetate, this compound was transformed into monofluorocitrate which competitively inhibited aconitase: this in turn improved the ratio of citric to isocitric acid to 85:15. In an effort to obtain a strain with low aconitase activity, a mutant unable to use citrate as carbon source was first isolated, and further mutagenized in order to obtain a fluoroacetate sensitive strain: this last strain yielded citric and isocitric acids in a ratio of 97:3, with a yield of 145% (w/w).
Tréton and Heslot  observed that on glucose or glycerol medium, the ratio of citrate to isocitrate was 92:8 vs 67:33 on n-paraffin. However the intracellular ratio of the two acids was roughly the same (90:10) on all media. These authors further showed that both acids were equally well secreted into the growth medium, but that only citric acid was reconsumed on paraffins, whereas neither was reused on glucose.
Use of mutants unable to use acetate or citrate as carbon sources has been evaluated by Wojatatowicz et al. . A direct selection of citrate overproducing strains has been described by McKay et al. .
Processes have been described for producing citric acid from glucose or n-paraffins in liquid cultures (for a review see Mattey, ) or with immobilized cells  but also from date coats , tapioca starch hydrolyzates  and glucose hydrol .
α-ketoglutarate and other organic acids
Y. lipolytica strains grown on n-alkanes in the presence of a limited thiamine supply (a vitamin not synthesized by this yeast but required for α-ketoglutarate dehydrogenase activity) secrete large amounts of α-ketoglutarate. Use of a diploid strain for α-ketoglutarate production has been patented (Maldonado and Gaillardin, 1972). Furthermore, a haploid strain selected as a suppressor mutant of a mutation in methionine biosynthesis secretes very large amounts of this organic acid during growth on glucose (patented by Weißbrodt et al., 1988).
Production of 2-hydroxyglutaric and 2-ketoglutaric acids from glucose has been described by Oogaki et al. . A process for very high yield production of isopropylmalic acid using a leucine auxotrophic strain has been patented .
Mutations affecting all 11 biosynthetic steps of lysine except one have been identified. A single locus, LYS1 encoding homocitrate synthase controls the first step [63, 64] . The existence of isoenzymes hitherto prevented characterization of such mutants in S. cerevisiae. A detailed analysis of the LYS1 gene identified homocitrate synthase as a major check point of the lysine biosynthetic pathway . The LYS1 gene has recently been cloned and sequenced . Mutations at the LYS11 locus, unlinked to LYS1, were initially isolated as suppressors of LYC1 mutations . They abolish completely homocitrate synthase activity. It is unknown presently if LYS11 defines a regulator of LYS1, or a second subunit of homocitrate synthase.
No mutant affecting homoaconitate hydratase has been identified so far. This activity could not be purified away from aconitate hydratase. LYS6 and LYS7 mutations are defective in the third step of the pathway (homoaconitase), thermosensitive mutants at the LYS6 locus identified it as homoaconitase structural gene. Homoisocitrate dehydrogenase was purified to homogeneity. Two tightly linked loci, LYS9 and LYS10, actually define the same structural gene encoding a bifunctional protein involved in successive dehydrogenation and decarboxylation of homoisocitrate . LYS8 mutants are unable to convert α-ketoadipic acid into α-aminoadipic acid suggesting that a single enzyme catalyzes this step in Y. lipolytica vs. probably two isoenzymes in S. cerevisiae. LYS2 and LYS3 are two closely linked loci, which control adenylylation of α-aminoadipic acid, a rate-limiting step in vivo and the only one in the pathway which is insensitive to the general control of amino acid biosynthesis . The two last steps of the pathway are controlled by LYS4 and LYS5, respectively. The latter gene, encoding saccharopine dehydrogenase, has been cloned and sequenced [68, 69] .
Y. lipolytica uses lysine both as a nitrogen and as a carbon source, via the N-6-acetyllysine-5 aminovalerate pathway, like Hansenula saturnus [70, 71] . The LYC1 gene encoding the first step of the pathway, N-6-lysine acetyl transferase (LAT), was cloned and sequenced  and shown to be induced by lysine.
Combining mutations desensitizing homocitrate synthase (LYS1.5), as well as mutations simultaneously preventing lysine catabolism and relieving the lysine effect on saccharopine dehydrogenase (lyc1.5), yielded strains accumulating 40 times more lysine than the wild type strain . No lysine secretion in the growth medium was observed, indicating that very efficient systems existed for uptake and/or retention within the cell.
Both a high and a low affinity transport system have been described for lysine uptake  and mutants affecting either or both systems were identified. Intracellular lysine is stored in the vacuole . Mutants affecting vacuolar storage of lysine were identified among strains selected for reduced content of polyphosphates in the vacuole [76, 77], although no direct relationship could be established between lysine and polyphosphate pools. A direct selection of mutations preventing lysine retention within the cell was carried out, starting with a LYS1.5, lyc1.5 strain: mutants displayed a complex phenotype including reversion of the lyc1.5 mutation, mating-type switch, and mating-type context dependent phenotypes .
Secretion of proteins
Extracellular alkaline protease (AEP)
AEP is encoded by the XPR2 gene, one of at least 11 genes controlling AEP synthesis, secretion and/or activity [81, 82] . The gene has been cloned and sequenced from three different strains [83–85] . AEP is a 32 kDa protease of the subtilisin family, which is intracellularly processed from a 55 kDa glycosylated precursor [84, 86] . Processing includes cleavage of a 15-amino acid presequence and glycosylation of a unique site on the propeptide (both occurring cotranslationally during translocation into the endoplasmic reticulum), processive hydrolysis of a tail of 9 N-terminal dipeptides (X-Ala or X-Pro) by a Golgi dipeptidyl aminopeptidase, followed by the cleavage of the propeptide at a Lys-Arg junction ahead of the mature part. The endopeptidase is encoded by the XPR6 gene, a KEX2 homolog which has recently been cloned and sequenced . The propeptide has been shown to be required for folding, transit and activation of the mature part [88, 89], a finding which has since been extended to various proenzymes in S. cerevisiae.
Regulation of AEP synthesis is complex and reflects the carbon, nitrogen, sulfur and pH status of the cell, among others. A study of the promoter has been initiated; internal deletions and in vivo footprints demonstrate that there are two main UASs located 700 bp (UAS1) and 40 bp (UAS2) upstream from the TATA box . Each UAS seems to contain targets for general transcription factors (RAP1- and possibly ABF1-like). These UAS are able to activate a minimal promoter consisting of LEU2 TATA box and initiation site (Madzak et al., to be published). Unlinked mutations affecting transcriptional control of XPR2 have been selected and identify at least four unlinked loci (Lambert et al., submitted). One of these identifies a zinc-finger transcriptional activator (YlRim1p) which synthesis and activity are controlled by ambient pH (Lambert et al., submitted).
Acid extracellular protease
On rich YPD medium at pH 4.0, an acid protease activity was detected in the growth medium. Three protein species were initially described, of 28, 32 and 36 kDa . More recent data suggest that there is only one acid protease, which may undergo partial proteolysis in some strains and/or under certain growth conditions . Interestingly, induction of the acid protease occurs under conditions very similar to that used for inducing AEP, except for the pH of the medium. The gene encoding acid protease has been cloned and sequenced (Young, unpublished), and its disruption abolishes acid protease activity. It encodes a preproenzyme with an unusually large 5′ upstream region. Interestingly, there seem to exist some conservation of the promoter elements of both acid and alkaline protease. Comparing the determinants of the regulation of these two genes might thus be quite revealing.
A secreted RNase activity is detected in Y. lipolytica cultures grown under conditions leading to alkaline protease secretion, the major 45 kDa protein being partially degraded by AEP into 43 kDa and 34 kDa proteins . RNase is probably synthesized as a 73 kDa glycosylated precursor , but definite confirmation of the maturation process awaits determination of the sequence of the gene.
A cell wall-bound acid phosphatase activity is induced if Y. lipolytica is grown on media depleted of inorganic phosphate sources. A glycosylated protein of 90–200 kDa has been purified, which is converted into a 60 kDa species upon endoglycosidase H treatment. Antibodies directed against it cross-react with S. cerevisiae major acid phosphatase . A kinetic study of this enzyme has been published .
Tréton et al.  tried to clone its structural gene by functional complementation of a pho5,pho3 mutant of S. cerevisiae. This approach failed to identify the gene searched for, but yielded PHO2, the gene of a secreted minor phosphatase. The PHO2 gene product has no homology to other acid phosphatases, it corresponds to an acid phosphatase activity with a narrow substrate spectrum, the synthesis of which is induced in cells starved for inorganic phosphate.
Extracellular lipase and esterase
Y. lipolytica strains display a lipase activity which acts preferentially on oleyl residues at positions 1 and 3 of the glyceride. Lipase activities were examined by different investigators on different strains. It is unknown whether the reported discrepancies reflect differences in the methodologies used, or the existence of different types of strains. Peters and Nelson [96, 97] described a single glucose repressible activity with a pH optimum around pH 6.2–6.5. Zviagintzeva et al.  also reported on a single cell wall-anchored lipase, whereas Kalle et al.  described two cell-bound activities: a constitutive, glucose insensitive lipase, and a second activity induced by sorbitan monooleate. Ota et al.  described both an extracellular activity in cultures supplemented with a protein-like fraction derived from soybean, and cell-bound lipases. The extracellular lipase required oleic acid as a stabilizer/activator, whereas the cell-bound lipases did not, and differed in several properties from the extracellular enzyme. Sugiura et al.  and Ota et al.  solubilized two cell-bound, monomeric lipases of 39 and 44 kDa, which differed in their pH optima. Gomi et al.  showed that the lipase(s) exist in the cell wall as an activator-bound complex, which is rapidly dissociated upon enzyme purification. The extracellular, activator dependent activity may thus reflect release of (up to 50% of) the cell-bound acivity, perhaps under conditions leading to extracellular protease(s) induction. Further details on Y. lipolytica lipase(s) may be found in a recent review .
Mutants unable to use tributyrin as a carbon source were isolated after nystatin enrichment , and were shown to define three complementation groups . A gene encoding a secretory protein highly homologous to fungal lipases has been recently cloned in an independent approach (A. Dominguez, personal communication): this should rapidly permit assessing if there is more than one lipase activity in Y. lipolytica.
Y. lipolytica lipase may be of interest for synthesis of 2,4-dimethylglutaric acid monoesters, transesterification of meso-cyclopentane diols  or for use in the leather industry or in cheese manufacturing (German patent DD-272867).
An extracellular thermostable esterase of low molecular weight (10 kDa), specific for the 1-position of triglycerides has recently been described .
Y. lipolytica grows well on oleic acid and this is accompanied by an extensive proliferation of peroxisomes. Antibodies directed against the targeting signal Ser-Lys-Leu COOH (anti-SKL) of peroxisomal proteins of several yeasts, react strongly with three peroxisomal proteins of Y. lipolytica. Using these two features, studies on peroxisome biosynthesis have recently begun with the identification of 17 oleic acid non-utilizer mutants. Three of them were shown to affect peroxisomal assembly (pay mutants) by a rapid immunofluorescence assay, using anti-SKL antibodies . The punctate labelling pattern observed with anti-SKL in wild type cells was lost in the mutants, no peroxisomal structures could be observed, and two peroxisomal marker enzymes (β-hydroxyacyl-CoA dehydrogenase and catalase) failed to localize to the particulate fraction in mutant cells. Genes complementing pay mutations (PAY2, PAY4, PAY5, PAY32) were isolated by complementation using a genomic library made in a replicating vector. The PAY2, PAY4, PAY5 and PAY32 genes have been sequenced [110–113] . Pay2p is a 42 kDa peroxisomal integral membrane protein. It is essential for matrix protein import and enlargement of peroxisomes . PAY4 encodes a 112 kDa hydrophilic protein, presumably localized in the cytoplasm, which is expressed at low levels on glucose containing medium and is strongly induced upon shifting to oleic acid. Pay4p exhibits an ATP binding site and appears closely related to Pas5p of Pichia pastoris (59% identity) and less to Pas1p of S. cerevisiae, both involved in peroxisome assembly in their respective hosts . PAY5 encodes a peroxisomal integrale membrane protein homologous to the mammalian peroxisome assembly factor PAF1 . PAY32 encodes an intraperoxisomal protein of the matrix protein translocation machinery .
Secretion and genes involved in of the secretory pathway
An essential gene called RYL1 has been isolated by hybridization , and may represent a homologue of SEC4, which encodes a small G protein involved in the control of secretory vesicle fusion with the plasma membrane. A homologue of S. cerevisiae SEC14 was identified both by PCR and by functional complementation of a S. cerevisiae sec14 mutant using a cDNA expression library of Y. lipolytica mRNA made in a S. cerevisiae expression vector : unexpectedly, the deletion of this gene in Y. lipolytica proved to be viable and did not affect protein secretion, although recent electron microscopy data suggest that (derivatives of) the Golgi apparatus look abnormal (Rambourg et al., unpublished). This deletion however prevented the yeast-mycelium transition characteristic of this species, suggesting the existence of some link between Golgi function(s) and cellular differentiation.
Several genes controlling early steps of the pathway have been characterized. A homologue of SEC61 (one of the subunits of the translocation pore into the endoplasmic reticulum) was cloned by PCR and shown to be highly conserved between S. cerevisiae and Y. lipolytica (68% identity at amino acid level). A homologue of SEC62 was isolated using the cDNA expression library to complement a sec62 thermosensitive mutant of S. cerevisiae. Its predicted product shows only 37% identity with its S. cerevisiae counterpart (D. Swennen, unpublished).
Rather close homologues of a signal recognition particle (SRP), involved in the targeting of secretory polypeptides towards the endoplasmic reticulum, have been observed in Y. lipolytica and in S. pombe. On the contrary, S. cerevisiae SRP seems to have diverged considerably and was identified later . Y. lipolytica SRP contains a typical 7S RNA, which can be immunoprecipitated from whole cell extracts by human anti SRP antibodies . This RNA is encoded by two unlinked genes, SCR1 and SCR2, whose products share 94% identity . Either gene can be deleted, but the double deletion is lethal , indicating that essential secretory proteins strictly require the SRP dependent pathway for secretion. This also seems to be the case for AEP, whose synthesis and secretion are severely depressed under non-permissive conditions in conditional mutants of SCR1 or SCR2 [120, 117] . Such mutants were created by site directed mutagenesis of conserved nucleotides in a loop binding Srp19p, one of the six polypeptides associated to mammalian 7S RNA. The genes encoding two of these polypeptides, Srp19 (SEC65) and Srp54 have been cloned and sequenced recently (M. Sanchez et al., unpublished; D. Ogrydziak, unpublished).
In an attempt to identify new genes whose product may (directly or not) interact with the SRP, mutations displaying colethality with conditional mutations within SCR2 were selected . One of these mutations identifies a gene (SLS1) whose product is predicted to be a resident protein in the lumen of the endoplasmic reticulum. Interestingly, deletion of this gene leads to a thermosensitive phenotype, which is further pronounced in the presence of a 7S RNA thermosensitive mutation, and results in a dramatic decrease of the amount of AEP precursor synthesis: this suggests that Sls1p may facilitate SRP-dependent translocation of secretory precursors from the lumenal side of the ER membrane.
Wild type strains of Y. lipolytica exhibit various colony shapes, ranging from smooth and glistening to heavily convoluted and mat. The colony morphology is determined both by growth conditions (aeration, carbon and nitrogen sources, pH etc.) and by the genetic background of the strain. Wild type isolates may give rise to various colonial types, each giving rise to the other types: the reason for this instability, which is not seen in laboratory strains, is unknown.
Y. lipolytica is a natural dimorphic fungus, which forms yeast cells, pseudohyphae and septate hyphae [122, 8] . True mycelium consists of septate hyphae 3 to 5 μm in width and up to several mm in length. Apical cells often exceed 100 μm, whereas segments are 50 to 70 μm long. There is a single nucleus per segments. Septa show a minute, ascomycete-type central pore, unusual for other filamentous yeasts, with endoplasmic reticulum extending through it from one segment to the next .
The proportion of the different cell forms depends on the strain used, probably accounting for the morphological differences observed at colony level. Certain conditions are known, however, which cause preferential formation of yeast cells, or induce mycelial development [124–126] . Mycelial development was inhibited by a deficiency of magnesium sulfate and ferric chloride or by the addition of cysteine or reduced glutathione. Rodriguez and Domiguez  could induce a complete and reproducible yeast-mycelium transition in minimal medium containing N-acetylglucosamine as sole carbon source, but this seems to be strain dependent, too. Guevara-Olvera et al.  enhanced the yeast-mycelial transition by heat shocking the cells during the inoculation in medium containing N-acetylglucosamine.
Some studies were done to detect physiological changes occurring during yeast-hyphae transition [127, 128, 126] . Comparison of the composition of yeast and hyphal cells have shown that hyphal walls exhibit a higher content of aminosugars and a reduced content of protein (Vega and Domiguez, 1986 ). Furthermore, ornithine decarboxylase activity and polyamine cell pools increased in hyphal cells grown on N-acetylglucosamine-containing medium .
Mutations in the genes SEC14, GPR1 (G. Barth, unpublished), and deletion of XPR6 (Ogrydziak, personal communication) have strong effects on the yeast to hyphae transition. However, it is not known whether the proteins encoded by these genes are directly involved in the regulation of this transition. Morphological mutants completely unable to form hyphae, can be easily selected after visual inspection of the colonies .
Mating and mating type alleles
All natural isolates of Y. lipolytica so far tested are heterothallic. The mating type is determined by the two alleles MATA and MATB. The MATA allele has been cloned . Similar to heterothallic strains of S. cerevisiae haploid and diploid cells are vegetatively stable. Kurischko and Weber  reported spontaneous haploidization of some diploid cells after formation of zygotes. Furthermore, it was observed in several laboratories that diploid strains spontaneously sporulate during prolonged storage on complete media. Therefore, it is recommended to prepare fresh diploid strains for genetic analysis.
Mating frequencies of natural isolates are always very low (1% viable zygotes/cell or less). Attempts to increase the mating response by inbreeding [134–136, 49] failed to raise mating frequency above 15% viable zygotes/cell. Mating frequency is dependent on several parameters besides the genetic background of strains: medium, cell density, growth phase, and temperature [134, 137] .
Y. lipolytica does not require nitrogen limitation for induction of sporulation in contrast to S. cerevisiae and some other yeasts. Diploid strains sporulate on solid or in liquid complete medium when glucose is exhausted. Highest sporulation frequencies were obtained with many strains if 1.5% sodium citrate was used as a carbon source  at temperatures of 20°C to 30°C. High sporulation frequencies may also be obtained in liquid or on solid YM and V8 media .
Some natural isolates of Y. lipolytica sporulate quite well after mating but only inbred strains form a high proportion of complete tetrads and have a satisfactory level of germinating spores.
Initial studies with Wickerham's strains were plagued by low mating frequencies, poor sporulation ability and low ascospore viability: these defects could be partially alleviated by inbreeding programmes, but no perfect set of strains could be obtained. The proportion of complete tetrads is high and the spore germination of these strains plateaued at 80%, which however makes tetrad analysis possible (summary ). A very efficient method for selection of spores for random spore analysis using nystatin has been developed .
Since genetic data are mainly gathered currently on strains of the E129–E150 series, it might be advisable to localize new genes on this map. In any case, the existence of efficient methodologies for gene transfer/replacement and for retrieval of chromosomal mutations once the gene has been cloned, makes it likely that exchange of markers between different inbred lines will be much easier than in the past.
Genetics and molecular biology
Mutagenesis and mutants
Several commonly used chemical and physical mutagens have been tested on wild type and inbred strains of Y. lipolytica, and many different mutants have been isolated and characterized. This is not the place to describe all these mutants, but some main classes and general aspects will be discussed. Results of extensive studies on cell inactivation, frequency of mutants or of certain classes of mutants, spontaneous mutability and frequency of reversions in different strains of Y. lipolytica have been published [131, 140, 134, 141, 135, 142] .
Many mutants have been described which show nutritional requirements for most of the amino acids, adenine, uracil, and the vitamins biotin, nicotinic acid, pantothenic acid, riboflavin, and thiamine [131, 134, 143, 135, 142, 4]). There are several colour mutants described which turn the colonies green, brown or red ([143, 144, 142]) but no red adenine requiring mutants have been isolated. Several mutants have been described which are resistant to antifungal agents or different analogues of amino acids and carbon sources [142, 145, 63, 64] . Copper resistance varies widely among natural isolates of Y. lipolytica. Two tightly linked metallothionein genes, transcribed by a bidirectional promoter, have been cloned and sequenced (Dominguez et al., unpublished).
Most of the isolated mutants reverted at the expected frequency of 10−7 to 10−8 or less. However, some of the mutants reverted at much higher frequencies [142, 78, 36] . The reason for this high revertibility has not been clarified in most cases. The cloning and sequencing of one of such a highly revertible allele (GPR1–112) resulted in the detection of a retrotransposon which is involved in the reversion process .
The nystatin method for enrichment of mutants [147, 148] has been successfully adapted to concentrate auxotrophic and citrate non-utilizing mutants of Y. lipolytica [134, 142] . Enrichment of mutants using inositol-death has been reported (DeZeeuw, personal communication).
A G+C content of 49.6–51.7% has been reported for Y. lipolytica [149–151] . Determination of deoxyadenosine content of exponentially growing haploid cells yielded an estimated genome size of 4×109 Da or about 11 Mb , which would be smaller than that of S. cerevisiae. It was later reported that different haploid strains of Y. lipolytica exhibit wide variations in their DNA content estimated by this method , but no absolute values were given. Chromosome separation by pulse field electrophoresis yielded much larger genome estimates in the range of 12.7 to 22.1 Mb for wild type or laboratory strains . A pronounced chromosome length polymorphism was observed between different isolates, as well as a variation of the number of chromosomal bands detected (4 to 6), which may reflect both gel artifacts (chromosome comigration) and aneuploidy (evidence for a duplication of the URA3 marker on separate chromosomes was observed in a strain displaying 6 bands). Natural isolates of Y. lipolytica appeared to have widely divergent genetic structures, which may account for the low spore viability observed in the initial inbreeding programs.
Genetic linkage groups and chromosome maps
Based on tetrad dissection and random spore analysis, a genetic map encompassing more than 60 markers was established for a certain inbred line [135, 153] . Linkage was observed 8 to 10 times more frequently between random markers than in S. cerevisiae, suggesting that the average recombination per kb was lower than in S. cerevisiae. Five linkage groups were defined at that time but linkage studies have proceeded at slow pace since that time . Linkage groups were also defined for German lines  and for industrial strains (De Zeeuw, personal communication), but since marker exchange between the different lines is difficult, no pooled set of data is available so far. No centromere linked marker has been conclusively identified, but strains carrying centromeres tagged with a marker have been recently constructed (Fournier et al., 1993 ) and should help in this regard.
Inbred strains of Y. lipolytica tend to display 5 chromosomes in the range of 2 to 5 Mb (P. Fournier, S. Casaregola, unpublished). It should be noted that at this stage 5 is still an estimate for the chromosome number of the species: comigration may occur in this range of size. Some bands are fuzzy, which may be due to the presence of rDNA repeats, so that specific chromosome breakage will be required in order to get definite results. As expected from the study of the initial isolates from which the inbred lines were derived, each line shows a completely different chromosome pattern. An average size of 15 to 18 Mb has been estimated for inbred lines (Fournier and Nguyen, unpublished). A highly inbred strain (E150) giving consistently 5 well resolved chromosomal bands has been selected to assign 36 cloned genes by hybridization. Ribosomal RNA genes were found on four of the five chromosomes. The linkage groups appear conserved among different independent lines.
ARS and centromeres
Three different ARS have been isolated up to now [129, 157], each carrying a centromere (CEN) and a nearby chromosomal origin of replication (ORI): ORI3018/CEN3, ORI1068/CEN1, ORI4002/CEN4. ORI1068 and ORI3018 have been shown by two dimensional gel electrophoresis to be included in regions of initiation of replication in the chromosome (Fournier and Vernis, unpublished), whereas all three CEN sequences are able to induce chromosome breakage when integrated at the LEU2 locus. CEN1 and CEN3 were shown additionally to be closely linked genetically to a chromosomal centromere, and to correspond to a strong pausing site of the polymerase. ORI and CEN functions are carried by two independent regions of the original ARS inserts, and can be exchanged between different ARS: a given CEN sequence can thus apparently confer ARS function to any chromosomal ORI, thus opening the way to the cloning of many ORI. Several new ORIs were recently isolated from a DNA library made in a CEN non-transforming vector (Chasles and Fournier, to be published). One of them (ORIX009) was shown to be repeated at several places in the genome, another one originated from rDNA spacer.
ORI and CEN sequences of Y. lipolytica show no homology to S. cerevisiae or K. lactis corresponding sequences (nor to S. pombe), and the existence of a clear functional consensus for these sequences in Y. lipolytica has still to be assessed.
Ribosomal RNA and other RNA genes
Genetic data and hybridization of separated chromosomes showed that most if not all wild type isolates of Y. lipolytica contained several clusters of rDNA units located on 1–4 [12, 158] . Up to five types of repeated units were identified in a single strain, differing in length and structure of the non-transcribed spacer DNA. All isolates so far tested showed this polymorphic rDNA structure, but one strain (ATCC 18944) was reported to contain a single type of the unit .
Genetic recombination within the clusters occurs during meiosis, leading to expansion or reduction of individual clusters. Thus, crossing individual strains within a given Y. lipolytica inbred line results in the progeny in a strain-specific reshuffling of rDNA genes supplied by each parent . Strain typing can thus be done by simple inspection of hyperdense bands on digests of total chromosomal DNA. The total number of repeated units varies around a mean value of 50–60 . The contribution of the rDNA repeats to interchromosome exchange seems to be low in inbred lines and in wild type isolates.
Small nuclear RNAs U1, U2, U4, U5 and U6 were analyzed by Roiha et al. . Their sizes appeared much closer to that of human snRNA than to those of S. cerevisiae.
A retrotransposon, called Ylt1, was detected in the genome of Y. lipolytica. It is 9.4 kb long and can transpose in the genome. This retrotransposon is bounded by a long terminal repeat (LTR), the zeta element, which is 714 bp long, highly conserved, and can exist also as a solo element. Ylt1 and solo zeta elements are flanked by a 4-bp directly repeated genomic sequence. The copy numbers of Ylt1 and solo zeta are dependent on the strain examined, but at least 35 copies of Ylt1 and more than 30 copies of the solo zeta element per haploid genome have been observed.
Ylt1 belongs to the Ty3/gypsy group of retrotransposons and contains two overlapping large open reading frames (YltA and YltB). The predicted YltA gene product shows high similarity to retroviral Gag encoded proteins. YltB encodes protein domains with some similarity to retroviral gene products encoded by the pol gene. In the YltB gene, these predicted products occur in the order protease, reverse transcriptase, RNase H and integrase (Barth et al., in preparation).
Features of protein encoding genes
Several motifs similar to consensus sequences for transcriptional factor binding in S. cerevisiae can be observed in Y. lipolytica promoters, including TATA boxes, CT rich blocks, TUF/RPG or ABF1 binding sites, or UASGCN and UASLEU. The function of these sequences has not been assessed in most cases, with one exception so far , and the corresponding transcription factors have not been isolated. One gene encoding a transcription factor has recently become available: CRF1. Transcription starting points have been mapped in a few cases and often occur within a CCAAA type of structure, 20–30 bp downstream from the TATA box.
Positions −1 and −3 are strongly conserved upstream of the initiator ATG: an A is observed in 12/15 cases at position −1 and in 11/15 cases at position −3. An A is found at both positions in strongly expressed genes like XPR2, PGK1, TPI1, PYK1, ICL1 and LEU2.
Introns have been detected in several Y. lipolytica genes (5 intron containing genes among 37 known sequences). Two genes show two introns (CDC42 and SEC14) separated by a short exon. The 5′ end of the intron (donor site) is GTGAGTPu in all cases. The 3′ internal consensus (branch site) is TACTAAC in all cases but one (cgCTAAC in the first intron of SEC14), and is separated by one or two nucleotides only from the 3′ end of the intron: CAG.
Most Y. lipolytica genes show a typical signal for transcription termination TAG.TA(T)GT.TTT, which is located upstream of the site of poly A addition .
Codon usage appears to be different from that of S. cerevisiae and similar to that of Aspergillus (A. Dominguez, unpublished).
Plasmids and VLPs
No DNA plasmid was detected in a systematic survey of 24 wild type isolates (Tréton, unpublished), but a linear dsRNA of 4.9 kb was observed in several strains [161, 162] . This RNA is encapsidated within virus-like particles of 50 nm diameter. The capsid seems to be composed of two major polypeptides of 83 and 77 kDa [162, 163] . Based on hybridization data, there seems to exist at least two types of dsRNAs in different strains, which show little homology to one another . An additional linear dsRNA molecule of about 6 kb was detected in some strains besides the smaller 4.9 kb long dsRNA (Barth, unpublished) similar to the situation in killer strains of S. cerevisiae. No homology was found compared with genomic DNA of Y. lipolytica, nor with dsRNAs from S. cerevisiae. No killer phenotype associated with these VLPs could be evidenced. Curing after UV treatment did not lead to any phenotypic change.
The mitochondrial genome of Y. lipolytica has a buoyant density of 1.687 g cm−3 and a GC content of 24.9% . It consists of a circular molecule of 14.5 mm and its restriction map was established by Wesolowski et al.  on strain W29. The arrangement of genes for ATPase subunits, rRNA and 4SRNA are conserved with respect to S. cerevisiae and K. lactis. Several mitochondrial genes, including those for ATPase subunit 6, 8 and 9, cytochrome oxidase subunit 3, NADH oxidoreductase (ubiquinone) subunit 4 and tRNA genes, have been cloned [165, 166] . The restriction map of mitDNA seems to be well conserved among different isolates, as judged from the mobility of the corresponding hyperdense bands observed on digests of total genomic DNA stained with ethidium bromide.
A single mitochondrial mutation leading to oligomycin resistance was described by Matsuoka et al. . It has been used in protoplast fusion experiments.
Nowadays many genetic tools are available which make this yeast to an excellent model organism for addressing several fundamental questions. Powerful methods have been developed and well characterized strains and plasmids are available now for transformation, expression and secretion of foreign genes. A detailed description of these methods as well as a summary of the available strains, plasmids, cloned genes etc. is given in the review by Barth and Gaillardin .
The ‘non-conventional’ yeast Y. lipolytica turns out to be a superb model organism, both exquisitely original and generally meaningful. With the recent development of powerful gene amplification systems , the identification of a retrotransposon , and the characterization of strongly regulated or constitutive promoters [52, 90] the stage is now set for concentrating on specific biological problems in this yeast. Investigation of mechanisms of gene expression and protein secretion for expression of heterologous genes are mainly done for biotechnological application. Fundamental studies are directed now on the structure and functioning of the genome which is in some aspects very different from what is seen in other yeasts: structure and maintenance of polymorphic and dispersed rRNA genes, absolute requirement for a centromeric function for the maintenance of extrachromosomal plasmids, and genetic structure of natural populations which consist of widely divergent haploid lines. Other aspects of Y. lipolytica biology have begun to be explored, like determinants of the mating type, control of the dimorphic transition from yeast to hyphae, alkane and fatty acid metabolism, structure and function of retrotransposons, glyoxylic pathway, peroxisome biogenesis, etc.
Note added in proof
Addition to Section 6.3. It has been shown that chromosome band no. 3 was a doublet. Therefore, the chromosome number is six and the genome size is 18–20 Mb instead of 15–18 Mb.