Characterization of a gene in the car cluster of Fusarium fujikuroi that codes for a protein of the carotenoid oxygenase family

ORIGINAL PAPER

S. Thewes Æ A. Prado-Cabrero Æ M. M. Prado

B. Tudzynski Æ J. Avalos

Characterization of a gene in the car cluster of Fusarium fujikuroi thatcodes for a protein of the carotenoid oxygenase family

Received: 15 November 2004 / Accepted: 19 May 2005 / Published online: 28 July 2005� Springer-Verlag 2005

Abstract The ascomycete Fusarium fujikuroi producescarotenoids by means of the enzymes encoded by threecar genes. The enzymes encoded by carRA and carB areresponsible of the synthesis of b-carotene and torulene,respectively, while the product encoded by carT cleavestorulene to produce the acidic xanthophyll neurospora-xanthin. carRA and carB are found in a cluster with athird gene, carO, which codes for an opsin-like protein.However, no information is available on the sequence orchromosomal location of carT, which has been identifiedonly by mutant analysis. Transcription of the threeclustered genes is stimulated by light and by mutationsin a regulatory gene, leading to overproduction of car-otenoids. We have now identified a fourth gene in thecar cluster, called carX, which codes for a protein similarto known carotenoid-cleaving oxygenases. carX is tran-scribed divergently from carRA, and exhibits the sametranscriptional pattern as carRA, carB and carO. Tar-geted deletion of carX resulted in a phenotype charac-terized by a significant increase in the overall carotenoidcontent. In the dark, the carX mutants accumulate atleast five times more carotenoids than the wild type, andexhibit partial derepression of carRA and carB tran-scription. The mutants also show more intense pig-mentation in the light, but the increase in the carotenoidcontent relative to the wild type is less than twofold.Under these conditions, the mutants also show a relative

increase in the amounts of phytoene and cyclic carote-noids formed, suggesting that CarRA activity is en-hanced.

Keywords Gibberella Æ Apocarotenoid Æ Dioxygenase ÆGene cluster

Introduction

Carotenoids are terpenoid pigments which are widelydistributed in nature and are characterized by theirbright yellow, orange and red colors (Britton et al.1998). They are synthesized in large amounts by plants,where they play a role as accessory pigments in photo-synthesis, and by many microorganisms, including bac-teria and fungi. Animals lack the ability to synthesizecarotenoids, and have to obtain them from nutrientsources. They use carotenoids for a variety of functions,including pigmentation (Meyers 1994), cell regulation(Morriss-Kay and Ward 1999) and light detection (vonLintig and Vogt 2004).

Several fungal species have been widely used as modelsystems to investigate carotenoid biosynthesis and itsregulation (reviewed by Sandmann and Misawa 2002).Outstanding examples are the zygomycetes Phycomycesblakesleeanus and Mucor circinelloides for b-carotene,the basidiomycete Xhanthophyllomyces dendrorhous forastaxhanthin, and the ascomycetes Neurospora crassaand Fusarium fujikuroi for neurosporaxanthin (Avalosand Cerda-Olmedo 2004). Formation of these com-pounds begins with the biosynthesis of the colorlessprecursor phytoene from two geranylgeranyl pyrophos-phate molecules by phytoene synthase (Fig. 1a), thesequential desaturation of several intermediates by adehydrogenase, and the introduction of end-rings by acyclase (producing b-carotene in zygomycetes). In fungi,the three reactions depend on two genes, one of whichcodes for a bifunctional protein (phytoene synthase andcarotene cyclase), and these genes have been cloned and

Communicated by J. Avalos

A. Prado-Cabrero Æ M. M. Prado Æ J. Avalos (&)Departamento de Genetica, Universidad de Sevilla,Apartado 1095, 41080 Sevilla, SpainE-mail: [email protected].: +34-954-557110Fax: +34-954-557104

S. Thewes Æ B. TudzynskiInstitut fur Botanik, Westfalische Wilhelms-Universitat,Schlossgarten 3, 48149 Munster, Germany

Present address: S. ThewesRobert-Koch Institut, Nordufer 20, 13353 Berlin, Germany

Mol Gen Genomics (2005) 274: 217–228DOI 10.1007/s00438-005-0015-6

characterized from several fungal species (Avalos andCerda-Olmedo 2004). Additional hydroxylase and ke-tolase activities in X. dendrorhous are required to pro-duce astaxanthin, while a torulene-cleaving enzyme forthe production of neurosporaxanthin is found inF. fujikuroi (Fig. 1a) and N. crassa.

Fusarium fujikuroi (Gibberella fujikuroi, mating pop-ulation C) is noteworthy for its ability to produce largeamounts of gibberellins (Avalos et al. 1999; Tudzynski1999), terpenoid plant hormones whose growth-pro-moting properties have found agronomic applications.The fact that F. fujikuroi produces both carotenoids andgibberellins, via pathways which share the initial bio-synthetic steps, makes this fungus an attractive modelsystem for research on terpenoid metabolism. The sevengenes coding for the enzyme activities of gibberellinbiosynthesis are clustered in the F. fujikuroi genome(Tudzynski and Holter 1998; Rojas et al. 2001; Tu-dzynski et al. 2001, 2002, 2003). The regulation of thepathway by nitrogen availability has been the subject ofdetailed attention (reviewed by Tudzynski 1999 andAvalos et al. 1999), and the genes responsible for thisregulation are currently under intensive investigation(Tudzynski et al. 1999; Mihlan et al. 2003; Teichertet al. 2004). Similarly, the genes responsible for thesynthesis of torulene (carRA and carB) and a gene thatcodes for an opsin-like protein (carO) are clustered, andare subject to a common mechanism of transcriptionalregulation (Linnemanstons et al. 2002; Prado et al.2004).

Neurosporaxanthin, the final product of the carot-enoid biosynthetic pathway in F. fujikuroi, is an apoca-rotenoid with a carboxylic end-group (Aasen and Jensen1965). Apocarotenoids are molecules that have fewer

than the usual 40 carbon atoms, and are derived from aprecursor carotenoid by at least one cleavage reaction.Well known examples include retinal, abscisic acid, andbixin, which result from the cleavage of b-carotene(Wyss 2004; von Lintig and Vogt 2004), zeaxanthin(Taylor et al. 2000; Tan et al. 2003) and lycopene(Bouvier et al. 2003a), respectively. Retinal is the light-absorbing prosthetic group of rhodopsins, a family ofproteins that act as light sensors and ion-pumps(Spudich et al. 2000).

The gene for the enzyme responsible for the cleavageof torulene to produce neurosporaxanthin has not beenidentified. However, a mutant that accumulates toru-lene, named carT, has been described in F. fujikuroi(Avalos and Cerda-Olmedo 1987), suggesting that asingle gene is responsible for this step in the pathway.We have now identified a gene which is linked to thecarRA-carB-carO cluster, and codes for a protein thatshows similarity to the carotenoid oxygenases responsi-ble for cleavage reactions leading to the production ofvarious apocarotenoids, such as retinal in animals andabscisic acid in plants. Here we describe the sequenceand functional analysis of this gene, which we havenamed carX.

Materials and methods

Strains and culture conditions

IMI58289 and M567 are wild-type strains of Fusariumfujikuroi (Gibberella fujikuroi mating population C;O’Donnell et al. 1998) obtained from the Imperial

Fig. 1 a, b Pathway ofneurosporaxanthin biosynthesisin F. fujikuroi (a) and map ofthe car gene cluster (b). a Theletters indicate enzymaticactivities (S, phytoene synthase,D, dehydrogenase, C, cyclase).The identity of the enzymeresponsible for the last reaction,cleavage of torulene, (shown inthe dotted box) is not known.b Map of the linked genescarRA, carB, carO and carX.The black boxes indicate introns

218

Mycological Institute (Kew, U.K.) and from the FungalCulture Collection (Weimar, Germany), respectively.SG1, SG22 and A06 are neurosporaxanthin-overpro-ducing strains obtained from IMI58289 (SG1 and SG22,Avalos and Cerda-Olmedo 1987) and from M567 (A06,B. Tudzynski, unpublished). SG68 is a torulene-over-producing carT mutant obtained from SG22 (Avalosand Cerda-Olmedo 1987).

For DNA isolation and protoplasting, F. fujikuroistrains were incubated in 100 ml of complete medium(CM, Pontecorvo et al. 1953) at 28�C on a rotary shaker(200 rpm) for 3 days and 18 h, respectively. For RNAisolation, mycelia were incubated as described by Pradoet al. (2004). For comparative analysis of carX andcarRA expression, the strains were grown in 250 ml ofminimal medium in 500-ml Erlenmeyer flasks (Avaloset al. 1985) for 4 days at 30�C on an orbital shaker in thedark, and the mycelia were recovered by filtrationthrough filter paper. The mycelial pads were either keptin the dark or illuminated for different times, and thenfrozen in liquid nitrogen. For comparative analyses ofcar gene expression in the wild type and the carX mu-tants, two samples of each strain were grown as sub-merged cultures in Petri dishes for 3 days in the dark at30�C, in the presence of 100 ml of minimal medium. Oneof the cultures was exposed to white light (25 W/m2) at30�C for 1 h and the other was used as a dark control;the mycelia were then removed from the medium andfrozen in liquid nitrogen. Manipulations of the controlculture were performed under red light.

For carotenoid analysis, the strains were grown onminimal agar (Avalos et al. 1985) for 4 days at 22�C or30�C, either in the dark or under illumination [5 W/m2

(white light) at 22�C, or 25 W/m2 at 30�C]. HPLCanalyses were performed on 7-day-old cultures. Retinaland retinol acetate (Sigma, St Louis, Mo.) were dis-solved in ethanol and Tween 80, and added to themedium after autoclaving at final concentrations of 8 mlethanol/4 ml Tween80 per l. The plates were inoculatedwith samples from 3-day-old colonies obtained fromconidia grown on the same medium at 30�C in the dark.

DNA sequencing and protein analysis

The carX gene was identified as a coding sequence lo-cated adjacent to the carRA gene. A 3.3-kb HindIIIfragment and a 3.1-kb overlapping SalI fragment from ak genomic library, encompassing both carX and thecarRA gene, were cloned into pUC19 to yield the plas-mids pcarX-Hind and pcarX-Sal. The carX sequence,including 5¢ and 3¢ non-coding regions, was obtainedfrom a set of subclones made from both plasmids.

In order to localize introns, segments of carX DNAwere obtained by PCR from a cDNA sample obtainedfrom the overproducing mutant SG22 using the TimeSaver cDNA synthesis kit (Pharmacia, Uppsala, Swe-den). Two different DNA segments were obtained usingthe primer pairs CLU-5F (5¢-ACTCG CAACAGGGTA

TCGTG-3¢) and CLU-4R (5¢-TCGTTTATCTGTAGTGCGCG-3¢), and CLU-6F (5¢-CAGTGATGGGAAAGTAGTCG-3¢) and CLU-4R, cloned into the pGEM-TEasy vector (Promega, Madison, Wis.) and sequencedfrom both ends by Newbiotechnic (Seville, Spain) withan ABI Prism 3100 Genetic Analyzer (Applied Biosys-tems, Foster City, Calif.).

Sequence comparisons were done with the Clustal Xmultiple alignment analysis program (Thompson et al.1997). The phylogenetic tree was generated with Tree-View 1.5, Macintosh version (D. M. Page, University ofGlasgow; available at http://taxonomy.zool-ogy.gla.ac.uk/rod/rod.html) from a Clustal X alignment,including gaps and correcting for multiple substitutions.Blast analyses were done via the NCBI server (http://www.ncbi.nlm.nih.gov/blast/). BlastP analyses of fungalgenomes were done on the Broad Institute server (MIT,Cambridge, Mass.; http://www.broad.mit.edu/annota-tion/fungi/).

Construction of the gene disruption vectorand transformation of F. fujikuroi

A gene disruption vector, pcarX-GR, was made byintroducing two carX flanking DNA segments into twodifferent locations in the polylinker flanking the hphexpression cassette in the plasmid pGPC1 (Desjardinset al. 1992). The two segments were obtained by PCRamplification with two primer sets containing appro-priate restriction sites. PCRs (50 ll) contained 25 ng ofDNA, 10 ng of each primer, 0.2 mM dNTPs and 2 U ofTaq polymerase (Red Taq, Sigma-Aldrich, Deisenhofen,Germany). The reactions were carried out at 94�C for4 min, followed by 30 cycles of 94�C for 1 min, 62�C for1 min and 72�C for 1 min. The primer sequences, eachcontaining a restriction site (underlined), were: GR1(SacI) (5¢-TACCCTGAGCTCAATGAACAAGGC-3¢),GR2 (BamHI) (5¢-TCACTTTCAGGATCCCCTGTTGCG3¢), GR3 (XbaI) (5¢-AAGTGAGCTTCGTGTTCTAGACCGC-3¢), and GR4 (SphI) (5¢-AAGAAGTCGGAGCATGCGTAGTGC-3¢).

F. fujikuroi protoplasts were prepared as described byTudzynski et al. (1996). Some 107 protoplasts of strainIMI58289 were incubated with 10 g of the SphI-it SacIfragment of the replacement vector pcarX-GR. Trans-formed protoplasts were regenerated at 28�C on com-plete regeneration agar [0.7 M sucrose, 0.05% yeastextract, 0.1% (NH4)2SO4] containing 120 g/ml hygro-mycin B (Calbiochem) for 6–7 days. Single conidialcultures were established from hygromycin B-resistanttransformants, and used for DNA isolation and South-ern analysis.

Southern and Northern analyses

Lyophilized mycelium was ground to a fine powder usingmortar and pestle. DNA for Southern hybridization

219

experiments was prepared following the protocol ofDoyle and Doyle (1990). DNA for PCR analysis wasobtained by dispersing ground samples in extractionbuffer as described by Cenis (1993). Plasmid DNA wasextracted using the Genomed plasmid extraction kit(Genomed, Germany).

RNA samples for Northern analysis were isolatedwith the RNAgents total RNA isolation kit (Promega).Northern hybridizations were performed according tothe protocol for the DIG labeling system (Roche Diag-nostics, Mannheim, Germany) following Di Pietro andRoncero (1998). After transfer to the nylon membranes,RNA samples were stained with methylene blue, andrRNA bands were used as load controls. Densitometryanalysis was performed with Image Gauge software(version 3.3, Fuji Photo Film).

Probes for the car genes were obtained by PCR usingthe following primer pairs: 5¢-CACAGATATCTTACCGGCAAC-3¢ and 5¢-GGTAGGTTCGAGAGAATGACG-3¢ for carX (PCR product 0.4 kb), 5¢-ATCTATGAATCTATGACCTC3¢ and 5¢-TCCGGCGCATTTCCTATC-3¢ for carRA (0.6 kb), 5¢-ACTTCTCTTGCCACGTGAAG-3¢ and 5¢-AGGTGGATTCCACAAGGTTAG-3¢ for carB (0.7 kb), and 5¢-GGAAAATGTGGGATTGAAGC-3¢ and 5¢-AACCTACAGAATGTCGTCAG-3¢ for carO (1.4 kb).

Carotenoid analysis

Carotenoids were extracted with acetone from lyophi-lized, sand-ground, mycelial samples and dried undervacuum. Total amounts of colored carotenoids wereestimated from absorption maxima in hexane, assumingan average maximal E (1 mg/l, 1-cm path length) valueof 200. Apolar and polar carotenoids were separated byadding the samples to a 1 ml pipette tip filled with a 1-cm layer of Al2O3 (grade II–III) in light petroleum. Theapolar carotenoids were eluted by the addition of 1 mlof diethyl ether, and the polar carotenoids were retainedat the top of the column. Spectra for the polar carot-enoid fractions were obtained by computer subtractionof the spectrum of apolar colored carotenoid fractionsfrom the total spectrum of each sample. The amounts ofapolar colored carotenoids were estimated from themaximal absorption in hexane using an average maximalE (1 mg/l, 1 cm) value of 250. High-pressure liquidchromatography was carried out as described by Arrachet al. (2002).

Results

The gene carX

Previous studies had shown that the three genes carRA,carB and carO are clustered in the F. fujikuroi genome(Fig. 1b) and show similar patterns of transcriptionalregulation (Prado et al. 2004). Sequence analysis of

flanking sequences was carried out on both sides of thecluster to determine whether the neighboring genesmight also be related to carotenoid biosynthesis.Downstream from carO we found an ORF coding for aprotein similar to ammonium permeases. Upstream ofcarRA (Linnemanstons et al. 2002) we found a diver-gently transcribed coding sequence (Fig. 1b), and thehomologies of its predicted product to known proteinssuggested a possible connection with the carotenoidpathway. The sequence of the gene, which we namedcarX, was determined from genomic DNA and fromcDNA (GenBank/EMBL Accession No. AJ854252).The gene consists of a 2,135-bp ORF interrupted by asingle 44-bp intron close to the 5¢ end, and codes for aprotein of 696 amino acids.

Blast analysis of the predicted CarX polypeptide re-vealed significant similarity to a variety of proteins fromthe carotenoid oxygenase family (representative exam-ples are shown in Fig. 2), many of which are known tocatalyze oxidative cleavage reactions leading to apoca-rotenoids. A phylogenetic tree of a representative col-lection of CarX-related proteins shows that most ofthem can be assigned to five major clusters (Fig. 3). Twoof these comprise enzymes involved in plant carotenoidmetabolism: those which cleave 9,10 and 9¢,10¢ doublecarbon-carbon bonds of different carotenoids (Schwartzet al. 2001), and the 9-cis-epoxycarotenoid dioxygenasesthat cleave epoxy-xanthophylls into the abscisic acidprecursor xanthoxin (Taylor et al. 2000). Two furthergroups consist of animal enzymes involved in retinalmetabolism: the b,b-carotene-15,15¢ dioxygenases con-vert b-carotene into two molecules of retinal (Fig. 3), thefirst reaction in the production of vitamin A from die-tary carotenoids (von Lintig and Wyss 2001), and pro-teins of the RPE65 group, also essential for normalvision and abundantly expressed in retina pigment epi-thelium (Redmond et al. 1998). This latter group alsoincludes animal enzymes that can cleave carotenoidsasymmetrically (Kiefer et al. 2001), which are not in-cluded in the tree. The last group is made up of bacteriallignostilbene-a,b-dioxygenases, enzymes which are ableto cleave the biphenolic lignin model lignostilbene andother related molecules (Kamoda et al. 2003). Thiscompound, which is chemically unrelated to carote-noids, has a central double bond similar to the onescleaved by carotenoid dioxygenases, albeit not in thecontext of a conjugated system (Fig. 3).

The lowest E-value of the Blast analysis is found witha novel carotenoid cleavage dioxygenase from A. thali-ana (Schwartz et al. 2001, CAA06712 in Figs. 2 and 3),identified from a wilt-responsive cDNA (Neill et al.1998), but phylogenetic analysis locates CarX closer toanimal and bacterial genes (Fig. 3). Its closest relative inthe tree is a small protein encoded by sim14, one of the38 genes of the simocyclinone biosynthetic gene clusterof Streptomyces antibioticus (Galm et al. 2002). Simo-cyclinone, a structurally complex antibiotic containingpolyketide and coumarin components, is producedthrough a long series of enzymatic reactions, many of

220

which remain to be characterized. The function of Sim14has not been elucidated, but the 456-amino acid proteinshows 36.8% identity when aligned with the equivalentinternal segment of CarX. This level of identity is similarto that found upon alignment of CarX with the 538-amino acid sequence of the CAA06712 dioxygenaseenzyme from A. thaliana (197 identical positions,36.6%).

The only fungal protein in the databases that showssignificant similarity to CarX (26.6% identity over asegment of 484 residues) is N-acetylglucosamine-1-phosphate transferase, an enzyme involved in tunica-mycin resistance in Coprinus cinereus (Hartog andBishop 1987, O13438 in Fig. 3). The reaction catalyzed

by this enzyme, the synthesis of N-acetyl-D-glucosami-nyl-diphosphodolichol from UDP-N-acetyl-D-glucosa-mine and dolichyl phosphate, has no obviousrelationship to the oxidative cleavage of carotenoids orbiphenolic compounds.

The gene carX is linked in the same way to carRA,carB and carO homologues in the genome of F. grami-nearum (Gibberella zeae), a close relative of F. fujikuroi.The F. graminearum proteome contains a second puta-tive carotenoid oxygenase protein, FG02625.1, whosesequence is more similar to animal enzymes that cleaveb-carotene to produce retinal than to CarX: clustalalignments reveal 148 identical positions betweenFG02625.1 and the human b,b-carotene 15,15¢-dioxy-genase (Accession No. Q9HAY6) compared to 124 be-tween CarX and the human protein. Interestingly, thetwo putative carotenoid oxygenases from Fusarium onlycoincide at 128 positions, suggesting that they play dif-ferent biological roles.

carX transcription

Carotenoid biosynthesis in F. fujikuroi is induced bylight (Avalos and Schrott 1990) and it is deregulated inoverproducing mutants (Avalos and Cerda-Olmedo

Fig. 2 Clustal X comparison of CarX with six proteins of thecarotenoid oxygenase family. Amino acids present in the sameposition in at least three of the proteins are shaded in grey; theblack boxes indicate positions that are conserved in at least six ofthe proteins. Hs: Homo sapiens (Hs1: retinal pigment epitheliumprotein, SPTREMBL Accession No. Q16518; Hs2: b,b–carotene15,15¢-dioxygenase, Q9HAY6). At: Arabidopsis thaliana (At1: 9-cis-epoxycarotenoid dioxygenase, Q9LRR7; At2: carotenoid 9,10–9¢,10¢ cleavage dioxygenase, CAA06712). Ss: Synechocystis sp.(lignostilbene-a,b dioxygenase, P74334). Cc: Coprinus cinereus(UDP-N-acetylglucosamine-dolichyl-phosphate-N-acetylglucos-aminephosphotransferase, O13438). Ff: Fusarium fujikuroi (CarXprotein)

221

1987). The sequence of carX and its linkage to the carcluster suggest a possible connection with the carotenoidpathway. To obtain more information on its biologicalfunction, we investigated the effect of light and carot-enoid deregulation on carX transcription.

We could not detect any carX transcripts in dark-grown cultures of the wild type. Exposure to light re-sulted in a rapid induction of carX transcription, withthe mRNA level reaching a maximum after 1 h ofexposure to light (Fig. 4, left panels). Longer exposuresresult in lower carX mRNA amounts, indicating a light-adaptation response similar to those observed in N.crassa (Schwerdtfeger and Linden 2001) and M. circi-nelloides (Velayos et al. 2000a, b).

The carS mutants of F. fujikuroi, such as SG22(Avalos and Cerda-Olmedo 1987), produce largeamounts of carotenoids under all culture conditions

tested. Expression of carX is deregulated in these mu-tants, and high transcript levels were obtained irrespec-tive of illumination or genetic background (Fig. 4,central panels). A similar deregulation was found in atorulene-accumulating mutant obtained from SG22(Fig. 4, right panels), defective in the oxidative cleavageof torulene to produce neurosporaxanthin (Avalos andCerda-Olmedo 1987).

The regulatory pattern of carX—light induction fol-lowed by light adaptation and dark expression incarotenoid overproducing mutants—is the same as thatexhibited by the linked genes carRA, carB and carO(Prado et al. 2004; carRA is shown in Fig. 4). Thus, theregulation of carX resembles that expected for a struc-tural gene of the carotenoid pathway, which is com-patible with the hypothesis that this gene might beresponsible for the oxidative cleavage of torulene.

Targeted deletion of carX

To construct the carX replacement vector pcarX-GR, 5¢and 3¢ flanking regions of the gene were amplified byPCR and cloned into the hygromycin resistance vectorpGPC1 (Fig. 5a; see Materials and methods). Trans-formation of the wild-type strain IMI58289 was per-formed using the SphI-SacI fragment of pcarX-GR. Insix out of 48 hygromycin-resistant transformants, thecoding region of the carX gene was replaced by the hy-gromycin resistance cassette as the result of a doublecross over event (Fig. 5b). Deletion of the carX gene wasconfirmed by Southern analysis (Fig. 5c), which revealeda hybridizing band with the size expected from the gene

Fig. 3 Phylogram of representative protein sequences of thecarotenoid oxygenase family. Each protein is referred to by itsSPTREMBL Accession No. and a species abbreviation. Ag,Ambystoma tigrinum; As, Anabaena sp.; At, Arabidopsis thaliana;Br, Brachydanio rerio; Bs, Bos taurus; Ca, Capsicum annuum; Cc,Coprinus cinereus; Cp, Cynops pyrrhogaster; Cs, Crocus sativus; Gg:Gallus gallus; Hs, Homo sapiens; Le, Lycopersicon esculentum; Mm,Mus musculus; Pa, Persea americana; Pp, Pseudomonas paucimobi-lis; Ps, Pisum sativum; Pv, Phaseolus vulgaris; Sa, Streptomycesantibioticus; Se, Synechococcus elongatus; Ss, Synechocystis sp.;Zm, Zea mays. The asterisks indicate the sequences displayed inFig. 2. The broken bar (labeled ’x3’) is three times longer thanshown. The scale bar represents 0.1 substitutions per site. Reactionscatalyzed by a representative plant carotenoid dioxygenase, b,b-carotene-9¢,10¢ dioxygenase, and lignostilbene-a,b-dioxygenase areindicated on the right. The double bond cleaved by each of theenzymes is shaded

222

replacement scheme depicted in Fig. 5b. Three trans-formants, 2.1, 5.2, and 17.1, were chosen for detailedphenotypic analysis.

The three transformants are very similar to the wildtype in growth and colony morphology, but they exhibitdifferences in pigmentation. Thus, agar cultures of thetransformants grown at two different temperatures inthe dark show a pale orange color that is not apparent inthe wild type. The pigmentation is due to a marked in-crease (�10-fold in transformant 2.1, fivefold in trans-formants 5.2 and 17.1) in the amounts of carotenoidsproduced by the mutants (Fig. 6a). As expected (Avalosand Cerda-Olmedo 1987), the carotenoid content of thewild type was higher at 30�C than at 22�C; this alsoholds for the transformants.

The transformants also accumulated more carote-noids than the wild type under illumination, irrespectiveof the temperature and light intensity used (Fig. 6a). Thedifference in this case was proportionately lower, but thecarotenoid levels were higher than in the dark. Again,the increase in the carotenoid content was more pro-nounced in the transformant 2.1.

Because of the higher amounts available, more de-tailed chemical analyses of the carotenoids were donewith illuminated cultures. Like the wild type, the threetransformants contained a mixture of polar and apolarcarotenoids (Fig. 6a). The spectra of the polar fractionsfrom all four strains are similar in shape and absorptionmaxima (spectra from samples at 30�C shown inFig. 6b), and coincide with that of neurosporaxanthin.The increased carotenoid content in the transformants ismainly due to greater accumulation of apolar forms,corresponding to the intermediates of the pathway.HPLC analysis of the apolar carotenoids accumulatedby these strains in the light shows an increase of at leasta fourfold in the phytoene and b-carotene contents(Fig. 6c). In addition, the c-carotene peak is moreprominent in the chromatograms from the transfor-mants than in the one from the wild type, indicating ahigher c-carotene/torulene ratio in the mutants. Thehigher proportions of phytoene, c-carotene and b-caro-tene suggest enhanced phytoene synthase and carotenecyclase activities in these strains, both activities areassociated with the bifunctional enzyme encoded by thegene carRA.

Fig. 4 Expression of the genes carX and carRA. Left panelsNorthern blots of total RNA isolated from the wild typeIMI58289 grown in the dark, or following incubation in the lightfor the indicated times. Central panelsNorthern blots of total RNAfrom the wild types IMI58289 (wt1) and M567 (wt2) and threeneurosporaxanthin-overproducing mutants grown in the dark. SG1and SG22 were both obtained from IMI58289 and A06 fromM567. Right panels Northern blots of total RNA from the wild-type IMI58289 (wt1), the neurosporaxanthin-overproducingmutant SG22, and the torulene-overproducing mutant SG68,obtained from SG22. The probes are indicated on the left. rRNAbands are shown below each panel as load controls. For carRA inthe left and central panels, see Prado et al. (2004). The bars belowthe panels show the ratios of the signal intensities of the indicatedgenes to the rRNA controls. In each panel, the values are expressedrelative to the maximum (indicated by the shaded bar in each case).Signal intensities were estimated by densitometric analysis

223

The expression of the genes carRA, carB and carO inthe wild-type and in two carX mutants was investigatedby Northern analysis (Fig. 7). The three strains containvery low amounts of mRNA for these genes in the dark,but transcripts accumulate to high levels after illumi-nation for 1 h. Interestingly, the genes carRA and carO(but not carB) were more strongly induced in the carXmutants than in the wild type (Fig. 7, left panels), pro-viding an explanation for the increased levels of phyto-ene and cyclized products in the mutants (Fig. 6c)

In agreement with the induced carotenoid productionin the dark, long exposure of the autoradiographs al-lowed us to detect low levels of carRA and carB mRNAsin mycelia from the transformants grown in the dark butnot in wild-type mycelium (Fig. 7, right panels), indi-cating some partial deregulation of these genes in thecarX mutants. In contrast, no such induction of the genecarO was seen under these conditions.

The occurrence in Fusarium of a second predictedprotein similar to retinal-producing enzymes makes sucha role for CarX rather unlikely, but does not exclude it.

To check whether retinal might act as a regulatory signalon carX or its product, the wild type and the mutantswere grown in the presence of 0.75 mM retinol or retinalin the dark and in the light at different temperatures.The presence of these retinoids in the medium did notchange the pigmentation of the wild type or the mutantsunder any of the conditions tested.

Discussion

In this work, we have described a novel fungal gene,carX, coding for a predicted polypeptide with sequencesimilarity to a family of proteins that include oxygenaseenzymes involved in the oxidative cleavage of carote-noids (Fig. 2). The gene was identified in F. fujikuroi onthe basis of its linkage to the genes carRA and carB,which code for the enzymes responsible for the steps incarotenoid biosynthesis that lead to the formation oftorulene (Fig. 1). It therefore seemed possible that carXis allelic to carT, a gene required for the cleavage oftorulene to yield the end-product neurosporaxanthin(Fig. 1); to date carT has only been identified bymutation (Avalos and Cerda-Olmedo 1987). The simi-larity between CarX and carotenoid oxygenases iscompatible with the hypothesis that carX is the generesponsible for cleavage of torulene. Moreover, like thestructural genes carRA and carB (Linnemanstons et al.2002; Prado et al. 2004), carX is induced by light and isderepressed in the dark in carotenoid-overproducingmutants. We therefore constructed targeted carX nullmutants by replacing carX with a selectable marker andexamined their phenotype. We found that these mu-tants retain the ability to synthesize neurosporaxanthin,

Fig. 5 a–c Generation and identification of carX deletion mutants.a Construction of the carX replacement vector pcarX-GRB. TwocarX-flanking PCR products containing artificially added restric-tion sites (see Materials and methods) were cloned into theindicated locations in plasmid pGPC1. Restriction sites: Sp, SphI,X, XbaI, B, BamHI, S, SalI. b Molecular events leading to thereplacement of carX in the F. fujikuroi genome by the hphexpression cassette from plasmid pcarX-GRB. BamHI restrictionsites relevant for interpretation of the southern blot are indicated. cSouthern analysis of genomic DNAs from the wild type (wt) andfive independent transformants that have undergone the expecteddouble recombination event. DNA samples were digested withBamHI, and hybridized with a HindIII-XbaI probe (indicated in b).Relevant fragment sizes are shown in kb

224

indicating that CarX is not necessary for the cleavageof torulene.

The chemical reactions mediated by known enzymesin the oxygenase family to which CarX belongs have incommon the ability to oxidatively cleave a double

carbon bond (shaded in Fig3), yielding two aldehydeproducts. This is also the case for the reactions catalyzedby novel enzymes from this group, such as a lycopenecleaving dioxygenase from Bixa orellana, which pro-duces bixin aldehyde in the bixin biosynthetic pathway(Bouvier et al. 2003a), or a zeaxanthin-cleaving dioxy-genase from Crocus sativus, which converts zeaxanthininto the saffron components crocetin dialdehyde andhydroxy-b-cylocitral (Bouvier et al. 2003b). The toru-lene-cleaving enzyme acts on a similar double carbonbond (Fig. 1), but the larger product of the reaction hasa carboxy end-group, while the smaller moleculeresulting from this reaction is unknown. Although thecarboxy group may be the result of an additional en-zyme activity, its occurrence suggests a different mode ofaction for the enzyme CarT, which could therefore be-long to a different protein family.

The similarity of CarX to carotenoid oxygenases re-mains a suggestive clue to its function, but the functionaldiversity of the enzymes in this family makes it difficultto discern its role. This diversity is expected to expand inthe future with the discovery of novel biological func-tions in plants. Thus, a Blast search for 9-cis-epoxyca-rotenoid dioxygenase-like coding sequences in the A.thaliana genome detects at least nine similar putativegenes (SPTREMBL Accession Nos. 065572, Q9M9F5,O49675, Q9C6Z1, Q9LRR7, O49505, Q9LRM7,Q8VY26). One of them has recently been found to codefor an enzyme involved in the generation of a newbranching-inhibiting carotenoid-derived hormone(Booker et al. 2004); no specific functions have yet beenattributed to the others.

In contrast to the case in plants, such enzymes areunderrepresented in fungi. BlastP searches find twogenes from this family in the genome of F. graminearum,the only Fusarium genome sequence yet available indatabases. One of them is the carX homologue inF. graminearum. The second one is more similar to thegenes that code for oxygenases that cleave b-carotene toproduce retinal. Such an enzymatic function is likely toexist in this fungus, assuming that the opsin-like CarOprotein contains the conserved residues required to bindretinal as a prosthetic group (Prado et al. 2004). Twosimilar carotenoid oxygenase genes are also encoded inthe genome of N. crassa, a fungus with a similar carot-enoid pathway (Arrach et al. 2002) in which retinalbinding has been demonstrated for an opsin protein(Bieszke et al. 1999). The functions of these two genes inN. crassa are currently under investigation.

The carX mutants accumulate more carotenoids thanthe wild type at the different conditions tested, suggest-ing a regulatory role for the CarX protein. Carotenebiosynthesis in F. fujikuroi is induced by light and re-pressed by the carS gene product (see Fig. 4, Prado et al.2004). In addition, the pathway exhibits a light-specificfeedback regulatory mechanism, as deduced from thefivefold increase in the carotene content of carB nullmutants when these are grown in the light (Fernandez-Martın et al. 2000, and our unpublished results). The

Fig. 6 a–c Carotenoid analysis of the wild type IMI58289 (wt) andthree carX deletion mutants. a Carotenoid content of strains grownon minimal medium either in the dark (black bars) or in the light(shaded bars). The data for illuminated cultures represent the sumof apolar (lower section of the bar) and polar carotenoids (uppersection of the bar). Each bar represents the average of twodeterminations. b Absorption spectra (in hexane) of the polarcarotenoid fractions obtained from the wild type and the threemutants in a representative experiment at 30�C in the light.Maximal absorption for the four samples occurred at about477 nm. c HPLC elution profiles at 462.5 nm (286.5 nm whererelevant) of the apolar carotenoids from the wild type IMI58289(wt) and the same transformants grown at 30�C in the light

225

expected occurrence of a carotenoid recognition sitemakes CarX a candidate to play a regulatory role as acarotenoid-sensing protein. However, the phenotype ofthe mutants makes a role for CarX in the feedbackregulatory mechanism unlikely. First, the increase incarotenoid content exhibited by the carX mutants is lesspronounced than that produced by the carB null muta-tion. Second, the carX mutants accumulate up to tentimes more carotenoids than the wild type when grownin the dark, while carB mutations have no effect underthese conditions.

The increase in the carotenoid content of the carXmutants in the light is accompanied by a greater

accumulation of phytoene, c-carotene and b-carotene,which is consistent with enhanced phytoene synthaseand carotene cyclase activities in relation to the dehy-drogenase. Accordingly, the amount of carRA mRNAsynthesized in response to illumination is notably higherin the two mutants analyzed than in the wild type. ThecarB transcript is unaffected, so this suggests a specificregulatory effect of the carX mutation on carRA tran-scription. In contrast, the increased carotenoid contentof the mutants grown in the dark correlates with partialtranscriptional derepression of both carRA and carB.Moreover, carO mRNA levels are higher in the carXmutants in the light, but not in the dark, indicatingdifferences in the molecular mechanisms that regulatethe transcription of the three car genes.

Our results do not rule out other biological functionsconnected with carotenoid metabolism for CarX. A rolefor carotenoids in protection against oxidative damagehas been proposed in other fungi, such as N. crassa(Michan et al. 2003; Yoshida and Hasunuma 2004) andPhaffia rhodozyma (Schroeder and Johnson 1995). CarXcould play a role in recycling oxidative carotenoidproducts in F. fujikuroi. An alternative hypothesis issuggested by the similarity of CarX to the animal pro-teins represented by RPE65 (Redmond et al. 1998). Forappropriate function, retinol (vitamin A) must beisomerized to 11-cis-retinal, the chromophore in rho-dopsin. Light induces the conversion of rhodopsin-bound 11-cis-retinal to the free all-trans isomer. RPE65produces 11-cis-retinal from trans isomers, closing thevisual cycle (Xue et al. 2004). A recycling role for CarXin association with a similar isomerization reaction inFusarium cannot be excluded. Improvements in ourknowledge of the chemistry of the carotenoids in Fusa-rium, and their oxidation derivatives, together with thepowerful tool offered by the carX deletion mutants,should help us to gain a better understanding of thebiological function of this novel fungal protein.

Acknowledgements We thank J. Schulte, S. Richter, C. Vallejo andL.Perez de Camino for technical assistance. This work was sup-ported by the German Government (DFG, grant number Tu101-7), the Spanish Government (INIA, project PB96-1336 and Min-isterio de Ciencia y Tecnologıa, project BIO2003-01548) and theEuropean Union (project QLK1-CT-2001-00780). This work hasbeen carried out in compliance with the current Spanish andGerman laws concerning genetic research

References

Aasen AJ, Jensen SL (1965) Fungal carotenoids II. The structure ofthe carotenoid acid neurosporaxanthin. Acta Chem Scand19:1843–1853

Arrach N, Schmidhauser TJ, Avalos J (2002) Mutants of the car-otene cyclase domain of al-2 from Neurospora crassa. MolGenet Genom 266:914–921

Avalos J, Cerda-Olmedo E (1987) Carotenoid mutants of Gibber-ella fujikuroi. Curr Genet 11:505–511

Avalos J, Cerda-Olmedo E (2004) Fungal carotenoid production.In: Arora DK (ed) Handbook of fungal biotechnology, 2ndedn. Marcel Dekker, New York, pp 367–378

Fig. 7 Expression of carRA, carB and carO in carX deletionstrains. Left panelsNorthern blots of total RNA from the wild typeIMI58289 (wt) and two carX mutants grown in the dark, or after a1-h incubation in the light. A short exposure of the film is shownfor better visualization of differences in mRNA levels. Right panelsNorthern blots of total RNA from the same strains grown in thedark. mRNA levels measured in the wild type after incubation for1 h in the light are shown for comparison with the short exposureshown on the left. rRNA bands are shown below each panel asloading controls. The bars below the panels show the ratios of thesignal intensities of the indicated genes to the rRNA controls. Ineach panel, the values are expressed relative to the maximum(indicated by the shaded bar). Signal intensities were estimated bydensitometric analysis

226

Avalos J, Schrott EL (1990) Photoinduction of carotenoid bio-synthesis in Gibberella fujikuroi. FEMS Microbiol Lett 66:295–298

Avalos J, Casadesus J, Cerda-Olmedo E (1985) Gibberella fujikuroimutants obtained with UV radiation and N-methyl-N¢-nitro-N-nitrosoguanidine. Appl Environ Microbiol 49:197–191

Avalos J, Fernandez-Martın R, Prado MM, Cerda-Olmedo E(1999) Gibberellin biosynthesis in Gibberella. J Bot Gall 146:55–65

Bieszke JA, Spudich EN, Scott KL, Borkovich KA, Spudich JL(1999) A eukaryotic protein, NOP-1, binds retinal to form anarchaeal rhodopsin-like photochemically reactive pigment.Biochemistry 38:14138–14145

Booker J, Auldridge M, Wills S, McCarty D, Klee H, Leyser O(2004) MAX3/CCD7 is a carotenoid cleavage dioxygenase re-quired for the synthesis of a novel plant signaling molecule.Curr Biol 14:1232–1238

Bouvier F, Dogbo O, Camara B (2003a) Biosynthesis of the foodand cosmetic plant pigment bixin (annatto). Science 300:2089–2091

Bouvier F, Suire C, Mutterer J, Camara B (2003b) Oxidativeremodeling of chromoplast carotenoids: identification of thecarotenoid dioxygenase CsCCD and CsZCD genes involved inCrocus secondary metabolite biogenesis. Plant Cell 15:47–62

Britton G, Liaaen-Jensen S, Pfander H (1998) Carotenoids, vols 1–3. Birkhauser Verlag, Basel

Cenis JL (1993) Rapid extraction of fungal DNA for PCR ampli-fication. Nucleic Acids Res 20:2380

Desjardins AE, Gardner HW, Weltring K-M (1992) Detoxificationof sesquiterpene phytoalexins by Gibberella pulicaris (Fusariumsambucinum) and its importance for virulence on potato tubers.J Ind Microbiol 9:201–211

Di Pietro A, Roncero MI (1998) Cloning, expression, and role inpathogenicity of pg1 encoding the major extracellular endo-polygalacturonase of the vascular wilt pathogen Fusariumoxysporum. Mol Plant Microbe Interact 11:91–98

Doyle JJ, Doyle JL (1990) Isolation of plant DNA from fresh tis-sue. Focus 12:13–15

Fernandez-Martın R, Cerda-Olmedo E, Avalos J (2000) Homolo-gous recombination and allele replacement in transformants ofFusarium fujikuroi. Mol Gen Genet 263:838–845

Galm U, Schimana J, Fiedler H-P, Schmidt J, Li S-M, Heide L(2002) Cloning and analysis of the simocyclinone biosyntheticgene cluster of Streptomyces antibioticus Tu 6040. ArchMicrobiol 178:102–114

Hartog KO, Bishop B (1987) Genomic sequence coding for tu-nicamycin resistance in yeast. Nucleic Acids Res 15:3627–3627

Kamoda S, Terada T, Saburi Y (2003) A common structure ofsubstrate shared by lignostilbenedioxygenase isozymes fromSphingomonas paucimobilis TMY1009. Biosc Biotechnol Bio-chem 67:1394–1396

Kiefer C, Hessel S, Lampert JM, Vogt K, Lederer MO, BreithauptDE, von Lintig J (2001) Identification and characterization of amammalian enzyme catalyzing the asymmetric oxidativecleavage of provitamin A. J Biol Chem 276:14110–14116

Linnemannstons P, Prado MM, Fernandez-Martın R, TudzynskiB, Avalos J (2002) A carotenoid biosynthesis gene cluster inFusarium fujikuroi: the genes carB and carRA. Mol Genet Ge-nom 267:593–602

Meyers SP (1994) Developments in world aquaculture, formula-tions, and the role of carotenoids. Pure Appl Chem 66:1069–1076

Michan S, Lledias F, Hansberg W (2003). Asexual development isincreased in Neurospora crassa cat-3-null mutant strains. Eu-karyot Cell 2:798–808

Mihlan M, Homann V, Liu TW, Tudzynski B (2003) AREA di-rectly mediates nitrogen regulation of gibberellin biosynthesis inGibberella fujikuroi, but its activity is not affected by NMR.Mol Microbiol 47:975–991

Morriss-Kay GM, Ward SJ (1999) Retinoids and mammaliandevelopment. Int Rev Cytol 188:73–131

Neill SJ, Burnett EC, Desikan R, Hancock JT (1998) Cloning of awilt-responsive cDNA from an Arabidopsis thaliana suspensionculture cDNA library that encodes a putative 9-cis-epoxy-carotenoid dioxygenase. J Exp Bot 49:1893–1894

O’Donnell K, Cigelnik E, Nirenberg HI (1998) Molecular system-atics and phylogeography of the Gibberella fujikuroi speciescomplex. Mycologia 90:465–493

Pontecorvo G, Roper JA, Hemmonns LM, MacDonald KD,Bufton AW (1953) The genetics of Aspergillus nidulans. AdvGenet 141:141–238

Prado MM, Prado-Cabrero A, Rafael Fernandez-Martın R, Ava-los J (2004) A gene of the opsin family in the carotenoid genecluster of Fusarium fujikuroi. Curr Genet 46:47–58

Redmond TM, Yu S, Lee E, Bok D, Hamasaki D, Chen N, GoletzP, Ma JX, Crouch RK, Pfeifer K (1998) Rpe65 is necessary forproduction of 11-it cis-vitamin A in the retinal visual cycle. NatGenet 20:344–351

Rojas MC, Hedden P, Gaskin P, Tudzynski B (2001) The P450-1gene of Gibberella fujikuroi encodes a multifunctional enzyme ingibberellin biosynthesis. Proc Natl Acad Sci USA 98:5838–5843

Sandmann G, Misawa N (2002) Fungal carotenoids. In: OsiewaczHD (ed) The Mycota X. Industrial applications. Springer,Berlin Heidelberg New York

Schroeder WA, Johnson EA (1995) Singlet oxygen and peroxylradicals regulate carotenoid biosynthesis in Phaffia rhodozyma.J Biol Chem 31:18374–18379

Schwartz SH, Qin X, Zeevaart JAD (2001) Characterization of anovel carotenoid cleavage dioxygenase from plants. J BiolChem 276:25208–25211

Schwerdtfeger C, Linden H (2001) Blue light adaptation anddesensitization of light signal transduction in Neurospora cras-sa. Mol Microbiol 39:1080–1087

Spudich JL, Yang CS, Jung KH, Spudich EN (2000) Retinylideneproteins: structures and functions from Archaea to humans.Annu Rev Cell Dev Biol 16:365–392

Tan BC, Joseph LM, Deng WT, Liu L, Li QB, Cline K, McCartyDR (2003) Molecular characterization of the Arabidopsis 9-cisepoxycarotenoid dioxygenase gene family. Plant J 35:44–56

Taylor IB, Burbidge A, Thompson AJ (2000) Control of abscisicacid biosynthesis. J Exp Bot 51:1563–1574

Teichert S, Schonig B, Richter S, Tudzynski B (2004) Deletion ofthe Gibberella fujikuroi glutamine synthetase gene has signifi-cant impact on transcriptional control of primary and second-ary metabolism. Mol Microbiol 53:1661–1675

Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, HigginsDG (1997) The CLUSTAL_X windows interface: flexiblestrategies for multiple sequence alignment aided by qualityanalysis tools. Nucleic Acids Res 25:4876–4882

Tudzynski B (1999) Biosynthesis of gibberellins in Gibberella fu-jikuroi: biomolecular aspects. Appl Microbiol Biotechnol52:298–310

Tudzynski B, Holter K (1998) Gibberellin biosynthetic pathway inGibberella fujikuroi: Evidence for a gene cluster. Fungal GenetBiol 25:157–170

Tudzynski B, Mende K, Weltring KM, Kinghorn JR, Unkles SE(1996) The Gibberella fujikuroi niaD gene encoding nitratereductase: isolation, sequence, homologous transformation andelectrophoretic karyotype location. Microbiology 142:533–539

Tudzynski B, Homann V, Feng B, Marzluf GA (1999) Isolation,characterization and disruption of the areA nitrogen regulatorygene of Gibberella fujikuroi. Mol Gen Genet 261:106–114

Tudzynski B,HeddenP,CarreraE,Gaskin P (2001) TheP450-4 geneof Gibberella fujikuroi encodes ent-kaurene oxidase in the gibber-ellin biosynthesis pathway.ApplEnvironMicrobiol 67:3514–3522

Tudzynski B, Rojas MC, Gaskin P, Hedden P (2002) The Gibber-ella fujikuroi gibberellin 20-oxidase is a multifunctional mono-oxygenase. J Biol Chem 277:21264–21253

Tudzynski B, Mihlan M, Rojas MC, Linnemannstons P, Gaskin P,Hedden P (2003) Characterization of the final two genes of thegibberellin biosynthesis gene cluster of Gibberella fujikuroi.J Biol Chem 278:28635–28643

227

Velayos A, Blasco JL, Alvarez MI, Iturriaga EA, Eslava AP(2000a) Blue-light regulation of phytoene dehydrogenase(carB) gene expression in Mucor circinelloides. Planta 210:938–946

Velayos A, Eslava AP, Iturriaga EA (2000b) A bifunctional enzymewith lycopene cyclase and phytoene synthase activities is en-coded by the carRP gene ofMucor circinelloides. Eur J Biochem267:5509–5519

Von Lintig J, Vogt K (2004) Vitamin A formation in animals:molecular identification and functional characterization ofcarotene cleaving enzymes. J Nutr 134:251S–256S

Von Lintig J, Wyss A (2001) Molecular analysis of vitamin Aformation: cloning and characterization of b-carotene 15,15¢-dioxygenases. Arch Biochem Biophys 385:47–52

Wyss A (2004) Carotene oxygenases: a new family of double bondcleavage enzymes. J Nutr 134:246S–250S

Xue L, Gollapalli DR, Maiti P, Jahng WJ, Rando RR (2004)Palmitoylation switch mechanism in the regulation of the visualcycle. Cell 117:761–771

Yoshida Y, Hasunuma K (2004) Reactive oxygen species affectphotomorphogenesis in Neurospora crassa. J Biol Chem279:6986–6993

228