Vhr1p, a New Transcription Factor from Budding Yeast, Regulates Biotin-dependent Expression of VHT1 and BIO5*

Transcription of the Saccharomyces cerevisiae vitamin H transporter gene VHT1 is enhanced by low extracellular biotin. Here we present the identification and characterization of Vhr1p as a transcriptional regulator of VHT1 (VHR1 (YIL056w); VHT1 regulator 1) and the identification of the cis-regulatory target sequences for Vhr1p in two yeast promoters. VHR1 was identified in a complementation screening of mutagenized yeast cells that had lost the capacity to express the gene of the green fluorescent protein (GFP) from the VHT1 promoter. Δvhr1 deletion mutants fail to induce VHT1 on low biotin concentrations. In yeast one-hybrid analyses performed with fusions of Vhr1p N-terminal and C-terminal fragments to the Gal4p activation domain or to the Gal4p DNA-binding domain, the Vhr1p N terminus mediated biotin-dependent DNA binding, and the Vhr1p C terminus triggered biotin-dependent transcriptional activation. The analyzed Vhr1p N-terminal fragment has previously been described as a domain of unknown function (DUF352). Deletion and linker scanning analyses of the VHT1 promoter revealed the palindromic 18-nucleotide sequence AATCA-N8-TGAYT as the vitamin H-responsive element. This sequence was identified also in the BIO5 promoter that is also transcriptionally activated on low biotin concentrations. Bio5p mediates the transport of 7-keto-8-aminopelargonic acid across the yeast plasma membrane, a compound that is used as a precursor in biotin biosynthesis. Δvhr1 deletion mutants fail to induce BIO5 on low biotin concentrations. The presented data characterize Vhr1p as an essential component of the biotin-dependent signal transduction cascade in yeast.

Transcription of the Saccharomyces cerevisiae vitamin H transporter gene VHT1 is enhanced by low extracellular biotin. Here we present the identification and characterization of Vhr1p as a transcriptional regulator of VHT1 (VHR1 (YIL056w); VHT1 regulator 1) and the identification of the cis-regulatory target sequences for Vhr1p in two yeast promoters. VHR1 was identified in a complementation screening of mutagenized yeast cells that had lost the capacity to express the gene of the green fluorescent protein (GFP) from the VHT1 promoter. ⌬vhr1 deletion mutants fail to induce VHT1 on low biotin concentrations. In yeast one-hybrid analyses performed with fusions of Vhr1p N-terminal and C-terminal fragments to the Gal4p activation domain or to the Gal4p DNA-binding domain, the Vhr1p N terminus mediated biotin-dependent DNA binding, and the Vhr1p C terminus triggered biotin-dependent transcriptional activation. The analyzed Vhr1p N-terminal fragment has previously been described as a domain of unknown function (DUF352). Deletion and linker scanning analyses of the VHT1 promoter revealed the palindromic 18-nucleotide sequence AATCA-N 8 -TGAYT as the vitamin H-responsive element. This sequence was identified also in the BIO5 promoter that is also transcriptionally activated on low biotin concentrations. Bio5p mediates the transport of 7-keto-8-aminopelargonic acid across the yeast plasma membrane, a compound that is used as a precursor in biotin biosynthesis. ⌬vhr1 deletion mutants fail to induce BIO5 on low biotin concentrations. The presented data characterize Vhr1p as an essential component of the biotin-dependent signal transduction cascade in yeast.
The water-soluble vitamin biotin (vitamin H) is a ubiquitous carrier of CO 2 in carboxylation, decarboxylation, and transcarboxylation reactions in all forms of life (1)(2)(3)(4). Many organisms, including most bacteria, fungi, and all higher plants, are able to synthesize biotin de novo. Many other organisms, however, including mammals and bakers' yeast (Saccharomyces cerevisiae) have lost this capability and depend on the supply of biotin from exogenous sources (4,5). In humans, biotin deficiency leads to a variety of clinical abnormalities, including neurological disorders, growth retardation, and dermal abnormalities (4,6).
Bakers' yeast cannot grow when biotin is absent from the medium, and due to this strict biotin requirement, yeast has been used in growth assays for the microbiological determination of biotin. The retarded growth of yeast cells on low biotin results from the lack of biotin for its covalent attachment to biotin-dependent enzymes. This step is cata-lyzed by the biotin-protein ligase Bpl1p (7), which transfers biotin to enzymes, such as acetyl-CoA carboxylase (Acc1p) or the two isoforms of pyruvate carboxylase (Pyc1p and Pyc2p). Acc1p catalyzes the first step of fatty acid biosynthesis (i.e. the carboxylation of acetyl-CoA to malonyl-CoA), and Pyc1p and Pyc2p generate oxaloacetate from pyruvate.
Rogers and Lichstein (5,8) showed for the first time that biotin uptake in yeast is a carrier-mediated and energy-requiring mechanism. They also discovered that biotin uptake is controlled by the biotin levels in the growth medium with high levels of biotin reducing the rate of biotin uptake. In 1999, Stolz et al. (9) cloned the gene of the yeast plasma membrane vitamin H transporter Vht1p by complementation of a yeast mutant that had been isolated in a screen for fatty acid auxotrophy. The Vht1p transporter belongs to the major facilitator superfamily (10) and is part of a subfamily of eight S. cerevisiae genes encoding proteins sharing 20 -30% identity on the amino acid level. This subfamily had been named the allantoate permease family (11) after the product of the DAL5 (degradation of allantoin 5) gene, which was the first member of this family that has been assigned a specific transporter function (12). Dal5p transports allantoate and ureidosuccinate across the yeast plasma membrane. Other members of this family were shown to encode transporters for pantothenate (Fen2p) (13) or for nicotinate (Tna1p) (14,15). Similar transporters were also identified in Schizosaccharomyces pombe. SpLIZ1 was shown to encode a transporter for pantothenate (16) and SpVHT1 to be the homolog of the bakers' yeast vitamin transporter gene VHT1 (17).
Little is known about the transcriptional regulation of this group of transporter genes. In some cases, expression is regulated by the substrate concentrations in the growth medium (9,(13)(14)(15). However, compounds that are not a substrate for the respective transporter were also shown to regulate expression of some genes of the DAL5 family. For example, low levels of p-aminobenzoate in the growth medium resulted in increased TNA1 mRNA levels (14), and low levels of thiamine caused enhanced expression of YLR004c, a so far uncharacterized member of the DAL5 family, although thiamine is not transported by the product of YLR004c (15).
Here we present detailed analyses of the transcriptional regulation of VHT1. The cis-regulatory, vitamin H-responsive element (VHRE) 2 that is necessary and sufficient for the biotin-dependent expression is identified, and the previously uncharacterized protein YIL056wp is characterized as a transcriptional regulator responsible for VHRE-dependent VHT1 expression. The YIL056w gene was named VHR1 (VHT1 regulator 1). Our analyses also demonstrate that Vhr1p is also involved in the transcriptional regulation of BIO5 (biotin biosynthesis 5).

EXPERIMENTAL PROCEDURES
Yeast Strains and Media-Yeast strains used in this study are listed in Table 1. Rich medium (yeast extract-peptone-dextrose) was prepared as described (18). Synthetic dextrose (SD) medium contained 0.67% bactoyeast nitrogen base without amino acids (Difco, Augsburg, Germany) and 2% glucose and was supplemented with the necessary amino acids and nucleobases to meet the growth requirements of the strains. According to the manufacturer's manual (Difco), the biotin concentration in this standard SD medium is 2 g/liter. SD medium with defined biotin concentrations was made from analytical grade chemicals according to the formula of Difco for yeast nitrogen base without amino acids. All vitamins were added to autoclaved media from filter-sterilized stock solutions. For most analyses, yeast cells were grown on biotin concentrations of 0.2 g/liter (low biotin) or 200 g/liter (high biotin).
Gene Replacements and Plasmid Constructs-For constitutive expression, the VHT1 gene was introduced into JSY⌬vht1 (9) by homologous recombination into the vht1::HIS3 locus. For this purpose, an ADH (alcohol dehydrogenase) promoter/VHT1-ORF/VHT1 terminator cassette was PCR-amplified from the plasmid pVHT1oe (9) with the primers VHT1oe-ki-5Ј and VHT1oe-ki-3Ј (all primer sequences are shown in Table 2). Transformants (AMY-VHT1gc) were selected for recovery of growth on low biotin. Gene replacement was confirmed by screening for loss of the HIS3 marker and by PCR with appropriate primers.
Deletion of VHR1 in the wild-type strain JS91. 15-23 (9) was carried out by PCR amplification of the HIS3 disruption cassette of pFA6a-HIS3MX6 (19) with the primers YIL056W-ko-5Ј and YIL056W-ko-3Ј. Disruption of VHR1 was confirmed by PCR with appropriate primers. The resulting strain was called MWY⌬vhr1.
A GFP reporter plasmid driving GFP expression under the control of 760 nucleotides of the VHT1 promoter (pMW-760; details for all plasmids generated in this work are given in Table 3) was constructed as follows. First, an XbaI fragment harboring one of two HindIII sites was removed from the PMA1 terminator of the yeast Escherichia coli shuttle plasmid NEV-N (20). This shortened NEV-N vector was digested with HindIII and NotI to replace the PMA1 promoter by a 741-bp HindIII/ NotI fragment from the plasmid pGFP-TYGpAK (21). This fragment represents the ORF of an enhanced GFP (GFP(S65T)). From the resulting plasmid, the HindIII/XbaI GFP/PMA1 terminator cassette was iso- lated and ligated into YEplac181 (22), yielding the plasmid pMW-GFP2. The VHT1 promoter was amplified by PCR from pVHT1sc (9) with the primers VHT1-5Ј and VHT1-3Ј that introduced an NcoI site at the VHT1 start ATG. This PCR product was digested with HindIII and NcoI, and the 760-bp fragment was ligated into the respective sites of pMW-GFP2. Shorter variants of the VHT1 promoter were generated with appropriate primers and cloned into the same sites. Linker scanning analyses were performed by introducing mutations into plasmids harboring nucleotides Ϫ284 to Ϫ238 or nucleotides Ϫ267 to Ϫ249 of the VHT1 promoter upstream from the GAL1 (galactose metabolism) minimal promoter (GAL1 min ; pMWmin plasmids). These plasmids were constructed by ligating annealed oligonucleotides as outlined in Fig. 1 into SpeI and XhoI sites that had been introduced upstream of GAL1 min . GAL1 min had been amplified from genomic DNA (nucleotides Ϫ228 to Ϫ3 of the GAL1 promoter; primers ScGAL1g-228f-HSX and ScGAL1g-3r-NcoI) (12), adding HindIII, SpeI, and XhoI sites to the 5Ј-end and an NcoI site to the 3Ј-end. The PCR product was digested with HindIII and NcoI and used to replace the VHT1 promoter in pMW-760. The resulting plasmid was called pMWmin-p.
For N-terminal and C-terminal fusions of Vhr1p to GFP, the ORF of VHR1 was amplified from genomic DNA (primers YIL056Wϩ1f-ClaIϩ and YIL056Wϩ1920r-SalIϩ), introducing new restriction sites upstream of the start ATG (ClaI) and at the 3Ј-end (SalI), thereby removing the VHR1 stop codon. These sites were used to clone the VHR1 ORF upstream of the GFP ORF into pUG23 (23) and downstream of the GFP ORF into pUG34 (23). The resulting plasmids carry VHR1-GFP (pUG23-1920) or GFP-VHR1 fusions (pUG34-1920).
For complementation of vhr1 mutants that were unable to express GFP under the control of the VHT1 promoter, the VHR1 gene was PCR-amplified from genomic DNA including 830 bp of 5Ј-flanking sequence and 507 bp of 3Ј-flanking sequence. This PCR product was ligated into the multicopy vector YEp24 (24) that had been cut with BamHI and blunted with Klenow fragment. vhr1 mutant strains were transformed with the resulting plasmid (YEp056) and screened for GFP fluorescence.
For further complementation analyses, the genomic fragment was excised from YEp056 with SphI and XmaI and inserted into the single copy plasmid YCplac33 (22). The resulting vector was named YCp056. Mutants were transformed with this plasmid and tested for GFP fluorescence.
For the yeast one-hybrid analyses, two reporter plasmids driving the expression of GFP under the control of different promoters were VHT1oe-ki-5Ј 5Ј-CAC TCA GGA AAA CAT TAT TGC TTT AAA TTG ACA TAG TGT TCT TAT GGT AGA ACT TCT TTT CTT  TTT TTT TCT TTT CTC-3Ј  VHT1oe-ki-3Ј  5Ј-AGG TGA CCT TTA TCA ACA TCG AG-3Ј  YIL056W-ko-5Ј  5Ј-TTC TAC CTA TTC TTC TTC GTT GCT TAT ATC CTA TAC ACA TCA GCT GAA GCT TCG TAC GC-3Ј  YIL056W-ko-3Ј  5Ј-AAT GTG TAT TAG GCA GTT ATG TAT GTA GAC AAT TTT GTT TGC ATA GGC CAC TAG TGG ATC TG-  constructed. One plasmid carried a triple repeat of the Ϫ284 to Ϫ238 fragment of the VHT1 promoter (VHT1 284 -238 ) upstream of GAL1 min . To this end a 1.5-kb fragment harboring this promoter sequence (VHT1 284 -238 ), GAL1 min , the GFP ORF, and the PMA1 terminator was excised from the respective pMWmin plasmid and cloned into pFL61 (25). The ampicillin resistance gene of this plasmid was replaced by the kanamycin resistance gene from pMN234 (26). Two additional VHT1 284 -238 copies were cloned upstream of the first VHT1 284 -238 copy, and the resulting plasmid with three VHT1 284 -238 copies was named pSCR1. For the second plasmid (pSCR2), GAL1 min in pMWmin-p was replaced by a 533-bp fragment carrying the complete GAL1 promoter, including its upstream activation sequence (UAS G ) (27). This fragment had been excised with SacI/HindIII from pJG4-5 (28) and blunted.
The function of the putative DNA-binding domain (DBD) of Vhr1p was analyzed with constructs encoding fusion proteins of the N-terminal domain of Vhr1p (Vhr1p 1-133 ) with the AD of Gal4p (yielding Vhr1p 1-133 /Gal4p-AD) or encoding Vhr1p 1-133 alone. These constructs were generated in the plasmid pGAD-C1 (29), which drives expression of these ORFs with the ADH1 promoter. As a control, fulllength VHR1 was cloned into the same plasmid, yielding the plasmid pG056S. To this end, the VHR1 ORF was amplified from genomic DNA with the primers Sc056gϩ1f/NX and RSc056ϩ1923r, which introduced an XbaI site upstream of the VHR1 start ATG and a NotI site downstream of the stop TGA, and was cloned into pCR-BluntII-TOPO (Invitrogen). The resulting plasmid had a second XbaI site (inherent in pCR-BluntII-TOPO) downstream of the VHR1 ORF and the NotI site. The VHR1 ORF was excised with XbaI, blunted, and used to replace GAL4-AD in pGAD-C1 that had been cut with HindIII and been blunted. For a second plasmid, pG056N-AD, the first 133 codons of the VHR1 ORF were fused to the 5Ј-end of GAL4-AD. To this end, a BglII/ NotI fragment was removed from pG056S and replaced by the GAL4-AD that had been PCR-amplified from pGAD-C1 (primers Gal4ADϩ1f-Bgl and Gal4ADϩ408rS-Xma). The third plasmid, pG056N-S, encoded only the first 133 amino acids of Vhr1p. For this construct, a stop TAG was introduced after the first 133 codons of VHR1 by ligating the annealed oligonucleotides 6ϫStop-NotBX-f and 6ϫStop-NotBE-r into pG056S that had been cut with BglII and NotI.
The function of the putative activation domain (AD) of Vhr1p was analyzed with constructs encoding a Gal4p-DBD/Vhr1p-AD fusion or the entire Vhr1p. For the Gal4p-DBD/Vhr1p-AD construct, codons 133-641 of VHR1 (including the stop codon) were cloned downstream of the GAL4-DBD sequence into pAS2-1 (Clontech, Heidelberg, Germany) by ligating the 1.5-kb fragment of pG056S (cut with NotI, blunted, and digested with BglII) into pAS2-1 that had been cut with PstI, treated with Klenow fragment, and cut with BamHI. The resulting plasmid was called pG056C-BD. As a control, plasmid pAS056S was generated that contained full-length VHR1 in the same vector backbone as pG056C-BD with the only exception that VHR1 was followed by the PMA1 terminator instead of the ADH1 terminator. The plasmid pAS056S was constructed as follows. First, the PMA1 terminator from NEV-N (20) was excised with NotI and SalI and ligated between the VHR1 coding sequence and the ADH1 terminator in pG056S that had been cut with NotI and XhoI. The ADH1p/VHR1/PMA1t cassette was excised from the resulting plasmid with SphI and ligated into pAS2-1 that had been cut with SphI, thereby replacing part of the ADH1 promoter, the GAL4-DBD, and the ADH1 terminator. By this ligation, the longer ADH1 promoter of pAS2-1 (conferred to pG056S) was regained. The resulting plasmid was called pAS056S. For the empty control plasmid, pAS2-1xs, the ADH1 promoter and the GAL4-DBD were removed from pAS2-1 by cutting this vector with XhoI and SalI and religating the 5.7-kb vector fragment.
Fluorescence Microscopy and Photography-GFP fluorescence of yeast colonies was monitored using an epifluorescence stereomicroscope (Leica MZFLIII, Leica Microsystems, Wetzlar, Germany) and a Color View II camera controlled by the analySIS Doku 3.2 imaging software (Soft Imaging Systems, Münster, Germany). GFP was excited with light of 460 -500 nm, and emitted fluorescence was detected at wavelengths longer than 510 nm. Petri dishes were photographed with a Nikon Coolpix 995 Digital Camera (Nikon, Düsseldorf, Germany). For fluorescence pictures, a 510-nm-long pass filter (HQ 510 LP, AF Analysentechnik, Tübingen, Germany) was put in front of the lens, and GFP was excited with a mercury lamp (HBO 103 W/2, Carl Zeiss, Jena, Germany).
In order to study the intracellular localization of GFP or of Vhr1p-GFP fusions, yeasts were grown on SD medium with 0.2 g/liter or 200 g/liter biotin, harvested, and resuspended in H 2 O. Nuclear DNA was stained for 15 min by the addition of 0.2 g/ml 4Ј,6-diamidino-2-phenylindole (DAPI). Yeast cells were immobilized on microscope slides by mixing the yeast suspension with a solution of 1% (w/v) low melting point agarose and 15% (v/v) glycerol. Microscopic analyses were performed on an epifluorescence microscope (Zeiss Axioskop, Carl Zeiss, Jena, Germany). GFP was detected as described above, and DAPI was excited with light of 365-395 nm and detected at wavelengths longer than 420 nm. Images were recorded with a Sony MC-3255P 3CCD color video camera (Sony, Köln, Germany) controlled by Imaging System KS200 software (Kontron Elektronik, München, Germany).
Images were processed in analySIS Doku 3.2 imaging software (Soft Imaging Systems) and in Adobe Photoshop Version 7.0 (Adobe Systems, München, Germany).
Expression Analysis of VHT1-For quantitative fluorescence measurements, cells were grown on SD plates, transferred to water and adjusted to an A 600 ϭ 5. Fluorescence was measured in a fluorescence microplate reader (Tecan, Crailsheim, Germany) with the following filters: excitation 485 Ϯ 10 nm, emission 510 Ϯ 10 nm. In parallel, the OD of the analyzed cells was measured at 595 nm.
For Northern blot analyses, yeast cells were suspended in TRIzol (Invitrogen) and vortexed for 15 min in the presence of 0.5-mm glass beads. RNA preparation with TRIzol was continued according to the manufacturer's instructions. RNA gel blot analysis was performed on agarose gels as described elsewhere (30). Each lane was loaded with 20 g of total RNA. A VHT1-specific probe representing the nucleotides 1196 -1529 of the VHR1 ORF was labeled with [␣-32 P]dCTP. Control hybridizations were performed with an ACT1 probe (nucleotides 438 -813 after the start ATG of the ACT1 genomic sequence).
Ethyl Methane Sulfonate Mutagenesis and Yeast Transformation-Stationary phase cells of MWY760gc were mutagenized with EMS (Sigma) to a survival rate of 30% (31). Subsequently, cells were grown in SD medium for four generations to establish the mutations. Cells were then plated onto SD medium containing 0.2 g/liter biotin and were screened for the loss of GFP fluorescence. For a high efficiency of transformation, mutant yeast strains were first grown overnight in SD medium, followed by incubation in yeast extract-peptone-dextrose for two generations. Cells were harvested during logarithmic growth, transformed with the genomic library (32) using the lithium acetate method (33), and plated on low biotin medium. After 3 days, colonies were screened for GFP fluorescence. All other yeast strains were transformed using a modified lithium acetate method (34).
Real Time Reverse Transcription-PCR-Real time reverse transcription-PCRs were performed on a RotorGene 2000 (Corbett Research, Sydney, Australia) with the QuantiTect SYBR Green PCR Master Mix from Qiagen (Hilden, Germany). Samples were standardized to the ACT1 mRNA levels. Oligonucleotide sequences used for the amplification were ScACT1g-4f and ScACT1gϩ527r for ACT1, ScVHT1gϩ607f and ScVHT1gϩ808r for VHT1, and ScBIO5gϩ1040f and ScBIO5gϩ1244r for BIO5.

RESULTS
Identification of a VHRE in the VHT1 Promoter-The VHT1 biotin transporter gene of bakers' yeast is expressed on low biotin concentrations in the medium (0.2 g/liter biotin) but not in media with high biotin concentrations (200 g/liter biotin) (9). To identify the upstream regulatory sequences in the VHT1 promoter that are responsible for this regulation, deletion analyses were performed. To this end, 10 different 5Ј-deleted promoter fragments (starting 760, 608, 457, 307, 284, 278, 248, 218, 188, or 158 bp upstream from the start ATG; VHT1 760 to VHT1 158 ) were cloned in front of the GFP ORF in the YEplac181-based (22) plasmid pMW-GFP2. The capacity of the resulting constructs to drive the expression of GFP was analyzed on low and high biotin concentrations in the yeast strain JS91. 15-23 (9), and the ratio of the GFP fluorescence was determined (low biotin/high biotin represents -fold induction). Strong induction ratios of GFP fluorescence were detected in the five longest promoter constructs (VHT1 760 to VHT1 278 ; VHT1 760 and VHT1 284 are shown in Fig. 1A). In all other constructs (VHT1 248 or shorter; VHT1 248 and VHT1 0 are shown in Fig. 1A), the GFP fluorescence ratios were very small. This suggested that the regulatory ciselements responsible for VHT1 expression on low biotin medium were located between bp Ϫ278 and Ϫ248.
For further analysis, a fragment (nucleotides Ϫ284 to Ϫ238 of the VHT1 promoter, VHT1 284 -238 ) covering these 30 nucleotides was cloned upstream of the GAL1 minimal promoter (GAL1 min ). This promoter fusion (VHT1 284 -238 -GAL1 min ) induced GFP expression to the same level as the VHT1 284 promoter (Fig. 1), confirming that VHT1 284 -238 contains the vitamin H-responsive element (VHRE). VHT1 284 -238 was then analyzed by linker scanning analyses. Fig. 1A shows that replacement of nucleotides Ϫ278 to Ϫ268 (VHT1 278 -268-M -GAL1 min ) had no effect and that the constructs VHT1 260 -254-M -GAL1 min and VHT1 248 -246-M -GAL1 min reduced GFP induction by 40 -50%. A complete loss of GFP induction was observed with the constructs VHT1 270 -260-M -GAL1 min and VHT1 262-258-M -GAL1 min . VHT1 254 -252-M -GAL1 min and VHT1 252-251-M -GAL1 min caused a very strong or almost complete loss of GFP induction. As expected, construct VHT1 278 -244-M -GAL1 min that had the entire sequence replaced resulted in a complete loss of GFP induction. These data suggested that the sequence TGAATCA might represent an essential part of the VHRE and that additional downstream sequences (ATGA) might contribute to this element (boxed areas in Fig. 1A).
This was confirmed with the construct VHT1 267-249 -GAL1 min , which contained only 19 nucleotides of the VHT1 promoter (Fig. 1A). Both a mutation with the TGAATCA heptanucleotide (VHT1 267-265-M -GAL1 min ) and a mutation in the residual sequence (VHT1 260 -250-M -GAL1 min ) resulted in a complete loss of GFP induction.
Finally, the promoter sequences of VHT1 and of the homologous genes from four other Saccharomyces species (Saccharomyces paradoxus, Saccharomyces mikatae, Saccharomyces bayanus, and Saccharomyces castellii) and from Eremothecium gossypii, also a yeast of the order Saccharomycetales, were aligned (Fig. 1B). Sequences similar to the predicted VHRE were found in similar distances upstream from the respective start ATGs in all VHT1 promoters. The regions predicted to be essential for VHRE function (Fig. 1A) are conserved and allowed the identification of a palindromic sequence (Fig. 1B) with four almost perfectly conserved pyrimidines (Y) and four purines (R) in the center.
Screening for Factors Involved in VHT1 Transcriptional Regulation-For the identification of trans-factors involved in the induction or derepression of VHT1 expression at low extracellular biotin concentrations, reporter strains were generated that allowed a complementation screening for such genes. To this end, the ORF of VHT1 and VHT1 3Ј-flanking sequences were put under control of the ADH promoter and inserted into the ⌬vht1 mutant locus of JSY⌬vht1 (9). The resulting strain (AMY-VHT1gc) shows constitutive expression of VHT1 and grows on low and high biotin concentrations. In a second step, AMY-VHT1gc was transformed with the YEplac181-based (22) plasmid pMW-760. This plasmid harbors a cassette that drives biotin-regulated expression of GFP under the control of the VHT1 promoter (760-bp fragment). The resulting strain (MWY760gc) grows on low and high extracellular biotin concentrations but shows GFP fluorescence only on low biotin (Fig. 2B).
MWY760gc cells were mutagenized with EMS and plated on low biotin. Resulting colonies (150,000) were screened for loss of GFP fluorescence on low biotin (GFP minus lines) under an epifluorescence stereomicroscope. These GFP minus lines were expected to carry mutations in genes for VHT1 regulatory proteins. The stability of the GFP minus phenotype and the introduced mutations was tested in 12 independently obtained GFP minus lines by replating at least 20,000 of each line on low biotin. Three GFP minus lines showing zero revertants were mated with the MATa strain BY4741 (EUROSCARF accession number Y00000) to test whether they were still able to express GFP in the presence of an unmutagenized genome ( Fig. 2A). Two GFP minus lines (MWYmut9.1 and MWYmut36.1) with stable, recessive mutations were complemented with a yeast genomic library (32), plated on low biotin, and screened for restored GFP fluorescence. From 500,000 screened colonies, 12 showed restored GFP fluorescence (two from MWYmut9.1 and 10 from MWYmut36.1).
The sequences obtained from these 12 complements could be assigned to three complementation groups. The smallest genomic fragment present in all seven members of the first complementation group harbored a single protein-encoding gene, YIL056w. The second complementation group contained four members with almost or completely identical genomic fragments that covered a region containing the four genes TFB3 (which encodes a subunit of TFIIH) (35), MFA1 (which encodes the a-factor mating pheromone precursor) (36), MRPL28 (which codes for a protein of mitochondrial ribosomes) (37), and STP1 (which encodes an activator of amino acid permease genes) (38). The third complementation group represented a single fragment harboring the overlapping ORFs of RHO2 (which encodes a nonessential Ras-like GTPase) (39) and the so far uncharacterized gene YNL089c.
For four of the seven identified genes, deletion mutants were ordered from EUROSCARF (Frankfurt/Main, Germany): ⌬stp1 (strain Y04297), ⌬rho2 (strain Y07230; represents also a deletion mutant for the overlapping YNL089c), and ⌬yil056w (strain Y01449). Deletions in TFB3 are known to be lethal (40), and the MFA1 (mating factor) and MRPL28 (ribosomal protein) genes were excluded as candidates for a VHT1 transcriptional regulator. The ⌬stp1, ⌬rho2/ynl089c, and ⌬yil056w mutant lines were transformed with the plasmid pMW-760, which contains the VHT1 promoter-GFP expression cassette, and were screened for GFP fluorescence on low biotin. The pMW-760-containing ⌬stp1 and ⌬rho2/ynl089c lines showed GFP fluorescence. This suggests that none of these genes encodes a regulator of VHT1 transcription. In contrast, no GFP fluorescence was observed in the pMW-760-containing ⌬yil056w mutant line, indicating that the protein encoded by YIL056w is likely to be involved in VHT1 regulation (not shown). For this reason, the YIL056w gene was named VHR1 (for VHT1 regulator 1).
To confirm this suggested role of Vhr1p as a regulator of VHT1 expression, we transformed the mutant lines MWYmut9.1 and MWYmut36.1 with multicopy (YEp056) or single copy (YCp056) plasmids carrying the entire VHR1 gene. Fig. 2A confirms that VHR1 does complement the no-fluorescence phenotype of the mutant lines both from a multicopy (top image) and from a single copy (middle image) plasmid. GFP fluorescence was regained on low biotin after complementation with the isolated gene but not after transformation with the empty vector. Finally, the two mutant lines were mated with yeast strains BY4741 (wild type (WT)) (41) and with the isogenic ⌬vhr1 mutant Y01449 (Fig. 2A, bottom image). The crosses with BY4741 but not with Y01449 showed GFP fluorescence on low biotin, confirming that only VHR1 and no other gene of the yeast genome can complement the mutations in MWYmut9.1 and MWYmut36.1.
The induction of the VHT1 promoter in mutant lines complemented with the isolated VHR1 gene was quantified. Fig. 2B demonstrates that activation of the VHT1 promoter is similar in the unmutagenized strain (MWY760gc) and in the two complemented mutant lines. Strong GFP fluorescence is seen only on low biotin (0.2 g/liter) but not on high biotin (200 g/liter). In mutants transformed with the empty vector, this enhanced GFP fluorescence on low biotin was absent.
Finally, the effect of a deletion of VHR1 on the expression of its natural target gene VHT1 was studied. Fig. 2C shows a Northern blot of total RNA preparations from the WT strain JS91.15-23 and from the isogenic ⌬vhr1 strain MWY⌬vhr1. Total RNA was isolated from both strains after growth on low or high biotin, and the blot was hybridized with probes for ACT1 and VHT1 mRNAs. VHT1 mRNA was detected only in the VHR1 WT strain on low biotin. This confirms that Vhr1p is an essential component of the VHT1 induction pathway.
We also determined the genomic VHR1 sequences in both mutant lines. As expected from the data shown above, the alleles, vhr1-9.1 and vhr1-36.1, were mutated, and the observed mutations are likely to affect the function of the resulting proteins. In vhr1-9.1, the cytosine at position 466 is replaced by a thymine, which changes the CAG codon for Gln 156 into a TAG stop codon. In vhr1-36.1, the guanine at position 1129 is replaced by an adenine, which changes Glu 377 into Lys 377 .
Vhr1p Shows Nuclear Localization-To test the subcellular localization of Vhr1p, we generated C-terminal and N-terminal fusions of Vhr1p to GFP in the plasmids pUG23 and pUG34 (23). The resulting constructs pUG23-1920 (GFP fused to the Vhr1p C-terminus) and pUG34 -1920 (GFP fused to the Vhr1p N terminus) were used to transform strain BY4742 (41) (EUROSCARF accession number Y10000). The fluorescence of GFP-Vhr1p and Vhr1p-GFP was then compared with the subcellular distribution of GFP and with the fluorescence of DAPI-stained nuclei. As seen in Fig. 3, both fusions showed preferred (GFP-Vhr1p) or exclusive (Vhr1p-GFP) accumulation in nuclei. This confirms previous results obtained in a global analysis of protein localization in budding yeast (42).
Search for Proteins Homologous to Vhr1p-Vhr1p is a protein with a predicted molecular mass of 71.42 kDa, is composed of 640 amino acids, and has an isoelectric point of 9.67. A BLAST search within the S. cerevisiae genome identified a single gene, YER064c, that shares significant homology with VHR1. Yer064cp has only 505 amino acids (56.64 kDa; isoelectric point 9.91), and the two proteins share 46.0% identical amino acids (55.9% similarity). BLAST searches within all publicly available data libraries identified additional 12 homologs from 11 other yeasts (Fig. 4, B and C). None of these genes had been functionally characterized.
In Fig. 4A, the protein sequence of Vhr1p is presented. The highest degree of sequence conservation between Vhr1p and the 13 homologous proteins is seen in the two highlighted domains. Domain I, which is 97 amino acids long in Vhr1p, has previously been described as domain of unknown function number 352 (DUF352; accession number PF04001; available on the World Wide Web at www.sanger.ac.uk/cgi-bin/Pfam/ getacc?PF04001 or http://www.ebi.ac.uk/interpro/IEntry?acϭIPR007147). It is located near the N terminus of Vhr1p, and the degree of identity between domain I from Vhr1p and the respective DUF352 regions of the other proteins varies from 100% (S. mikatae protein) to 60.7% (Yarrowia lipolytica protein) (Fig. 4B). Domain II, which is 250 amino acids long in Vhr1p, is located in the center of the protein (Fig. 4A). The sequence conservation between the domain II regions of the different proteins is less pronounced than for domain I (Fig. 4C). This results in part from different lengths of these domain II regions (237-251 residues in most proteins and only 129, 142, or 144 residues in the proteins from Y. lipolytica and from C. glabrata and in Yer064cp, the second S. cerevisiae protein) but also from variations within similarly sized domain II sequences (e.g. 96% identity between the domain II of Vhr1p and the respective sequence from S. paradoxus and only 31% identity between the domain II of Vhr1p and the respective sequence from D. hansenii) (Fig. 4C).
Interestingly, our BLAST searches identified these homologous proteins only in closely related fungal species. No homologous proteins were found in mammals, plants, or bacteria, suggesting that this group of transcriptional regulators is unique to the order of Saccharomycetales within the phylum Ascomycota.
Vhr1p Is a Transcription Factor with an N-terminal DNA-binding Domain and a C-terminal Activation Domain-As a transcriptional regulator, Vhr1p should have specific interaction sites, such as a DBD and/or an AD, for interactions with the transcriptional machinery. Based on the sequence comparisons shown in Fig. 4 and based on in silico structural analyses, we speculated that the N-terminal part of Vhr1p, which is highly conserved in all related proteins and which has predicted helix-turn-helix motifs (e.g. amino acids 59 -102 (predicted with PHDsec at the PredictProtein server) (43)(44)(45) or amino acids 62-99 (predicted with SOPM) (46)) known to mediate DNA binding in other transcription factors (47,48), might represent the Vhr1p DNAbinding domain.
To test this hypothesis, we generated fusion proteins of the N-terminal domain I of Vhr1p (Vhr1p 1-133 ) with the AD of Gal4p (yielding Vhr1p 1-133 /Gal4p-AD), one of the best characterized yeast transcription factors (49). The Gal4p-DBD and the Gal4p-AD are widely used in yeast two-hybrid analyses (50). We analyzed the capacity of these fusions to mediate GFP expression in a strain harboring a plasmid (pSCR1) that carries GFP under the control of the GAL1 min and three repeats of the regulatory cis-element identified in the VHT1 promoter (3 ϫ VHT1 284 -238 ; see Fig. 1A). In a second approach, we generated fusion proteins of the Gal4p-DBD and the C-terminal part of Vhr1p (Vhr1p 133-640 3 Gal4p-DBD/Vhr1p 133-640 ) and analyzed the capacity of these fusions to mediate GFP expression from a complete GAL1 promoter (i.e. including the upstream activating sequence that is recognized by Gal4p (UAS G )) (27). Controls with the intact Vhr1p, with Gal4p-DBD, Gal4p-AD, and Vhr1p 1-133 alone, or with the empty vector were included. The GFP fluorescence obtained with these controls was compared with the fluorescence obtained in strains harboring the Vhr1p 1-133 /Gal4p-AD or the Gal4p-DBD/Vhr1p 133-640 .   5 summarizes the results obtained in these yeast one-hybrid analyses. As expected, intact Vhr1p mediates expression of GFP from the (3 ϫ VHT1 284 -238 )-GAL1 min , and, in contrast, neither Gal4p-AD alone nor Vhr1p 1-133 alone cause any detectable GFP expression from this reporter construct (Fig. 5A). However, the Vhr1p 1-133 /Gal4p-AD fusion yielded strong GFP expression on low biotin (about 50% of the Vhr1p signal), and a significantly lower level of Vhr1p 1-133 /Gal4p-AD-dependent GFP fluorescence was also seen on high biotin medium. This suggests that the activity of Vhr1p 1-133 is modulated by the biotin concentration. Moreover, this demonstrates that Vhr1p 1-133 can functionally replace the Gal4p-DBD and can mediate the interaction of the fusion protein with the (3 ϫ VHT1 284 -238 ) cis-element. Finally, the DUF352 domain of unknown function that represents the major part (amino acids 11-107) of Vhr1p 1-133 is likely to be responsible for this function.  This biotin-dependent regulation of both the DNA recognition and the transcriptional activation by Vhr1p may allow a fine tuning of the VHT1 promoter over a wide range of biotin concentrations in the medium. Therefore, we determined the activity of the VHT1 promoter between 40 and 0.05 g/liter (Fig. 6). The VHT1 promoter is "off " between 40 and 2 g/liter and is, in fact, gradually activated between 1 and 0.05 g/liter (i.e. over 2 orders of biotin concentrations).
Vhr1p Is Also Involved in the Regulation of BIO5 Expression-After the characterization of Vhr1p as a biotin-dependent transcriptional regulator, and after the identification of the VHRE we asked if this transfactor/cis-element combination may also be responsible for the regulation of other genes encoding proteins of the yeast biotin metabolism. When we screened S. cerevisiae promoters for the presence of the AATCA-N 8 -TGA sequence conserved in the VHRE characterized in Fig. 1B, we found this sequence in the promoter of the BIO5 gene at a very similar position (Ϫ283 to Ϫ268) as in the VHT1 promoter (Ϫ265 to Ϫ250). BIO5 encodes a plasma membrane transporter for 7-keto-8aminopelargonic acid, a compound that can be used as a biotin biosynthetic precursor (51).
We studied BIO5 mRNA levels on low and high biotin, both in a VHR1 WT strain and in a ⌬vhr1 mutant. Fig. 7 shows that in the VHR1 WT strain, BIO5 mRNA levels are about 10-fold higher on low than on high biotin, a difference that is quite similar to that of VHT1 mRNAs. Moreover, Fig. 7 shows that the biotin dependence of BIO5 expression has disappeared in the ⌬vhr1 mutant. This confirms that Vhr1p is responsible for the transcriptional regulation of at least one additional gene coding for a biotin metabolic protein.

DISCUSSION
This paper presents a detailed analysis of cis-regulatory and transregulatory elements involved in biotin-dependent gene expression in budding yeast. It presents the identification of a VHRE in the VHT1 and BIO5 promoters and the characterization of Vhr1p as a transcriptional   Fig. 3). The capacity of intact Vhr1p, of Gal4p-AD, of Vhr1p 1-133 , or of Vhr1p 1-133 /Gal4p-AD to mediate GFP expression from pSCR1 is presented Ϯ S.D. Intact Vhr1p mediates the expected, strong GFP expression on low biotin, whereas no expression is seen with Gal4p-AD or with Vhr1p 1-133 . In contrast, Vhr1p 1-133 /Gal4p-AD causes a strong expression of GFP on low biotin and a much lower expression on high biotin (n ϭ 3). The gray region shows the background activity obtained with Vhr1p on high biotin. B, characterization of the C-terminal part of Vhr1p (amino acids 133-640) that harbors the conserved domain II (see Fig. 3). The capacity of Gal4p-DBD, of intact Vhr1p, or of Gal4p-DBD/ Vhr1p 133-640 to mediate GFP expression from pSCR2 is shown Ϯ S.D. An empty vector with no ORF for a potential transcription factor was included as a control (gray region shows the background). Intact Vhr1p mediates the same background GFP expression as the vector control, which is independent from the biotin concentration. Gal4p-DBD, which can bind to the GAL1 promoter, causes a slight increase of the expression, but again this is independent of the biotin concentration. In contrast, Gal4p-DBD/Vhr1p 133-640 causes a strong expression of GFP on low biotin that is not detectable on high biotin (n ϭ 4).
activator of these promoters on low biotin concentrations. Vhr1p is encoded by YIL056w, contains the DUF352 domain of unknown function in its N-terminal region, and has so far been described as a protein of unknown function.
The VHRE Consensus Sequence-Deletion and linker scanning analyses of the VHT1 promoter identified a 19-nucleotide fragment that was essential for VHT1 expression on low biotin (Fig. 1A). When cloned upstream of GAL1 min , these 19 nucleotides mediated biotin-dependent expression of GFP only on low (0.2 g/liter) but not on high (200 g/liter) biotin (Fig. 5). This sequence was highly conserved in the VHT1 promoters of five other yeast species, and an alignment of all six sequences suggested the 18-nucleotide palindromic sequence AATCAY 4 R 4 TGAYT as the Vhr1p target site (Fig. 1B). The identification of BIO5, the gene of a plasma membrane transporter for the biotin precursor 7-keto-8-aminopelargonic acid, as a second target gene for Vht1p (Fig. 7) by an in silico search for conserved AATCA-N 8 -TGA sequences in yeast promoters confirmed the role of this cis-element. Alignments of BIO5 promoter sequences from seven different yeast species (S. bayanus, S. cerevisiae, Saccharomyces kluyveri, Saccharomyces kudriavzevii, S. mikatae, S. paradoxus, and Caenorhabditis albicans; data not shown) yielded a slightly different consensus sequence, AATCAYRYNNNNYTGAYN for the seven BIO5 promoters, and a combination of the 13 sequences from all VHT1 and BIO5 VHREs resulted in the consensus motif AAT-CAYNYNNNNNTGAYN. The inverted triplet TCA-N 8 -TGA seems to represent the core portion of the VHRE. In addition, however, the perfect conservation of the AA nucleotides at the 5Ј-end of this sequence and the almost complete loss of GFP expression in a mutation affecting one of these two nucleotides (Fig. 1A) suggest an essential function also for this part of the sequence.
Similar cis-regulatory elements (i.e. inverted triplets that are separated by a less conserved spacer) have been described for other yeast transcription factors. Examples are CCG-N 4 -CGG for Leu3p (activation of metabolism of branched amino acids) (52), CGG-N 10 -CCG for Put3p (activation of proline metabolism) (53), or CGG-N 11 -CCG for Gal4p (activation of galactose metabolism) (54). Hap1p, which regulates genes required for cellular respiration, recognizes a sequence with a direct repeat (CGG-N 6 -CGG), and it has been shown that two nucleotides outside this sequence (CGG-N 6 -CGG-N 2 -TA) are also important for the binding of Hap1p to its target sequence (55). This resembles the situation observed for the VHRE that has two highly conserved and essential nucleotides (AA) adjacent to the 5Ј-end of the inverted TCA-N 8 -TGA repeat (Fig. 1, A and B). However, the triplet sequences recognized by the proteins listed above are quite conserved (CCG, CGG, or CCA) and clearly different from the inverted triplet in TCA-N 8 -TGA. To our knowledge, the AA-TCA-N 8 -TGA sequence identified in this work represents a new, previously unknown cis-regulatory element.
Vhr1p Represents the Prototype of a New Class of Transcriptional Regulators-All of the above mentioned transcription factors are characterized by the presence of a zinc finger containing a cluster of 6 cysteines (consensus sequence: CX 2 CX 6 CX 5-9 CX 2 CX 6 -8 C). Proteins with such a cysteine cluster bind two zinc atoms and are referred to as Zn 2 Cys 6 binuclear cluster proteins (56,57). Conserved cysteine residues that might form such a cluster are not found in Vhr1p. Zn 2 Cys 6 -type proteins and also transcription factors from many other groups bind to their target sequences as homo-or heterodimers (58,59). Typically, this dimerization occurs via a coiled-coil (56,60) or a leucine zipper (59). We could not identify potential leucine zippers (available on the World Wide Web at 2zip.molgen.mpg.de/) or coiled-coil sequences (available on the World Wide Web at bioweb.pasteur.fr/seqanal/ interfaces/pepcoil.html) in the Vhr1p protein.
Vhr1p-GFP fusions were located to the nucleus (Fig. 3), confirming data from global analysis of protein localization in yeast (42). We tested if this subcellular distribution of the fusion proteins is modulated by the biotin concentration, but we could not detect any differences (not shown). A nuclear localization has been predicted (42) (available on the World Wide Web at db.yeastgenome.org/) also for Yer064cp, the S. cerevisiae homolog of Vhr1p. For the homologs from other species, similar information is lacking. The nuclear localization sequence of Vhr1p has not been determined in the present work, but the basic region between amino acid residues 97 and 107 in Vhr1p, which is the most highly conserved part of all Vhr1p homologs, might act as a nuclear localization sequence (Fig. 4B).
DNA Binding and Transcriptional Activation by Vhr1p Are Biotindependent-The function of Vhr1p as transcriptional regulator is doubtlessly proven by the yeast one-hybrid analyses presented in Fig. 5. In these analyses, the N-terminal 133 amino acids (containing domain I from Fig. 4; DUF352) or the C-terminal 508 amino acids (containing domain II from Fig. 4) were used to replace the DBD or the AD of Gal4p (50). The generated fusions, Vhr1p 1-133 /Gal4p-AD and Gal4p-DBD/ Vhr1p 133-640 , were able to drive GFP expression from the respective promoters ((3 ϫ VHT1 284 -238 )-GAL1 min or GAL1 promoter), confirming the independent functions of the two Vhr1p parts analyzed.
Surprisingly, both DNA binding by Vhr1p 1-133 and transcriptional activation by Vhr1p 133-640 seem to be biotin-dependent. The GFP fluorescence was 3-fold higher on low than on high biotin when DNA binding was mediated by Vhr1p 1-133 (Fig. 5A) and 10-fold higher when transcriptional activation was mediated by Vhr1p 133-640 (Fig. 5B). Therefore, either both domains (Vht1p-DBD and Vht1p-AD) are modulated independently by the biotin concentration, or an additional protein interacts with both parts of Vhr1p to modulate its DNA binding and/or its activity as transcriptional activator. This dual regulation may be the reason for the observed fine tuning of the VHT1 promoter (Fig. 6), which is increasingly activated over 2 orders of biotin concentrations. In a simple model, Vhr1p alone could be responsible for the regulation of the VHT1 promoter. In this case, it would represent a soluble biotin sensor that regulates VHT1 expression simply upon biotin interaction and by interaction-dependent conformational changes. In fact, repression of biotin biosynthetic genes in Escherichia coli is mediated via such a mechanism. A bifunctional protein (BirA) acts as a proteinbiotin ligase or as a sensor-like transcriptional regulator (7), and after complexation of BirA with biotin-5Ј-AMP (61) or with biotin (62), BirA monomers bind cooperatively to the biotin operator. However, also eukaryotic transcription factors are known to act as sensors by directly interacting with a specific metabolite and by regulating the transcription of target genes in a concentration-dependent way (63). An example is Put3p, which is tethered to its target DNAs in the absence and presence of proline, its activating metabolite. When proline is the sole nitrogen source and when proline binds to Put3p, a conformational change is induced, and Put3p-regulated promoters become transcriptionally active (reviewed in Ref. 63).
Alternatively, the transcriptional activation by Vhr1p might depend on the activity of another protein (e.g. for phosphorylation). Phosphorylation-dependent regulation was described for numerous transcription factors (e.g. for the transcriptional activator Pho4p, which regulates PHO gene expression in bakers' yeast) (64). Several phosphorylation sites in Pho4p independently regulate nuclear import, nuclear export, and the interaction with the transcription factor Pho2p (65).
Vhr1p May Be a General Regulator of Biotin-dependent Gene Expression-The observation that Vhr1p regulates not only the expression of VHT1 but also that of BIO5 (Fig. 7) demonstrates that Vhr1p is a more general regulator of biotin-dependent transcription. Nevertheless, there is a difference between the Vhr1p-dependent regulation of VHT1 and BIO5 expression that becomes obvious only in the ⌬vhr1 mutant (Fig.  7). Whereas the VHT1 mRNA levels in this mutant are low, both on low and high biotin (with a lack of induction), the mutant BIO5 mRNA levels are between the low biotin and high biotin levels observed in the VHR1 WT (with deregulation). This suggests that BIO5 is not only induced by Vhr1p on low biotin but is also repressed on high biotin, which could possibly be mediated by another protein.
A link between the regulation of VHT1 and BIO5 has already been reported (66). In this paper, an increased expression of both genes in a ⌬grx5 mutant has been shown. GRX5 encodes a mitochondrial glutaredoxin that is involved in the formation of iron/sulfur clusters. Deletion of this gene results in iron accumulation, in the enhanced expression of numerous genes involved in iron homeostasis, and also in the enhanced expression of VHT1 and of BIO5. However, the signal that causes the enhanced expression of the latter genes may result from a reduction of intracellular biotin levels. Bio2p, the protein catalyzing the final step in budding yeast biotin synthesis contains iron/sulfur centers (67); its activity would thus be low in ⌬grx5 cells and result in reduced biotin levels. The resulting induction of BIO5 that is shown in Fig. 7 has not been observed in a previous paper (66), but this may result from the different ways the gene was analyzed (Northern analyses (in Ref. 66) versus reverse transcription-PCR (this paper)).
In future studies, it will be necessary to study the mechanism of Vhr1p binding to its target DNA in detail and to reveal similarities with and differences from other groups of transcription factors. Moreover, it will be important to find whether Vhr1p regulates the expression of more genes, and candidate genes are, among others, BIO3 (which encodes a 7,8-diaminopelargonic acid aminotransferase) and BIO4 (which encodes a dethiobiotin synthetase) or BPL1, the gene of the biotin protein ligase.