Functional analysis of yeast gene families involved in metabolism of vitamins B1 and B6


  • Susana Rodríguez-Navarro,

    1. Departamento de Bioquímica y Biología Molecular, Universitat de València, C/Dr Moliner 50, E-46100, Burjassot, Spain
    Current affiliation:
    1. Biochemie-Zentrum-Heidelberg (BZH), Im Neuenheimer Feld 328, D-69120 Heidelberg, Germany.
    Search for more papers by this author
    • These authors contributed equally to this work

  • Bertrand Llorente,

    1. Unité de Génétique Moléculaire des Levures (URA 2171 du CNRS, UFR 927 Université Pierre et Marie Curie), Institut Pasteur, Paris, France
    Current affiliation:
    1. Department of Microbiology and Institute of Cancer Research, Columbia University College of Physicians and Surgeons, 701 West 168th Street, New York, NY 10032, USA.
    Search for more papers by this author
    • These authors contributed equally to this work

  • María Teresa Rodríguez-Manzaneque,

    1. Departament de Ciències Mèdiques Bàsiques, Facultat de Medicina, Universitat de Lleida, Rovira Roure 44, E-25198 Lleida, Spain
    Search for more papers by this author
  • Anna Ramne,

    1. Department of Cell and Molecular Biology, Lundberg Laboratory, Göteborg University, PO Box 462, S-40530 Göteborg, Sweden
    Current affiliation:
    1. Department of Molecular and Cellular Engineering, University of Pennsylvania School of Medicine, Philadelphia, PA 19104-6160, USA.
    Search for more papers by this author
  • Genoveva Uber,

    1. Departamento de Bioquímica y Biología Molecular, Universitat de València, C/Dr Moliner 50, E-46100, Burjassot, Spain
    Search for more papers by this author
  • Denis Marchesan,

    1. Department of Cell and Molecular Biology, Lundberg Laboratory, Göteborg University, PO Box 462, S-40530 Göteborg, Sweden
    Current affiliation:
    1. Department of Medical Biochemistry, Göteborg University, PO Box 440, S-40530 Göteborg, Sweden.
    Search for more papers by this author
  • Bernard Dujon,

    1. Unité de Génétique Moléculaire des Levures (URA 2171 du CNRS, UFR 927 Université Pierre et Marie Curie), Institut Pasteur, Paris, France
    Search for more papers by this author
  • Enrique Herrero,

    1. Departament de Ciències Mèdiques Bàsiques, Facultat de Medicina, Universitat de Lleida, Rovira Roure 44, E-25198 Lleida, Spain
    Search for more papers by this author
  • Per Sunnerhagen,

    1. Department of Cell and Molecular Biology, Lundberg Laboratory, Göteborg University, PO Box 462, S-40530 Göteborg, Sweden
    Search for more papers by this author
  • José E. Pérez-Ortín

    Corresponding author
    1. Departamento de Bioquímica y Biología Molecular, Universitat de València, C/Dr Moliner 50, E-46100, Burjassot, Spain
    • Departamento de Bioquímica y Biología Molecular, Universitat de Valéncia, C/Dr Moliner 50, E-46100, Burjassot, Spain.
    Search for more papers by this author


In order to clarify their physiological functions, we have undertaken a characterization of the three-membered gene families SNZ1–3 and SNO1–3. In media lacking vitamin B6, SNZ1 and SNO1 were both required for growth in certain conditions, but neither SNZ2, SNZ3, SNO2 nor SNO3 were required. Copies 2 and 3 of the gene products have, in spite of their extremely close sequence similarity, slightly different functions in the cell. We have also found that copies 2 and 3 are activated by the lack of thiamine and that the Snz proteins physically interact with the thiamine biosynthesis Thi5 protein family. Whereas copy 1 is required for conditions in which B6 is essential for growth, copies 2 and 3 seem more related with B1 biosynthesis during the exponential phase. Copyright © 2002 John Wiley & Sons, Ltd.


Paralogous gene families, which cover the three domains of life (Galperin, 2001; Goffeau et al., 1996; Henikoff et al., 1997; Rubin et al., 2000) represent an important part of all genomes sequenced so far. They comprise ca. 40% of the yeast genome (Blandin et al., 2000). This gene redundancy can be associated with an exact functional redundancy (Brookfield, 1997). Examples are the rDNA genes, the histone genes in species having a rapid early embryonic development, such as the sea urchins (Tartof, 1975) and the CUP1 gene from Saccharomyces cerevisiae. However, in many cases, paralogous genes have undergone functional specializations and are only partially redundant or even functionally non-redundant. The HSP70 paralogous gene family of S. cerevisiae is an interesting example (Boorstein et al., 1994). On the one hand, some of its members have overlapping functions, such as SSA1, SSA2 and SSA3. None of them is essential, but the triple deletion mutant exhibits a synthetic phenotype and is not viable. On the other hand, other members have acquired essential functions, such as KAR2 and SSC1, although they encode proteins very similar to the Ssa proteins. To obtain an exhaustive overview of these relationships in S. cerevisiae, we and others have undertaken a systematic functional characterization of 35 gene families containing two to four members with uncharacterized or poorly characterized functions (Dujon et al., in preparation).

Here we report a study of the SNZ and SNO gene families of S. cerevisiae, each consisting of three members (called 1, 2 and 3), located adjacently in chromosomes XIII, XIV and VI, respectively (Figure 1). SNO2 and SNO3 nucleotide sequences are almost identical (99%), as well as SNZ2 and SNZ3 sequences. SNO1 and SNZ1 sequences are more divergent from their respective counterparts (around 81% identical). Copies 2 and 3 are located within large subtelomeric duplicated regions that encompass other genes, including two members of a family of thiamine (vitamin B1) putative biosynthetic enzymes: THI5 and THI12. Homologues of SNZ and SNO genes have been found in a wide range of microorganisms and plants (Galperin, 2001; Mittenhuber, 2001), thereby making their functional analysis of general interest. The SNZ genes in yeast were originally discovered as expressed in stationary phase (Braun et al., 1996), and the SNO genes were found as proximal and coordinately regulated with the SNZ genes (Padilla et al., 1998). Three different studies have revealed that homologues of the SNZ and SNO genes from Aspergillus nidulans (Osmani et al., 1999), Neurospora crassa (Bean et al., 2001) and Cercospora nicotiniae (Ehrenshaft et al., 1999; Ehrenshaft and Daub, 2001) were related to the biosynthesis of pyridoxal (vitamin B6).

Figure 1.

Chromosomal organization of SNZ, SNO and THI5/12 families. Chromosomal regions from chromosomes VI, XIV (both subtelomeric) and XIII, including the gene families, are shown. Genes transcribed in ‘Watson’ orientation are shown in the upper line and those transcribed from the ‘Crick’ strand in the lower one. Relevant genes are named. The black bar between chromosomes VI and XIV marks a region of 6350 bp that has only 45 nucleotide changes, mostly single nucleotide transitions. The dotted bars between chromosomes mark regions with partial similarity (50–90%). Genes THI11 and THI13 (members of the THI5 family) are both subtelomeric on chromosomes X and IV, respectively, but do not have neighbouring SNZ/SNO genes

In this study we demonstrate that SNO1 and SNZ1 are required for growth of yeast in the presence of low level of intracellular vitamin B6. We also show that transcripts of SNO2, SNO3, SNZ2 and SNZ3 are accumulated in the absence of external thiamine, as well as THI5 and THI11 transcripts, and that Snz proteins can interact with Thi5 and Thi12 proteins.

Materials and Methods

Yeast strains

The yeast strains used or constructed in this work are listed in Table 1 BY4741, BY4742, CML235 and CML236 are the wild-type strains from which the snz and sno mutants were obtained. All strains, except for W303-1A, are from the S288c genetic background.

Table 1. Yeast strains used and constructed
StrainGenotypeSource or reference
FY1679MATa/α, ura3-52/ura3-52, leu2-Δ1/LEU2, trp1-Thierry and Dujon,
 63/TRP1, his3-Δ200/HIS31992
BY4741MATa, leu2-Δ0, his3-Δ1, met15-Δ0, ura3-Δ0Brachmann et al., 1998
BY4742MATα, leu2-Δ0, his3-Δ1, lys2-Δ0, ura3-Δ0Brachmann et al., 1998
CML235MATa, ura3-52, leu2-Δ1, his3-Δ200Spore from FY1679
CML236MATα, ura3-52, leu2-Δ1, his3-Δ200Spore from FY1679
W303-1AMATa, ade2-1, leu2-2, 122, ura3-1, his3-11, trp1-1aH. Ronne
PJ69-4aMATa, ade2, trp1-109, leu2-3, 112, ura3-52, his3-200James et al., 1996
 gal4Δ, gal80Δ, GAL2:ADE2, LYS2::GAL1:HIS3 
PJ69-4αMATα, ade2, trp1-109, leu2-3, 112, ura3-52, his3-200James et al., 1996
 gal4Δ, gal80Δ, GAL2::ADE2, LYS2::GAL1:HIS3, 
BQS1029(BY4742) snz1-Δ0::LEU2This work
BQS1037(BY4741) snz1-Δ0::MET15This work
BQS1067(BY4741) snz3-Δ0::LEU2This work
BQS1068(BY4742) snz2-Δ0::URA3This work
BQS1148(BY4742) snz1-Δ0::KanMX4, snz2-Δ0::URA3This work
BQS1149(BY4741) snz1-Δ0::KanMX4, snz3-Δ0::LEU2This work
BQS1060(BY4742) snz3-Δ0::LEU2, snz2-Δ0::URA3This work
BQS1073(BY4742) snz3-Δ0::LEU2, snz2-Δ0::URA3, snz1-This work
FYBL1-8BMATa, ura3-Δ851, leu2-Δ1, his3-Δ200, lys2-Δ202Fairhead et al., 1996
FYBL119-MATα, ura3-Δ851, leu2-Δ1, his3-Δ200, yol055c-Llorente et al., 1999
5BΔ::KanMX2, ypl258c-Δ::KanMX2, ypr121w- 
FYBL1-MATa, ura3-Δ851, leu2-Δ1, his3-Δ200, lys2-Δ202,Llorente et al., 1999
FYBL1-MATa, ura3-Δ851, leu2-Δ1, his3-Δ200, lys2-Δ202,Llorente et al., 1999
MML21(CML235) sno1-Δ0::KanMX4This work
MML23(CML235) sno2-Δ0::KanMX4This work
MML25(CML235) sno3-Δ0::KanMX4This work
MML27(CML235) sno1-Δ0::KanMX4 sno2-Δ0::KanMX4This work
MML49(CML236) sno1-Δ0::KanMX4 sno2-Δ0::KanMX4This work
MML50(CML236) sno2-Δ0::KanMX4 sno3-Δ0::KanMX4This work
MML259(CML235) sno1-Δ0::KanMX4 sno3-Δ0::KanMX4This work

Generation of multiple mutants

Single null mutations in the genes were generated by disruption with the cassettes described by Wach et al. (1994) or Brachman et al. (1998). Multiple mutants were generated by successive transformations with different markers or by conventional crosses. Disruptions were tested by analytical PCR using the adequate oligonucleotide sets (Rodríguez-Navarro et al., 1999; Wach et al., 1994).

In the cases of SNZ2 (YNL333W) and SNZ3 (YFL059W), their almost identical sequences precluded targeted disruption of each copy. SNZ2 and SNZ3 are located in different chromosomes (see Figure 1). We designed a ‘blind’ disruption and crossing strategy using two different markers, URA3 and LEU2, in both MATa and MATα. Crosses between randomly selected a and α clones with different marker were made and diploids were sporulated to obtain, in some cases, Leu+Ura+ spores that, therefore, corresponded to a snz2 snz3 double mutant. Thus, the parental strains that had been used for that particular cross should correspond to single snz2 and snz3 mutants. The identity of the single and double mutants was corroborated by Southern blot after pulsed-field electrophoresis (not shown). Double and triple mutants, including Δsnz1 deletion, were made from the single or the snz2 snz3 double mutants by disrupting SNZ1 with the KanMX4 cassette (Wach et al., 1994).

In the case of SNO2 and SNO3, the KanMX4 cassette was used to disrupt both genes. Individual mutants in each of them were distinguished by Southern analysis after ScaI-XhoI digestion of genomic DNA (a XhoI site is present upstream of SNO3 that is absent in the corresponding SNO2 region).

Growth conditions

The S. cerevisiae strains were routinely grown on YPD (1% yeast extract, 2% peptone and 2% glucose), minimal SD medium [0.67% yeast nitrogen base (YNB without amino acids, DIFCO), 2% glucose, supplemented with auxotrophic requirements]; or minimal SC medium [0.67% yeast nitrogen base (YNB without amino acids, DIFCO), 2% glucose, supplemented with Drop-out mix (DIFCO)]. Vitamin B6-deficient medium (SC-B6) was prepared by substituting the pre-mixed YNB for a mixture of the same components except from vitamin B6 [biotin, pantothenic acid, nicotinic acid, thiamine, inositol, H3BO3, CuSO4, KI, MnSO4, NaMoO3, ZnSO4, H2KPO4, (NH)2SO4, MgSO4, CaCl2, FeCl3, Na2MoO4] at the same concentrations as the DIFCO medium in 0.5 M, pH 6, 2-[N-morpholine]ethanesulphonic acid buffer. For control experiments this medium was supplemented with vitamin B6 to 2 µg/ml. For complementation analysis with pCM plasmids, SC-B6 in the absence (derepressing conditions) or in the presence (repressing conditions) of doxycycline was used.

Thiamine-deficient medium was prepared as described (Llorente et al., 1999). It is identical to the B6-deficient medium but without thiamine and plus vitamin B6 to 2 µg/ml. Geneticin-resistant strains were grown on YPD plates containing 200 mg/l geneticin (Gibco BRL). Meiosis induction was carried out by growing cells in YPD to saturation and then in YPA (1% yeast extract, 2% peptone and 2% acetate) to 2 × 107 cells/ml. After washing cells twice with water they were resuspended in sporulation medium (0.5% potassium acetate).

For phenotypic analysis, the growth of haploid mutants was checked on YP 2% glycerol and in YPD containing 1.2 M NaCl, 0.8 M KCl, or 1.8 M sorbitol. Cells were grown at 15° C, 28° C and 37° C for 2–3 days or longer when necessary.

Effects of overexpression of SNZ1 and SNZ2/3 genes on growth curves were made after transformation of strain W303-1A to uracil prototrophy with plasmids pCM262SNZ1, pCM262SNZ2 or empty pCM262 vector. At all times before the actual experiment, expression from the tetO7 promoter was turned off by the presence of 2 µg/ml doxycycline in the culture medium. To measure growth curves, transformed cells were initially cultivated in liquid SC medium lacking uracil and containing doxycycline. The culture was then split in two aliquots, one of which was rinsed free of doxycycline and then cultured without doxycycline. Growth of cells at 30° C in 400 µl microchambers with continuous shaking was measured as turbidity in a Labsystems Bioscreen C Microbiology Workstation, using a wide-band visible light filter.

Analysis of sensitivity to menadione

Cell cultures in YPD medium at 30° C and at the indicated growth stage (exponential or post-diauxic) were directly added with 20 or 40 mM menadione. After the indicated times, 1 : 5 serial dilutions were made and drops spotted onto YPD plates. Growth was recorded after 2 days of incubation at 30° C.

Northern analysis

Isolation of total RNA, electrophoresis, radioactive or non-radioactive probe labelling, hybridization and signal detection were all done as previously described (Llorente et al., 1999; Gallego et al., 1997). Probes were generated by PCR amplification from genomic DNA, using oligonucleotides designed to amplify the entire ORF without adjacent sequences (Table 2).

Table 2. Oligonucleotide primers used for PCR and cloning
NameSequence 5′-3′


A library of genomic S. cerevisiae DNA from the his3 ade2 gal4 strain JB974 in the Gal4 activation domain (Gal4-AD) fusion vector pACT2 (13) was used. The construction of this library will be described elsewhere (Ramne A, Sunnerhagen P et al., in preparation).

Bait clones encoding fusions of Gal4 DNA-binding domains (Gal4-DB) and proteins of interest were constructed using homologueous recombination in vivo (Muhlrad et al., 1992). Briefly, full-length coding sequences of genes were PCR amplified from total genomic S. cerevisiae FY1679 DNA, using the Roche Expand High Fidelity system and hybrid primers with 30 nucleotides of homology to the gene and 17–21 nucleotides of homology to sequences flanking the cloning site (Table 2) of the Gal4-DB vector pGBT9 (Bartel et al., 1993). PCR products were co-transformed with pGBT9 restricted with BamHI and EcoR1 into S. cerevisiae PJ69-4α (James et al., 1996). In our hands, >90% of plasmids from tryptophan prototrophs obtained in this manner contained an insert of the correct size.

Transformants (50–100, picked at random from each transformation) were pooled and used for subsequent mass mating with PJ69-4a transformed with the S. cerevisiae genomic DNA Gal4-AD fusion library. Diploid cells with a functional two-hybrid interaction were selected on medium lacking tryptophan, leucine and histidine, and containing 3 mM 3-aminotriazole (3-AT) and 2 mg/l adenine. The identity and reading frame of genes in prey plasmids was verified by partial sequencing.

Plasmid construction

Recombinant clones containing SNZ1 or SNZ2 open reading frames in the pCM262 plasmid vector were constructed by gap repair (see Table 2) in S. cerevisiae similarly to the construction of two-hybrid bait clones. pCM262 is an episomal plasmid derived from pCM190 (Garí et al., 1997), designed to overexpress genes tagged at the C-terminus with three haemaglutinin (HA) epitopes and six histidine residues in tandem, under the control of the tetO7 promoter. The synthetic 3HA-6His cassette was introduced as a PstI-AscI fragment in the polylinker of pCM190 (Rodríguez-Manzaneque MT, Herrero E, to be described elsewhere). After co-transformation into FY1679 of pCM262 (restricted with PstI and NotI) and PCR products containing the respective ORFs, uracil prototrophs were picked and checked for expression of full-length protein product by Western analysis, using anti-HA antibodies. Plasmids were then recovered into E. coli from such yeast transformants and the correctness of their restriction patterns verified.

Macroarray analysis

We used the hybridization membranes produced by J. Hoheisel (Hauser et al., 1998) and followed his recommendations for use. Briefly, single-stranded α33P dCTP-radiolabelled complex cDNA samples were synthesized by reverse transcription of the same RNA extracts as for Northern blots. An equimolar mix of the 12 anchored 17-mer oligonucleotides dT15(A, C, G)N was used to prime for the reverse transcription reaction. 1/20th of the sample was run on a 5% denaturing polyacrylamide gel and then exposed for 30 min onto X-ray film (Kodak) to check the efficiency of labelling and the extent of the reverse transcription reaction. Samples that gave good results displayed a smear ranging approximately from 80 to >600 nucleotides. The samples were used to hybridize the membranes in the same conditions as for Northern blots. Hybridization signals were revealed with a Phosphorimager (Molecular Dynamics 445SI) after 24 h exposure. Images were analysed using the XDotsReader program commercialized by COSE. For each hybridization, the intensities were normalized by the mean intensity of the membrane, for comparative purposes. Only genes that showed more than a three-fold increase ratio in the absence of thiamine vs. its presence, and that were differentially expressed in several experiments, were considered.


Requirement for vitamin B6 of the different mutants

The SNZ and SNO homologues SOR1 and PDX2, respectively, from C. nicotiniae (Ehrenshaft et al., 1999; Ehrenshaft and Daub 2001) and pyroA from A. nidulans (Osmani et al., 1999) have been described as required for vitamin B6 biosynthesis in those organisms, suggesting that the SNZ and SNO genes could be related to the same pathway in yeast. To check this hypothesis, we constructed all the combinations of single, double and triple deletion mutants for each family. All these mutants grew as well as the control strain on YP or complete synthetic media at either 15° C, 28° C or 37° C and with glucose or glycerol as sole carbon sources (not shown). Because vitamin B6 is a common compound of standard complete and minimum culture media for yeast, we assayed the growth in a synthetic medium without it. We did not see any major growth defect when the inoculum was pre-cultured in YPD or SC. However, when those cells were pre-grown in SC-B6 medium, single snz1, sno1 mutants and snz triple mutants showed a strong growth defect (Figure 2A, B, C). This defect was more acute in snz1 than in the sno1 mutant. Neither single snz2, snz3, sno2 or sno3 mutants (not shown) or double snz2 snz3 mutants showed any growth defects under these conditions. Triple sno mutant behaved similarly to snz triple mutant (not shown).

Figure 2.

Growth of snz and sno mutant strains in a vitamin B6-lacking medium. Serial four-fold dilutions of exponentially (Exp) or stationary (Sta) phase cultures (inoculum) were spotted on to YPD (control) and SC-B6 plates. Plates were incubated at 28° C for 24–36 h (all YPD plates), 2 days [SC plates (A, D)], 7 days [SC plates (B)] or 14 days [SC plates (C)]. Complementation of the auxotrophy for vitamin B6 in the snz1 mutant by a pCM259/262–SNZ1 or pCM259/262–SNZ2 plasmids is shown in (D). Two independent transformants for each plasmid are shown. The relevant genetic background of the strains is indicated

Defects of the snz1 and sno1 mutants were further analysed with regard to the growth phase of the pre-culture. When the culture time was extended to 7 days, the snz1 mutant was more severely defective than the sno1 mutant and the defect worsened if the pre-culture proceeded from stationary phase (Figure 2B). Although, at first sight, the triple mutant snz1 snz2 snz3 behaved identically to the single snz1 mutant (not shown), very long incubation times (2 weeks, Figure 2C) revealed a stronger growth defect for the triple mutant. It is worth noting that the growth level seen in YPD control plates is similar for the wild-type and for all the mutants tested, suggesting that the viability of the cells is unaffected by the absence of those gene products during stationary phase.

The growth defect of snz1 mutant can be complemented by the overexpression of both SNZ1 and SNZ2 (see Figure 2D). This result suggests that SNZ2 and SNZ3 code for a protein with a similar activity to Snz1p. As expected, the addition of vitamin B6 alone also restores the growth (data not shown).

In conclusion, SNZ1 and SNO1 are both required for growth when cells are depleted in vitamin B6. The residual growth in SC-B6 observed for snz1 snz2 snz3 triple mutants pre-cultured until the exponential phase probably reflects traces of vitamin B6 in the cells. On the other hand, SNZ2 and SNZ3 have no complete functional redundancy with SNZ1, despite similar biochemical properties of their products.

Menadione sensitivity

It has been shown previously that some of the snz and sno mutants are sensitive to methylene blue, a generator of singlet oxygen (Padilla et al., 1998), one of the most active ROS (reactive oxygen species). Similarly, the SNZ-homologous genes pyroA and SOR1 genes are known to protect A. nidulans and C. nicotiniae, respectively, against singlet oxygen (Ehrenshaft et al., 1999; Osmani et al., 1999).

We tested the sensitivity of snz and sno mutants to the superoxide generator, menadione. Menadione sensitivities of all the mutants were similar to that of the control strain when treated during the exponential phase for 30 min (Figure 3) or longer (not shown). However, when treated during the post-diauxic phase for 180 min with 20 mM menadione, the triple sno1 sno2 sno3 and, especially, snz1 snz2 snz3 mutants, were extremely sensitive to it. By using a higher drug concentration (40 mM) it was possible to observe that the single snz1 or snz3 and, not surprisingly, the double snz1 snz3 and snz2 snz3 mutants were more sensitive than other single or double mutants. All mutants, as well as wild-type cells, were highly resistant to both drug concentrations when treated during the stationary phase (not shown).

Figure 3.

Menadione sensitivity of snz and sno mutant strains. Sensitivity of snz and sno mutants to menadione treatment. Cells were grown in YPD medium at 30° C, to either exponential phase (2 × 107cells/ml) or 20 h later (post-diauxic phase) and were treated with the indicated menadione concentrations for 30 min (exponential cells) or 180 min (post-diauxic cells). After treatments, cells were diluted in fresh YPD medium (1 : 5 serial dilutions) and 2 µl drops were spotted on YPD plates. Growth was recorded after 2 days of incubation at 30° C. The apparent higher resistance of the snz1 snz2 mutant compared with snz1 single mutant is due to a higher cell number in this particular experiment. It was not observed in other experiments

These results confirm that SNO, and especially SNZ genes, confer resistance to ROS to the cells. These results also show that the protection effect is more dependent on SNZ1 and, surprisingly, SNZ3 than on SNZ2, and that this effect is masked by the intrinsic resistance of advanced stationary cells to environmental stresses (Werner-Washburne et al., 1996).

Vitamin B6 effects on expression of SNZ genes

The role of SNZ and SNO genes in vitamin B6 biosynthesis suggested possible regulation of their expression by this vitamin. Figure 4 shows that SNZ1 expression in exponential phase is not dependent on either the absence or the presence of a high vitamin B6 concentration (0.2 mg/ml), although it is repressed in YPD. SNZ2–3, however, are expressed approximately at the same rate in all the four conditions.

Figure 4.

Influence of vitamin B6 in SNZ expression. Total RNA from log phase cultures in YPD, SC (vitamin B6 is 2 µg/ml), SC − B6 and SC + B6 (vitamin B6 is 0.2 mg/ml) was analysed in a wild-type strain with a Northern experiment using SNZ1 and SNZ2/3 probes (made with oligonucleotides SNZ2A1 and SNZ2A4). Total rRNA is included as loading control


In the course of these experiments, we noted that overexpression (from the doxycycline-regulated tetO7 promoter) of Snz1–3HA–6His or Snz2–3HA–6His in S. cerevisiae strain W303-1A caused a marked delay in recovery from stationary phase (Figure 5). A similar response was seen in strain FY1679 (not shown). When cells overexpressing these proteins were diluted 1 : 100 from saturated overnight cultures into fresh medium, resumption of logarithmic growth occurred up to 16 h later than for cells not overexpressing either protein. This effect was clearly stronger for the Snz2 fusion protein (Figure 5C) than for Snz1p (not shown). Once logarithmic growth had resumed, only a minor effect on growth rate was seen after the longest periods in stationary phase. Density at saturation was also largely unaffected. As can be seen (Figure 5D), this effect is clearly dependent on gene expression: at a high concentration of doxycycline, 2 µg/ml (transcription repressed), no delay was observed. At intermediate concentrations (0.05 or 0.5 µg/ml), the delay was less pronounced (not shown). The addition of vitamin B6 at any stage of the experiment did not change the results (not shown).

Figure 5.

Growth of SNZ2 overexpressing strains after dilution in fresh medium. W303-1A cells transformed with empty vector pCM262 (inverted triangles), pCM262-SNZ1 (not shown), or pCM262-SNZ2 (triangles) were grown in liquid SC − ura medium in the presence or absence of 2 µg/ml doxycycline, as indicated in Materials and methods. From the logarithmic pre-culture (A600 = 0.5), the cells were either transferred directly (A), or allowed to proceed to stationary phase (B, C, D). For B, C and D, cells were then kept for the indicated times at stationary phase. Next, the cell suspension was diluted to OD = 0.1 in fresh medium (with or without doxycycline, as indicated), and growth with shaking and aeration in 400 µl chambers was recorded. Values are the mean of two independent experiments; for clarity, error bars are omitted. Absorbance values (y axis) are shown on a logarithmic scale

Two-hybrid screens

In order to find more clues to the specific roles of SNZ genes, we conducted two-hybrid screens. In the first one, with the SNZ1 bait, out of 10 positive clones obtained that encoded bona fide in-frame proteins, three contained sequences of the YHR198c ORF. These represented two independent clones with a common overlapping segment, encoding a short central portion of the Yhr198c protein (amino acids 167–180).

When SNZ2 was used as the bait, YJR156c (THI11) was found as reactive prey in the correct reading frame. The THI11 gene product is probably involved in the biosynthesis of the pyrimidine precursor of thiamine (Hohmann and Meacock, 1998), and it is homologous to the Schizosaccharomyces pombe Nmt1 protein (Van Dyck et al., 1995). Thi11p has three paralogues in the yeast genome, with almost identical sequences, Thi5p, Thi12p and Thi13p. In similar screens using SNO1 or SNO2/3 as baits, no preys were found that corresponded to proteins involved in vitamin biosynthesis (not shown).

Vitamin B1 effects on gene expression

The close proximity on chromosomes VI and XIV of the putative thiamine biosynthetic genes THI5 and THI12, respectively, to the SNZ3/SNO3 and SNZ2/SNO2 loci (see Figure 1) together with the results of our two-hybrid experiments (see above) may suggest a possible functional link between these two classes of genes. Therefore, we investigated the effects of vitamin B1 depletion on the transcription of the SNZ and SNO families, and on the prototrophy of the corresponding null mutants.

It is known that the concentration of exogenous thiamine influences the transcript amounts of several genes involved in its metabolism, such as THI4, 5, 6, 10, 11, 12, 13, 20, 21, 22 and PHO3. Moreover, this regulation is under the positive control of one or both of the two regulators Thi2p and Thi3p (Hohmann and Meacock, 1998). In order to have an exhaustive list of genes whose transcripts are regulated by the extracellular concentration of thiamine, we used macroarrays of genes produced by the J. Hoheisel laboratory (Hauser et al., 1998). Hybridizations were performed using complex cDNA samples synthesized from RNAs of the wild-type strain FYBL1-8B, grown in the presence of high concentration of extracellular thiamine (1 µM) and in the absence of extracellular thiamine. We also performed hybridizations using complex cDNA samples synthesized from RNAs of the strains FYBL138 and FYBL142 deleted for THI2 and THI3, respectively, grown in the presence of low concentration of extracellular thiamine (10−2 µM). All the genes we found to be upregulated in the absence of exogenous thiamine are indicated in Table 3. We confirmed the already published upregulation of THI4, 5, 6, 10, 11, 12, 13, 20, 21, 22, PET18, YLR004c and PHO3. But, in addition, we found that transcripts of SNO2/3, SNZ2/3, THI2 and ECM15 also accumulated in the absence of exogenous thiamine. Since Thi2p and Thi3p do not regulate the transcription of ECM15, this gene has not been studied further. The accumulation of the other transcripts is dependent on the presence of either Thi2p or Thi3p or both. These results were confirmed by Northern blot experiments, as shown in Figure 6 for SNO2/3 and SNZ2/3. Other known examples of thiamine regulation, such as PDC5 and PDC1 (Muller et al., 1999), are not listed here because of the very stringent criteria used for significance levels (see Materials and methods) but were detected as induced by Northern blot. It is impossible to conclude whether both the transcripts of SNO2 and SNO3 are regulated in the same way because their nucleotide sequences are nearly identical, which must generate cross-hybridization. This is also the case for SNZ2 and SNZ3, as well as for the PHO gene family, of which only the transcripts of PHO3 have been described as being regulated by extracellular thiamine concentration (Nishimura et al., 1992).

The Northern blot experiments shown in Figure 6 illustrate that the SNO1 transcripts are undetectable with this approach during exponential growth phase, and remain unaffected by the extracellular concentration of thiamine. SNO2–3 transcripts are detectable during the exponential growth phase only in the absence of extracellular thiamine, in a THI2–3-dependent manner. SNZ1 transcripts are slightly more abundant during the exponential growth phase in the presence of low concentration of extracellular thiamine with respect to high concentration, but this accumulation is not dependent on either Thi2p or Thi3p. SNZ2–3 transcripts accumulation occurs clearly in the absence of extracellular thiamine, and is completely dependent on Thi2p and only partially dependent on Thi3p.

Figure 6.

Gene expression analysis in thiamine-limiting conditions. 10 µg total RNA extracted from cells grown in SC to exponential phase in the presence of high (H) or low (L) extracellular thiamine concentration were loaded and run on a 1.5% agarose gel containing 0.6% formaldehyde, transferred to Hybond N+ membranes (Amersham) and hybridized with specific probes (upper panel; made by PCR with oligonucleotide pairs called ‘lo’ and ‘up’ for each case; see Table 2) indicated at the bottom of the figure. H, 1 µ M extracellular thiamine; L, no thiamine for the wild-type strain (WT) FYBL1-8B, and 10−2 µ M for the auxotrophic strains Δthi2 (FYBL1-8B/BL138) and Δthi3 (FYBL1-8B/BL142). ACT1 was used as loading control for all the samples

Table 3. Yeast transcripts regulated by extracellular thiamine
Genes InductionΔthi2Δthi3NorthernConclusion
  1. This table indicates the results from macroarray hybridization experiments. The Induction column represents the relative levels of transcripts observed for the wild-type strain FYBL1-8B grown in the absence of extracellular thiamine with respect to those found in the presence of extracellular thiamine (only ratios above or equal to a three-fold increase have been considered). Columns Δthi2 and Δthi3 represent the same ratios but for the strains FYBL138 and FYBL142 deleted for THI2 and THI3, respectively. These strains are auxotrophic for thiamine and have thus been grown in the presence of a low extracellular thiamine concentration (10−8M) instead of no thiamine. ‘Yes’ indicates that the regulation has been confirmed by Northern blot experiments. NT, non-tested by Northern blot experiment.

YLR004c +++YesConfirmation
YBR240cTHI2+++ YesNew


Role of SNZ and SNO genes in vitamin B6 biosynthesis

Most microorganisms and plants possess at least one pathway leading to pyridoxine and pyridoxal 5′-phosphate synthesis. The Escherichia coli pathway has been extensively studied (see Drewke and Leistner, 2001) but the corresponding pathway in yeast is not well defined. Nevertheless, the fact that the L-[amide-15N]glutamine label is incorporated efficiently into pyridoxine in S. cerevisiae but not in E. coli indicates that vitamin B6 biosynthesis must be significantly different in either organism (Tazuya et al., 1995).

Other eukaryotic organisms, such as A. nidulans, Mucor racemosus and N. crassa and prokaryotes, such as Staphylococcus aureus and Bacillus subtilis, may have biosynthetic pathways for pyridoxine similar to that of S. cerevisiae (Tanaka et al., 2000). It has recently been shown that the pyroA gene from A. nidulans (Osmani et al., 1999) and the PDX1 (SOR1) gene from C. nicotiniae (Ehrenshaft et al., 1999) are involved in de novo biosynthesis of vitamin B6 in those fungi. It has been suggested that mutations in pdx-1 and pdx-2 (SNZ and SNO homologues, respectively) cause pyridoxine auxotrophy in N. crassa (Bean et al., 2001). A more recent study shows an involvement of PDX2 of C. nicotianae in the pyridoxine biosynthesis pathway (Ehrenshaft and Daub, 2001). The fungal genes are 58–67% (SNZ) and 36–38% (SNO) identical to the yeast genes. The SNZ and SNO families are widely represented in eubacteria, archaea and eukaryotes (Braun et al., 1996; Ehrenshaft et al., 1999; Galperin, 2001; Mittenhuber, 2001). Our results indicate that the SNZ and SNO gene families are also involved in the vitamin B6 biosynthesis in S. cerevisiae. We identified a functional specialization within these two gene families, since the absence of copy 1 leads to a more severe growth phenotype than the absence of copy 2 and 3 when cells are grown in SC-B6. Although the SNZ2–3 genes seem to be dispensable for any condition tested, they should code for proteins with a similar activity to Snz1p because they can complement, at least as efficiently as SNZ1 itself (Figure 2D), the snz1 null phenotype when overexpressed. This suggests that all the three SNZ genes code for enzymes involved in vitamin B6 biosynthesis, probably at the last step, the ring closure, as suggested by Ehrenshaft et al. (1999). Vitamin B6, however, seems not to be a regulator of SNZ gene transcription (Figure 4), although the high transcription levels already present in synthetic medium might mask the effect of the absence of the vitamin.

The phenotypes of sno mutants are less pronounced than that of the corresponding snz mutants. It has been suggested that SNO genes are involved in the first step of pyridoxal biosynthesis (Ehrenshaft and Daub, 2001; Osmani et al., 1999). Thus, it is conceivable that, if the SNZ genes code for pyridoxine biosynthetic enzymes, the substrate for Snz proteins could be produced in the absence of Sno proteins, although in very minor amounts, by alternative pathways.

SNZ and SNO genes have an interesting effect on sensitivity to oxidative radicals. Padilla et al. (1998) have shown that sno1 or snz1 mutations are very sensitive to the singlet oxygen generator methylene blue. Furthermore, it has been shown that B6 vitamers are efficient quenchers of singlet oxygen in vitro (Bilski et al., 2000; Ehrenshaft and Daub, 2001) and that externally added vitamin B6 effectively suppresses the toxicity of methylene blue in A. nidulans (Osmani et al., 1999). We have shown here that a different kind of ROS, the superoxide radical, produced by menadione, has a similar effect on snz/sno mutants. It has been argued that B6 can act in active oxygen resistance and that such as a protective effect is more necessary in stationary phase, when cells are subjected to increased oxidative stress (Ehrenshaft et al., 1999). Vitamin B6 may be just an antioxidant or, perhaps, its destruction by ROS causes deficiency of B6 vitamers that are necessary for other metabolic uses (Osmani et al., 1999). This last hypothesis is supported by the fact that neither H2O2 nor menadione induce the transcription of any of these genes (Gasch et al., 2000). Furthermore, preliminary data from the groups of Joaquim Ros and Enrique Herrero (University of Lleida, Spain) indicate that addition of vitamin B6 reduces the level of protein carbonylation (a parameter measuring protein oxidation) induced by addition of oxidants such as menadione or hydrogen peroxide. Finally, the striking difference between menadione sensitivity of snz2 and snz3 (see Figure 3) suggests that the two genes have somewhat different roles, in spite of their high sequence similarity.

Integration of biosynthetic pathways for vitamins B1 and B6

It is known that in some prokaryotes (e.g. E. coli), some precursors and enzymes (e.g. the pdxK gene product) are shared by biosynthetic pathways for the vitamins B1 and B6 (Begley et al., 1999; Mittenhuber, 2001). In S. cerevisiae the fact that the transcription of the SNZ2-3 and SNO2-3 genes is induced by the absence of thiamine in a Thi2p/Thi3p-dependent manner, suggests that those genes have a function related with the biosynthesis of vitamin B1 as well. Two additional facts support this view.

First, the finding of interactions between Thi and Snz proteins by two-hybrid analysis (see below) provides an independent suggestion for a common functional pathway for Snz and Thi proteins, i.e. a putative role for the Snz, and also Sno, proteins in thiamine biosynthesis, and corroborates the idea of Hohmann and Meacock (1998). Our results suggest, however, that Snz2-3 and Sno2-3 proteins, but not Snz1p and Sno1p, are related to vitamin B1 biosynthesis: SNZ1 and SNO1 transcription is not induced by thiamine depletion. Although all the members of Snz and Sno protein families can reasonably be considered as putative enzymes acting on the same reactions, it cannot be dismissed that they may act at different times or cellular compartments, similarly to other cases of isoenzymes (Matthews et al., 2000). Several pathways for thiamine biosynthesis should exist because none of the single, double and triple deletion mutants of the SNO and the SNZ gene families display auxotrophy for thiamine (Llorente B and Peréz-Ortín JE, not shown). This is consistent with the fact that under anaerobic conditions the pyrimidine moiety of thiamine is not derived from pyridoxal, therefore a different pathway must act instead (Tanaka et al., 2000).

Second, there is close map proximity between these genes and some members of the THI5/11/12/13 family (see Figure 1). These THI genes are highly similar: the proteins differ only in one amino acid out of 340. The corresponding protein is only detected in the absence of thiamine (Muller et al., 1999), THI5 and THI12 transcripts accumulate in the absence of extracellular thiamine (Meacock PA, personal communication) and they putatively code for an enzyme involved in the biosynthesis of the pyrimidine moiety of the thiamine molecule (Hohmann and Meacock, 1998). It is therefore interesting that three consecutive genes repeated twice in the genome have a common transcriptional response. The three couples SNO/SNZ share divergently transcribed promoters, something that supports the notion of common regulatory sequences (Padilla et al., 1998). This head-to-head arrangement has been also found in the yeasts Candida albicans and S. kluivery (Llorente B, unpublished), in the fungus N. crassa (Bean et al., 2001), and in the sponge Suberites domunculata (Seack et al., 2001) that suggests an ancient origin for SNO-SNZ co-regulation. The close association of THI genes with SNZ-SNO, however, is not so common. We have analysed the genomes of Schizosaccharomyces pombe, Candida albicans and the hemiascomycetes group of the Genolevures project (Souciet et al., 2000). Most of them possess only one orthologue of SNZ, SNO and THI5. None of the THI5 orthologues has been found to be syntenic with the SNO-SNZ orthologues. In addition, copies 2 and 3 of the SNZ-SNO genes seem to be the product of a recent subtelomeric chromosome duplication, since they are not detected in some S. cerevisiae strains (Padilla et al., 1998). The functional specialization of copies 2 and 3 and their association with THI5 locus may thus have developed recently.

A protein complex of Snz, Sno and Thi proteins

As discussed above, our two-hybrid results show that Snz1p interacts with Yhr198p and Snz2/3p with Thi11p. Previous two-hybrid analysis showed interactions between the Snz and Sno proteins themselves and with other proteins (DIP database, 2001; Ito et al., 2000; Padilla et al., 1998; Uetz et al., 2000). A scheme of the putative interactions is shown in Figure 7.

Figure 7.

Summary of two-hybrid interactions of Snz, Sno and Thi families. Protein preys detected in this work are represented in degraded grey. Interacting proteins a linked by bars and arrows. Sno and Snz proteins are highlighted. See text for further discussion

Examples of multi-enzymatic complexes comprising enzymes that catalyse successive steps in a metabolic pathway are well known (Matthews et al., 2000). It seems that every one of the Snz and Sno proteins can interact with itself and with the other members. Because the transcription profiles of copies 2 and 3, on the one hand, and copy 1, on the other hand, are so different that it is possible that the protein complex changes depending on the physiological circumstances of the cell. In fact, Padilla et al. (1998) described a 230 kDa complex dependent on Snz1p and which only appears during stationary phase. Because the predicted molecular weight for these proteins is between 25 kDa (Sno) and 32 kDa (Snz), the complex found should include several copies of them and/or additional proteins. The two-hybrid experiments show candidates for those interacting proteins.

The first candidate is Yhr198p. No definitive function has yet been assigned to YHR198c, but it carries a purine/pyrimidine phosphoribosyl transferase signature, indicative of a role in the purine/pyrimidine salvage pathway. It has been described to interact with Temlp by two-hybrid analysis which itself, in turn, interacts with Snz2/3p (Uetz et al., 2000). Another protein that has been described to interact with the Sno-Snz proteins is Ymr322p (DIP database, 2001).

The existence of a multiprotein complex is also supported by our overexpression studies. We found that Snz1p and, specifically, Snz2p when overexpressed as 3HA-6His-tagged fusions in a non-regulated way, caused a delay in the recovery from the stationary phase. The most direct explanation for the delay is that some metabolite(s) that are exhausted during this phase but necessary to resume growth are, in this condition, more difficult to synthesize. However, the supplementary addition of the obvious candidate, pyridoxal, does not reduce the delay (Sunnerhagen P, not shown). Furthermore, the delay occurred after growth in standard (i.e. not vitamin-depleted medium). The defect caused by extra copies of tagged Snz proteins may be caused by an imbalance of the protein subunits of the complex due to of the withdrawal of some important subunit(s), which is caused by the excess subunits of Snz protein or by the 3HA tag. It has been argued that an imbalance between the putative subunits of the complex explains the dominant-negative effect of the snz1-Δ2 mutation in 6-AU sensitivity (Padilla et al., 1998).

In summary, our results show that these two families, SNZ and SNO, contain genes that are only partially redundant in function in spite of their high sequence similarity. This seems to be another case in which gene duplicates have different roles in yeast (Blandin et al., 2000; Brookfield, 1997; Llorente et al., 1999).


We thank Dr J. Hoheisel for the gift of yeast macroarrays, Dr S. Hohmann for his critical reading of the manuscript and A. Llopis for her excellent technical work. B.D. is a member of the Institut Universitaire de France. This work was funded by the European Commission as part of the Eurofan 2 project under contract BIO4-CT97-2294 to B.D., J.E.P.-O., E.H. and P.S. Funding was also provided to P.S. by the Swedish Natural Science Research Council (2000-5471); to J.E.P-O. by the Spanish C.I.C.Y.T. (BIO98-1316-CE), and to E.H. from Generalitat de Catalunya (1999SGR 00170).